[dpdk-dev] [PATCH v2 0/3] refactoring ring library
Do some work for ring refactoring, which includes:
1. add rte prefix before update tail API
2. Move all private APIs into new header files
v2:
1. add rte_ prefixes for private files (Konstantin)
Feifei Wang (3):
test/ring: reduce iteration numbers to make test duration shorter
ring: add rte prefix before update tail API
ring: refactor ring library
app/test/test_ring_perf.c | 2 +-
lib/librte_ring/meson.build | 11 +-
...{rte_ring_c11_mem.h => rte_ring_c11_pvt.h} | 10 +-
lib/librte_ring/rte_ring_elem.h | 374 +
lib/librte_ring/rte_ring_elem_pvt.h | 385 ++
..._ring_generic.h => rte_ring_generic_pvt.h} | 10 +-
lib/librte_ring/rte_ring_hts.h| 84 +---
..._hts_c11_mem.h => rte_ring_hts_elem_pvt.h} | 88 +++-
lib/librte_ring/rte_ring_peek.h | 71 +---
...eek_c11_mem.h => rte_ring_peek_elem_pvt.h} | 75 +++-
lib/librte_ring/rte_ring_peek_zc.h| 2 +-
lib/librte_ring/rte_ring_rts.h| 84 +---
..._rts_c11_mem.h => rte_ring_rts_elem_pvt.h} | 88 +++-
13 files changed, 649 insertions(+), 635 deletions(-)
rename lib/librte_ring/{rte_ring_c11_mem.h => rte_ring_c11_pvt.h} (96%)
create mode 100644 lib/librte_ring/rte_ring_elem_pvt.h
rename lib/librte_ring/{rte_ring_generic.h => rte_ring_generic_pvt.h} (95%)
rename lib/librte_ring/{rte_ring_hts_c11_mem.h => rte_ring_hts_elem_pvt.h}
(60%)
rename lib/librte_ring/{rte_ring_peek_c11_mem.h => rte_ring_peek_elem_pvt.h}
(61%)
rename lib/librte_ring/{rte_ring_rts_c11_mem.h => rte_ring_rts_elem_pvt.h}
(62%)
--
2.25.1
[dpdk-dev] [PATCH v2 1/3] test/ring: reduce iteration numbers to make test duration shorter
When testing ring performance in the case that multiple lcores are mapped to
the same physical core, e.g. --lcores '(0-3)@10', it takes a very long time
to wait for the "enqueue_dequeue_bulk_helper" to finish. This is because
too much iteration numbers and extremely low efficiency for enqueue and
dequeue with this kind of core mapping. Following are the test results to
show the above phenomenon:
x86-Intel(R) Xeon(R) Gold 6240:
$sudo ./app/test/dpdk-test --lcores '(0-1)@25'
Testing using two hyperthreads(bulk (size: 8):)
iter_shift: 3 5 7 9 11 13*15 17 19
21 23
run time: 7s7s7s8s9s 16s47s170s 660s
>0.5h >1h
legacy APIs: SP/SC: 37116 40525 40525 40209 40367 40407 40541
NoData NoData
legacy APIs: MP/MC: 56141150657 40526 40526 40526 40625 40585
NoData NoData
aarch64-n1sdp:
$sudo ./app/test/dpdk-test --lcore '(0-1)@1'
Testing using two hyperthreads(bulk (size: 8):)
iter_shift: 3 5 7 9 11 13*15 17 19
21 23
run time: 8s8s8s9s9s 14s34s111s 418s
25min >1h
legacy APIs: SP/SC: 0.4 0.2 0.1 488 488488488488489
489 NoData
legacy APIs: MP/MC: 0.4 0.3 0.2 488 488488488490489
489 NoData
As the number of iterations increases, so does the time which is required to
run the program. Currently (iter_shift = 23), it will take more than 1 hour
to wait for the test to finish. To fix this, the "iter_shift" should decrease
and ensure enough iterations to keep the test data stable. In order to achieve
this, we also test with "-l" EAL argument:
x86-Intel(R) Xeon(R) Gold 6240:
$sudo ./app/test/dpdk-test -l 25-26
Testing using two NUMA nodes(bulk (size: 8):)
iter_shift: 3 5 7 9 11 13*15 17 19
21 23
run time: 6s6s6s6s6s 6s 6s 7s 8s
11s 27s
legacy APIs: SP/SC: 4720132254 83 91 73 81
75 95
legacy APIs: MP/MC: 441818240 245270250249252
250 253
aarch64-n1sdp:
$sudo ./app/test/dpdk-test -l 1-2
Testing using two physical cores(bulk (size: 8):)
iter_shift: 3 5 7 9 11 13*15 17 19
21 23
run time: 8s8s8s8s8s 8s 8s 9s 9s
11s 23s
legacy APIs: SP/SC: 0.7 0.4 1.2 1.8 2.02.02.02.02.0
2.0 2.0
legacy APIs: MP/MC: 0.3 0.4 1.3 1.9 2.92.92.92.92.9
2.9 2.9
According to above test data, when "iter_shift" is set as "15", the test run
time is reduced to less than 1 minute and the test result can keep stable
in x86 and aarch64 servers.
Fixes: 1fa5d0099efc ("test/ring: add custom element size performance tests")
Cc: honnappa.nagaraha...@arm.com
Cc: sta...@dpdk.org
Signed-off-by: Feifei Wang
Reviewed-by: Honnappa Nagarahalli
Reviewed-by: Ruifeng Wang
Acked-by: Konstantin Ananyev
---
app/test/test_ring_perf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/app/test/test_ring_perf.c b/app/test/test_ring_perf.c
index e63e25a86..fd82e2041 100644
--- a/app/test/test_ring_perf.c
+++ b/app/test/test_ring_perf.c
@@ -178,7 +178,7 @@ enqueue_dequeue_bulk_helper(const unsigned int flag, const
int esize,
struct thread_params *p)
{
int ret;
- const unsigned int iter_shift = 23;
+ const unsigned int iter_shift = 15;
const unsigned int iterations = 1 << iter_shift;
struct rte_ring *r = p->r;
unsigned int bsize = p->size;
--
2.25.1
[dpdk-dev] [PATCH v2 2/3] ring: add rte prefix before update tail API
Add __rte prefix before update_tail API because it is a internal
function.
Signed-off-by: Feifei Wang
Reviewed-by: Honnappa Nagarahalli
Reviewed-by: Ruifeng Wang
Acked-by: Konstantin Ananyev
---
lib/librte_ring/rte_ring_c11_mem.h | 4 ++--
lib/librte_ring/rte_ring_elem.h| 4 ++--
lib/librte_ring/rte_ring_generic.h | 4 ++--
3 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/lib/librte_ring/rte_ring_c11_mem.h
b/lib/librte_ring/rte_ring_c11_mem.h
index 0fb73a337..7f5eba262 100644
--- a/lib/librte_ring/rte_ring_c11_mem.h
+++ b/lib/librte_ring/rte_ring_c11_mem.h
@@ -11,8 +11,8 @@
#define _RTE_RING_C11_MEM_H_
static __rte_always_inline void
-update_tail(struct rte_ring_headtail *ht, uint32_t old_val, uint32_t new_val,
- uint32_t single, uint32_t enqueue)
+__rte_ring_update_tail(struct rte_ring_headtail *ht, uint32_t old_val,
+ uint32_t new_val, uint32_t single, uint32_t enqueue)
{
RTE_SET_USED(enqueue);
diff --git a/lib/librte_ring/rte_ring_elem.h b/lib/librte_ring/rte_ring_elem.h
index 7034d29c0..57344d47d 100644
--- a/lib/librte_ring/rte_ring_elem.h
+++ b/lib/librte_ring/rte_ring_elem.h
@@ -423,7 +423,7 @@ __rte_ring_do_enqueue_elem(struct rte_ring *r, const void
*obj_table,
__rte_ring_enqueue_elems(r, prod_head, obj_table, esize, n);
- update_tail(&r->prod, prod_head, prod_next, is_sp, 1);
+ __rte_ring_update_tail(&r->prod, prod_head, prod_next, is_sp, 1);
end:
if (free_space != NULL)
*free_space = free_entries - n;
@@ -470,7 +470,7 @@ __rte_ring_do_dequeue_elem(struct rte_ring *r, void
*obj_table,
__rte_ring_dequeue_elems(r, cons_head, obj_table, esize, n);
- update_tail(&r->cons, cons_head, cons_next, is_sc, 0);
+ __rte_ring_update_tail(&r->cons, cons_head, cons_next, is_sc, 0);
end:
if (available != NULL)
diff --git a/lib/librte_ring/rte_ring_generic.h
b/lib/librte_ring/rte_ring_generic.h
index 953cdbbd5..37c62b8d6 100644
--- a/lib/librte_ring/rte_ring_generic.h
+++ b/lib/librte_ring/rte_ring_generic.h
@@ -11,8 +11,8 @@
#define _RTE_RING_GENERIC_H_
static __rte_always_inline void
-update_tail(struct rte_ring_headtail *ht, uint32_t old_val, uint32_t new_val,
- uint32_t single, uint32_t enqueue)
+__rte_ring_update_tail(struct rte_ring_headtail *ht, uint32_t old_val,
+ uint32_t new_val, uint32_t single, uint32_t enqueue)
{
if (enqueue)
rte_smp_wmb();
--
2.25.1
[dpdk-dev] [PATCH v2 3/3] ring: refactor ring library
For legacy modes, rename ring_generic/c11 to ring_generic/c11_pvt.
Furthermore, add new file ring_elem_pvt.h which includes ring_do_eq/deq
and ring element copy/delete APIs.
For other modes, rename xx_c11_mem to xx_elem_pvt. Move all private APIs
into these new header files.
Finally, the external APIs and internal APIs will be separated from each
other. This can remind users not to use internal APIs and make ring
library easier to maintain.
Suggested-by: Honnappa Nagarahalli
Signed-off-by: Feifei Wang
Reviewed-by: Honnappa Nagarahalli
Reviewed-by: Ruifeng Wang
---
lib/librte_ring/meson.build | 11 +-
...{rte_ring_c11_mem.h => rte_ring_c11_pvt.h} | 6 +-
lib/librte_ring/rte_ring_elem.h | 374 +
lib/librte_ring/rte_ring_elem_pvt.h | 385 ++
..._ring_generic.h => rte_ring_generic_pvt.h} | 6 +-
lib/librte_ring/rte_ring_hts.h| 84 +---
..._hts_c11_mem.h => rte_ring_hts_elem_pvt.h} | 88 +++-
lib/librte_ring/rte_ring_peek.h | 71 +---
...eek_c11_mem.h => rte_ring_peek_elem_pvt.h} | 75 +++-
lib/librte_ring/rte_ring_peek_zc.h| 2 +-
lib/librte_ring/rte_ring_rts.h| 84 +---
..._rts_c11_mem.h => rte_ring_rts_elem_pvt.h} | 88 +++-
12 files changed, 644 insertions(+), 630 deletions(-)
rename lib/librte_ring/{rte_ring_c11_mem.h => rte_ring_c11_pvt.h} (98%)
create mode 100644 lib/librte_ring/rte_ring_elem_pvt.h
rename lib/librte_ring/{rte_ring_generic.h => rte_ring_generic_pvt.h} (97%)
rename lib/librte_ring/{rte_ring_hts_c11_mem.h => rte_ring_hts_elem_pvt.h}
(60%)
rename lib/librte_ring/{rte_ring_peek_c11_mem.h => rte_ring_peek_elem_pvt.h}
(61%)
rename lib/librte_ring/{rte_ring_rts_c11_mem.h => rte_ring_rts_elem_pvt.h}
(62%)
diff --git a/lib/librte_ring/meson.build b/lib/librte_ring/meson.build
index 36fdcb6a5..f007893b5 100644
--- a/lib/librte_ring/meson.build
+++ b/lib/librte_ring/meson.build
@@ -5,12 +5,13 @@ sources = files('rte_ring.c')
headers = files('rte_ring.h',
'rte_ring_core.h',
'rte_ring_elem.h',
- 'rte_ring_c11_mem.h',
- 'rte_ring_generic.h',
+ 'rte_ring_elem_pvt.h',
+ 'rte_ring_c11_pvt.h',
+ 'rte_ring_generic_pvt.h',
'rte_ring_hts.h',
- 'rte_ring_hts_c11_mem.h',
+ 'rte_ring_hts_elem_pvt.h',
'rte_ring_peek.h',
- 'rte_ring_peek_c11_mem.h',
+ 'rte_ring_peek_elem_pvt.h',
'rte_ring_peek_zc.h',
'rte_ring_rts.h',
- 'rte_ring_rts_c11_mem.h')
+ 'rte_ring_rts_elem_pvt.h')
diff --git a/lib/librte_ring/rte_ring_c11_mem.h
b/lib/librte_ring/rte_ring_c11_pvt.h
similarity index 98%
rename from lib/librte_ring/rte_ring_c11_mem.h
rename to lib/librte_ring/rte_ring_c11_pvt.h
index 7f5eba262..759192f4c 100644
--- a/lib/librte_ring/rte_ring_c11_mem.h
+++ b/lib/librte_ring/rte_ring_c11_pvt.h
@@ -7,8 +7,8 @@
* Used as BSD-3 Licensed with permission from Kip Macy.
*/
-#ifndef _RTE_RING_C11_MEM_H_
-#define _RTE_RING_C11_MEM_H_
+#ifndef _RTE_RING_C11_PVT_H_
+#define _RTE_RING_C11_PVT_H_
static __rte_always_inline void
__rte_ring_update_tail(struct rte_ring_headtail *ht, uint32_t old_val,
@@ -178,4 +178,4 @@ __rte_ring_move_cons_head(struct rte_ring *r, int is_sc,
return n;
}
-#endif /* _RTE_RING_C11_MEM_H_ */
+#endif /* _RTE_RING_C11_PVT_H_ */
diff --git a/lib/librte_ring/rte_ring_elem.h b/lib/librte_ring/rte_ring_elem.h
index 57344d47d..98c5495e0 100644
--- a/lib/librte_ring/rte_ring_elem.h
+++ b/lib/librte_ring/rte_ring_elem.h
@@ -21,6 +21,7 @@ extern "C" {
#endif
#include
+#include
/**
* Calculate the memory size needed for a ring with given element size
@@ -105,379 +106,6 @@ ssize_t rte_ring_get_memsize_elem(unsigned int esize,
unsigned int count);
struct rte_ring *rte_ring_create_elem(const char *name, unsigned int esize,
unsigned int count, int socket_id, unsigned int flags);
-static __rte_always_inline void
-__rte_ring_enqueue_elems_32(struct rte_ring *r, const uint32_t size,
- uint32_t idx, const void *obj_table, uint32_t n)
-{
- unsigned int i;
- uint32_t *ring = (uint32_t *)&r[1];
- const uint32_t *obj = (const uint32_t *)obj_table;
- if (likely(idx + n < size)) {
- for (i = 0; i < (n & ~0x7); i += 8, idx += 8) {
- ring[idx] = obj[i];
- ring[idx + 1] = obj[i + 1];
- ring[idx + 2] = obj[i + 2];
- ring[idx + 3] = obj[i + 3];
- ring[idx + 4] = obj[i + 4];
- ring[idx + 5] = obj[i + 5];
- ring[idx + 6] = obj[i + 6];
- ring[idx + 7] = obj[i + 7];
- }
- switch (n & 0x7) {
- case 7:
-
Re: [dpdk-dev] [PATCH v2] app/testpmd: avoid exit without terminal restore
Acked-by: Xiaoyun Li
> -Original Message-
> From: Yu, DapengX
> Sent: Tuesday, January 26, 2021 15:44
> To: Li, Xiaoyun ; Xing, Beilei ;
> Lu,
> Wenzhuo ; Iremonger, Bernard
> ; step...@networkplumber.org
> Cc: dev@dpdk.org; sta...@dpdk.org
> Subject: RE: [PATCH v2] app/testpmd: avoid exit without terminal restore
>
> @Xiaoyun,
> Add one more explanation.
> even if registers prompt_exit failed, when users type "quit" and then break
> from
> cmdline_interact(), the ret check will not prevent cmdline_stdin_exit() to be
> called to restore terminal settings in this patch.
>
> @Beilei
> Can you give comment on this patch v2, so I can move on to resolve this issue
> related with this patch.
>
> -Original Message-
> From: Yu, DapengX
> Sent: Tuesday, January 26, 2021 3:13 PM
> To: Li, Xiaoyun ; Xing, Beilei ;
> Lu,
> Wenzhuo ; Iremonger, Bernard
> ; step...@networkplumber.org
> Cc: dev@dpdk.org; sta...@dpdk.org
> Subject: RE: [PATCH v2] app/testpmd: avoid exit without terminal restore
>
> Hi Xiaoyun,
>
> For this situation: registering prompt_exit succeed, and user type "quit"
> command to exit testpmd.
>
> If the second call to cmdline_stdin_exit is not excluded from the code
> conditionally, the cmdline_stdin_exit will be called twice;
>
> 1. the first time: in prompt()
> #0 cmdline_stdin_exit (cl=0x3638470)
> at ../lib/librte_cmdline/cmdline_socket.c:56
> #1 0x0051fb62 in prompt () at ../app/test-pmd/cmdline.c:17134
> #2 0x005a4d0f in main (argc=2, argv=0x7fffda70) at
> ../app/test-
> pmd/testpmd.c:3849 2. the second time: in exit() .
> #0 cmdline_stdin_exit (cl=0x3638470)
> at ../lib/librte_cmdline/cmdline_socket.c:56
> #1 0x0051fb95 in prompt_exit () at ../app/test-
> pmd/cmdline.c:17142
> #2 0x7630fe9c in __run_exit_handlers () from /lib64/libc.so.6
> #3 0x7630ffd0 in exit () from /lib64/libc.so.6
> #4 0x762f96aa in __libc_start_main () from /lib64/libc.so.6
> #5 0x004ead2e in _start ()
>
> -Original Message-
> From: Li, Xiaoyun
> Sent: Tuesday, January 26, 2021 2:33 PM
> To: Yu, DapengX ; Xing, Beilei ;
> Lu, Wenzhuo ; Iremonger, Bernard
> ; step...@networkplumber.org
> Cc: dev@dpdk.org; Yu, DapengX ; sta...@dpdk.org
> Subject: RE: [PATCH v2] app/testpmd: avoid exit without terminal restore
>
> Hi
>
> > -Original Message-
> > From: dapengx...@intel.com
> > Sent: Monday, January 25, 2021 11:30
> > To: Xing, Beilei ; Li, Xiaoyun
> > ; Lu, Wenzhuo ; Iremonger,
> > Bernard ; step...@networkplumber.org
> > Cc: dev@dpdk.org; Yu, DapengX ; sta...@dpdk.org
> > Subject: [PATCH v2] app/testpmd: avoid exit without terminal restore
> >
> > From: Dapeng Yu
> >
> > In interactive mode, if testpmd exit by calling rte_exit without
> > restore terminal attributes, terminal will not echo keyboard input.
> >
> > register a function with atexit() in prompt(), when exit() in
> > rte_exit() is called, the registered function restores terminal attributes.
> >
> > Fixes: 5a8fb55c48ab ("app/testpmd: support unidirectional
> > configuration")
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: Dapeng Yu
> > ---
> > app/test-pmd/cmdline.c | 13 +++--
> > 1 file changed, 11 insertions(+), 2 deletions(-)
> >
> > diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c index
> > 89034c8b7..f7e18ba3d 100644
> > --- a/app/test-pmd/cmdline.c
> > +++ b/app/test-pmd/cmdline.c
> > @@ -17116,6 +17116,7 @@ cmdline_read_from_file(const char *filename)
> > void
> > prompt(void)
> > {
> > +int ret;
> > /* initialize non-constant commands */ cmd_set_fwd_mode_init();
> > cmd_set_fwd_retry_mode_init(); @@ -17123,15 +17124,23 @@ prompt(void)
> > testpmd_cl = cmdline_stdin_new(main_ctx, "testpmd> "); if (testpmd_cl
> > == NULL) return;
> > +
> > +ret = atexit(prompt_exit);
> > +if (ret != 0)
> > +printf("Cannot set exit function for cmdline\n");
> > +
> > cmdline_interact(testpmd_cl);
> > -cmdline_stdin_exit(testpmd_cl);
> > +if (ret != 0)
>
> Why do you need to check ret here?
> Even if registers prompt_exit failed, when users type "quit" and then break
> from
> cmdline_interact(), cmdline_stdin_exit() should be called to restore terminal
> settings.
>
> > +cmdline_stdin_exit(testpmd_cl);
> > }
> >
> > void
> > prompt_exit(void)
> > {
> > -if (testpmd_cl != NULL)
> > +if (testpmd_cl != NULL) {
> > cmdline_quit(testpmd_cl);
> > +cmdline_stdin_exit(testpmd_cl);
> > +}
> > }
> >
> > static void
> > --
> > 2.27.0
>
>
Re: [dpdk-dev] [PATCH] app/testpmd: add IPv6 DSCP option for modify_field action
26/01/2021 05:40, Alexander Kozyrev:
> Support IPv6 DSCP modification via the modify_field action.
> Add the ipv6_dscp option for the corresponding header field.
>
> Fixes: 24e1a5e39df ("app/testpmd: support modify field flow action")
>
> Signed-off-by: Alexander Kozyrev
> ---
> app/test-pmd/cmdline_flow.c | 2 +-
It should be squashed with the patch adding the field in ethdev.
[dpdk-dev] [PATCH] net/octeontx: fix max Rx packet length corruption
From: Sunil Kumar Kori
Maximum Rx packet length is getting updated twice which
corrupts actual value.
Fixes: 3151e6a687a3 ("net/octeontx: support MTU")
Cc: sta...@dpdk.org
Signed-off-by: Sunil Kumar Kori
---
drivers/net/octeontx/octeontx_ethdev.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/net/octeontx/octeontx_ethdev.c
b/drivers/net/octeontx/octeontx_ethdev.c
index 81779885d..5836dbe09 100644
--- a/drivers/net/octeontx/octeontx_ethdev.c
+++ b/drivers/net/octeontx/octeontx_ethdev.c
@@ -867,7 +867,6 @@ octeontx_dev_info(struct rte_eth_dev *dev,
dev_info->max_mac_addrs =
octeontx_bgx_port_mac_entries_get(nic->port_id);
- dev_info->max_rx_pktlen = PKI_MAX_PKTLEN;
dev_info->max_rx_queues = 1;
dev_info->max_tx_queues = PKO_MAX_NUM_DQ;
dev_info->min_rx_bufsize = 0;
--
2.25.1
Re: [dpdk-dev] [PATCH v5] app/testpmd: fix setting maximum packet length
Hi,
> -Original Message-
> From: Ferruh Yigit
> Sent: Monday, January 25, 2021 8:16 PM
> To: Wenzhuo Lu ; Xiaoyun Li
> ; Bernard Iremonger
> ; Steve Yang
> Cc: Ferruh Yigit ; dev@dpdk.org; sta...@dpdk.org;
> lance.richard...@broadcom.com; ouli...@huawei.com; Wisam Monther
> ; lihuis...@huawei.com
> Subject: [PATCH v5] app/testpmd: fix setting maximum packet length
>
> From: Steve Yang
>
> "port config all max-pkt-len" command fails because it doesn't set the
> 'DEV_RX_OFFLOAD_JUMBO_FRAME' offload flag properly.
>
> Commit in the fixes line moved the 'DEV_RX_OFFLOAD_JUMBO_FRAME'
> offload flag update from 'cmd_config_max_pkt_len_parsed()' to
> 'init_config()'.
> 'init_config()' function is only called during testpmd startup, but the flag
> status needs to be calculated whenever 'max_rx_pkt_len' changes.
>
> The issue can be reproduce as [1], where the 'max-pkt-len' reduced and
> 'DEV_RX_OFFLOAD_JUMBO_FRAME' offload flag should be cleared but it
> didn't.
>
> Adding the 'update_jumbo_frame_offload()' helper function to update
> 'DEV_RX_OFFLOAD_JUMBO_FRAME' offload flag and 'max_rx_pkt_len'. This
> function is called both by 'init_config()' and
> 'cmd_config_max_pkt_len_parsed()'.
>
> Default 'max-pkt-len' value set to zero, 'update_jumbo_frame_offload()'
> updates it to "RTE_ETHER_MTU + PMD specific Ethernet overhead" when it is
> zero.
> If '--max-pkt-len=N' argument provided, it will be used instead.
> And with each "port config all max-pkt-len" command, the
> 'DEV_RX_OFFLOAD_JUMBO_FRAME' offload flag, 'max-pkt-len' and MTU is
> updated.
>
> [1]
> --
> dpdk-testpmd -c 0xf -n 4 -- -i --max-pkt-len=9000 --tx-offloads=0x8000
> --rxq=4 --txq=4 --disable-rss
> testpmd> set verbose 3
> testpmd> port stop all
> testpmd> port config all max-pkt-len 1518 port start all
>
> // Got fail error info without this patch Configuring Port 0 (socket 1) Ethdev
> port_id=0 rx_queue_id=0, new added offloads 0x800 must be within per-
> queue offload capabilities 0x0 in rte_eth_rx_queue_setup() Fail to configure
> port 0 rx queues //<-- Fail error info;
> --
>
> Fixes: 761c4d66900f ("app/testpmd: fix max Rx packet length for VLAN
> packets")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Steve Yang
> Signed-off-by: Ferruh Yigit
> ---
>
> v5:
> * 'update_jumbo_frame_offload()' helper updated
> * check zero 'max-pkt-len' value
> * Update how queue offload flags updated
> * Update MTU if JUMBO_FRAME flag is not set
> * Default testpmd 'max-pkt-len' value set to zero
>
> Cc: lance.richard...@broadcom.com
> Cc: ouli...@huawei.com
> Cc: wis...@mellanox.com
> Cc: lihuis...@huawei.com
> ---
It fixed this bug indeed: https://bugs.dpdk.org/show_bug.cgi?id=625
Thanks
Acked-by: Wisam Jaddo
BRs,
Wisam Jaddo
Re: [dpdk-dev] DPDK Release Status Meeting 21/01/2021
On 1/21/2021 12:04 PM, Ferruh Yigit wrote: Meeting minutes of 21 January 2021 -- <...> Release Dates - * v21.02 dates * -rc1 is released on Tuesday, 19 January 2021 * http://inbox.dpdk.org/dev/4846307.Pfn0FrbqUJ@thomas/ * -rc2 Friday, 29 January 2021 * -rc3 Friday, 5 February 2021 * Release pushed to *Wednesday, 17 February 2021* * Release date may conflict with Chinese New Year, we need to discuss and define the release date offline in the mail list, please comment. Hi all, Following is the status of the Intel validation team, If the -rc2 & -rc3 quality is good, the Intel validation team can finish the -rc3 testing before holiday starts (11 February) so that release can be done on 17 February. If the code changes need to happen after 11 February, or if there are more than expected defects on the -rc3 testing, pushing the release to 26 February is the safest option to complete all testing after holidays. Taking into account that -rc3 is on 5 February, this gives us a little time to code freeze to make release on 17 February.
Re: [dpdk-dev] [PATCH v3 44/44] net/virtio: handle Virtio-user setup failure properly
On 1/26/21 7:36 AM, Xia, Chenbo wrote:
> Hi Maxime,
>
>> -Original Message-
>> From: Maxime Coquelin
>> Sent: Tuesday, January 26, 2021 1:15 AM
>> To: dev@dpdk.org; Xia, Chenbo ; olivier.m...@6wind.com;
>> amore...@redhat.com; david.march...@redhat.com
>> Cc: Maxime Coquelin
>> Subject: [PATCH v3 44/44] net/virtio: handle Virtio-user setup failure
>> properly
>>
>> This patch fixes virtio_user_dev_setup() error path,
>> by cleaning all resources it allocates. It introduces
>> virtio_user_dev_uninit_notify() that cleans all open
>> FDs. It implies assigning all FDs to -1 at init time.
>>
>> With these changes done, virtio_user_dev_init_notify()
>> can be simplified.
>>
>> Suggested-by: Adrian Moreno
>> Signed-off-by: Maxime Coquelin
>> ---
>> .../net/virtio/virtio_user/virtio_user_dev.c | 68 +--
>> 1 file changed, 47 insertions(+), 21 deletions(-)
>>
>> diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> index a1e32158bb..7c60022a26 100644
>> --- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> +++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> @@ -283,13 +283,7 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
>> int callfd;
>> int kickfd;
>>
>> -for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
>> -if (i >= dev->max_queue_pairs * 2) {
>> -dev->kickfds[i] = -1;
>> -dev->callfds[i] = -1;
>> -continue;
>> -}
>> -
>> +for (i = 0; i < dev->max_queue_pairs * 2; i++) {
>> /* May use invalid flag, but some backend uses kickfd and
>> * callfd as criteria to judge if dev is alive. so finally we
>> * use real event_fd.
>> @@ -297,25 +291,48 @@ virtio_user_dev_init_notify(struct virtio_user_dev
>> *dev)
>> callfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
>> if (callfd < 0) {
>> PMD_DRV_LOG(ERR, "(%s) callfd error, %s", dev->path,
>> strerror(errno));
>> -break;
>> +goto err;
>> }
>> kickfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
>> if (kickfd < 0) {
>> close(callfd);
>> PMD_DRV_LOG(ERR, "(%s) kickfd error, %s", dev->path,
>> strerror(errno));
>> -break;
>> +goto err;
>> }
>> dev->callfds[i] = callfd;
>> dev->kickfds[i] = kickfd;
>> }
>>
>> -if (i < VIRTIO_MAX_VIRTQUEUES) {
>> -for (j = 0; j < i; ++j) {
>> -close(dev->callfds[j]);
>> +return 0;
>> +err:
>> +for (j = 0; j < i; j++) {
>> +if (dev->kickfds[j] >= 0) {
>> close(dev->kickfds[j]);
>> +dev->kickfds[j] = -1;
>> }
>> +if (dev->callfds[j] >= 0) {
>> +close(dev->callfds[j]);
>> +dev->callfds[j] = -1;
>> +}
>> +}
>>
>> -return -1;
>> +return -1;
>> +}
>> +
>> +static int
>
> Use void?
yes.
>> +virtio_user_dev_uninit_notify(struct virtio_user_dev *dev)
>> +{
>> +uint32_t i;
>> +
>> +for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
>
> Do we need to loop for VIRTIO_MAX_VIRTQUEUES? Kick/call fds would only be set
> valid in virtio_user_dev_init_notify, and since virtio_user_dev_init_notify
> only
> loop for dev->max_queue_pairs * 2, this function will also loop for
> dev->max_queue_pairs * 2?
> Other fds should be -1, right?
Yes, given max_queue_pairs is never updated during the lifecycle, it is
safe to loop for max_queue_pairs.
I'll fix it in v4.
Thanks,
Maxime
> Thanks,
> Chenbo
>
>> +if (dev->kickfds[i] >= 0) {
>> +close(dev->kickfds[i]);
>> +dev->kickfds[i] = -1;
>> +}
>> +if (dev->callfds[i] >= 0) {
>> +close(dev->callfds[i]);
>> +dev->callfds[i] = -1;
>> +}
>> }
>>
>> return 0;
>> @@ -427,15 +444,22 @@ virtio_user_dev_setup(struct virtio_user_dev *dev)
>>
>> if (virtio_user_dev_init_notify(dev) < 0) {
>> PMD_INIT_LOG(ERR, "(%s) Failed to init notifiers\n", dev->path);
>> -return -1;
>> +goto destroy;
>> }
>>
>> if (virtio_user_fill_intr_handle(dev) < 0) {
>> PMD_INIT_LOG(ERR, "(%s) Failed to init interrupt handler\n",
>> dev-
>>> path);
>> -return -1;
>> +goto uninit;
>> }
>>
>> return 0;
>> +
>> +uninit:
>> +virtio_user_dev_uninit_notify(dev);
>> +destroy:
>> +dev->ops->destroy(dev);
>> +
>> +return -1;
>> }
>>
>> /* Use below macro to filter features from vhost backend */
>> @@ -466,9 +490,16 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
>> *path, int queues,
>> enum v
[dpdk-dev] [PATCH v1] net/virtio: fix compiling issue for vectorized NEON path
In file included from ../drivers/net/virtio/virtio_rxtx_packed.c:22:0:
../drivers/net/virtio/virtio_rxtx_packed_neon.h: In function
‘virtqueue_enqueue_batch_packed_vec’:
../drivers/net/virtio/virtio_rxtx_packed_neon.h:74:2: warning:
implicit declaration of function ‘vreinterpretq_p128_u32’
[-Wimplicit-function-declaration]
poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg, def_ref_seg));
^
The message shows ‘vreinterpretq_p128_u32’ instrisic is not supported
because an old version gcc (gcc 4.8.5) was used. So fix the issue with
implementing the logic with other intrinsics.
Bugzilla ID: 621
Fixes: 530887469350 ("net/virtio: add vectorized packed ring NEON Tx")
Fixes: 5971ce5e2a59 ("net/virtio: add vectorized packed ring NEON Rx")
Cc: sta...@dpdk.org
Signed-off-by: Joyce Kong
---
drivers/net/virtio/virtio_rxtx_packed_neon.h | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/virtio/virtio_rxtx_packed_neon.h
b/drivers/net/virtio/virtio_rxtx_packed_neon.h
index 01c77b712..00dd04277 100644
--- a/drivers/net/virtio/virtio_rxtx_packed_neon.h
+++ b/drivers/net/virtio/virtio_rxtx_packed_neon.h
@@ -71,8 +71,8 @@ virtqueue_enqueue_batch_packed_vec(struct virtnet_tx *txvq,
uint32x4_t def_ref_seg = vdupq_n_u32(0x10001);
/* Check refcnt and nb_segs. */
uint32x4_t ref_seg = vreinterpretq_u32_u8(vqtbl2q_u8(mbuf,
ref_seg_msk));
- poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg,
def_ref_seg));
- if (unlikely(cmp1))
+ uint64x2_t cmp1 = vreinterpretq_u64_u32(~vceqq_u32(ref_seg,
def_ref_seg));
+ if (unlikely(vgetq_lane_u64(cmp1, 0) || vgetq_lane_u64(cmp1, 1)))
return -1;
/* Check headroom is enough. */
@@ -225,10 +225,10 @@ virtqueue_dequeue_batch_packed_vec(struct virtnet_rx
*rxvq,
if (vq->vq_packed.used_wrap_counter)
v_used_flag = vdupq_n_u32(PACKED_FLAGS_MASK);
- poly128_t desc_stats = vreinterpretq_p128_u32(~vceqq_u32(v_flag,
v_used_flag));
+ uint64x2_t desc_stats = vreinterpretq_u64_u32(~vceqq_u32(v_flag,
v_used_flag));
/* Check all descs are used. */
- if (desc_stats)
+ if (unlikely(vgetq_lane_u64(desc_stats, 0) ||
vgetq_lane_u64(desc_stats, 1)))
return -1;
/* Load 2 mbuf pointers per time. */
--
2.30.0
Re: [dpdk-dev] [PATCH v3 02/44] bus/vdev: add driver IOVA VA mode requirement
Hi Maxime,
> -Original Message-
> From: Maxime Coquelin
> Sent: Tuesday, January 26, 2021 1:14 AM
> To: dev@dpdk.org; Xia, Chenbo ; olivier.m...@6wind.com;
> amore...@redhat.com; david.march...@redhat.com
> Cc: Maxime Coquelin
> Subject: [PATCH v3 02/44] bus/vdev: add driver IOVA VA mode requirement
>
> This patch adds driver flag in vdev bus driver so that
> vdev drivers can require VA IOVA mode to be used, which
> for example the case of Virtio-user PMD.
>
> The patch implements the .get_iommu_class() callback, that
> is called before devices probing to determine the IOVA mode
> to be used.
>
> It also adds a check right before the device is probed to
> ensure compatible IOVa mode has been selected.
s/IOVa/IOVA
>
> Signed-off-by: Maxime Coquelin
> ---
> drivers/bus/vdev/rte_bus_vdev.h | 4
> drivers/bus/vdev/vdev.c | 29 +
> 2 files changed, 33 insertions(+)
>
> diff --git a/drivers/bus/vdev/rte_bus_vdev.h b/drivers/bus/vdev/rte_bus_vdev.h
> index f99a41f825..c8b41e649c 100644
> --- a/drivers/bus/vdev/rte_bus_vdev.h
> +++ b/drivers/bus/vdev/rte_bus_vdev.h
> @@ -113,8 +113,12 @@ struct rte_vdev_driver {
> rte_vdev_remove_t *remove; /**< Virtual device remove function. */
> rte_vdev_dma_map_t *dma_map; /**< Virtual device DMA map function.
> */
> rte_vdev_dma_unmap_t *dma_unmap; /**< Virtual device DMA unmap function.
> */
> + uint32_t drv_flags;/**< Flags RTE_VDEV_DRV_*. */
I remember David mentioned that the comment above should be consistent with
others, which
also makes sense to me
Thanks,
Chenbo
> };
>
> +/** Device driver needs IOVA as VA and cannot work with IOVA as PA */
> +#define RTE_VDEV_DRV_NEED_IOVA_AS_VA 0x0001
> +
> /**
> * Register a virtual device driver.
> *
> diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
> index acfd78828f..9a673347ae 100644
> --- a/drivers/bus/vdev/vdev.c
> +++ b/drivers/bus/vdev/vdev.c
> @@ -189,6 +189,7 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
> {
> const char *name;
> struct rte_vdev_driver *driver;
> + enum rte_iova_mode iova_mode;
> int ret;
>
> if (rte_dev_is_probed(&dev->device))
> @@ -199,6 +200,14 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
>
> if (vdev_parse(name, &driver))
> return -1;
> +
> + iova_mode = rte_eal_iova_mode();
> + if ((driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA) && (iova_mode ==
> RTE_IOVA_PA)) {
> + VDEV_LOG(ERR, "%s requires VA IOVA mode but current mode is PA,
> not initializing",
> + name);
> + return -1;
> + }
> +
> ret = driver->probe(dev);
> if (ret == 0)
> dev->device.driver = &driver->driver;
> @@ -594,6 +603,25 @@ vdev_unplug(struct rte_device *dev)
> return rte_vdev_uninit(dev->name);
> }
>
> +static enum rte_iova_mode
> +vdev_get_iommu_class(void)
> +{
> + const char *name;
> + struct rte_vdev_device *dev;
> + struct rte_vdev_driver *driver;
> +
> + TAILQ_FOREACH(dev, &vdev_device_list, next) {
> + name = rte_vdev_device_name(dev);
> + if (vdev_parse(name, &driver))
> + continue;
> +
> + if (driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA)
> + return RTE_IOVA_VA;
> + }
> +
> + return RTE_IOVA_DC;
> +}
> +
> static struct rte_bus rte_vdev_bus = {
> .scan = vdev_scan,
> .probe = vdev_probe,
> @@ -603,6 +631,7 @@ static struct rte_bus rte_vdev_bus = {
> .parse = vdev_parse,
> .dma_map = vdev_dma_map,
> .dma_unmap = vdev_dma_unmap,
> + .get_iommu_class = vdev_get_iommu_class,
> .dev_iterate = rte_vdev_dev_iterate,
> };
>
> --
> 2.29.2
Re: [dpdk-dev] [PATCH v3 02/44] bus/vdev: add driver IOVA VA mode requirement
On 1/26/21 10:23 AM, Xia, Chenbo wrote:
> Hi Maxime,
>
>> -Original Message-
>> From: Maxime Coquelin
>> Sent: Tuesday, January 26, 2021 1:14 AM
>> To: dev@dpdk.org; Xia, Chenbo ; olivier.m...@6wind.com;
>> amore...@redhat.com; david.march...@redhat.com
>> Cc: Maxime Coquelin
>> Subject: [PATCH v3 02/44] bus/vdev: add driver IOVA VA mode requirement
>>
>> This patch adds driver flag in vdev bus driver so that
>> vdev drivers can require VA IOVA mode to be used, which
>> for example the case of Virtio-user PMD.
>>
>> The patch implements the .get_iommu_class() callback, that
>> is called before devices probing to determine the IOVA mode
>> to be used.
>>
>> It also adds a check right before the device is probed to
>> ensure compatible IOVa mode has been selected.
>
> s/IOVa/IOVA
Just spotted it while reworking the commit message :)
>>
>> Signed-off-by: Maxime Coquelin
>> ---
>> drivers/bus/vdev/rte_bus_vdev.h | 4
>> drivers/bus/vdev/vdev.c | 29 +
>> 2 files changed, 33 insertions(+)
>>
>> diff --git a/drivers/bus/vdev/rte_bus_vdev.h
>> b/drivers/bus/vdev/rte_bus_vdev.h
>> index f99a41f825..c8b41e649c 100644
>> --- a/drivers/bus/vdev/rte_bus_vdev.h
>> +++ b/drivers/bus/vdev/rte_bus_vdev.h
>> @@ -113,8 +113,12 @@ struct rte_vdev_driver {
>> rte_vdev_remove_t *remove; /**< Virtual device remove function. */
>> rte_vdev_dma_map_t *dma_map; /**< Virtual device DMA map function.
>> */
>> rte_vdev_dma_unmap_t *dma_unmap; /**< Virtual device DMA unmap function.
>> */
>> +uint32_t drv_flags;/**< Flags RTE_VDEV_DRV_*. */
>
> I remember David mentioned that the comment above should be consistent with
> others, which
> also makes sense to me
>
> Thanks,
> Chenbo
>
>> };
>>
>> +/** Device driver needs IOVA as VA and cannot work with IOVA as PA */
>> +#define RTE_VDEV_DRV_NEED_IOVA_AS_VA 0x0001
>> +
>> /**
>> * Register a virtual device driver.
>> *
>> diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
>> index acfd78828f..9a673347ae 100644
>> --- a/drivers/bus/vdev/vdev.c
>> +++ b/drivers/bus/vdev/vdev.c
>> @@ -189,6 +189,7 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
>> {
>> const char *name;
>> struct rte_vdev_driver *driver;
>> +enum rte_iova_mode iova_mode;
>> int ret;
>>
>> if (rte_dev_is_probed(&dev->device))
>> @@ -199,6 +200,14 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
>>
>> if (vdev_parse(name, &driver))
>> return -1;
>> +
>> +iova_mode = rte_eal_iova_mode();
>> +if ((driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA) && (iova_mode ==
>> RTE_IOVA_PA)) {
>> +VDEV_LOG(ERR, "%s requires VA IOVA mode but current mode is PA,
>> not initializing",
>> +name);
>> +return -1;
>> +}
>> +
>> ret = driver->probe(dev);
>> if (ret == 0)
>> dev->device.driver = &driver->driver;
>> @@ -594,6 +603,25 @@ vdev_unplug(struct rte_device *dev)
>> return rte_vdev_uninit(dev->name);
>> }
>>
>> +static enum rte_iova_mode
>> +vdev_get_iommu_class(void)
>> +{
>> +const char *name;
>> +struct rte_vdev_device *dev;
>> +struct rte_vdev_driver *driver;
>> +
>> +TAILQ_FOREACH(dev, &vdev_device_list, next) {
>> +name = rte_vdev_device_name(dev);
>> +if (vdev_parse(name, &driver))
>> +continue;
>> +
>> +if (driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA)
>> +return RTE_IOVA_VA;
>> +}
>> +
>> +return RTE_IOVA_DC;
>> +}
>> +
>> static struct rte_bus rte_vdev_bus = {
>> .scan = vdev_scan,
>> .probe = vdev_probe,
>> @@ -603,6 +631,7 @@ static struct rte_bus rte_vdev_bus = {
>> .parse = vdev_parse,
>> .dma_map = vdev_dma_map,
>> .dma_unmap = vdev_dma_unmap,
>> +.get_iommu_class = vdev_get_iommu_class,
>> .dev_iterate = rte_vdev_dev_iterate,
>> };
>>
>> --
>> 2.29.2
>
Re: [dpdk-dev] [dpdk-stable] [PATCH v1] net/virtio: fix compiling issue for vectorized NEON path
On Tue, Jan 26, 2021 at 10:20 AM Joyce Kong wrote:
>
> In file included from ../drivers/net/virtio/virtio_rxtx_packed.c:22:0:
> ../drivers/net/virtio/virtio_rxtx_packed_neon.h: In function
> ‘virtqueue_enqueue_batch_packed_vec’:
> ../drivers/net/virtio/virtio_rxtx_packed_neon.h:74:2: warning:
> implicit declaration of function ‘vreinterpretq_p128_u32’
> [-Wimplicit-function-declaration]
> poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg, def_ref_seg));
> ^
>
> The message shows ‘vreinterpretq_p128_u32’ instrisic is not supported
intrinsic*
> because an old version gcc (gcc 4.8.5) was used. So fix the issue with
> implementing the logic with other intrinsics.
>
> Bugzilla ID: 621
> Fixes: 530887469350 ("net/virtio: add vectorized packed ring NEON Tx")
> Fixes: 5971ce5e2a59 ("net/virtio: add vectorized packed ring NEON Rx")
> Cc: sta...@dpdk.org
No reason to copy sta...@dpdk.org.
Cc: reporter.
--
David Marchand
Re: [dpdk-dev] [PATCH] ethdev: add IPv6 DSCP option for modify field action
On Mon, Jan 25, 2021 at 9:21 PM Alexander Kozyrev wrote:
>
> > From: Stephen Hemminger on Monday, January 25,
> > 2021 22:44
> >
> > On Tue, 26 Jan 2021 03:38:24 +
> > Alexander Kozyrev wrote:
> >
> > > IPv6 DSCP field ID is missing from the original list of Field IDs
> > > for MODIFY_FIELD action. Add it to support IPv6 header fully.
> > >
> > > Fixes: 73b68f4c54a ("ethdev: introduce generic modify flow action")
> > >
> > > Signed-off-by: Alexander Kozyrev
> > > ---
> > > lib/librte_ethdev/rte_flow.h | 1 +
> > > 1 file changed, 1 insertion(+)
> > >
> > > diff --git a/lib/librte_ethdev/rte_flow.h b/lib/librte_ethdev/rte_flow.h
> > > index 46e8ee70ab..68c68cdd6c 100644
> > > --- a/lib/librte_ethdev/rte_flow.h
> > > +++ b/lib/librte_ethdev/rte_flow.h
> > > @@ -2842,6 +2842,7 @@ enum rte_flow_field_id {
> > > RTE_FLOW_FIELD_IPV4_TTL,
> > > RTE_FLOW_FIELD_IPV4_SRC,
> > > RTE_FLOW_FIELD_IPV4_DST,
> > > + RTE_FLOW_FIELD_IPV6_DSCP,
> > > RTE_FLOW_FIELD_IPV6_HOPLIMIT,
> > > RTE_FLOW_FIELD_IPV6_SRC,
> > > RTE_FLOW_FIELD_IPV6_DST,
> >
> > Adding field in middle of enum will break ABI.
>
> I added the rte_flow_field_id enum a week ago into 20.11-rc1.
21.02-rc1
> I believe it is not too late to make it right without breaking ABI, don't you
> think so?
Re: [dpdk-dev] [PATCH] ethdev: add IPv6 DSCP option for modify field action
On Mon, Jan 25, 2021 at 7:38 PM Alexander Kozyrev wrote:
>
> IPv6 DSCP field ID is missing from the original list of Field IDs
> for MODIFY_FIELD action. Add it to support IPv6 header fully.
>
> Fixes: 73b68f4c54a ("ethdev: introduce generic modify flow action")
>
> Signed-off-by: Alexander Kozyrev
Acked-by: Ajit Khaparde
> ---
> lib/librte_ethdev/rte_flow.h | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/lib/librte_ethdev/rte_flow.h b/lib/librte_ethdev/rte_flow.h
> index 46e8ee70ab..68c68cdd6c 100644
> --- a/lib/librte_ethdev/rte_flow.h
> +++ b/lib/librte_ethdev/rte_flow.h
> @@ -2842,6 +2842,7 @@ enum rte_flow_field_id {
> RTE_FLOW_FIELD_IPV4_TTL,
> RTE_FLOW_FIELD_IPV4_SRC,
> RTE_FLOW_FIELD_IPV4_DST,
> + RTE_FLOW_FIELD_IPV6_DSCP,
> RTE_FLOW_FIELD_IPV6_HOPLIMIT,
> RTE_FLOW_FIELD_IPV6_SRC,
> RTE_FLOW_FIELD_IPV6_DST,
> --
> 2.24.1
>
[dpdk-dev] [PATCH] net/i40e: fix register setting for hash enable
From: Dapeng Yu
The original code causes wrong value to be set into PFQF_HENA
register because unnecessary calling to get translated pctype
value for X722 NIC. The result is RSS cannot work.
So remove the unnecessary translation.
Fixes: ef4c16fd9148 ("net/i40e: refactor RSS flow")
Cc: sta...@dpdk.org
Signed-off-by: Dapeng Yu
---
drivers/net/i40e/i40e_hash.c | 4
1 file changed, 4 deletions(-)
diff --git a/drivers/net/i40e/i40e_hash.c b/drivers/net/i40e/i40e_hash.c
index 9271797a7..b1cb24f43 100644
--- a/drivers/net/i40e/i40e_hash.c
+++ b/drivers/net/i40e/i40e_hash.c
@@ -678,10 +678,6 @@ i40e_hash_enable_pctype(struct i40e_hw *hw,
{
uint32_t reg, reg_val, mask;
- /* For X722, get translated pctype in fd pctype register */
- if (hw->mac.type == I40E_MAC_X722)
- pctype = i40e_read_rx_ctl(hw, I40E_GLQF_FD_PCTYPES(pctype));
-
if (pctype < 32) {
mask = BIT(pctype);
reg = I40E_PFQF_HENA(0);
--
2.27.0
Re: [dpdk-dev] [PATCH] event/octeontx2: enhance Tx path cache locality
On Mon, Nov 23, 2020 at 2:48 AM wrote:
>
> From: Pavan Nikhilesh
>
> Enhance Tx path cache locality, remove current tag type and group
> stores from datapath to conserve store buffers.
>
> Signed-off-by: Pavan Nikhilesh
Applied to dpdk-next-net-eventdev/for-main. Thanks
> drivers/event/octeontx2/otx2_evdev.c | 15 ++
> drivers/event/octeontx2/otx2_evdev.h | 24 -
> drivers/event/octeontx2/otx2_worker.c | 42 +--
> drivers/event/octeontx2/otx2_worker.h | 32 +--
> drivers/event/octeontx2/otx2_worker_dual.c | 63 ++
> drivers/event/octeontx2/otx2_worker_dual.h | 2 -
> drivers/net/octeontx2/otx2_ethdev_sec_tx.h | 9 ++--
> 7 files changed, 74 insertions(+), 113 deletions(-)
>
> diff --git a/drivers/event/octeontx2/otx2_evdev.c
> b/drivers/event/octeontx2/otx2_evdev.c
> index 0fe014c24..14f16a68f 100644
> --- a/drivers/event/octeontx2/otx2_evdev.c
> +++ b/drivers/event/octeontx2/otx2_evdev.c
> @@ -833,10 +833,12 @@ sso_configure_dual_ports(const struct rte_eventdev
> *event_dev)
> ws->port = i;
> base = dev->bar2 + (RVU_BLOCK_ADDR_SSOW << 20 | vws << 12);
> sso_set_port_ops((struct otx2_ssogws *)&ws->ws_state[0],
> base);
> + ws->base[0] = base;
> vws++;
>
> base = dev->bar2 + (RVU_BLOCK_ADDR_SSOW << 20 | vws << 12);
> sso_set_port_ops((struct otx2_ssogws *)&ws->ws_state[1],
> base);
> + ws->base[1] = base;
> vws++;
>
> gws_cookie = ssogws_get_cookie(ws);
> @@ -909,6 +911,7 @@ sso_configure_ports(const struct rte_eventdev *event_dev)
> ws->port = i;
> base = dev->bar2 + (RVU_BLOCK_ADDR_SSOW << 20 | i << 12);
> sso_set_port_ops(ws, base);
> + ws->base = base;
>
> gws_cookie = ssogws_get_cookie(ws);
> gws_cookie->event_dev = event_dev;
> @@ -1447,20 +1450,12 @@ sso_cleanup(struct rte_eventdev *event_dev, uint8_t
> enable)
> ws = event_dev->data->ports[i];
> ssogws_reset((struct otx2_ssogws *)&ws->ws_state[0]);
> ssogws_reset((struct otx2_ssogws *)&ws->ws_state[1]);
> - ws->swtag_req = 0;
> ws->vws = 0;
> - ws->ws_state[0].cur_grp = 0;
> - ws->ws_state[0].cur_tt = SSO_SYNC_EMPTY;
> - ws->ws_state[1].cur_grp = 0;
> - ws->ws_state[1].cur_tt = SSO_SYNC_EMPTY;
> } else {
> struct otx2_ssogws *ws;
>
> ws = event_dev->data->ports[i];
> ssogws_reset(ws);
> - ws->swtag_req = 0;
> - ws->cur_grp = 0;
> - ws->cur_tt = SSO_SYNC_EMPTY;
> }
> }
>
> @@ -1479,8 +1474,6 @@ sso_cleanup(struct rte_eventdev *event_dev, uint8_t
> enable)
> otx2_write64(enable, ws->grps_base[i] +
> SSO_LF_GGRP_QCTL);
> }
> - ws->ws_state[0].cur_grp = 0;
> - ws->ws_state[0].cur_tt = SSO_SYNC_EMPTY;
> } else {
> struct otx2_ssogws *ws = event_dev->data->ports[0];
>
> @@ -1492,8 +1485,6 @@ sso_cleanup(struct rte_eventdev *event_dev, uint8_t
> enable)
> otx2_write64(enable, ws->grps_base[i] +
> SSO_LF_GGRP_QCTL);
> }
> - ws->cur_grp = 0;
> - ws->cur_tt = SSO_SYNC_EMPTY;
> }
>
> /* reset SSO GWS cache */
> diff --git a/drivers/event/octeontx2/otx2_evdev.h
> b/drivers/event/octeontx2/otx2_evdev.h
> index 0513cb81c..e381b9e52 100644
> --- a/drivers/event/octeontx2/otx2_evdev.h
> +++ b/drivers/event/octeontx2/otx2_evdev.h
> @@ -80,6 +80,8 @@
>
> #define OTX2_SSOW_GET_BASE_ADDR(_GW)((_GW) - SSOW_LF_GWS_OP_GET_WORK)
> #define OTX2_SSOW_TT_FROM_TAG(x) (((x) >> 32) & SSO_TT_EMPTY)
> +#define OTX2_SSOW_GRP_FROM_TAG(x) (((x) >> 36) & 0x3ff)
> +#define OTX2_SSOW_SWTAG_PEND(x)((x) & BIT_ULL(62))
>
> #define NSEC2USEC(__ns)((__ns) / 1E3)
> #define USEC2NSEC(__us) ((__us) * 1E3)
> @@ -169,25 +171,23 @@ struct otx2_sso_evdev {
> uintptr_t wqp_op;
> \
> uintptr_t swtag_flush_op;
> \
> uintptr_t swtag_norm_op;
> \
> - uintptr_t swtag_desched_op;
> \
> - uint8_t cur_tt;
> \
> - uint8_t cur_grp
> + uintptr_t swtag_desched_op;
>
> /* E
[dpdk-dev] [PATCH v2] net/virtio: fix compiling issue for vectorized NEON path
In file included from ../drivers/net/virtio/virtio_rxtx_packed.c:22:0:
../drivers/net/virtio/virtio_rxtx_packed_neon.h: In function
‘virtqueue_enqueue_batch_packed_vec’:
../drivers/net/virtio/virtio_rxtx_packed_neon.h:74:2: warning:
implicit declaration of function ‘vreinterpretq_p128_u32’
[-Wimplicit-function-declaration]
poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg, def_ref_seg));
^
The message shows ‘vreinterpretq_p128_u32’ intrinsic is not supported
because an old version gcc (gcc 4.8.5) is used. So fix the issue with
implementing the logic with other intrinsics.
Bugzilla ID: 621
Fixes: 530887469350 ("net/virtio: add vectorized packed ring NEON Tx")
Fixes: 5971ce5e2a59 ("net/virtio: add vectorized packed ring NEON Rx")
Reported-by: Subhi Masri
Signed-off-by: Joyce Kong
---
drivers/net/virtio/virtio_rxtx_packed_neon.h | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/virtio/virtio_rxtx_packed_neon.h
b/drivers/net/virtio/virtio_rxtx_packed_neon.h
index 01c77b712..00dd04277 100644
--- a/drivers/net/virtio/virtio_rxtx_packed_neon.h
+++ b/drivers/net/virtio/virtio_rxtx_packed_neon.h
@@ -71,8 +71,8 @@ virtqueue_enqueue_batch_packed_vec(struct virtnet_tx *txvq,
uint32x4_t def_ref_seg = vdupq_n_u32(0x10001);
/* Check refcnt and nb_segs. */
uint32x4_t ref_seg = vreinterpretq_u32_u8(vqtbl2q_u8(mbuf,
ref_seg_msk));
- poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg,
def_ref_seg));
- if (unlikely(cmp1))
+ uint64x2_t cmp1 = vreinterpretq_u64_u32(~vceqq_u32(ref_seg,
def_ref_seg));
+ if (unlikely(vgetq_lane_u64(cmp1, 0) || vgetq_lane_u64(cmp1, 1)))
return -1;
/* Check headroom is enough. */
@@ -225,10 +225,10 @@ virtqueue_dequeue_batch_packed_vec(struct virtnet_rx
*rxvq,
if (vq->vq_packed.used_wrap_counter)
v_used_flag = vdupq_n_u32(PACKED_FLAGS_MASK);
- poly128_t desc_stats = vreinterpretq_p128_u32(~vceqq_u32(v_flag,
v_used_flag));
+ uint64x2_t desc_stats = vreinterpretq_u64_u32(~vceqq_u32(v_flag,
v_used_flag));
/* Check all descs are used. */
- if (desc_stats)
+ if (unlikely(vgetq_lane_u64(desc_stats, 0) ||
vgetq_lane_u64(desc_stats, 1)))
return -1;
/* Load 2 mbuf pointers per time. */
--
2.30.0
[dpdk-dev] [PATCH v4 00/44] net/virtio: Virtio PMD rework
This V3 fixes comments from Chenbo on patch 44 and implements the ABI exception in patch 2. This series significantly rework Virtio PMD to improve the Virtio-user PMD and its backends integration. First part of the series removes the dependency of Virtio-user ethdev on Virtio PCI, by creating generic files, adding per-bus meta data, ... Main (if not single) functionnal change of this first part is to remove the hack for Virtio-user to work in IOVA as PA mode, this hack being very fragile. Second part of the series reworks Virtio-user internal, by reworking the requests handling so that vDPA and Kernel backends no more hack into being Vhost-user backend. It implies implementing new ops for all the request types. Also, all the backend specific actions are moved from the virtio_user_dev.c and virtio_user_ethdev.c to their backend files. Only functionnal change in this second part is making the Vhost-user server mode blocking at init time, as long as a client is not connected. The goal of this change is to make the Vhost-user support much more robust, as without blocking, the driver has to assume features that are going to be supported by the client, which is very fragile and error prone. As a side-effect, it also simplifies the logic nin several place of the virtio-user PMD. Main changes in v4: - Add ABI exception (David) - Close FDs only up to max_queue_pairs - virtio_user_dev_uninit_notify() to return void Main changes in v3: - Rename .intr_event to .intr_detect - Rework last patch, properly clean allocated resources on failure. - Rebase on top of latest net-next/main - Minor typo fixes in comments and log improvements Main changes in v2: === - Introduce vdev driver flag for drivers to require IOVA VA mode - Rebase on top of -rc1 changes - Fix regressions introduced in V1 (vhost-kernel broken, vhost-user reconnect...) - Various minor issues & typos fixed - Fix status feature issue introduced in v20.11, only reproduceable now that server mode is made blocking - Improve failure handling in Virtio-user - Improve logging Testing coverage (All passed) = - Virtio-pci PMD * Virtio PMD in guest with Vhost-user backend in host * Virtio PMD in guest with Vhost-kernel backend in host - Virtio-user PMD with Vhost-user backend * Vhost-user PMD server <-> Virtio-user client PMD IO loopback * Vhost-user PMD client <-> Virtio-user server PMD IO loopback * Vhost-user PMD client <-> Virtio-user server PMD reconnect - Virtio-user PMD with Vhost-kernel backend * iperf test case * Txonly testpmd - Virtio-user PMD with Vhost-vDPA backend * vdpa-sim (IO loopback) * CX-6 DX Kernel vDPA (Tx only) Maxime Coquelin (44): bus/vdev: add helper to get vdev from ethdev bus/vdev: add driver IOVA VA mode requirement net/virtio: fix getting old status on reconnect net/virtio: introduce Virtio bus type net/virtio: refactor virtio-user device net/virtio: introduce PCI device metadata net/virtio: move PCI device init in dedicated file net/virtio: move PCI specific dev init to PCI ethdev init net/virtio: move MSIX detection to PCI ethdev net/virtio: force IOVA as VA mode for Virtio-user net/virtio: store PCI type in Virtio device metadata net/virtio: add callback for device closing net/virtio: validate features at bus level net/virtio: remove bus type enum net/virtio: move PCI-specific fields to PCI device net/virtio: pack virtio HW struct net/virtio: move legacy IO to Virtio PCI net/virtio: introduce generic virtio header net/virtio: move features definition to generic header net/virtio: move virtqueue defines in generic header net/virtio: move config definitions to generic header net/virtio: make interrupt handling more generic net/virtio: move vring alignment to generic header net/virtio: remove last PCI refs in non-PCI code net/virtio: make Vhost-user request sender consistent net/virtio: add Virtio-user ops to set owner net/virtio: add Virtio-user features ops net/virtio: add Virtio-user protocol features ops net/virtio: add Virtio-user memory tables ops net/virtio: add Virtio-user vring setting ops net/virtio: add Virtio-user vring file ops net/virtio: add Virtio-user vring address ops net/virtio: add Virtio-user status ops net/virtio: remove useless request ops net/virtio: improve Virtio-user errors handling net/virtio: move Vhost-user requests to Vhost-user backend net/virtio: make server mode blocking net/virtio: move protocol features to Vhost-user net/virtio: introduce backend data net/virtio: move Vhost-user specifics to its backend net/virtio: move Vhost-kernel data to its backend net/virtio: move Vhost-vDPA data to its backend net/virtio: improve Vhost-user error logging net/virtio: handle Virtio-user setup failure properly devtools/libabigail.abignore | 2 + drivers/bus/vdev/rte_bus_vdev.h | 6 + drivers/bus/vdev/vdev.c
[dpdk-dev] [PATCH v4 01/44] bus/vdev: add helper to get vdev from ethdev
This patch adds an helper macro to get the rte_vdev_device
pointer from a rte_eth_dev pointer.
This is similar to RTE_ETH_DEV_TO_PCI().
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/bus/vdev/rte_bus_vdev.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/bus/vdev/rte_bus_vdev.h b/drivers/bus/vdev/rte_bus_vdev.h
index d14eeb41b0..f99a41f825 100644
--- a/drivers/bus/vdev/rte_bus_vdev.h
+++ b/drivers/bus/vdev/rte_bus_vdev.h
@@ -34,6 +34,8 @@ struct rte_vdev_device {
#define RTE_DEV_TO_VDEV_CONST(ptr) \
container_of(ptr, const struct rte_vdev_device, device)
+#define RTE_ETH_DEV_TO_VDEV(eth_dev) RTE_DEV_TO_VDEV((eth_dev)->device)
+
static inline const char *
rte_vdev_device_name(const struct rte_vdev_device *dev)
{
--
2.29.2
[dpdk-dev] [PATCH v4 02/44] bus/vdev: add driver IOVA VA mode requirement
This patch adds driver flag in vdev bus driver so that
vdev drivers can require VA IOVA mode to be used, which
for example the case of Virtio-user PMD.
The patch implements the .get_iommu_class() callback, that
is called before devices probing to determine the IOVA mode
to be used, and adds a check right before the device is
probed to ensure compatible IOVA mode has been selected.
It also adds a ABI exception rule to accommodate with an
update on the driver registration API
Signed-off-by: Maxime Coquelin
Signed-off-by: David Marchand
---
devtools/libabigail.abignore| 2 ++
drivers/bus/vdev/rte_bus_vdev.h | 4
drivers/bus/vdev/vdev.c | 29 +
3 files changed, 35 insertions(+)
diff --git a/devtools/libabigail.abignore b/devtools/libabigail.abignore
index 1dc84fa74b..170304c876 100644
--- a/devtools/libabigail.abignore
+++ b/devtools/libabigail.abignore
@@ -11,6 +11,8 @@
; Explicit ignore for driver-only ABI
[suppress_type]
name = eth_dev_ops
+[suppress_function]
+name_regexp = rte_vdev_(|un)register
; Ignore fields inserted in cacheline boundary of rte_cryptodev
[suppress_type]
diff --git a/drivers/bus/vdev/rte_bus_vdev.h b/drivers/bus/vdev/rte_bus_vdev.h
index f99a41f825..fc315d10fa 100644
--- a/drivers/bus/vdev/rte_bus_vdev.h
+++ b/drivers/bus/vdev/rte_bus_vdev.h
@@ -113,8 +113,12 @@ struct rte_vdev_driver {
rte_vdev_remove_t *remove; /**< Virtual device remove function. */
rte_vdev_dma_map_t *dma_map; /**< Virtual device DMA map function.
*/
rte_vdev_dma_unmap_t *dma_unmap; /**< Virtual device DMA unmap
function. */
+ uint32_t drv_flags; /**< Flags RTE_VDEV_DRV_*. */
};
+/** Device driver needs IOVA as VA and cannot work with IOVA as PA */
+#define RTE_VDEV_DRV_NEED_IOVA_AS_VA 0x0001
+
/**
* Register a virtual device driver.
*
diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
index acfd78828f..9a673347ae 100644
--- a/drivers/bus/vdev/vdev.c
+++ b/drivers/bus/vdev/vdev.c
@@ -189,6 +189,7 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
{
const char *name;
struct rte_vdev_driver *driver;
+ enum rte_iova_mode iova_mode;
int ret;
if (rte_dev_is_probed(&dev->device))
@@ -199,6 +200,14 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
if (vdev_parse(name, &driver))
return -1;
+
+ iova_mode = rte_eal_iova_mode();
+ if ((driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA) && (iova_mode ==
RTE_IOVA_PA)) {
+ VDEV_LOG(ERR, "%s requires VA IOVA mode but current mode is PA,
not initializing",
+ name);
+ return -1;
+ }
+
ret = driver->probe(dev);
if (ret == 0)
dev->device.driver = &driver->driver;
@@ -594,6 +603,25 @@ vdev_unplug(struct rte_device *dev)
return rte_vdev_uninit(dev->name);
}
+static enum rte_iova_mode
+vdev_get_iommu_class(void)
+{
+ const char *name;
+ struct rte_vdev_device *dev;
+ struct rte_vdev_driver *driver;
+
+ TAILQ_FOREACH(dev, &vdev_device_list, next) {
+ name = rte_vdev_device_name(dev);
+ if (vdev_parse(name, &driver))
+ continue;
+
+ if (driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA)
+ return RTE_IOVA_VA;
+ }
+
+ return RTE_IOVA_DC;
+}
+
static struct rte_bus rte_vdev_bus = {
.scan = vdev_scan,
.probe = vdev_probe,
@@ -603,6 +631,7 @@ static struct rte_bus rte_vdev_bus = {
.parse = vdev_parse,
.dma_map = vdev_dma_map,
.dma_unmap = vdev_dma_unmap,
+ .get_iommu_class = vdev_get_iommu_class,
.dev_iterate = rte_vdev_dev_iterate,
};
--
2.29.2
[dpdk-dev] [PATCH v4 03/44] net/virtio: fix getting old status on reconnect
This patch fixes getting reset status from the restarted
vhost-user backend in case of reconnection, instead of the
status at the time of the disconnection.
This issue was not spotted earlier because Vhost-user
protocol status feature was disabled in server mode.
Fixes: 47235f16505f ("net/virtio-user: set status on socket reconnect")
Cc: sta...@dpdk.org
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user_ethdev.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/virtio/virtio_user_ethdev.c
b/drivers/net/virtio/virtio_user_ethdev.c
index 241808cd8f..7abba1f1ee 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -77,7 +77,7 @@ virtio_user_server_reconnect(struct virtio_user_dev *dev)
return -1;
dev->vhostfd = connectfd;
- old_status = vtpci_get_status(hw);
+ old_status = dev->status;
vtpci_reset(hw);
--
2.29.2
[dpdk-dev] [PATCH v4 05/44] net/virtio: refactor virtio-user device
This patch moves the virtio_hw structure into the virtio_user_dev
structure, with the goal of making virtio_hw bus-agnostic.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c| 2 +-
drivers/net/virtio/virtio_pci.h | 1 -
.../net/virtio/virtio_user/virtio_user_dev.h | 1 +
drivers/net/virtio/virtio_user_ethdev.c | 64 ---
4 files changed, 29 insertions(+), 39 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 0ba10c274a..c8a01a46df 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -747,7 +747,7 @@ virtio_dev_close(struct rte_eth_dev *dev)
#ifdef RTE_VIRTIO_USER
if (hw->bus_type == VIRTIO_BUS_USER)
- virtio_user_dev_uninit(hw->virtio_user_dev);
+ virtio_user_dev_uninit(dev->data->dev_private);
else
#endif
if (dev->device) {
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 6388f0a74d..b35a596169 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -277,7 +277,6 @@ struct virtio_hw {
uint16_t*notify_base;
struct virtio_pci_common_cfg *common_cfg;
struct virtio_net_config *dev_cfg;
- void*virtio_user_dev;
/*
* App management thread and virtio interrupt handler thread
* both can change device state, this lock is meant to avoid
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.h
b/drivers/net/virtio/virtio_user/virtio_user_dev.h
index 3b5b6bc3ae..b3776cc7a0 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.h
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.h
@@ -24,6 +24,7 @@ struct virtio_user_queue {
};
struct virtio_user_dev {
+ struct virtio_hw hw;
enum virtio_user_backend_type backend_type;
/* for vhost_user backend */
int vhostfd;
diff --git a/drivers/net/virtio/virtio_user_ethdev.c
b/drivers/net/virtio/virtio_user_ethdev.c
index 441721b6ca..620ecb126c 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -26,13 +26,13 @@
#include "virtio_user/virtio_user_dev.h"
#include "virtio_user/vhost.h"
-#define virtio_user_get_dev(hw) \
- ((struct virtio_user_dev *)(hw)->virtio_user_dev)
+#define virtio_user_get_dev(hwp) container_of(hwp, struct virtio_user_dev, hw)
static void
-virtio_user_reset_queues_packed(struct rte_eth_dev *dev)
+virtio_user_reset_queues_packed(struct rte_eth_dev *eth_dev)
{
- struct virtio_hw *hw = dev->data->dev_private;
+ struct virtio_user_dev *dev = eth_dev->data->dev_private;
+ struct virtio_hw *hw = &dev->hw;
struct virtnet_rx *rxvq;
struct virtnet_tx *txvq;
uint16_t i;
@@ -48,14 +48,14 @@ virtio_user_reset_queues_packed(struct rte_eth_dev *dev)
rte_delay_ms(1);
/* Vring reset for each Tx queue and Rx queue. */
- for (i = 0; i < dev->data->nb_rx_queues; i++) {
- rxvq = dev->data->rx_queues[i];
+ for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
+ rxvq = eth_dev->data->rx_queues[i];
virtqueue_rxvq_reset_packed(rxvq->vq);
- virtio_dev_rx_queue_setup_finish(dev, i);
+ virtio_dev_rx_queue_setup_finish(eth_dev, i);
}
- for (i = 0; i < dev->data->nb_tx_queues; i++) {
- txvq = dev->data->tx_queues[i];
+ for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
+ txvq = eth_dev->data->tx_queues[i];
virtqueue_txvq_reset_packed(txvq->vq);
}
@@ -69,7 +69,7 @@ virtio_user_server_reconnect(struct virtio_user_dev *dev)
{
int ret, connectfd, old_status;
struct rte_eth_dev *eth_dev = &rte_eth_devices[dev->port_id];
- struct virtio_hw *hw = eth_dev->data->dev_private;
+ struct virtio_hw *hw = &dev->hw;
uint64_t protocol_features;
connectfd = accept(dev->listenfd, NULL, NULL);
@@ -605,21 +605,15 @@ virtio_user_eth_dev_alloc(struct rte_vdev_device *vdev)
struct virtio_hw *hw;
struct virtio_user_dev *dev;
- eth_dev = rte_eth_vdev_allocate(vdev, sizeof(*hw));
+ eth_dev = rte_eth_vdev_allocate(vdev, sizeof(*dev));
if (!eth_dev) {
PMD_INIT_LOG(ERR, "cannot alloc rte_eth_dev");
return NULL;
}
data = eth_dev->data;
- hw = eth_dev->data->dev_private;
-
- dev = rte_zmalloc(NULL, sizeof(*dev), 0);
- if (!dev) {
- PMD_INIT_LOG(ERR, "malloc virtio_user_dev failed");
- rte_eth_dev_release_port(eth_dev);
- return NULL;
- }
+ dev = eth_dev->data->dev_private;
+ hw = &dev->hw;
hw->port_id = data->port_id;
dev->port_id = data->port_id;
@@ -634,17 +628,13 @@ virtio_user_
[dpdk-dev] [PATCH v4 04/44] net/virtio: introduce Virtio bus type
This patch is preliminary work for introducing a bus layer
in Virtio PMD, in order to improve Virtio-user integration.
A new bus type is added to provide a unified way to distinguish
which bus type is used (PCI modern, PCI legacy or Virtio-user).
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c | 40 +
drivers/net/virtio/virtio_pci.c | 4 +--
drivers/net/virtio/virtio_pci.h | 9 +-
drivers/net/virtio/virtio_user_ethdev.c | 2 +-
4 files changed, 32 insertions(+), 23 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 92a3d4efa5..0ba10c274a 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -592,9 +592,9 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
* we use virtual address. And we need properly set _offset_, please see
* VIRTIO_MBUF_DATA_DMA_ADDR in virtqueue.h for more information.
*/
- if (!hw->virtio_user_dev)
+ if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN) {
vq->offset = offsetof(struct rte_mbuf, buf_iova);
- else {
+ } else if (hw->bus_type == VIRTIO_BUS_USER) {
vq->vq_ring_mem = (uintptr_t)mz->addr;
vq->offset = offsetof(struct rte_mbuf, buf_addr);
if (queue_type == VTNET_TQ)
@@ -746,13 +746,13 @@ virtio_dev_close(struct rte_eth_dev *dev)
virtio_free_queues(hw);
#ifdef RTE_VIRTIO_USER
- if (hw->virtio_user_dev)
+ if (hw->bus_type == VIRTIO_BUS_USER)
virtio_user_dev_uninit(hw->virtio_user_dev);
else
#endif
if (dev->device) {
rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(dev));
- if (!hw->modern)
+ if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY)
rte_pci_ioport_unmap(VTPCI_IO(hw));
}
@@ -1299,7 +1299,7 @@ virtio_intr_unmask(struct rte_eth_dev *dev)
if (rte_intr_ack(dev->intr_handle) < 0)
return -1;
- if (!hw->virtio_user_dev)
+ if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
return 0;
@@ -1313,7 +1313,7 @@ virtio_intr_enable(struct rte_eth_dev *dev)
if (rte_intr_enable(dev->intr_handle) < 0)
return -1;
- if (!hw->virtio_user_dev)
+ if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
return 0;
@@ -1327,7 +1327,7 @@ virtio_intr_disable(struct rte_eth_dev *dev)
if (rte_intr_disable(dev->intr_handle) < 0)
return -1;
- if (!hw->virtio_user_dev)
+ if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
return 0;
@@ -1368,13 +1368,13 @@ virtio_negotiate_features(struct virtio_hw *hw,
uint64_t req_features)
PMD_INIT_LOG(DEBUG, "features after negotiate = %" PRIx64,
hw->guest_features);
- if (hw->modern && !vtpci_with_feature(hw, VIRTIO_F_VERSION_1)) {
+ if (hw->bus_type == VIRTIO_BUS_PCI_MODERN && !vtpci_with_feature(hw,
VIRTIO_F_VERSION_1)) {
PMD_INIT_LOG(ERR,
"VIRTIO_F_VERSION_1 features is not enabled.");
return -1;
}
- if (hw->modern || hw->virtio_user_dev) {
+ if (hw->bus_type == VIRTIO_BUS_PCI_MODERN || hw->bus_type ==
VIRTIO_BUS_USER) {
vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_FEATURES_OK);
if (!(vtpci_get_status(hw) & VIRTIO_CONFIG_STATUS_FEATURES_OK))
{
PMD_INIT_LOG(ERR,
@@ -1709,7 +1709,7 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
hw->weak_barriers = !vtpci_with_feature(hw, VIRTIO_F_ORDER_PLATFORM);
- if (!hw->virtio_user_dev)
+ if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
/* If host does not support both status and MSI-X then disable LSC */
@@ -1856,7 +1856,7 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
static int
virtio_remap_pci(struct rte_pci_device *pci_dev, struct virtio_hw *hw)
{
- if (hw->modern) {
+ if (hw->bus_type == VIRTIO_BUS_PCI_MODERN) {
/*
* We don't have to re-parse the PCI config space, since
* rte_pci_map_device() makes sure the mapped address
@@ -1872,7 +1872,7 @@ virtio_remap_pci(struct rte_pci_device *pci_dev, struct
virtio_hw *hw)
PMD_INIT_LOG(DEBUG, "failed to map pci device!");
[dpdk-dev] [PATCH v4 06/44] net/virtio: introduce PCI device metadata
This patch initiate refactoring of Virtio PCI, by introducing
a new device structure for PCI-specific metadata.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c | 2 +-
drivers/net/virtio/virtio_pci.c| 2 ++
drivers/net/virtio/virtio_pci.h| 5 +
3 files changed, 8 insertions(+), 1 deletion(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index c8a01a46df..aafba15ac2 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -2154,7 +2154,7 @@ static int eth_virtio_pci_probe(struct rte_pci_driver
*pci_drv __rte_unused,
if (vdpa == 1)
return 1;
- return rte_eth_dev_pci_generic_probe(pci_dev, sizeof(struct virtio_hw),
+ return rte_eth_dev_pci_generic_probe(pci_dev, sizeof(struct
virtio_pci_dev),
eth_virtio_dev_init);
}
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 1692268f30..345d73f868 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -689,6 +689,8 @@ virtio_read_caps(struct rte_pci_device *dev, struct
virtio_hw *hw)
int
vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw)
{
+ RTE_BUILD_BUG_ON(offsetof(struct virtio_pci_dev, hw) != 0);
+
/*
* Try if we can succeed reading virtio pci caps, which exists
* only on modern pci device. If failed, we fallback to legacy
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index b35a596169..59f6688218 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -289,6 +289,11 @@ struct virtio_hw {
struct virtqueue **vqs;
};
+struct virtio_pci_dev {
+ struct virtio_hw hw;
+};
+
+#define virtio_pci_get_dev(hwp) container_of(hwp, struct virtio_pci_dev, hw)
/*
* While virtio_hw is stored in shared memory, this structure stores
--
2.29.2
[dpdk-dev] [PATCH v4 07/44] net/virtio: move PCI device init in dedicated file
This patch moves the PCI Ethernet device registration bits
in a dedicated patch. In following patches, more code will
be moved there, with the goal of making virtio_ethdev.c
truly bus-agnostic.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/meson.build | 1 +
drivers/net/virtio/virtio_ethdev.c | 117 +--
drivers/net/virtio/virtio_ethdev.h | 2 +
drivers/net/virtio/virtio_pci_ethdev.c | 149 +
4 files changed, 157 insertions(+), 112 deletions(-)
create mode 100644 drivers/net/virtio/virtio_pci_ethdev.c
diff --git a/drivers/net/virtio/meson.build b/drivers/net/virtio/meson.build
index 0e78d1243b..07e085376b 100644
--- a/drivers/net/virtio/meson.build
+++ b/drivers/net/virtio/meson.build
@@ -8,6 +8,7 @@ if is_windows
endif
sources += files('virtio_ethdev.c',
+ 'virtio_pci_ethdev.c',
'virtio_pci.c',
'virtio_rxtx.c',
'virtio_rxtx_simple.c',
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index aafba15ac2..ad7e9efd5f 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -38,17 +38,14 @@
#include "virtio_rxtx.h"
#include "virtio_user/virtio_user_dev.h"
-static int eth_virtio_dev_uninit(struct rte_eth_dev *eth_dev);
static int virtio_dev_configure(struct rte_eth_dev *dev);
static int virtio_dev_start(struct rte_eth_dev *dev);
-static int virtio_dev_stop(struct rte_eth_dev *dev);
static int virtio_dev_promiscuous_enable(struct rte_eth_dev *dev);
static int virtio_dev_promiscuous_disable(struct rte_eth_dev *dev);
static int virtio_dev_allmulticast_enable(struct rte_eth_dev *dev);
static int virtio_dev_allmulticast_disable(struct rte_eth_dev *dev);
static uint32_t virtio_dev_speed_capa_get(uint32_t speed);
static int virtio_dev_devargs_parse(struct rte_devargs *devargs,
- int *vdpa,
uint32_t *speed,
int *vectorized);
static int virtio_dev_info_get(struct rte_eth_dev *dev,
@@ -89,15 +86,6 @@ static int virtio_dev_queue_stats_mapping_set(
static void virtio_notify_peers(struct rte_eth_dev *dev);
static void virtio_ack_link_announce(struct rte_eth_dev *dev);
-/*
- * The set of PCI devices this driver supports
- */
-static const struct rte_pci_id pci_id_virtio_map[] = {
- { RTE_PCI_DEVICE(VIRTIO_PCI_VENDORID, VIRTIO_PCI_LEGACY_DEVICEID_NET) },
- { RTE_PCI_DEVICE(VIRTIO_PCI_VENDORID, VIRTIO_PCI_MODERN_DEVICEID_NET) },
- { .vendor_id = 0, /* sentinel */ },
-};
-
struct rte_virtio_xstats_name_off {
char name[RTE_ETH_XSTATS_NAME_SIZE];
unsigned offset;
@@ -714,7 +702,7 @@ virtio_alloc_queues(struct rte_eth_dev *dev)
static void virtio_queues_unbind_intr(struct rte_eth_dev *dev);
-static int
+int
virtio_dev_close(struct rte_eth_dev *dev)
{
struct virtio_hw *hw = dev->data->dev_private;
@@ -1932,8 +1920,7 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
return 0;
}
- ret = virtio_dev_devargs_parse(eth_dev->device->devargs,
-NULL, &speed, &vectorized);
+ ret = virtio_dev_devargs_parse(eth_dev->device->devargs, &speed,
&vectorized);
if (ret < 0)
return ret;
hw->speed = speed;
@@ -1995,36 +1982,6 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
return ret;
}
-static int
-eth_virtio_dev_uninit(struct rte_eth_dev *eth_dev)
-{
- int ret;
- PMD_INIT_FUNC_TRACE();
-
- if (rte_eal_process_type() == RTE_PROC_SECONDARY)
- return 0;
-
- ret = virtio_dev_stop(eth_dev);
- virtio_dev_close(eth_dev);
-
- PMD_INIT_LOG(DEBUG, "dev_uninit completed");
-
- return ret;
-}
-
-
-static int vdpa_check_handler(__rte_unused const char *key,
- const char *value, void *ret_val)
-{
- if (strcmp(value, "1") == 0)
- *(int *)ret_val = 1;
- else
- *(int *)ret_val = 0;
-
- return 0;
-}
-
-
static uint32_t
virtio_dev_speed_capa_get(uint32_t speed)
{
@@ -2062,10 +2019,8 @@ static int vectorized_check_handler(__rte_unused const
char *key,
}
#define VIRTIO_ARG_SPEED "speed"
-#define VIRTIO_ARG_VDPA "vdpa"
#define VIRTIO_ARG_VECTORIZED "vectorized"
-
static int
link_speed_handler(const char *key __rte_unused,
const char *value, void *ret_val)
@@ -2084,8 +2039,7 @@ link_speed_handler(const char *key __rte_unused,
static int
-virtio_dev_devargs_parse(struct rte_devargs *devargs, int *vdpa,
- uint32_t *speed, int *vectorized)
+virtio_dev_devargs_parse(struct rte_devargs *devargs, uint32_t *speed, int
*vectorized)
{
struct rte_kvargs *kvlist;
int ret = 0;
@@ -2098,18 +2052,7 @@ virtio_dev_devargs_parse(struct rte_devargs *devargs,
int *vdpa,
PMD_INIT_LOG(ERR, "error when parsing param");
return 0;
}
- if (vdpa && rte_kvargs_count(kvlist, VIRTIO_A
[dpdk-dev] [PATCH v4 08/44] net/virtio: move PCI specific dev init to PCI ethdev init
This patch moves the PCI specific initialization from
eth_virtio_dev_init() to eth_virtio_pci_init().
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio_ethdev.c | 63 +--
drivers/net/virtio/virtio_pci_ethdev.c | 71 +-
2 files changed, 71 insertions(+), 63 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index ad7e9efd5f..a3e81f336d 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1676,7 +1676,6 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
struct virtio_hw *hw = eth_dev->data->dev_private;
struct virtio_net_config *config;
struct virtio_net_config local_config;
- struct rte_pci_device *pci_dev = NULL;
int ret;
/* Reset the device although not necessary at startup */
@@ -1697,9 +1696,6 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
hw->weak_barriers = !vtpci_with_feature(hw, VIRTIO_F_ORDER_PLATFORM);
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
- pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
-
/* If host does not support both status and MSI-X then disable LSC */
if (vtpci_with_feature(hw, VIRTIO_NET_F_STATUS) &&
hw->use_msix != VIRTIO_MSIX_NONE)
@@ -1828,45 +1824,9 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
vtpci_reinit_complete(hw);
- if (pci_dev)
- PMD_INIT_LOG(DEBUG, "port %d vendorID=0x%x deviceID=0x%x",
- eth_dev->data->port_id, pci_dev->id.vendor_id,
- pci_dev->id.device_id);
-
return 0;
}
-/*
- * Remap the PCI device again (IO port map for legacy device and
- * memory map for modern device), so that the secondary process
- * could have the PCI initiated correctly.
- */
-static int
-virtio_remap_pci(struct rte_pci_device *pci_dev, struct virtio_hw *hw)
-{
- if (hw->bus_type == VIRTIO_BUS_PCI_MODERN) {
- /*
-* We don't have to re-parse the PCI config space, since
-* rte_pci_map_device() makes sure the mapped address
-* in secondary process would equal to the one mapped in
-* the primary process: error will be returned if that
-* requirement is not met.
-*
-* That said, we could simply reuse all cap pointers
-* (such as dev_cfg, common_cfg, etc.) parsed from the
-* primary process, which is stored in shared memory.
-*/
- if (rte_pci_map_device(pci_dev)) {
- PMD_INIT_LOG(DEBUG, "failed to map pci device!");
- return -1;
- }
- } else if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY) {
- if (rte_pci_ioport_map(pci_dev, 0, VTPCI_IO(hw)) < 0)
- return -1;
- }
-
- return 0;
-}
static void
virtio_set_vtpci_ops(struct virtio_hw *hw)
@@ -1909,17 +1869,11 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
eth_dev->rx_descriptor_done = virtio_dev_rx_queue_done;
if (rte_eal_process_type() == RTE_PROC_SECONDARY) {
- if (hw->bus_type != VIRTIO_BUS_USER) {
- ret = virtio_remap_pci(RTE_ETH_DEV_TO_PCI(eth_dev), hw);
- if (ret)
- return ret;
- }
-
virtio_set_vtpci_ops(hw);
set_rxtx_funcs(eth_dev);
-
return 0;
}
+
ret = virtio_dev_devargs_parse(eth_dev->device->devargs, &speed,
&vectorized);
if (ret < 0)
return ret;
@@ -1936,15 +1890,6 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
}
hw->port_id = eth_dev->data->port_id;
- /* For virtio_user case the hw->virtio_user_dev is populated by
-* virtio_user_eth_dev_alloc() before eth_virtio_dev_init() is called.
-*/
- if (hw->bus_type != VIRTIO_BUS_USER) {
- ret = vtpci_init(RTE_ETH_DEV_TO_PCI(eth_dev), hw);
- if (ret)
- goto err_vtpci_init;
- }
-
rte_spinlock_init(&hw->state_lock);
/* reset device and negotiate default features */
@@ -1971,12 +1916,6 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
return 0;
err_virtio_init:
- if (hw->bus_type == VIRTIO_BUS_PCI_MODERN || hw->bus_type ==
VIRTIO_BUS_PCI_LEGACY) {
- rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(eth_dev));
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY)
- rte_pci_ioport_unmap(VTPCI_IO(hw));
- }
-err_vtpci_init:
rte_free(eth_dev->data->mac_addrs);
eth_dev->data->mac_addrs = NULL;
[dpdk-dev] [PATCH v4 09/44] net/virtio: move MSIX detection to PCI ethdev
This patch introduces a new callback to notify the bus
driver some interrupt related operation was done.
This is used by Virtio PCI driver to check msix status.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c | 12 +--
drivers/net/virtio/virtio_pci.c| 120 ++---
drivers/net/virtio/virtio_pci.h| 6 +-
drivers/net/virtio/virtio_pci_ethdev.c | 2 +
4 files changed, 82 insertions(+), 58 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index a3e81f336d..52eb878c42 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1287,8 +1287,8 @@ virtio_intr_unmask(struct rte_eth_dev *dev)
if (rte_intr_ack(dev->intr_handle) < 0)
return -1;
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
- hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
+ if (VTPCI_OPS(hw)->intr_detect)
+ VTPCI_OPS(hw)->intr_detect(hw);
return 0;
}
@@ -1301,8 +1301,8 @@ virtio_intr_enable(struct rte_eth_dev *dev)
if (rte_intr_enable(dev->intr_handle) < 0)
return -1;
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
- hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
+ if (VTPCI_OPS(hw)->intr_detect)
+ VTPCI_OPS(hw)->intr_detect(hw);
return 0;
}
@@ -1315,8 +1315,8 @@ virtio_intr_disable(struct rte_eth_dev *dev)
if (rte_intr_disable(dev->intr_handle) < 0)
return -1;
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN)
- hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
+ if (VTPCI_OPS(hw)->intr_detect)
+ VTPCI_OPS(hw)->intr_detect(hw);
return 0;
}
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 345d73f868..556be1e3da 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -47,6 +47,56 @@ check_vq_phys_addr_ok(struct virtqueue *vq)
return 1;
}
+#define PCI_MSIX_ENABLE 0x8000
+
+static enum virtio_msix_status
+vtpci_msix_detect(struct rte_pci_device *dev)
+{
+ uint8_t pos;
+ int ret;
+
+ ret = rte_pci_read_config(dev, &pos, 1, PCI_CAPABILITY_LIST);
+ if (ret != 1) {
+ PMD_INIT_LOG(DEBUG,
+"failed to read pci capability list, ret %d", ret);
+ return VIRTIO_MSIX_NONE;
+ }
+
+ while (pos) {
+ uint8_t cap[2];
+
+ ret = rte_pci_read_config(dev, cap, sizeof(cap), pos);
+ if (ret != sizeof(cap)) {
+ PMD_INIT_LOG(DEBUG,
+"failed to read pci cap at pos: %x ret %d",
+pos, ret);
+ break;
+ }
+
+ if (cap[0] == PCI_CAP_ID_MSIX) {
+ uint16_t flags;
+
+ ret = rte_pci_read_config(dev, &flags, sizeof(flags),
+ pos + sizeof(cap));
+ if (ret != sizeof(flags)) {
+ PMD_INIT_LOG(DEBUG,
+"failed to read pci cap at pos:"
+" %x ret %d", pos + 2, ret);
+ break;
+ }
+
+ if (flags & PCI_MSIX_ENABLE)
+ return VIRTIO_MSIX_ENABLED;
+ else
+ return VIRTIO_MSIX_DISABLED;
+ }
+
+ pos = cap[1];
+ }
+
+ return VIRTIO_MSIX_NONE;
+}
+
/*
* Since we are in legacy mode:
* http://ozlabs.org/~rusty/virtio-spec/virtio-0.9.5.pdf
@@ -241,6 +291,12 @@ legacy_notify_queue(struct virtio_hw *hw, struct virtqueue
*vq)
VIRTIO_PCI_QUEUE_NOTIFY);
}
+static void
+legacy_intr_detect(struct virtio_hw *hw)
+{
+ hw->use_msix = vtpci_msix_detect(VTPCI_DEV(hw));
+}
+
const struct virtio_pci_ops legacy_ops = {
.read_dev_cfg = legacy_read_dev_config,
.write_dev_cfg = legacy_write_dev_config,
@@ -255,6 +311,7 @@ const struct virtio_pci_ops legacy_ops = {
.setup_queue= legacy_setup_queue,
.del_queue = legacy_del_queue,
.notify_queue = legacy_notify_queue,
+ .intr_detect= legacy_intr_detect,
};
static inline void
@@ -446,6 +503,14 @@ modern_notify_queue(struct virtio_hw *hw, struct virtqueue
*vq)
rte_write32(notify_data, vq->notify_addr);
}
+
+
+static void
+modern_intr_detect(struct virtio_hw *hw)
+{
+ hw->use_msix = vtpci_msix_detect(VTPCI_DEV(hw));
+}
+
const struct virtio_pci_ops modern_ops = {
[dpdk-dev] [PATCH v4 10/44] net/virtio: force IOVA as VA mode for Virtio-user
At least Vhost-user backend of Virtio-user PMD requires
IOVA as VA mode. Until now, it was implemented as a hack
by forcing to use mbuf's buf_addr field instead of buf_iova.
This patch removes all this logic and just fails probing
if IOVA as VA mode is not selected. It simplifies the
code overall, and removes some bus-specific logic from
generic virtio_ethdev.c.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c | 15 -
drivers/net/virtio/virtio_rxtx.c | 34
drivers/net/virtio/virtio_rxtx_packed.h | 2 +-
drivers/net/virtio/virtio_rxtx_packed_avx.h | 8 ++---
drivers/net/virtio/virtio_rxtx_packed_neon.h | 8 ++---
drivers/net/virtio/virtio_rxtx_simple.h | 3 +-
drivers/net/virtio/virtio_user_ethdev.c | 1 +
drivers/net/virtio/virtqueue.h | 25 +-
8 files changed, 26 insertions(+), 70 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 52eb878c42..fb789460e8 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -576,21 +576,6 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
hw->cvq = cvq;
}
- /* For virtio_user case (that is when hw->virtio_user_dev is not NULL),
-* we use virtual address. And we need properly set _offset_, please see
-* VIRTIO_MBUF_DATA_DMA_ADDR in virtqueue.h for more information.
-*/
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY || hw->bus_type ==
VIRTIO_BUS_PCI_MODERN) {
- vq->offset = offsetof(struct rte_mbuf, buf_iova);
- } else if (hw->bus_type == VIRTIO_BUS_USER) {
- vq->vq_ring_mem = (uintptr_t)mz->addr;
- vq->offset = offsetof(struct rte_mbuf, buf_addr);
- if (queue_type == VTNET_TQ)
- txvq->virtio_net_hdr_mem = (uintptr_t)hdr_mz->addr;
- else if (queue_type == VTNET_CQ)
- cvq->virtio_net_hdr_mem = (uintptr_t)hdr_mz->addr;
- }
-
if (queue_type == VTNET_TQ) {
struct virtio_tx_region *txr;
unsigned int i;
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index 622d4bf201..6875c8fbee 100644
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -271,13 +271,10 @@ virtqueue_enqueue_refill_inorder(struct virtqueue *vq,
dxp->cookie = (void *)cookies[i];
dxp->ndescs = 1;
- start_dp[idx].addr =
- VIRTIO_MBUF_ADDR(cookies[i], vq) +
- RTE_PKTMBUF_HEADROOM - hw->vtnet_hdr_size;
- start_dp[idx].len =
- cookies[i]->buf_len -
- RTE_PKTMBUF_HEADROOM +
- hw->vtnet_hdr_size;
+ start_dp[idx].addr = cookies[i]->buf_iova +
+ RTE_PKTMBUF_HEADROOM - hw->vtnet_hdr_size;
+ start_dp[idx].len = cookies[i]->buf_len -
+ RTE_PKTMBUF_HEADROOM + hw->vtnet_hdr_size;
start_dp[idx].flags = VRING_DESC_F_WRITE;
vq_update_avail_ring(vq, idx);
@@ -313,12 +310,10 @@ virtqueue_enqueue_recv_refill(struct virtqueue *vq,
struct rte_mbuf **cookie,
dxp->cookie = (void *)cookie[i];
dxp->ndescs = 1;
- start_dp[idx].addr =
- VIRTIO_MBUF_ADDR(cookie[i], vq) +
+ start_dp[idx].addr = cookie[i]->buf_iova +
RTE_PKTMBUF_HEADROOM - hw->vtnet_hdr_size;
- start_dp[idx].len =
- cookie[i]->buf_len - RTE_PKTMBUF_HEADROOM +
- hw->vtnet_hdr_size;
+ start_dp[idx].len = cookie[i]->buf_len -
+ RTE_PKTMBUF_HEADROOM + hw->vtnet_hdr_size;
start_dp[idx].flags = VRING_DESC_F_WRITE;
vq->vq_desc_head_idx = start_dp[idx].next;
vq_update_avail_ring(vq, idx);
@@ -355,10 +350,10 @@ virtqueue_enqueue_recv_refill_packed(struct virtqueue *vq,
dxp->cookie = (void *)cookie[i];
dxp->ndescs = 1;
- start_dp[idx].addr = VIRTIO_MBUF_ADDR(cookie[i], vq) +
- RTE_PKTMBUF_HEADROOM - hw->vtnet_hdr_size;
- start_dp[idx].len = cookie[i]->buf_len - RTE_PKTMBUF_HEADROOM
- + hw->vtnet_hdr_size;
+ start_dp[idx].addr = cookie[i]->buf_iova +
+ RTE_PKTMBUF_HEADROOM - hw->vtnet_hdr_size;
+ start_dp[idx].len = cookie[i]->buf_len -
+ RTE_PKTMBUF_HEADROOM + hw->vtnet_hdr_size;
vq->vq_desc_head_idx = dxp->next;
if (vq->vq_desc_head_idx == VQ_RING_DESC_CHAIN_
[dpdk-dev] [PATCH v4 11/44] net/virtio: store PCI type in Virtio device metadata
Going further in making the Virtio ethdev layer bus agnostic,
this patch adds a boolean in the Virtio PCI device metadata.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio_pci.c| 18 +++---
drivers/net/virtio/virtio_pci.h| 3 ++-
drivers/net/virtio/virtio_pci_ethdev.c | 12 +++-
3 files changed, 20 insertions(+), 13 deletions(-)
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 556be1e3da..6d9c712fd5 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -751,8 +751,10 @@ virtio_read_caps(struct rte_pci_device *dev, struct
virtio_hw *hw)
* Return 0 on success.
*/
int
-vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw)
+vtpci_init(struct rte_pci_device *pci_dev, struct virtio_pci_dev *dev)
{
+ struct virtio_hw *hw = &dev->hw;
+
RTE_BUILD_BUG_ON(offsetof(struct virtio_pci_dev, hw) != 0);
/*
@@ -760,19 +762,20 @@ vtpci_init(struct rte_pci_device *dev, struct virtio_hw
*hw)
* only on modern pci device. If failed, we fallback to legacy
* virtio handling.
*/
- if (virtio_read_caps(dev, hw) == 0) {
+ if (virtio_read_caps(pci_dev, hw) == 0) {
PMD_INIT_LOG(INFO, "modern virtio pci detected.");
virtio_hw_internal[hw->port_id].vtpci_ops = &modern_ops;
hw->bus_type = VIRTIO_BUS_PCI_MODERN;
+ dev->modern = true;
goto msix_detect;
}
PMD_INIT_LOG(INFO, "trying with legacy virtio pci.");
- if (rte_pci_ioport_map(dev, 0, VTPCI_IO(hw)) < 0) {
- rte_pci_unmap_device(dev);
- if (dev->kdrv == RTE_PCI_KDRV_UNKNOWN &&
- (!dev->device.devargs ||
-dev->device.devargs->bus !=
+ if (rte_pci_ioport_map(pci_dev, 0, VTPCI_IO(hw)) < 0) {
+ rte_pci_unmap_device(pci_dev);
+ if (pci_dev->kdrv == RTE_PCI_KDRV_UNKNOWN &&
+ (!pci_dev->device.devargs ||
+pci_dev->device.devargs->bus !=
rte_bus_find_by_name("pci"))) {
PMD_INIT_LOG(INFO,
"skip kernel managed virtio device.");
@@ -783,6 +786,7 @@ vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw)
virtio_hw_internal[hw->port_id].vtpci_ops = &legacy_ops;
hw->bus_type = VIRTIO_BUS_PCI_LEGACY;
+ dev->modern = false;
msix_detect:
VTPCI_OPS(hw)->intr_detect(hw);
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index b29bbb8074..4c22692414 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -292,6 +292,7 @@ struct virtio_hw {
struct virtio_pci_dev {
struct virtio_hw hw;
+ bool modern;
};
#define virtio_pci_get_dev(hwp) container_of(hwp, struct virtio_pci_dev, hw)
@@ -371,7 +372,7 @@ vtpci_packed_queue(struct virtio_hw *hw)
/*
* Function declaration from virtio_pci.c
*/
-int vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw);
+int vtpci_init(struct rte_pci_device *pci_dev, struct virtio_pci_dev *dev);
void vtpci_reset(struct virtio_hw *);
void vtpci_reinit_complete(struct virtio_hw *);
diff --git a/drivers/net/virtio/virtio_pci_ethdev.c
b/drivers/net/virtio/virtio_pci_ethdev.c
index 045b134ef2..076a5dbced 100644
--- a/drivers/net/virtio/virtio_pci_ethdev.c
+++ b/drivers/net/virtio/virtio_pci_ethdev.c
@@ -39,9 +39,11 @@ static const struct rte_pci_id pci_id_virtio_map[] = {
* could have the PCI initiated correctly.
*/
static int
-virtio_remap_pci(struct rte_pci_device *pci_dev, struct virtio_hw *hw)
+virtio_remap_pci(struct rte_pci_device *pci_dev, struct virtio_pci_dev *dev)
{
- if (hw->bus_type == VIRTIO_BUS_PCI_MODERN) {
+ struct virtio_hw *hw = &dev->hw;
+
+ if (dev->modern) {
/*
* We don't have to re-parse the PCI config space, since
* rte_pci_map_device() makes sure the mapped address
@@ -57,7 +59,7 @@ virtio_remap_pci(struct rte_pci_device *pci_dev, struct
virtio_hw *hw)
PMD_INIT_LOG(DEBUG, "failed to map pci device!");
return -1;
}
- } else if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY) {
+ } else {
if (rte_pci_ioport_map(pci_dev, 0, VTPCI_IO(hw)) < 0)
return -1;
}
@@ -76,13 +78,13 @@ eth_virtio_pci_init(struct rte_eth_dev *eth_dev)
VTPCI_DEV(hw) = pci_dev;
if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
- ret = vtpci_init(RTE_ETH_DEV_TO_PCI(eth_dev), hw);
+ ret = vtpci_init(RTE_ETH_DEV_TO_PCI(eth_dev), dev);
if (ret) {
PMD_INIT_LOG(ERR, "Failed to init PCI device\n");
return -1;
[dpdk-dev] [PATCH v4 12/44] net/virtio: add callback for device closing
This patch introduces a new callback for device closing,
making virtio_dev_close() bus-agnostic.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/meson.build | 2 --
drivers/net/virtio/virtio_ethdev.c | 13 +
drivers/net/virtio/virtio_pci.c | 25 +
drivers/net/virtio/virtio_pci.h | 2 ++
drivers/net/virtio/virtio_user_ethdev.c | 11 +++
5 files changed, 39 insertions(+), 14 deletions(-)
diff --git a/drivers/net/virtio/meson.build b/drivers/net/virtio/meson.build
index 07e085376b..f2873d6180 100644
--- a/drivers/net/virtio/meson.build
+++ b/drivers/net/virtio/meson.build
@@ -44,8 +44,6 @@ elif arch_subdir == 'arm' and
host_machine.cpu_family().startswith('aarch64')
endif
if is_linux
- dpdk_conf.set('RTE_VIRTIO_USER', 1)
-
sources += files('virtio_user_ethdev.c',
'virtio_user/vhost_kernel.c',
'virtio_user/vhost_kernel_tap.c',
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index fb789460e8..84edcd4724 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -718,18 +718,7 @@ virtio_dev_close(struct rte_eth_dev *dev)
virtio_dev_free_mbufs(dev);
virtio_free_queues(hw);
-#ifdef RTE_VIRTIO_USER
- if (hw->bus_type == VIRTIO_BUS_USER)
- virtio_user_dev_uninit(dev->data->dev_private);
- else
-#endif
- if (dev->device) {
- rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(dev));
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY)
- rte_pci_ioport_unmap(VTPCI_IO(hw));
- }
-
- return 0;
+ return VTPCI_OPS(hw)->dev_close(hw);
}
static int
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 6d9c712fd5..ea4ab381a6 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -297,6 +297,17 @@ legacy_intr_detect(struct virtio_hw *hw)
hw->use_msix = vtpci_msix_detect(VTPCI_DEV(hw));
}
+static int
+legacy_dev_close(struct virtio_hw *hw)
+{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ rte_pci_unmap_device(dev->pci_dev);
+ rte_pci_ioport_unmap(VTPCI_IO(hw));
+
+ return 0;
+}
+
const struct virtio_pci_ops legacy_ops = {
.read_dev_cfg = legacy_read_dev_config,
.write_dev_cfg = legacy_write_dev_config,
@@ -312,6 +323,7 @@ const struct virtio_pci_ops legacy_ops = {
.del_queue = legacy_del_queue,
.notify_queue = legacy_notify_queue,
.intr_detect= legacy_intr_detect,
+ .dev_close = legacy_dev_close,
};
static inline void
@@ -511,6 +523,16 @@ modern_intr_detect(struct virtio_hw *hw)
hw->use_msix = vtpci_msix_detect(VTPCI_DEV(hw));
}
+static int
+modern_dev_close(struct virtio_hw *hw)
+{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ rte_pci_unmap_device(dev->pci_dev);
+
+ return 0;
+}
+
const struct virtio_pci_ops modern_ops = {
.read_dev_cfg = modern_read_dev_config,
.write_dev_cfg = modern_write_dev_config,
@@ -526,6 +548,7 @@ const struct virtio_pci_ops modern_ops = {
.del_queue = modern_del_queue,
.notify_queue = modern_notify_queue,
.intr_detect= modern_intr_detect,
+ .dev_close = modern_dev_close,
};
@@ -757,6 +780,8 @@ vtpci_init(struct rte_pci_device *pci_dev, struct
virtio_pci_dev *dev)
RTE_BUILD_BUG_ON(offsetof(struct virtio_pci_dev, hw) != 0);
+ dev->pci_dev = pci_dev;
+
/*
* Try if we can succeed reading virtio pci caps, which exists
* only on modern pci device. If failed, we fallback to legacy
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 4c22692414..0515bbb247 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -240,6 +240,7 @@ struct virtio_pci_ops {
void (*del_queue)(struct virtio_hw *hw, struct virtqueue *vq);
void (*notify_queue)(struct virtio_hw *hw, struct virtqueue *vq);
void (*intr_detect)(struct virtio_hw *hw);
+ int (*dev_close)(struct virtio_hw *hw);
};
struct virtio_net_config;
@@ -292,6 +293,7 @@ struct virtio_hw {
struct virtio_pci_dev {
struct virtio_hw hw;
+ struct rte_pci_device *pci_dev;
bool modern;
};
diff --git a/drivers/net/virtio/virtio_user_ethdev.c
b/drivers/net/virtio/virtio_user_ethdev.c
index 241fe373b9..3cbf310c03 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -462,6 +462,16 @@ virtio_user_notify_queue(struct virtio_hw *hw, struct
virtqueue *vq)
strerror(errno));
}
+static int
+virtio_user_dev_close(struct virtio_hw *hw)
+{
+ struct virtio_user_dev *dev = virtio_user_get_dev(hw)
[dpdk-dev] [PATCH v4 13/44] net/virtio: validate features at bus level
This patch provides a new callback for the bus type
to validate negotiated features are compatible with it.
Only user for now is PCI modern bus type, which implies
that the device supports Virtio 1.0+.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c | 11 ---
drivers/net/virtio/virtio_pci.c | 19 +++
drivers/net/virtio/virtio_pci.h | 1 +
drivers/net/virtio/virtio_user_ethdev.c | 7 +++
4 files changed, 31 insertions(+), 7 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 84edcd4724..72f527144e 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1330,17 +1330,14 @@ virtio_negotiate_features(struct virtio_hw *hw,
uint64_t req_features)
PMD_INIT_LOG(DEBUG, "features after negotiate = %" PRIx64,
hw->guest_features);
- if (hw->bus_type == VIRTIO_BUS_PCI_MODERN && !vtpci_with_feature(hw,
VIRTIO_F_VERSION_1)) {
- PMD_INIT_LOG(ERR,
- "VIRTIO_F_VERSION_1 features is not enabled.");
+ if (VTPCI_OPS(hw)->features_ok(hw) < 0)
return -1;
- }
- if (hw->bus_type == VIRTIO_BUS_PCI_MODERN || hw->bus_type ==
VIRTIO_BUS_USER) {
+ if (vtpci_with_feature(hw, VIRTIO_F_VERSION_1)) {
vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_FEATURES_OK);
+
if (!(vtpci_get_status(hw) & VIRTIO_CONFIG_STATUS_FEATURES_OK))
{
- PMD_INIT_LOG(ERR,
- "failed to set FEATURES_OK status!");
+ PMD_INIT_LOG(ERR, "Failed to set FEATURES_OK status!");
return -1;
}
}
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index ea4ab381a6..1c8d4b7ddd 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -201,6 +201,12 @@ legacy_set_features(struct virtio_hw *hw, uint64_t
features)
VIRTIO_PCI_GUEST_FEATURES);
}
+static int
+legacy_features_ok(struct virtio_hw *hw __rte_unused)
+{
+ return 0;
+}
+
static uint8_t
legacy_get_status(struct virtio_hw *hw)
{
@@ -315,6 +321,7 @@ const struct virtio_pci_ops legacy_ops = {
.set_status = legacy_set_status,
.get_features = legacy_get_features,
.set_features = legacy_set_features,
+ .features_ok= legacy_features_ok,
.get_isr= legacy_get_isr,
.set_config_irq = legacy_set_config_irq,
.set_queue_irq = legacy_set_queue_irq,
@@ -389,6 +396,17 @@ modern_set_features(struct virtio_hw *hw, uint64_t
features)
&hw->common_cfg->guest_feature);
}
+static int
+modern_features_ok(struct virtio_hw *hw)
+{
+ if (!vtpci_with_feature(hw, VIRTIO_F_VERSION_1)) {
+ PMD_INIT_LOG(ERR, "Version 1+ required with modern devices\n");
+ return -1;
+ }
+
+ return 0;
+}
+
static uint8_t
modern_get_status(struct virtio_hw *hw)
{
@@ -540,6 +558,7 @@ const struct virtio_pci_ops modern_ops = {
.set_status = modern_set_status,
.get_features = modern_get_features,
.set_features = modern_set_features,
+ .features_ok= modern_features_ok,
.get_isr= modern_get_isr,
.set_config_irq = modern_set_config_irq,
.set_queue_irq = modern_set_queue_irq,
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 0515bbb247..6aacc942fc 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -227,6 +227,7 @@ struct virtio_pci_ops {
uint64_t (*get_features)(struct virtio_hw *hw);
void (*set_features)(struct virtio_hw *hw, uint64_t features);
+ int (*features_ok)(struct virtio_hw *hw);
uint8_t (*get_isr)(struct virtio_hw *hw);
diff --git a/drivers/net/virtio/virtio_user_ethdev.c
b/drivers/net/virtio/virtio_user_ethdev.c
index 3cbf310c03..bf958de571 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -327,6 +327,12 @@ virtio_user_set_features(struct virtio_hw *hw, uint64_t
features)
dev->features = features & dev->device_features;
}
+static int
+virtio_user_features_ok(struct virtio_hw *hw __rte_unused)
+{
+ return 0;
+}
+
static uint8_t
virtio_user_get_isr(struct virtio_hw *hw __rte_unused)
{
@@ -479,6 +485,7 @@ const struct virtio_pci_ops virtio_user_ops = {
.set_status = virtio_user_set_status,
.get_features = virtio_user_get_features,
.set_features = virtio_user_set_features,
+ .features_ok= virtio_user_features_ok,
.get_isr= virtio_user_get_isr,
.set_config_irq = virtio_user_set_config_irq,
.set_queue_irq = virtio_user_set_queue_irq,
--
2.29.2
[dpdk-dev] [PATCH v4 14/44] net/virtio: remove bus type enum
Bus type awareness at the generic ethdev level is no
more needed as previous patches have made it bus-agnostic.
This patch removes it from struct virtio_hw.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio_ethdev.c | 18 --
drivers/net/virtio/virtio_pci.c | 2 --
drivers/net/virtio/virtio_pci.h | 8
drivers/net/virtio/virtio_pci_ethdev.c | 7 ++-
drivers/net/virtio/virtio_user_ethdev.c | 5 -
5 files changed, 10 insertions(+), 30 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 72f527144e..c46fe4adf6 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1798,23 +1798,6 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
return 0;
}
-
-static void
-virtio_set_vtpci_ops(struct virtio_hw *hw)
-{
-#ifdef RTE_VIRTIO_USER
- if (hw->bus_type == VIRTIO_BUS_USER)
- VTPCI_OPS(hw) = &virtio_user_ops;
- else
-#endif
- if (hw->bus_type == VIRTIO_BUS_PCI_MODERN)
- VTPCI_OPS(hw) = &modern_ops;
- else if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY)
- VTPCI_OPS(hw) = &legacy_ops;
-
- return;
-}
-
/*
* This function is based on probe() function in virtio_pci.c
* It returns 0 on success.
@@ -1840,7 +1823,6 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
eth_dev->rx_descriptor_done = virtio_dev_rx_queue_done;
if (rte_eal_process_type() == RTE_PROC_SECONDARY) {
- virtio_set_vtpci_ops(hw);
set_rxtx_funcs(eth_dev);
return 0;
}
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 1c8d4b7ddd..20599774a7 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -809,7 +809,6 @@ vtpci_init(struct rte_pci_device *pci_dev, struct
virtio_pci_dev *dev)
if (virtio_read_caps(pci_dev, hw) == 0) {
PMD_INIT_LOG(INFO, "modern virtio pci detected.");
virtio_hw_internal[hw->port_id].vtpci_ops = &modern_ops;
- hw->bus_type = VIRTIO_BUS_PCI_MODERN;
dev->modern = true;
goto msix_detect;
}
@@ -829,7 +828,6 @@ vtpci_init(struct rte_pci_device *pci_dev, struct
virtio_pci_dev *dev)
}
virtio_hw_internal[hw->port_id].vtpci_ops = &legacy_ops;
- hw->bus_type = VIRTIO_BUS_PCI_LEGACY;
dev->modern = false;
msix_detect:
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 6aacc942fc..2cede4a100 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -246,15 +246,7 @@ struct virtio_pci_ops {
struct virtio_net_config;
-enum virtio_bus_type {
- VIRTIO_BUS_UNKNOWN,
- VIRTIO_BUS_PCI_LEGACY,
- VIRTIO_BUS_PCI_MODERN,
- VIRTIO_BUS_USER,
-};
-
struct virtio_hw {
- enum virtio_bus_type bus_type;
struct virtnet_ctl *cvq;
uint64_treq_guest_features;
uint64_tguest_features;
diff --git a/drivers/net/virtio/virtio_pci_ethdev.c
b/drivers/net/virtio/virtio_pci_ethdev.c
index 076a5dbced..8cc6561914 100644
--- a/drivers/net/virtio/virtio_pci_ethdev.c
+++ b/drivers/net/virtio/virtio_pci_ethdev.c
@@ -84,6 +84,11 @@ eth_virtio_pci_init(struct rte_eth_dev *eth_dev)
return -1;
}
} else {
+ if (dev->modern)
+ VTPCI_OPS(hw) = &modern_ops;
+ else
+ VTPCI_OPS(hw) = &legacy_ops;
+
ret = virtio_remap_pci(RTE_ETH_DEV_TO_PCI(eth_dev), dev);
if (ret < 0) {
PMD_INIT_LOG(ERR, "Failed to remap PCI device\n");
@@ -105,7 +110,7 @@ eth_virtio_pci_init(struct rte_eth_dev *eth_dev)
err_unmap:
rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(eth_dev));
- if (hw->bus_type == VIRTIO_BUS_PCI_LEGACY)
+ if (!dev->modern)
rte_pci_ioport_unmap(VTPCI_IO(hw));
return ret;
diff --git a/drivers/net/virtio/virtio_user_ethdev.c
b/drivers/net/virtio/virtio_user_ethdev.c
index bf958de571..61880a8e02 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -641,7 +641,6 @@ virtio_user_eth_dev_alloc(struct rte_vdev_device *vdev)
* Here just pretend that we support msix.
*/
hw->use_msix = 1;
- hw->bus_type = VIRTIO_BUS_USER;
hw->use_vec_rx = 0;
hw->use_vec_tx = 0;
hw->use_inorder_rx = 0;
@@ -691,6 +690,10 @@ virtio_user_pmd_probe(struct rte_vdev_device *vdev)
return -1;
}
+ dev = eth_dev->data->dev_private;
+ hw = &dev->hw;
+ VTPCI_OPS(hw) = &virtio_user_ops;
+
if (eth_virtio_dev_init(eth_dev)
[dpdk-dev] [PATCH v4 15/44] net/virtio: move PCI-specific fields to PCI device
This patch moves the fields from virtio_hw structure that
are PCI-specific to virtio_pci_dev_struct.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio_pci.c | 139 ++--
drivers/net/virtio/virtio_pci.h | 10 +--
2 files changed, 85 insertions(+), 64 deletions(-)
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 20599774a7..b347e5fbc0 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -344,18 +344,19 @@ static void
modern_read_dev_config(struct virtio_hw *hw, size_t offset,
void *dst, int length)
{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
int i;
uint8_t *p;
uint8_t old_gen, new_gen;
do {
- old_gen = rte_read8(&hw->common_cfg->config_generation);
+ old_gen = rte_read8(&dev->common_cfg->config_generation);
p = dst;
for (i = 0; i < length; i++)
- *p++ = rte_read8((uint8_t *)hw->dev_cfg + offset + i);
+ *p++ = rte_read8((uint8_t *)dev->dev_cfg + offset + i);
- new_gen = rte_read8(&hw->common_cfg->config_generation);
+ new_gen = rte_read8(&dev->common_cfg->config_generation);
} while (old_gen != new_gen);
}
@@ -363,23 +364,25 @@ static void
modern_write_dev_config(struct virtio_hw *hw, size_t offset,
const void *src, int length)
{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
int i;
const uint8_t *p = src;
for (i = 0; i < length; i++)
- rte_write8((*p++), (((uint8_t *)hw->dev_cfg) + offset + i));
+ rte_write8((*p++), (((uint8_t *)dev->dev_cfg) + offset + i));
}
static uint64_t
modern_get_features(struct virtio_hw *hw)
{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
uint32_t features_lo, features_hi;
- rte_write32(0, &hw->common_cfg->device_feature_select);
- features_lo = rte_read32(&hw->common_cfg->device_feature);
+ rte_write32(0, &dev->common_cfg->device_feature_select);
+ features_lo = rte_read32(&dev->common_cfg->device_feature);
- rte_write32(1, &hw->common_cfg->device_feature_select);
- features_hi = rte_read32(&hw->common_cfg->device_feature);
+ rte_write32(1, &dev->common_cfg->device_feature_select);
+ features_hi = rte_read32(&dev->common_cfg->device_feature);
return ((uint64_t)features_hi << 32) | features_lo;
}
@@ -387,13 +390,15 @@ modern_get_features(struct virtio_hw *hw)
static void
modern_set_features(struct virtio_hw *hw, uint64_t features)
{
- rte_write32(0, &hw->common_cfg->guest_feature_select);
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ rte_write32(0, &dev->common_cfg->guest_feature_select);
rte_write32(features & ((1ULL << 32) - 1),
- &hw->common_cfg->guest_feature);
+ &dev->common_cfg->guest_feature);
- rte_write32(1, &hw->common_cfg->guest_feature_select);
+ rte_write32(1, &dev->common_cfg->guest_feature_select);
rte_write32(features >> 32,
- &hw->common_cfg->guest_feature);
+ &dev->common_cfg->guest_feature);
}
static int
@@ -410,46 +415,59 @@ modern_features_ok(struct virtio_hw *hw)
static uint8_t
modern_get_status(struct virtio_hw *hw)
{
- return rte_read8(&hw->common_cfg->device_status);
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ return rte_read8(&dev->common_cfg->device_status);
}
static void
modern_set_status(struct virtio_hw *hw, uint8_t status)
{
- rte_write8(status, &hw->common_cfg->device_status);
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ rte_write8(status, &dev->common_cfg->device_status);
}
static uint8_t
modern_get_isr(struct virtio_hw *hw)
{
- return rte_read8(hw->isr);
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ return rte_read8(dev->isr);
}
static uint16_t
modern_set_config_irq(struct virtio_hw *hw, uint16_t vec)
{
- rte_write16(vec, &hw->common_cfg->msix_config);
- return rte_read16(&hw->common_cfg->msix_config);
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ rte_write16(vec, &dev->common_cfg->msix_config);
+ return rte_read16(&dev->common_cfg->msix_config);
}
static uint16_t
modern_set_queue_irq(struct virtio_hw *hw, struct virtqueue *vq, uint16_t vec)
{
- rte_write16(vq->vq_queue_index, &hw->common_cfg->queue_select);
- rte_write16(vec, &hw->common_cfg->queue_msix_vector);
- return rte_read16(&hw->common_cfg->queue_msix_vector);
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ rte_write16(vq->vq_queue_index, &dev->common_cfg->queue_select);
+ rte
[dpdk-dev] [PATCH v4 16/44] net/virtio: pack virtio HW struct
This patch improves the virtio_hw struct packing,
going from 88 down to 80 bytes with a 6 bytes hole in
the end of the first cacheline. Fields only used in the
slow path are placed in the end, so that hot path only
uses the first cacheline.
The patch also changes booleans fields to uint8_t type, and
fix inconsistencies in their assignments.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_ethdev.c | 12
drivers/net/virtio/virtio_pci.h| 45 +++---
drivers/net/virtio/virtqueue.h | 2 +-
3 files changed, 29 insertions(+), 30 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index c46fe4adf6..c1d7b14dda 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -699,7 +699,7 @@ virtio_dev_close(struct rte_eth_dev *dev)
if (!hw->opened)
return 0;
- hw->opened = false;
+ hw->opened = 0;
/* reset the NIC */
if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
@@ -1864,7 +1864,7 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
}
}
- hw->opened = true;
+ hw->opened = 1;
return 0;
@@ -1973,7 +1973,7 @@ virtio_dev_devargs_parse(struct rte_devargs *devargs,
uint32_t *speed, int *vect
return ret;
}
-static bool
+static uint8_t
rx_offload_enabled(struct virtio_hw *hw)
{
return vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM) ||
@@ -1981,7 +1981,7 @@ rx_offload_enabled(struct virtio_hw *hw)
vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6);
}
-static bool
+static uint8_t
tx_offload_enabled(struct virtio_hw *hw)
{
return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
@@ -2267,7 +2267,7 @@ virtio_dev_start(struct rte_eth_dev *dev)
}
set_rxtx_funcs(dev);
- hw->started = true;
+ hw->started = 1;
/* Initialize Link state */
virtio_dev_link_update(dev, 0);
@@ -2336,7 +2336,7 @@ virtio_dev_stop(struct rte_eth_dev *dev)
rte_spinlock_lock(&hw->state_lock);
if (!hw->started)
goto out_unlock;
- hw->started = false;
+ hw->started = 0;
if (intr_conf->lsc || intr_conf->rxq) {
virtio_intr_disable(dev);
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 4f3690032b..15f8144fc6 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -247,26 +247,25 @@ struct virtio_pci_ops {
struct virtio_net_config;
struct virtio_hw {
- struct virtnet_ctl *cvq;
- uint64_treq_guest_features;
- uint64_tguest_features;
- uint32_tmax_queue_pairs;
- boolstarted;
- uint16_tmax_mtu;
- uint16_tvtnet_hdr_size;
- uint8_t vlan_strip;
- uint8_t use_msix;
- uint8_t use_vec_rx;
- uint8_t use_vec_tx;
- uint8_t use_inorder_rx;
- uint8_t use_inorder_tx;
- uint8_t weak_barriers;
- boolhas_tx_offload;
- boolhas_rx_offload;
- uint16_tport_id;
- uint8_t mac_addr[RTE_ETHER_ADDR_LEN];
- uint32_tspeed; /* link speed in MB */
- uint8_t duplex;
+ struct virtqueue **vqs;
+ uint64_t guest_features;
+ uint16_t vtnet_hdr_size;
+ uint8_t started;
+ uint8_t weak_barriers;
+ uint8_t vlan_strip;
+ uint8_t has_tx_offload;
+ uint8_t has_rx_offload;
+ uint8_t use_vec_rx;
+ uint8_t use_vec_tx;
+ uint8_t use_inorder_rx;
+ uint8_t use_inorder_tx;
+ uint8_t opened;
+ uint16_t port_id;
+ uint8_t mac_addr[RTE_ETHER_ADDR_LEN];
+ uint32_t speed; /* link speed in MB */
+ uint8_t duplex;
+ uint8_t use_msix;
+ uint16_t max_mtu;
/*
* App management thread and virtio interrupt handler thread
* both can change device state, this lock is meant to avoid
@@ -274,9 +273,9 @@ struct virtio_hw {
*/
rte_spinlock_t state_lock;
struct rte_mbuf **inject_pkts;
- boolopened;
-
- struct virtqueue **vqs;
+ uint16_t max_queue_pairs;
+ uint64_t req_guest_features;
+ struct virtnet_ctl *cvq;
};
struct virtio_pci_dev {
diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
index 7611317581..3a9ce29069 100644
--- a/drivers/net/virtio/virtqueue.h
+++ b/drivers/net/virtio/virtqueue.h
@@ -615,7 +615,7 @@ virtqueue_notify(struct virtqueue *vq)
static inline void
virtqueue_xmit_offload(struct virtio_net_hdr *hdr,
struct rte_mbuf *cookie,
- bool offload)
+ uint8_t offload)
{
if (offload) {
if (cookie->ol_flags & PKT_TX_TCP_SEG)
--
2.29.2
[dpdk-dev] [PATCH v4 17/44] net/virtio: move legacy IO to Virtio PCI
This patch moves Virtio PCI legacy IO handling to
virtio_pci.c. Two functions are created so that
virtio_pci_ethdev does not have to care about it.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio_pci.c| 28 --
drivers/net/virtio/virtio_pci.h| 9 +++--
drivers/net/virtio/virtio_pci_ethdev.c | 6 ++
3 files changed, 31 insertions(+), 12 deletions(-)
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index b347e5fbc0..3fe0631a30 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -31,6 +31,15 @@
#define VIRTIO_PCI_CONFIG(hw) \
(((hw)->use_msix == VIRTIO_MSIX_ENABLED) ? 24 : 20)
+
+struct virtio_pci_internal {
+ struct rte_pci_ioport io;
+};
+
+#define VTPCI_IO(hw) (&virtio_pci_internal[(hw)->port_id].io)
+
+struct virtio_pci_internal virtio_pci_internal[RTE_MAX_ETHPORTS];
+
static inline int
check_vq_phys_addr_ok(struct virtqueue *vq)
{
@@ -300,7 +309,9 @@ legacy_notify_queue(struct virtio_hw *hw, struct virtqueue
*vq)
static void
legacy_intr_detect(struct virtio_hw *hw)
{
- hw->use_msix = vtpci_msix_detect(VTPCI_DEV(hw));
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ hw->use_msix = vtpci_msix_detect(dev->pci_dev);
}
static int
@@ -558,7 +569,9 @@ modern_notify_queue(struct virtio_hw *hw, struct virtqueue
*vq)
static void
modern_intr_detect(struct virtio_hw *hw)
{
- hw->use_msix = vtpci_msix_detect(VTPCI_DEV(hw));
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ hw->use_msix = vtpci_msix_detect(dev->pci_dev);
}
static int
@@ -857,3 +870,14 @@ vtpci_init(struct rte_pci_device *pci_dev, struct
virtio_pci_dev *dev)
return 0;
}
+void vtpci_legacy_ioport_unmap(struct virtio_hw *hw)
+{
+ rte_pci_ioport_unmap(VTPCI_IO(hw));
+}
+
+int vtpci_legacy_ioport_map(struct virtio_hw *hw)
+{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
+
+ return rte_pci_ioport_map(dev->pci_dev, 0, VTPCI_IO(hw));
+}
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 15f8144fc6..def8faca72 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -298,18 +298,12 @@ struct virtio_pci_dev {
*/
struct virtio_hw_internal {
const struct virtio_pci_ops *vtpci_ops;
- struct rte_pci_ioport io;
- struct rte_pci_device *dev;
};
#define VTPCI_OPS(hw) (virtio_hw_internal[(hw)->port_id].vtpci_ops)
-#define VTPCI_IO(hw) (&virtio_hw_internal[(hw)->port_id].io)
-#define VTPCI_DEV(hw) (virtio_hw_internal[(hw)->port_id].dev)
-
extern struct virtio_hw_internal virtio_hw_internal[RTE_MAX_ETHPORTS];
-
/*
* This structure is just a reference to read
* net device specific config space; it just a chodu structure
@@ -382,6 +376,9 @@ void vtpci_read_dev_config(struct virtio_hw *, size_t, void
*, int);
uint8_t vtpci_isr(struct virtio_hw *);
+void vtpci_legacy_ioport_unmap(struct virtio_hw *hw);
+int vtpci_legacy_ioport_map(struct virtio_hw *hw);
+
extern const struct virtio_pci_ops legacy_ops;
extern const struct virtio_pci_ops modern_ops;
extern const struct virtio_pci_ops virtio_user_ops;
diff --git a/drivers/net/virtio/virtio_pci_ethdev.c
b/drivers/net/virtio/virtio_pci_ethdev.c
index 8cc6561914..9fe59feb51 100644
--- a/drivers/net/virtio/virtio_pci_ethdev.c
+++ b/drivers/net/virtio/virtio_pci_ethdev.c
@@ -60,7 +60,7 @@ virtio_remap_pci(struct rte_pci_device *pci_dev, struct
virtio_pci_dev *dev)
return -1;
}
} else {
- if (rte_pci_ioport_map(pci_dev, 0, VTPCI_IO(hw)) < 0)
+ if (vtpci_legacy_ioport_map(hw) < 0)
return -1;
}
@@ -75,8 +75,6 @@ eth_virtio_pci_init(struct rte_eth_dev *eth_dev)
struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
int ret;
- VTPCI_DEV(hw) = pci_dev;
-
if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
ret = vtpci_init(RTE_ETH_DEV_TO_PCI(eth_dev), dev);
if (ret) {
@@ -111,7 +109,7 @@ eth_virtio_pci_init(struct rte_eth_dev *eth_dev)
err_unmap:
rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(eth_dev));
if (!dev->modern)
- rte_pci_ioport_unmap(VTPCI_IO(hw));
+ vtpci_legacy_ioport_unmap(hw);
return ret;
}
--
2.29.2
[dpdk-dev] [PATCH v4 18/44] net/virtio: introduce generic virtio header
This patch moves virtio_hw and virtio callbacks into
a generic virtio header, now that they have been
curated from PCI references.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio.h | 74 ++
drivers/net/virtio/virtio_ethdev.c | 32 +-
drivers/net/virtio/virtio_pci.c | 28 -
drivers/net/virtio/virtio_pci.h | 84 ++---
drivers/net/virtio/virtio_pci_ethdev.c | 5 +-
drivers/net/virtio/virtio_user_ethdev.c | 8 +--
drivers/net/virtio/virtqueue.h | 2 +-
7 files changed, 116 insertions(+), 117 deletions(-)
create mode 100644 drivers/net/virtio/virtio.h
diff --git a/drivers/net/virtio/virtio.h b/drivers/net/virtio/virtio.h
new file mode 100644
index 00..e55e3329f9
--- /dev/null
+++ b/drivers/net/virtio/virtio.h
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2014 Intel Corporation
+ * Copyright(c) 2021 Red Hat, Inc.
+ */
+
+#ifndef _VIRTIO_H_
+#define _VIRTIO_H_
+
+#include
+
+struct virtio_hw {
+ struct virtqueue **vqs;
+ uint64_t guest_features;
+ uint16_t vtnet_hdr_size;
+ uint8_t started;
+ uint8_t weak_barriers;
+ uint8_t vlan_strip;
+ uint8_t has_tx_offload;
+ uint8_t has_rx_offload;
+ uint8_t use_vec_rx;
+ uint8_t use_vec_tx;
+ uint8_t use_inorder_rx;
+ uint8_t use_inorder_tx;
+ uint8_t opened;
+ uint16_t port_id;
+ uint8_t mac_addr[RTE_ETHER_ADDR_LEN];
+ uint32_t speed; /* link speed in MB */
+ uint8_t duplex;
+ uint8_t use_msix;
+ uint16_t max_mtu;
+ /*
+* App management thread and virtio interrupt handler thread
+* both can change device state, this lock is meant to avoid
+* such a contention.
+*/
+ rte_spinlock_t state_lock;
+ struct rte_mbuf **inject_pkts;
+ uint16_t max_queue_pairs;
+ uint64_t req_guest_features;
+ struct virtnet_ctl *cvq;
+};
+
+struct virtio_ops {
+ void (*read_dev_cfg)(struct virtio_hw *hw, size_t offset, void *dst,
int len);
+ void (*write_dev_cfg)(struct virtio_hw *hw, size_t offset, const void
*src, int len);
+ uint8_t (*get_status)(struct virtio_hw *hw);
+ void (*set_status)(struct virtio_hw *hw, uint8_t status);
+ uint64_t (*get_features)(struct virtio_hw *hw);
+ void (*set_features)(struct virtio_hw *hw, uint64_t features);
+ int (*features_ok)(struct virtio_hw *hw);
+ uint8_t (*get_isr)(struct virtio_hw *hw);
+ uint16_t (*set_config_irq)(struct virtio_hw *hw, uint16_t vec);
+ uint16_t (*set_queue_irq)(struct virtio_hw *hw, struct virtqueue *vq,
uint16_t vec);
+ uint16_t (*get_queue_num)(struct virtio_hw *hw, uint16_t queue_id);
+ int (*setup_queue)(struct virtio_hw *hw, struct virtqueue *vq);
+ void (*del_queue)(struct virtio_hw *hw, struct virtqueue *vq);
+ void (*notify_queue)(struct virtio_hw *hw, struct virtqueue *vq);
+ void (*intr_detect)(struct virtio_hw *hw);
+ int (*dev_close)(struct virtio_hw *hw);
+};
+
+/*
+ * This structure stores per-process data. Only virtio_ops for now.
+ */
+struct virtio_hw_internal {
+ const struct virtio_ops *virtio_ops;
+};
+
+#define VIRTIO_OPS(hw) (virtio_hw_internal[(hw)->port_id].virtio_ops)
+
+extern struct virtio_hw_internal virtio_hw_internal[RTE_MAX_ETHPORTS];
+
+
+#endif /* _VIRTIO_H_ */
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index c1d7b14dda..4b22622d2d 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -446,7 +446,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
* Read the virtqueue size from the Queue Size field
* Always power of 2 and if 0 virtqueue does not exist
*/
- vq_size = VTPCI_OPS(hw)->get_queue_num(hw, vtpci_queue_idx);
+ vq_size = VIRTIO_OPS(hw)->get_queue_num(hw, vtpci_queue_idx);
PMD_INIT_LOG(DEBUG, "vq_size: %u", vq_size);
if (vq_size == 0) {
PMD_INIT_LOG(ERR, "virtqueue does not exist");
@@ -608,7 +608,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
}
}
- if (VTPCI_OPS(hw)->setup_queue(hw, vq) < 0) {
+ if (VIRTIO_OPS(hw)->setup_queue(hw, vq) < 0) {
PMD_INIT_LOG(ERR, "setup_queue failed");
return -EINVAL;
}
@@ -703,7 +703,7 @@ virtio_dev_close(struct rte_eth_dev *dev)
/* reset the NIC */
if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
- VTPCI_OPS(hw)->set_config_irq(hw, VIRTIO_MSI_NO_VECTOR);
+ VIRTIO_OPS(hw)->set_config_irq(hw, VIRTIO_MSI_NO_VECTOR);
if (intr_conf->rxq)
virtio_queues_unbind_intr(dev);
@@ -718,7 +718,7 @@ virtio_dev_close(struct rte_eth_dev *dev)
virtio_dev_free_mb
[dpdk-dev] [PATCH v4 19/44] net/virtio: move features definition to generic header
This patch moves all the Virtio definition to the generic
header. It also renames some helpers to no more reference
PCI.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/meson.build| 3 +-
drivers/net/virtio/virtio.c | 22
drivers/net/virtio/virtio.h | 94 +++
drivers/net/virtio/virtio_ethdev.c| 114 +-
drivers/net/virtio/virtio_pci.c | 21 +---
drivers/net/virtio/virtio_pci.h | 90 --
drivers/net/virtio/virtio_ring.h | 2 +-
drivers/net/virtio/virtio_rxtx.c | 38 +++---
drivers/net/virtio/virtio_rxtx_packed.h | 6 +-
.../net/virtio/virtio_user/vhost_kernel_tap.c | 2 +-
drivers/net/virtio/virtio_user_ethdev.c | 6 +-
drivers/net/virtio/virtqueue.c| 4 +-
drivers/net/virtio/virtqueue.h| 8 +-
13 files changed, 211 insertions(+), 199 deletions(-)
create mode 100644 drivers/net/virtio/virtio.c
diff --git a/drivers/net/virtio/meson.build b/drivers/net/virtio/meson.build
index f2873d6180..d595cfdcab 100644
--- a/drivers/net/virtio/meson.build
+++ b/drivers/net/virtio/meson.build
@@ -7,7 +7,8 @@ if is_windows
subdir_done()
endif
-sources += files('virtio_ethdev.c',
+sources += files('virtio.c',
+ 'virtio_ethdev.c',
'virtio_pci_ethdev.c',
'virtio_pci.c',
'virtio_rxtx.c',
diff --git a/drivers/net/virtio/virtio.c b/drivers/net/virtio/virtio.c
new file mode 100644
index 00..d8d6bf7add
--- /dev/null
+++ b/drivers/net/virtio/virtio.c
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2014 Intel Corporation
+ * Copyright(c) 2020 Red Hat, Inc.
+ */
+
+#include "virtio.h"
+
+uint64_t
+virtio_negotiate_features(struct virtio_hw *hw, uint64_t host_features)
+{
+ uint64_t features;
+
+ /*
+* Limit negotiated features to what the driver, virtqueue, and
+* host all support.
+*/
+ features = host_features & hw->guest_features;
+ VIRTIO_OPS(hw)->set_features(hw, features);
+
+ return features;
+}
+
diff --git a/drivers/net/virtio/virtio.h b/drivers/net/virtio/virtio.h
index e55e3329f9..cdfee5d182 100644
--- a/drivers/net/virtio/virtio.h
+++ b/drivers/net/virtio/virtio.h
@@ -8,6 +8,86 @@
#include
+/* The feature bitmap for virtio net */
+#define VIRTIO_NET_F_CSUM 0 /* Host handles pkts w/ partial csum */
+#define VIRTIO_NET_F_GUEST_CSUM1 /* Guest handles pkts w/
partial csum */
+#define VIRTIO_NET_F_MTU 3 /* Initial MTU advice. */
+#define VIRTIO_NET_F_MAC 5 /* Host has given MAC address. */
+#define VIRTIO_NET_F_GUEST_TSO47 /* Guest can handle TSOv4 in. */
+#define VIRTIO_NET_F_GUEST_TSO68 /* Guest can handle TSOv6 in. */
+#define VIRTIO_NET_F_GUEST_ECN 9 /* Guest can handle TSO[6] w/ ECN in. */
+#define VIRTIO_NET_F_GUEST_UFO 10 /* Guest can handle UFO in. */
+#define VIRTIO_NET_F_HOST_TSO4 11 /* Host can handle TSOv4 in. */
+#define VIRTIO_NET_F_HOST_TSO6 12 /* Host can handle TSOv6 in. */
+#define VIRTIO_NET_F_HOST_ECN 13 /* Host can handle TSO[6] w/ ECN in. */
+#define VIRTIO_NET_F_HOST_UFO 14 /* Host can handle UFO in. */
+#define VIRTIO_NET_F_MRG_RXBUF 15 /* Host can merge receive buffers. */
+#define VIRTIO_NET_F_STATUS16 /* virtio_net_config.status available */
+#define VIRTIO_NET_F_CTRL_VQ 17 /* Control channel available */
+#define VIRTIO_NET_F_CTRL_RX 18 /* Control channel RX mode support */
+#define VIRTIO_NET_F_CTRL_VLAN 19 /* Control channel VLAN filtering */
+#define VIRTIO_NET_F_CTRL_RX_EXTRA 20 /* Extra RX mode control support */
+#define VIRTIO_NET_F_GUEST_ANNOUNCE 21 /* Guest can announce device on the
network */
+#define VIRTIO_NET_F_MQ22 /* Device supports Receive Flow
Steering */
+#define VIRTIO_NET_F_CTRL_MAC_ADDR 23 /* Set MAC address */
+
+/*
+ * Do we get callbacks when the ring is completely used,
+ * even if we've suppressed them?
+ */
+#define VIRTIO_F_NOTIFY_ON_EMPTY 24
+
+/* Can the device handle any descriptor layout? */
+#define VIRTIO_F_ANY_LAYOUT27
+
+/* We support indirect buffer descriptors */
+#define VIRTIO_RING_F_INDIRECT_DESC28
+
+#define VIRTIO_F_VERSION_1 32
+#define VIRTIO_F_IOMMU_PLATFORM33
+#define VIRTIO_F_RING_PACKED 34
+
+/*
+ * Some VirtIO feature bits (currently bits 28 through 31) are
+ * reserved for the transport being used (eg. virtio_ring), the
+ * rest are per-device feature bits.
+ */
+#define VIRTIO_TRANSPORT_F_START 28
+#define VIRTIO_TRANSPORT_F_END 34
+
+/*
+ * Inorder feature indicates that all buffers are used by the device
+ * in the same order in which they have been made available.
+ */
+#define VIRTIO_F_IN_ORDER 35
+
+/*
+ * This feature indicates t
[dpdk-dev] [PATCH v4 20/44] net/virtio: move virtqueue defines in generic header
This patch moves the virtqueues defines from PCI header
to the generic one.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio.h| 18 ++
drivers/net/virtio/virtio_ethdev.h | 4 +++-
drivers/net/virtio/virtio_pci.h| 17 -
.../net/virtio/virtio_user/virtio_user_dev.h | 3 ++-
4 files changed, 23 insertions(+), 19 deletions(-)
diff --git a/drivers/net/virtio/virtio.h b/drivers/net/virtio/virtio.h
index cdfee5d182..0c2b3525d5 100644
--- a/drivers/net/virtio/virtio.h
+++ b/drivers/net/virtio/virtio.h
@@ -88,6 +88,24 @@
#define VIRTIO_NET_S_LINK_UP 1 /* Link is up */
#define VIRTIO_NET_S_ANNOUNCE 2 /* Announcement is needed */
+/*
+ * Each virtqueue indirect descriptor list must be physically contiguous.
+ * To allow us to malloc(9) each list individually, limit the number
+ * supported to what will fit in one page. With 4KB pages, this is a limit
+ * of 256 descriptors. If there is ever a need for more, we can switch to
+ * contigmalloc(9) for the larger allocations, similar to what
+ * bus_dmamem_alloc(9) does.
+ *
+ * Note the sizeof(struct vring_desc) is 16 bytes.
+ */
+#define VIRTIO_MAX_INDIRECT ((int)(PAGE_SIZE / 16))
+
+/*
+ * Maximum number of virtqueues per device.
+ */
+#define VIRTIO_MAX_VIRTQUEUE_PAIRS 8
+#define VIRTIO_MAX_VIRTQUEUES (VIRTIO_MAX_VIRTQUEUE_PAIRS * 2 + 1)
+
struct virtio_hw {
struct virtqueue **vqs;
uint64_t guest_features;
diff --git a/drivers/net/virtio/virtio_ethdev.h
b/drivers/net/virtio/virtio_ethdev.h
index 13395937c8..d82856df9b 100644
--- a/drivers/net/virtio/virtio_ethdev.h
+++ b/drivers/net/virtio/virtio_ethdev.h
@@ -7,7 +7,9 @@
#include
-#include "virtio_pci.h"
+#include
+
+#include "virtio.h"
#ifndef PAGE_SIZE
#define PAGE_SIZE 4096
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 3d93e74f36..2c27c906fb 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -67,23 +67,6 @@ struct virtnet_ctl;
#define VIRTIO_CONFIG_STATUS_DEV_NEED_RESET0x40
#define VIRTIO_CONFIG_STATUS_FAILED0x80
-/*
- * Each virtqueue indirect descriptor list must be physically contiguous.
- * To allow us to malloc(9) each list individually, limit the number
- * supported to what will fit in one page. With 4KB pages, this is a limit
- * of 256 descriptors. If there is ever a need for more, we can switch to
- * contigmalloc(9) for the larger allocations, similar to what
- * bus_dmamem_alloc(9) does.
- *
- * Note the sizeof(struct vring_desc) is 16 bytes.
- */
-#define VIRTIO_MAX_INDIRECT ((int) (PAGE_SIZE / 16))
-
-/*
- * Maximum number of virtqueues per device.
- */
-#define VIRTIO_MAX_VIRTQUEUE_PAIRS 8
-#define VIRTIO_MAX_VIRTQUEUES (VIRTIO_MAX_VIRTQUEUE_PAIRS * 2 + 1)
/* Common configuration */
#define VIRTIO_PCI_CAP_COMMON_CFG 1
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.h
b/drivers/net/virtio/virtio_user/virtio_user_dev.h
index b3776cc7a0..ab62463a5b 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.h
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.h
@@ -7,7 +7,8 @@
#include
#include
-#include "../virtio_pci.h"
+
+#include "../virtio.h"
#include "../virtio_ring.h"
enum virtio_user_backend_type {
--
2.29.2
[dpdk-dev] [PATCH v4 21/44] net/virtio: move config definitions to generic header
This patch moves config and status definitions from the PCI
header to the generic one.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio.c | 43 +++
drivers/net/virtio/virtio.h | 50 ++
drivers/net/virtio/virtio_ethdev.c | 36
drivers/net/virtio/virtio_pci.c | 44
drivers/net/virtio/virtio_pci.h | 55 -
drivers/net/virtio/virtio_user_ethdev.c | 10 ++---
6 files changed, 116 insertions(+), 122 deletions(-)
diff --git a/drivers/net/virtio/virtio.c b/drivers/net/virtio/virtio.c
index d8d6bf7add..ba3203e68b 100644
--- a/drivers/net/virtio/virtio.c
+++ b/drivers/net/virtio/virtio.c
@@ -20,3 +20,46 @@ virtio_negotiate_features(struct virtio_hw *hw, uint64_t
host_features)
return features;
}
+
+void
+virtio_read_dev_config(struct virtio_hw *hw, size_t offset,
+ void *dst, int length)
+{
+ VIRTIO_OPS(hw)->read_dev_cfg(hw, offset, dst, length);
+}
+
+void
+virtio_write_dev_config(struct virtio_hw *hw, size_t offset,
+ const void *src, int length)
+{
+ VIRTIO_OPS(hw)->write_dev_cfg(hw, offset, src, length);
+}
+
+void
+virtio_reset(struct virtio_hw *hw)
+{
+ VIRTIO_OPS(hw)->set_status(hw, VIRTIO_CONFIG_STATUS_RESET);
+ /* flush status write */
+ VIRTIO_OPS(hw)->get_status(hw);
+}
+
+void
+virtio_reinit_complete(struct virtio_hw *hw)
+{
+ virtio_set_status(hw, VIRTIO_CONFIG_STATUS_DRIVER_OK);
+}
+
+void
+virtio_set_status(struct virtio_hw *hw, uint8_t status)
+{
+ if (status != VIRTIO_CONFIG_STATUS_RESET)
+ status |= VIRTIO_OPS(hw)->get_status(hw);
+
+ VIRTIO_OPS(hw)->set_status(hw, status);
+}
+
+uint8_t
+virtio_get_status(struct virtio_hw *hw)
+{
+ return VIRTIO_OPS(hw)->get_status(hw);
+}
diff --git a/drivers/net/virtio/virtio.h b/drivers/net/virtio/virtio.h
index 0c2b3525d5..0ac5328c57 100644
--- a/drivers/net/virtio/virtio.h
+++ b/drivers/net/virtio/virtio.h
@@ -106,6 +106,50 @@
#define VIRTIO_MAX_VIRTQUEUE_PAIRS 8
#define VIRTIO_MAX_VIRTQUEUES (VIRTIO_MAX_VIRTQUEUE_PAIRS * 2 + 1)
+/* VirtIO device IDs. */
+#define VIRTIO_ID_NETWORK 0x01
+#define VIRTIO_ID_BLOCK0x02
+#define VIRTIO_ID_CONSOLE 0x03
+#define VIRTIO_ID_ENTROPY 0x04
+#define VIRTIO_ID_BALLOON 0x05
+#define VIRTIO_ID_IOMEMORY 0x06
+#define VIRTIO_ID_9P 0x09
+
+/* Status byte for guest to report progress. */
+#define VIRTIO_CONFIG_STATUS_RESET 0x00
+#define VIRTIO_CONFIG_STATUS_ACK 0x01
+#define VIRTIO_CONFIG_STATUS_DRIVER0x02
+#define VIRTIO_CONFIG_STATUS_DRIVER_OK 0x04
+#define VIRTIO_CONFIG_STATUS_FEATURES_OK 0x08
+#define VIRTIO_CONFIG_STATUS_DEV_NEED_RESET0x40
+#define VIRTIO_CONFIG_STATUS_FAILED0x80
+
+/*
+ * This structure is just a reference to read net device specific
+ * config space; it is just a shadow structure.
+ *
+ */
+struct virtio_net_config {
+ /* The config defining mac address (if VIRTIO_NET_F_MAC) */
+ uint8_tmac[RTE_ETHER_ADDR_LEN];
+ /* See VIRTIO_NET_F_STATUS and VIRTIO_NET_S_* above */
+ uint16_t status;
+ uint16_t max_virtqueue_pairs;
+ uint16_t mtu;
+ /*
+* speed, in units of 1Mb. All values 0 to INT_MAX are legal.
+* Any other value stands for unknown.
+*/
+ uint32_t speed;
+ /*
+* 0x00 - half duplex
+* 0x01 - full duplex
+* Any other value stands for unknown.
+*/
+ uint8_t duplex;
+
+} __rte_packed;
+
struct virtio_hw {
struct virtqueue **vqs;
uint64_t guest_features;
@@ -182,5 +226,11 @@ virtio_with_packed_queue(struct virtio_hw *hw)
}
uint64_t virtio_negotiate_features(struct virtio_hw *hw, uint64_t
host_features);
+uint8_t virtio_get_status(struct virtio_hw *hw);
+void virtio_set_status(struct virtio_hw *hw, uint8_t status);
+void virtio_write_dev_config(struct virtio_hw *hw, size_t offset, const void
*src, int length);
+void virtio_read_dev_config(struct virtio_hw *hw, size_t offset, void *dst,
int length);
+void virtio_reset(struct virtio_hw *hw);
+void virtio_reinit_complete(struct virtio_hw *hw);
#endif /* _VIRTIO_H_ */
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 1771844fe7..dfd8d29f60 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -714,7 +714,7 @@ virtio_dev_close(struct rte_eth_dev *dev)
dev->intr_handle->intr_vec = NULL;
}
- vtpci_reset(hw);
+ virtio_reset(hw);
virtio_dev_free_mbufs(dev);
virtio_free_queues(hw);
@@ -1096,7 +1096,7 @@ virtio_dev_stats_reset(struct rte_eth_dev *dev)
static void
virtio_set_hwaddr(struct virtio_hw *hw)
{
- vtpci_write_dev_config(hw,
+ virtio_write_dev_config(hw,
of
[dpdk-dev] [PATCH v4 22/44] net/virtio: make interrupt handling more generic
This patch aims at isolating MSIX notion into PCI
layer.
Signed-off-by: Maxime Coquelin
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio.c | 6
drivers/net/virtio/virtio.h | 11 +--
drivers/net/virtio/virtio_ethdev.c | 7 ++---
drivers/net/virtio/virtio_pci.c | 38 -
drivers/net/virtio/virtio_pci.h | 24 +---
drivers/net/virtio/virtio_user_ethdev.c | 9 ++
6 files changed, 46 insertions(+), 49 deletions(-)
diff --git a/drivers/net/virtio/virtio.c b/drivers/net/virtio/virtio.c
index ba3203e68b..7e1e77797f 100644
--- a/drivers/net/virtio/virtio.c
+++ b/drivers/net/virtio/virtio.c
@@ -63,3 +63,9 @@ virtio_get_status(struct virtio_hw *hw)
{
return VIRTIO_OPS(hw)->get_status(hw);
}
+
+uint8_t
+virtio_get_isr(struct virtio_hw *hw)
+{
+ return VIRTIO_OPS(hw)->get_isr(hw);
+}
diff --git a/drivers/net/virtio/virtio.h b/drivers/net/virtio/virtio.h
index 0ac5328c57..a7629ad16b 100644
--- a/drivers/net/virtio/virtio.h
+++ b/drivers/net/virtio/virtio.h
@@ -124,6 +124,13 @@
#define VIRTIO_CONFIG_STATUS_DEV_NEED_RESET0x40
#define VIRTIO_CONFIG_STATUS_FAILED0x80
+/* The bit of the ISR which indicates a device has an interrupt. */
+#define VIRTIO_ISR_INTR 0x1
+/* The bit of the ISR which indicates a device configuration change. */
+#define VIRTIO_ISR_CONFIG 0x2
+/* Vector value used to disable MSI for queue. */
+#define VIRTIO_MSI_NO_VECTOR 0x
+
/*
* This structure is just a reference to read net device specific
* config space; it is just a shadow structure.
@@ -168,7 +175,7 @@ struct virtio_hw {
uint8_t mac_addr[RTE_ETHER_ADDR_LEN];
uint32_t speed; /* link speed in MB */
uint8_t duplex;
- uint8_t use_msix;
+ uint8_t intr_lsc;
uint16_t max_mtu;
/*
* App management thread and virtio interrupt handler thread
@@ -232,5 +239,5 @@ void virtio_write_dev_config(struct virtio_hw *hw, size_t
offset, const void *sr
void virtio_read_dev_config(struct virtio_hw *hw, size_t offset, void *dst,
int length);
void virtio_reset(struct virtio_hw *hw);
void virtio_reinit_complete(struct virtio_hw *hw);
-
+uint8_t virtio_get_isr(struct virtio_hw *hw);
#endif /* _VIRTIO_H_ */
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index dfd8d29f60..51eb5aebf3 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1455,13 +1455,13 @@ virtio_interrupt_handler(void *param)
uint16_t status;
/* Read interrupt status which clears interrupt */
- isr = vtpci_isr(hw);
+ isr = virtio_get_isr(hw);
PMD_DRV_LOG(INFO, "interrupt status = %#x", isr);
if (virtio_intr_unmask(dev) < 0)
PMD_DRV_LOG(ERR, "interrupt enable failed");
- if (isr & VIRTIO_PCI_ISR_CONFIG) {
+ if (isr & VIRTIO_ISR_CONFIG) {
if (virtio_dev_link_update(dev, 0) == 0)
rte_eth_dev_callback_process(dev,
RTE_ETH_EVENT_INTR_LSC,
@@ -1668,8 +1668,7 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t
req_features)
hw->weak_barriers = !virtio_with_feature(hw, VIRTIO_F_ORDER_PLATFORM);
/* If host does not support both status and MSI-X then disable LSC */
- if (virtio_with_feature(hw, VIRTIO_NET_F_STATUS) &&
- hw->use_msix != VIRTIO_MSIX_NONE)
+ if (virtio_with_feature(hw, VIRTIO_NET_F_STATUS) && hw->intr_lsc)
eth_dev->data->dev_flags |= RTE_ETH_DEV_INTR_LSC;
else
eth_dev->data->dev_flags &= ~RTE_ETH_DEV_INTR_LSC;
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index baadc79c43..905534ac72 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -28,8 +28,8 @@
* The remaining space is defined by each driver as the per-driver
* configuration space.
*/
-#define VIRTIO_PCI_CONFIG(hw) \
- (((hw)->use_msix == VIRTIO_MSIX_ENABLED) ? 24 : 20)
+#define VIRTIO_PCI_CONFIG(dev) \
+ (((dev)->msix_status == VIRTIO_MSIX_ENABLED) ? 24 : 20)
struct virtio_pci_internal {
@@ -121,6 +121,7 @@ static void
legacy_read_dev_config(struct virtio_hw *hw, size_t offset,
void *dst, int length)
{
+ struct virtio_pci_dev *dev = virtio_pci_get_dev(hw);
#ifdef RTE_ARCH_PPC_64
int size;
@@ -128,17 +129,17 @@ legacy_read_dev_config(struct virtio_hw *hw, size_t
offset,
if (length >= 4) {
size = 4;
rte_pci_ioport_read(VTPCI_IO(hw), dst, size,
- VIRTIO_PCI_CONFIG(hw) + offset);
+ VIRTIO_PCI_CONFIG(dev) + offset);
*(uint32_t *)dst = rte_be_to_cpu_32(*(uint32_t *)dst);
} else if (length >= 2) {
[dpdk-dev] [PATCH v4 23/44] net/virtio: move vring alignment to generic header
This patch moves vring alignment define to the generic
Virtio header.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio.h | 3 +++
drivers/net/virtio/virtio_ethdev.c | 10 +-
drivers/net/virtio/virtio_pci.c | 2 +-
drivers/net/virtio/virtio_pci.h | 3 ---
drivers/net/virtio/virtio_user_ethdev.c | 4 ++--
5 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/drivers/net/virtio/virtio.h b/drivers/net/virtio/virtio.h
index a7629ad16b..21d54904e7 100644
--- a/drivers/net/virtio/virtio.h
+++ b/drivers/net/virtio/virtio.h
@@ -131,6 +131,9 @@
/* Vector value used to disable MSI for queue. */
#define VIRTIO_MSI_NO_VECTOR 0x
+/* The alignment to use between consumer and producer parts of vring. */
+#define VIRTIO_VRING_ALIGN 4096
+
/*
* This structure is just a reference to read net device specific
* config space; it is just a shadow structure.
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 51eb5aebf3..a2195bdda8 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -407,12 +407,12 @@ virtio_init_vring(struct virtqueue *vq)
memset(vq->vq_descx, 0, sizeof(struct vq_desc_extra) * vq->vq_nentries);
if (virtio_with_packed_queue(vq->hw)) {
vring_init_packed(&vq->vq_packed.ring, ring_mem,
- VIRTIO_PCI_VRING_ALIGN, size);
+ VIRTIO_VRING_ALIGN, size);
vring_desc_init_packed(vq, size);
} else {
struct vring *vr = &vq->vq_split.ring;
- vring_init_split(vr, ring_mem, VIRTIO_PCI_VRING_ALIGN, size);
+ vring_init_split(vr, ring_mem, VIRTIO_VRING_ALIGN, size);
vring_desc_init_split(vr->desc, size);
}
/*
@@ -497,14 +497,14 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
/*
* Reserve a memzone for vring elements
*/
- size = vring_size(hw, vq_size, VIRTIO_PCI_VRING_ALIGN);
- vq->vq_ring_size = RTE_ALIGN_CEIL(size, VIRTIO_PCI_VRING_ALIGN);
+ size = vring_size(hw, vq_size, VIRTIO_VRING_ALIGN);
+ vq->vq_ring_size = RTE_ALIGN_CEIL(size, VIRTIO_VRING_ALIGN);
PMD_INIT_LOG(DEBUG, "vring_size: %d, rounded_vring_size: %d",
size, vq->vq_ring_size);
mz = rte_memzone_reserve_aligned(vq_name, vq->vq_ring_size,
numa_node, RTE_MEMZONE_IOVA_CONTIG,
- VIRTIO_PCI_VRING_ALIGN);
+ VIRTIO_VRING_ALIGN);
if (mz == NULL) {
if (rte_errno == EEXIST)
mz = rte_memzone_lookup(vq_name);
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 905534ac72..d27e9eb515 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -492,7 +492,7 @@ modern_setup_queue(struct virtio_hw *hw, struct virtqueue
*vq)
avail_addr = desc_addr + vq->vq_nentries * sizeof(struct vring_desc);
used_addr = RTE_ALIGN_CEIL(avail_addr + offsetof(struct vring_avail,
ring[vq->vq_nentries]),
- VIRTIO_PCI_VRING_ALIGN);
+ VIRTIO_VRING_ALIGN);
rte_write16(vq->vq_queue_index, &dev->common_cfg->queue_select);
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 68f36dc0dd..6cdf2c55dd 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -122,9 +122,6 @@ struct virtio_pci_dev {
*/
#define VIRTIO_PCI_QUEUE_ADDR_SHIFT 12
-/* The alignment to use between consumer and producer parts of vring. */
-#define VIRTIO_PCI_VRING_ALIGN 4096
-
/*
* Function declaration from virtio_pci.c
*/
diff --git a/drivers/net/virtio/virtio_user_ethdev.c
b/drivers/net/virtio/virtio_user_ethdev.c
index c3dc3908d6..91794b8645 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -388,7 +388,7 @@ virtio_user_setup_queue_packed(struct virtqueue *vq,
sizeof(struct vring_packed_desc);
used_addr = RTE_ALIGN_CEIL(avail_addr +
sizeof(struct vring_packed_desc_event),
- VIRTIO_PCI_VRING_ALIGN);
+ VIRTIO_VRING_ALIGN);
vring->num = vq->vq_nentries;
vring->desc = (void *)(uintptr_t)desc_addr;
vring->driver = (void *)(uintptr_t)avail_addr;
@@ -410,7 +410,7 @@ virtio_user_setup_queue_split(struct virtqueue *vq, struct
virtio_user_dev *dev)
avail_addr = desc_addr + vq->vq_nentries * sizeof(struct vring_desc);
used_addr = RTE_ALIGN_CEIL(avail_addr + offsetof(struct vring_avail,
ring[vq->v
[dpdk-dev] [PATCH v4 24/44] net/virtio: remove last PCI refs in non-PCI code
This patch finalizes the bus isolation part of this
refactoring.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
Reviewed-by: David Marchand
---
drivers/net/virtio/virtio_ethdev.c | 21 +---
drivers/net/virtio/virtio_rxtx.c | 18 -
drivers/net/virtio/virtio_rxtx_packed.h | 2 +-
drivers/net/virtio/virtio_rxtx_packed_avx.h | 2 +-
drivers/net/virtio/virtio_rxtx_packed_neon.h | 2 +-
drivers/net/virtio/virtio_user/vhost.h | 4 +++-
drivers/net/virtio/virtqueue.c | 2 +-
drivers/net/virtio/virtqueue.h | 8
8 files changed, 29 insertions(+), 30 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index a2195bdda8..5f7ea390c8 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -9,14 +9,11 @@
#include
#include
-#include
#include
#include
#include
#include
#include
-#include
-#include
#include
#include
#include
@@ -32,7 +29,7 @@
#include
#include "virtio_ethdev.h"
-#include "virtio_pci.h"
+#include "virtio.h"
#include "virtio_logs.h"
#include "virtqueue.h"
#include "virtio_rxtx.h"
@@ -422,7 +419,7 @@ virtio_init_vring(struct virtqueue *vq)
}
static int
-virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
+virtio_init_queue(struct rte_eth_dev *dev, uint16_t queue_idx)
{
char vq_name[VIRTQUEUE_MAX_NAME_SZ];
char vq_hdr_name[VIRTQUEUE_MAX_NAME_SZ];
@@ -435,18 +432,18 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
struct virtqueue *vq;
size_t sz_hdr_mz = 0;
void *sw_ring = NULL;
- int queue_type = virtio_get_queue_type(hw, vtpci_queue_idx);
+ int queue_type = virtio_get_queue_type(hw, queue_idx);
int ret;
int numa_node = dev->device->numa_node;
PMD_INIT_LOG(INFO, "setting up queue: %u on NUMA node %d",
- vtpci_queue_idx, numa_node);
+ queue_idx, numa_node);
/*
* Read the virtqueue size from the Queue Size field
* Always power of 2 and if 0 virtqueue does not exist
*/
- vq_size = VIRTIO_OPS(hw)->get_queue_num(hw, vtpci_queue_idx);
+ vq_size = VIRTIO_OPS(hw)->get_queue_num(hw, queue_idx);
PMD_INIT_LOG(DEBUG, "vq_size: %u", vq_size);
if (vq_size == 0) {
PMD_INIT_LOG(ERR, "virtqueue does not exist");
@@ -459,7 +456,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
}
snprintf(vq_name, sizeof(vq_name), "port%d_vq%d",
-dev->data->port_id, vtpci_queue_idx);
+dev->data->port_id, queue_idx);
size = RTE_ALIGN_CEIL(sizeof(*vq) +
vq_size * sizeof(struct vq_desc_extra),
@@ -481,10 +478,10 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
PMD_INIT_LOG(ERR, "can not allocate vq");
return -ENOMEM;
}
- hw->vqs[vtpci_queue_idx] = vq;
+ hw->vqs[queue_idx] = vq;
vq->hw = hw;
- vq->vq_queue_index = vtpci_queue_idx;
+ vq->vq_queue_index = queue_idx;
vq->vq_nentries = vq_size;
if (virtio_with_packed_queue(hw)) {
vq->vq_packed.used_wrap_counter = 1;
@@ -527,7 +524,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t
vtpci_queue_idx)
if (sz_hdr_mz) {
snprintf(vq_hdr_name, sizeof(vq_hdr_name), "port%d_vq%d_hdr",
-dev->data->port_id, vtpci_queue_idx);
+dev->data->port_id, queue_idx);
hdr_mz = rte_memzone_reserve_aligned(vq_hdr_name, sz_hdr_mz,
numa_node, RTE_MEMZONE_IOVA_CONTIG,
RTE_CACHE_LINE_SIZE);
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index 728085e99b..b5657d5c83 100644
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -27,7 +27,7 @@
#include "virtio_logs.h"
#include "virtio_ethdev.h"
-#include "virtio_pci.h"
+#include "virtio.h"
#include "virtqueue.h"
#include "virtio_rxtx.h"
#include "virtio_rxtx_simple.h"
@@ -628,9 +628,9 @@ virtio_dev_rx_queue_setup(struct rte_eth_dev *dev,
const struct rte_eth_rxconf *rx_conf,
struct rte_mempool *mp)
{
- uint16_t vtpci_queue_idx = 2 * queue_idx + VTNET_SQ_RQ_QUEUE_IDX;
+ uint16_t vq_idx = 2 * queue_idx + VTNET_SQ_RQ_QUEUE_IDX;
struct virtio_hw *hw = dev->data->dev_private;
- struct virtqueue *vq = hw->vqs[vtpci_queue_idx];
+ struct virtqueue *vq = hw->vqs[vq_idx];
struct virtnet_rx *rxvq;
uint16_t rx_free_thresh;
@@ -678,9 +678,9 @@ virtio_dev_rx_queue_setup(struct rte_eth_dev *dev,
int
virtio_dev_rx_queue_setup_finish(struct rte_eth_dev *dev, u
[dpdk-dev] [PATCH v4 26/44] net/virtio: add Virtio-user ops to set owner
This patch implements a dedicated callback for
sending owner request. All the requests will be
converted that way so that backends other than
Vhost-user don't have to work around being it.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 1 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 28 +++--
drivers/net/virtio/virtio_user/vhost_user.c | 21 +++--
drivers/net/virtio/virtio_user/vhost_vdpa.c | 30 ---
.../net/virtio/virtio_user/virtio_user_dev.c | 3 +-
5 files changed, 72 insertions(+), 11 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 8ec3a6a62c..5413ec6778 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -109,6 +109,7 @@ struct virtio_user_dev;
struct virtio_user_backend_ops {
int (*setup)(struct virtio_user_dev *dev);
+ int (*set_owner)(struct virtio_user_dev *dev);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 2c805077af..b79dcad179 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -6,6 +6,7 @@
#include
#include
#include
+#include
#include
@@ -55,8 +56,28 @@ get_vhost_kernel_max_regions(void)
close(fd);
}
+static int
+vhost_kernel_ioctl(int fd, uint64_t request, void *arg)
+{
+ int ret;
+
+ ret = ioctl(fd, request, arg);
+ if (ret) {
+ PMD_DRV_LOG(ERR, "Vhost-kernel ioctl %"PRIu64" failed (%s)",
+ request, strerror(errno));
+ return -1;
+ }
+
+ return 0;
+}
+
+static int
+vhost_kernel_set_owner(struct virtio_user_dev *dev)
+{
+ return vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_OWNER, NULL);
+}
+
static uint64_t vhost_req_user_to_kernel[] = {
- [VHOST_USER_SET_OWNER] = VHOST_SET_OWNER,
[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
[VHOST_USER_SET_FEATURES] = VHOST_SET_FEATURES,
[VHOST_USER_GET_FEATURES] = VHOST_GET_FEATURES,
@@ -175,7 +196,7 @@ tap_support_features(void)
}
static int
-vhost_kernel_ioctl(struct virtio_user_dev *dev,
+vhost_kernel_send_request(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg)
{
@@ -385,6 +406,7 @@ vhost_kernel_enable_queue_pair(struct virtio_user_dev *dev,
struct virtio_user_backend_ops virtio_ops_kernel = {
.setup = vhost_kernel_setup,
- .send_request = vhost_kernel_ioctl,
+ .set_owner = vhost_kernel_set_owner,
+ .send_request = vhost_kernel_send_request,
.enable_qp = vhost_kernel_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 55c81a..ea3bd4ca10 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -125,6 +125,24 @@ vhost_user_read(int fd, struct vhost_user_msg *msg)
return -1;
}
+static int
+vhost_user_set_owner(struct virtio_user_dev *dev)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = VHOST_USER_SET_OWNER,
+ .flags = VHOST_USER_VERSION,
+ };
+
+ ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to set owner");
+ return -1;
+ }
+
+ return 0;
+}
+
struct walk_arg {
struct vhost_memory *vm;
int *fds;
@@ -230,7 +248,6 @@ prepare_vhost_memory_user(struct vhost_user_msg *msg, int
fds[])
static struct vhost_user_msg m;
const char * const vhost_msg_strings[] = {
- [VHOST_USER_SET_OWNER] = "VHOST_SET_OWNER",
[VHOST_USER_RESET_OWNER] = "VHOST_RESET_OWNER",
[VHOST_USER_SET_FEATURES] = "VHOST_SET_FEATURES",
[VHOST_USER_GET_FEATURES] = "VHOST_GET_FEATURES",
@@ -308,7 +325,6 @@ vhost_user_sock(struct virtio_user_dev *dev,
msg.size = sizeof(m.payload.u64);
break;
- case VHOST_USER_SET_OWNER:
case VHOST_USER_RESET_OWNER:
break;
@@ -519,6 +535,7 @@ vhost_user_enable_queue_pair(struct virtio_user_dev *dev,
struct virtio_user_backend_ops virtio_ops_user = {
.setup = vhost_user_setup,
+ .set_owner = vhost_user_set_owner,
.send_request = vhost_user_sock,
.enable_qp = vhost_user_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_vdpa.c
b/drivers/net/virtio/virtio_user/vhost_vdpa.c
index 004802b9eb..d9bc213e0d 100644
--- a/drivers/net/virtio/virtio_user/vhost_vdpa.c
+++ b/drivers/net/virtio/virtio_user/vhost_vdpa.c
@@ -39,7 +39,6 @@
#define VH
[dpdk-dev] [PATCH v4 25/44] net/virtio: make Vhost-user request sender consistent
This patch makes vhost_user_write() consistent with
vhost_user_read(), by passing a Vhost-user message
pointer instead of a buffer pointer and its length, which
is now calculated in the function.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost_user.c | 10 --
1 file changed, 4 insertions(+), 6 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 350eed4182..55c81a 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -51,7 +51,7 @@ struct vhost_user_msg {
(sizeof(struct vhost_user_msg) - VHOST_USER_HDR_SIZE)
static int
-vhost_user_write(int fd, void *buf, int len, int *fds, int fd_num)
+vhost_user_write(int fd, struct vhost_user_msg *msg, int *fds, int fd_num)
{
int r;
struct msghdr msgh;
@@ -63,8 +63,8 @@ vhost_user_write(int fd, void *buf, int len, int *fds, int
fd_num)
memset(&msgh, 0, sizeof(msgh));
memset(control, 0, sizeof(control));
- iov.iov_base = (uint8_t *)buf;
- iov.iov_len = len;
+ iov.iov_base = (uint8_t *)msg;
+ iov.iov_len = VHOST_USER_HDR_SIZE + msg->size;
msgh.msg_iov = &iov;
msgh.msg_iovlen = 1;
@@ -259,7 +259,6 @@ vhost_user_sock(struct virtio_user_dev *dev,
int has_reply_ack = 0;
int fds[VHOST_MEMORY_MAX_NREGIONS];
int fd_num = 0;
- int len;
int vhostfd = dev->vhostfd;
RTE_SET_USED(m);
@@ -364,8 +363,7 @@ vhost_user_sock(struct virtio_user_dev *dev,
return -1;
}
- len = VHOST_USER_HDR_SIZE + msg.size;
- if (vhost_user_write(vhostfd, &msg, len, fds, fd_num) < 0) {
+ if (vhost_user_write(vhostfd, &msg, fds, fd_num) < 0) {
PMD_DRV_LOG(ERR, "%s failed: %s",
vhost_msg_strings[req], strerror(errno));
return -1;
--
2.29.2
[dpdk-dev] [PATCH v4 27/44] net/virtio: add Virtio-user features ops
This patch introduces new callbacks for getting
and setting Virtio features, and implements them
for the different backend types.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 2 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 150 +-
.../net/virtio/virtio_user/vhost_kernel_tap.c | 23 +++
.../net/virtio/virtio_user/vhost_kernel_tap.h | 1 +
drivers/net/virtio/virtio_user/vhost_user.c | 64 +++-
drivers/net/virtio/virtio_user/vhost_vdpa.c | 38 +++--
.../net/virtio/virtio_user/virtio_user_dev.c | 5 +-
drivers/net/virtio/virtio_user_ethdev.c | 3 +-
8 files changed, 190 insertions(+), 96 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 5413ec6778..13a88c7671 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -110,6 +110,8 @@ struct virtio_user_dev;
struct virtio_user_backend_ops {
int (*setup)(struct virtio_user_dev *dev);
int (*set_owner)(struct virtio_user_dev *dev);
+ int (*get_features)(struct virtio_user_dev *dev, uint64_t *features);
+ int (*set_features)(struct virtio_user_dev *dev, uint64_t features);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index b79dcad179..e46039e649 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -38,6 +38,28 @@ struct vhost_memory_kernel {
#define VHOST_SET_VRING_ERR _IOW(VHOST_VIRTIO, 0x22, struct vhost_vring_file)
#define VHOST_NET_SET_BACKEND _IOW(VHOST_VIRTIO, 0x30, struct vhost_vring_file)
+/* with below features, vhost kernel does not need to do the checksum and TSO,
+ * these info will be passed to virtio_user through virtio net header.
+ */
+#define VHOST_KERNEL_GUEST_OFFLOADS_MASK \
+ ((1ULL << VIRTIO_NET_F_GUEST_CSUM) |\
+(1ULL << VIRTIO_NET_F_GUEST_TSO4) |\
+(1ULL << VIRTIO_NET_F_GUEST_TSO6) |\
+(1ULL << VIRTIO_NET_F_GUEST_ECN) |\
+(1ULL << VIRTIO_NET_F_GUEST_UFO))
+
+/* with below features, when flows from virtio_user to vhost kernel
+ * (1) if flows goes up through the kernel networking stack, it does not need
+ * to verify checksum, which can save CPU cycles;
+ * (2) if flows goes through a Linux bridge and outside from an interface
+ * (kernel driver), checksum and TSO will be done by GSO in kernel or even
+ * offloaded into real physical device.
+ */
+#define VHOST_KERNEL_HOST_OFFLOADS_MASK\
+ ((1ULL << VIRTIO_NET_F_HOST_TSO4) | \
+(1ULL << VIRTIO_NET_F_HOST_TSO6) | \
+(1ULL << VIRTIO_NET_F_CSUM))
+
static uint64_t max_regions = 64;
static void
@@ -77,10 +99,57 @@ vhost_kernel_set_owner(struct virtio_user_dev *dev)
return vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_OWNER, NULL);
}
+static int
+vhost_kernel_get_features(struct virtio_user_dev *dev, uint64_t *features)
+{
+ int ret;
+ unsigned int tap_features;
+
+ ret = vhost_kernel_ioctl(dev->vhostfds[0], VHOST_GET_FEATURES,
features);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to get features");
+ return -1;
+ }
+
+ ret = tap_support_features(&tap_features);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to get TAP features");
+ return -1;
+ }
+
+ /* with tap as the backend, all these features are supported
+* but not claimed by vhost-net, so we add them back when
+* reporting to upper layer.
+*/
+ if (tap_features & IFF_VNET_HDR) {
+ *features |= VHOST_KERNEL_GUEST_OFFLOADS_MASK;
+ *features |= VHOST_KERNEL_HOST_OFFLOADS_MASK;
+ }
+
+ /* vhost_kernel will not declare this feature, but it does
+* support multi-queue.
+*/
+ if (tap_features & IFF_MULTI_QUEUE)
+ *features |= (1ull << VIRTIO_NET_F_MQ);
+
+ return 0;
+}
+
+static int
+vhost_kernel_set_features(struct virtio_user_dev *dev, uint64_t features)
+{
+ /* We don't need memory protection here */
+ features &= ~(1ULL << VIRTIO_F_IOMMU_PLATFORM);
+ /* VHOST kernel does not know about below flags */
+ features &= ~VHOST_KERNEL_GUEST_OFFLOADS_MASK;
+ features &= ~VHOST_KERNEL_HOST_OFFLOADS_MASK;
+ features &= ~(1ULL << VIRTIO_NET_F_MQ);
+
+ return vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_FEATURES,
&features);
+}
+
static uint64_t vhost_req_user_to_kernel[] = {
[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
- [VHOST_USER_SET_FEATURES] = VHOST_SET_FEATURES,
- [VHOST_USER_GET_FEATURES] = VHOST_GET_FEATURES,
[VHOST_
[dpdk-dev] [PATCH v4 29/44] net/virtio: add Virtio-user memory tables ops
This patch implements a dedicated callback for
preparing and sending memory table to the backends.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 1 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 60 ++---
drivers/net/virtio/virtio_user/vhost_user.c | 86 ++-
drivers/net/virtio/virtio_user/vhost_vdpa.c | 8 +-
.../net/virtio/virtio_user/virtio_user_dev.c | 4 +-
5 files changed, 101 insertions(+), 58 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index d805526130..463801f563 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -114,6 +114,7 @@ struct virtio_user_backend_ops {
int (*set_features)(struct virtio_user_dev *dev, uint64_t features);
int (*get_protocol_features)(struct virtio_user_dev *dev, uint64_t
*features);
int (*set_protocol_features)(struct virtio_user_dev *dev, uint64_t
features);
+ int (*set_memory_table)(struct virtio_user_dev *dev);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index e46039e649..69f932bede 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -148,17 +148,6 @@ vhost_kernel_set_features(struct virtio_user_dev *dev,
uint64_t features)
return vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_FEATURES,
&features);
}
-static uint64_t vhost_req_user_to_kernel[] = {
- [VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
- [VHOST_USER_SET_VRING_CALL] = VHOST_SET_VRING_CALL,
- [VHOST_USER_SET_VRING_NUM] = VHOST_SET_VRING_NUM,
- [VHOST_USER_SET_VRING_BASE] = VHOST_SET_VRING_BASE,
- [VHOST_USER_GET_VRING_BASE] = VHOST_GET_VRING_BASE,
- [VHOST_USER_SET_VRING_ADDR] = VHOST_SET_VRING_ADDR,
- [VHOST_USER_SET_VRING_KICK] = VHOST_SET_VRING_KICK,
- [VHOST_USER_SET_MEM_TABLE] = VHOST_SET_MEM_TABLE,
-};
-
static int
add_memseg_list(const struct rte_memseg_list *msl, void *arg)
{
@@ -193,16 +182,17 @@ add_memseg_list(const struct rte_memseg_list *msl, void
*arg)
* have much more memory regions. Below function will treat each
* contiguous memory space reserved by DPDK as one region.
*/
-static struct vhost_memory_kernel *
-prepare_vhost_memory_kernel(void)
+static int
+vhost_kernel_set_memory_table(struct virtio_user_dev *dev)
{
struct vhost_memory_kernel *vm;
+ int ret;
vm = malloc(sizeof(struct vhost_memory_kernel) +
max_regions *
sizeof(struct vhost_memory_region));
if (!vm)
- return NULL;
+ goto err;
vm->nregions = 0;
vm->padding = 0;
@@ -211,14 +201,34 @@ prepare_vhost_memory_kernel(void)
* The memory lock has already been taken by memory subsystem
* or virtio_user_start_device().
*/
- if (rte_memseg_list_walk_thread_unsafe(add_memseg_list, vm) < 0) {
- free(vm);
- return NULL;
- }
+ ret = rte_memseg_list_walk_thread_unsafe(add_memseg_list, vm);
+ if (ret < 0)
+ goto err_free;
+
+ ret = vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_MEM_TABLE, vm);
+ if (ret < 0)
+ goto err_free;
- return vm;
+ free(vm);
+
+ return 0;
+err_free:
+ free(vm);
+err:
+ PMD_DRV_LOG(ERR, "Failed to set memory table");
+ return -1;
}
+static uint64_t vhost_req_user_to_kernel[] = {
+ [VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
+ [VHOST_USER_SET_VRING_CALL] = VHOST_SET_VRING_CALL,
+ [VHOST_USER_SET_VRING_NUM] = VHOST_SET_VRING_NUM,
+ [VHOST_USER_SET_VRING_BASE] = VHOST_SET_VRING_BASE,
+ [VHOST_USER_GET_VRING_BASE] = VHOST_GET_VRING_BASE,
+ [VHOST_USER_SET_VRING_ADDR] = VHOST_SET_VRING_ADDR,
+ [VHOST_USER_SET_VRING_KICK] = VHOST_SET_VRING_KICK,
+};
+
static int
vhost_kernel_send_request(struct virtio_user_dev *dev,
enum vhost_user_request req,
@@ -227,7 +237,6 @@ vhost_kernel_send_request(struct virtio_user_dev *dev,
int ret = -1;
unsigned int i;
uint64_t req_kernel;
- struct vhost_memory_kernel *vm = NULL;
int vhostfd;
unsigned int queue_sel;
@@ -235,13 +244,6 @@ vhost_kernel_send_request(struct virtio_user_dev *dev,
req_kernel = vhost_req_user_to_kernel[req];
- if (req_kernel == VHOST_SET_MEM_TABLE) {
- vm = prepare_vhost_memory_kernel();
- if (!vm)
- return -1;
- arg = (void *)vm;
- }
-
switch (req_kernel) {
case VHOST_SET_VRING_NUM:
case VHOS
[dpdk-dev] [PATCH v4 28/44] net/virtio: add Virtio-user protocol features ops
This patch introduces new callbacks for getting
and setting Vhost-user protocol features.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 2 +
drivers/net/virtio/virtio_user/vhost_user.c | 64 +--
drivers/net/virtio/virtio_user/vhost_vdpa.c | 14
.../net/virtio/virtio_user/virtio_user_dev.c | 14 ++--
drivers/net/virtio/virtio_user_ethdev.c | 14 ++--
5 files changed, 82 insertions(+), 26 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 13a88c7671..d805526130 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -112,6 +112,8 @@ struct virtio_user_backend_ops {
int (*set_owner)(struct virtio_user_dev *dev);
int (*get_features)(struct virtio_user_dev *dev, uint64_t *features);
int (*set_features)(struct virtio_user_dev *dev, uint64_t features);
+ int (*get_protocol_features)(struct virtio_user_dev *dev, uint64_t
*features);
+ int (*set_protocol_features)(struct virtio_user_dev *dev, uint64_t
features);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 8e5bf332d6..4877574f72 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -201,6 +201,62 @@ vhost_user_set_features(struct virtio_user_dev *dev,
uint64_t features)
return 0;
}
+static int
+vhost_user_get_protocol_features(struct virtio_user_dev *dev, uint64_t
*features)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = VHOST_USER_GET_PROTOCOL_FEATURES,
+ .flags = VHOST_USER_VERSION,
+ };
+
+ ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
+ if (ret < 0)
+ goto err;
+
+ ret = vhost_user_read(dev->vhostfd, &msg);
+ if (ret < 0)
+ goto err;
+
+ if (msg.request != VHOST_USER_GET_PROTOCOL_FEATURES) {
+ PMD_DRV_LOG(ERR, "Unexpected request type (%d)", msg.request);
+ goto err;
+ }
+
+ if (msg.size != sizeof(*features)) {
+ PMD_DRV_LOG(ERR, "Unexpected payload size (%u)", msg.size);
+ goto err;
+ }
+
+ *features = msg.payload.u64;
+
+ return 0;
+err:
+ PMD_DRV_LOG(ERR, "Failed to get backend protocol features");
+
+ return -1;
+}
+
+static int
+vhost_user_set_protocol_features(struct virtio_user_dev *dev, uint64_t
features)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = VHOST_USER_SET_PROTOCOL_FEATURES,
+ .flags = VHOST_USER_VERSION,
+ .size = sizeof(features),
+ .payload.u64 = features,
+ };
+
+ ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to set protocol features");
+ return -1;
+ }
+
+ return 0;
+}
+
struct walk_arg {
struct vhost_memory *vm;
int *fds;
@@ -315,8 +371,6 @@ const char * const vhost_msg_strings[] = {
[VHOST_USER_SET_VRING_KICK] = "VHOST_SET_VRING_KICK",
[VHOST_USER_SET_MEM_TABLE] = "VHOST_SET_MEM_TABLE",
[VHOST_USER_SET_VRING_ENABLE] = "VHOST_SET_VRING_ENABLE",
- [VHOST_USER_GET_PROTOCOL_FEATURES] = "VHOST_USER_GET_PROTOCOL_FEATURES",
- [VHOST_USER_SET_PROTOCOL_FEATURES] = "VHOST_USER_SET_PROTOCOL_FEATURES",
[VHOST_USER_SET_STATUS] = "VHOST_SET_STATUS",
[VHOST_USER_GET_STATUS] = "VHOST_GET_STATUS",
};
@@ -354,8 +408,6 @@ vhost_user_sock(struct virtio_user_dev *dev,
(!(dev->protocol_features &
(1ULL << VHOST_USER_PROTOCOL_F_STATUS
return -ENOTSUP;
- /* Fallthrough */
- case VHOST_USER_GET_PROTOCOL_FEATURES:
need_reply = 1;
break;
@@ -368,7 +420,6 @@ vhost_user_sock(struct virtio_user_dev *dev,
if (has_reply_ack)
msg.flags |= VHOST_USER_NEED_REPLY_MASK;
/* Fallthrough */
- case VHOST_USER_SET_PROTOCOL_FEATURES:
case VHOST_USER_SET_LOG_BASE:
msg.payload.u64 = *((__u64 *)arg);
msg.size = sizeof(m.payload.u64);
@@ -454,7 +505,6 @@ vhost_user_sock(struct virtio_user_dev *dev,
switch (req) {
case VHOST_USER_GET_STATUS:
- case VHOST_USER_GET_PROTOCOL_FEATURES:
if (msg.size != sizeof(m.payload.u64)) {
PMD_DRV_LOG(ERR, "Received bad msg size");
return -1;
@@ -592,6 +642,8 @@ struct virtio_user_backend_ops v
[dpdk-dev] [PATCH v4 30/44] net/virtio: add Virtio-user vring setting ops
This patch introduces new callbacks for setting
and getting vring state.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 3 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 49 +++-
drivers/net/virtio/virtio_user/vhost_user.c | 113 +-
drivers/net/virtio/virtio_user/vhost_vdpa.c | 41 +--
.../net/virtio/virtio_user/virtio_user_dev.c | 9 +-
5 files changed, 165 insertions(+), 50 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 463801f563..2d433d5966 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -115,6 +115,9 @@ struct virtio_user_backend_ops {
int (*get_protocol_features)(struct virtio_user_dev *dev, uint64_t
*features);
int (*set_protocol_features)(struct virtio_user_dev *dev, uint64_t
features);
int (*set_memory_table)(struct virtio_user_dev *dev);
+ int (*set_vring_num)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
+ int (*set_vring_base)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
+ int (*get_vring_base)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 69f932bede..f6e7cf4717 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -219,12 +219,49 @@ vhost_kernel_set_memory_table(struct virtio_user_dev *dev)
return -1;
}
+static int
+vhost_kernel_set_vring(struct virtio_user_dev *dev, uint64_t req, struct
vhost_vring_state *state)
+{
+ int ret, fd;
+ unsigned int index = state->index;
+
+ /* Convert from queue index to queue-pair & offset */
+ fd = dev->vhostfds[state->index / 2];
+ state->index %= 2;
+
+ ret = vhost_kernel_ioctl(fd, req, state);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to set vring (request %" PRIu64 ")",
req);
+ return -1;
+ }
+
+ /* restore index back to queue index */
+ state->index = index;
+
+ return 0;
+}
+
+static int
+vhost_kernel_set_vring_num(struct virtio_user_dev *dev, struct
vhost_vring_state *state)
+{
+ return vhost_kernel_set_vring(dev, VHOST_SET_VRING_NUM, state);
+}
+
+static int
+vhost_kernel_set_vring_base(struct virtio_user_dev *dev, struct
vhost_vring_state *state)
+{
+ return vhost_kernel_set_vring(dev, VHOST_SET_VRING_BASE, state);
+}
+
+static int
+vhost_kernel_get_vring_base(struct virtio_user_dev *dev, struct
vhost_vring_state *state)
+{
+ return vhost_kernel_set_vring(dev, VHOST_GET_VRING_BASE, state);
+}
+
static uint64_t vhost_req_user_to_kernel[] = {
[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
[VHOST_USER_SET_VRING_CALL] = VHOST_SET_VRING_CALL,
- [VHOST_USER_SET_VRING_NUM] = VHOST_SET_VRING_NUM,
- [VHOST_USER_SET_VRING_BASE] = VHOST_SET_VRING_BASE,
- [VHOST_USER_GET_VRING_BASE] = VHOST_GET_VRING_BASE,
[VHOST_USER_SET_VRING_ADDR] = VHOST_SET_VRING_ADDR,
[VHOST_USER_SET_VRING_KICK] = VHOST_SET_VRING_KICK,
};
@@ -245,10 +282,7 @@ vhost_kernel_send_request(struct virtio_user_dev *dev,
req_kernel = vhost_req_user_to_kernel[req];
switch (req_kernel) {
- case VHOST_SET_VRING_NUM:
case VHOST_SET_VRING_ADDR:
- case VHOST_SET_VRING_BASE:
- case VHOST_GET_VRING_BASE:
case VHOST_SET_VRING_KICK:
case VHOST_SET_VRING_CALL:
queue_sel = *(unsigned int *)arg;
@@ -403,6 +437,9 @@ struct virtio_user_backend_ops virtio_ops_kernel = {
.get_features = vhost_kernel_get_features,
.set_features = vhost_kernel_set_features,
.set_memory_table = vhost_kernel_set_memory_table,
+ .set_vring_num = vhost_kernel_set_vring_num,
+ .set_vring_base = vhost_kernel_set_vring_base,
+ .get_vring_base = vhost_kernel_get_vring_base,
.send_request = vhost_kernel_send_request,
.enable_qp = vhost_kernel_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 84765d5fdf..8d4c5cce77 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -417,17 +417,94 @@ vhost_user_set_memory_table(struct virtio_user_dev *dev)
return -1;
}
+static int
+vhost_user_set_vring(struct virtio_user_dev *dev, enum vhost_user_request req,
+ struct vhost_vring_state *state)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = req,
+ .flags = VHOST_USER_VERSION,
+ .size = size
[dpdk-dev] [PATCH v4 31/44] net/virtio: add Virtio-user vring file ops
This patch introduces new callbacks for setting
vring files (kick and call).
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 2 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 41 +++--
drivers/net/virtio/virtio_user/vhost_user.c | 46 +--
drivers/net/virtio/virtio_user/vhost_vdpa.c | 18 ++--
.../net/virtio/virtio_user/virtio_user_dev.c | 4 +-
5 files changed, 97 insertions(+), 14 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 2d433d5966..94c524547e 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -118,6 +118,8 @@ struct virtio_user_backend_ops {
int (*set_vring_num)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
int (*set_vring_base)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
int (*get_vring_base)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
+ int (*set_vring_call)(struct virtio_user_dev *dev, struct
vhost_vring_file *file);
+ int (*set_vring_kick)(struct virtio_user_dev *dev, struct
vhost_vring_file *file);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index f6e7cf4717..bdef436595 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -259,11 +259,44 @@ vhost_kernel_get_vring_base(struct virtio_user_dev *dev,
struct vhost_vring_stat
return vhost_kernel_set_vring(dev, VHOST_GET_VRING_BASE, state);
}
+static int
+vhost_kernel_set_vring_file(struct virtio_user_dev *dev, uint64_t req,
+ struct vhost_vring_file *file)
+{
+ int ret, fd;
+ unsigned int index = file->index;
+
+ /* Convert from queue index to queue-pair & offset */
+ fd = dev->vhostfds[file->index / 2];
+ file->index %= 2;
+
+ ret = vhost_kernel_ioctl(fd, req, file);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to set vring file (request %" PRIu64
")", req);
+ return -1;
+ }
+
+ /* restore index back to queue index */
+ file->index = index;
+
+ return 0;
+}
+
+static int
+vhost_kernel_set_vring_kick(struct virtio_user_dev *dev, struct
vhost_vring_file *file)
+{
+ return vhost_kernel_set_vring_file(dev, VHOST_SET_VRING_KICK, file);
+}
+
+static int
+vhost_kernel_set_vring_call(struct virtio_user_dev *dev, struct
vhost_vring_file *file)
+{
+ return vhost_kernel_set_vring_file(dev, VHOST_SET_VRING_CALL, file);
+}
+
static uint64_t vhost_req_user_to_kernel[] = {
[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
- [VHOST_USER_SET_VRING_CALL] = VHOST_SET_VRING_CALL,
[VHOST_USER_SET_VRING_ADDR] = VHOST_SET_VRING_ADDR,
- [VHOST_USER_SET_VRING_KICK] = VHOST_SET_VRING_KICK,
};
static int
@@ -283,8 +316,6 @@ vhost_kernel_send_request(struct virtio_user_dev *dev,
switch (req_kernel) {
case VHOST_SET_VRING_ADDR:
- case VHOST_SET_VRING_KICK:
- case VHOST_SET_VRING_CALL:
queue_sel = *(unsigned int *)arg;
vhostfd = dev->vhostfds[queue_sel / 2];
*(unsigned int *)arg = queue_sel % 2;
@@ -440,6 +471,8 @@ struct virtio_user_backend_ops virtio_ops_kernel = {
.set_vring_num = vhost_kernel_set_vring_num,
.set_vring_base = vhost_kernel_set_vring_base,
.get_vring_base = vhost_kernel_get_vring_base,
+ .set_vring_call = vhost_kernel_set_vring_call,
+ .set_vring_kick = vhost_kernel_set_vring_kick,
.send_request = vhost_kernel_send_request,
.enable_qp = vhost_kernel_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 8d4c5cce77..ccb40484f4 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -498,13 +498,51 @@ vhost_user_get_vring_base(struct virtio_user_dev *dev,
struct vhost_vring_state
return -1;
}
+static int
+vhost_user_set_vring_file(struct virtio_user_dev *dev, enum vhost_user_request
req,
+ struct vhost_vring_file *file)
+{
+ int ret;
+ int fd = file->fd;
+ int num_fd = 0;
+ struct vhost_user_msg msg = {
+ .request = req,
+ .flags = VHOST_USER_VERSION,
+ .size = sizeof(msg.payload.u64),
+ .payload.u64 = file->index & VHOST_USER_VRING_IDX_MASK,
+ };
+
+ if (fd >= 0)
+ num_fd++;
+ else
+ msg.payload.u64 |= VHOST_USER_VRING_NOFD_MASK;
+
+ ret = vhost_user_write(dev->vhostfd, &msg, &fd, num_fd);
+ if (ret
[dpdk-dev] [PATCH v4 32/44] net/virtio: add Virtio-user vring address ops
This patch introduces a new callback for setting
vrings addresses.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 1 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 32 +--
drivers/net/virtio/virtio_user/vhost_user.c | 29 +
drivers/net/virtio/virtio_user/vhost_vdpa.c | 8 -
.../net/virtio/virtio_user/virtio_user_dev.c | 2 +-
5 files changed, 55 insertions(+), 17 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 94c524547e..19040dfdc2 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -120,6 +120,7 @@ struct virtio_user_backend_ops {
int (*get_vring_base)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
int (*set_vring_call)(struct virtio_user_dev *dev, struct
vhost_vring_file *file);
int (*set_vring_kick)(struct virtio_user_dev *dev, struct
vhost_vring_file *file);
+ int (*set_vring_addr)(struct virtio_user_dev *dev, struct
vhost_vring_addr *addr);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index bdef436595..48eeb77aec 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -294,9 +294,30 @@ vhost_kernel_set_vring_call(struct virtio_user_dev *dev,
struct vhost_vring_file
return vhost_kernel_set_vring_file(dev, VHOST_SET_VRING_CALL, file);
}
+static int
+vhost_kernel_set_vring_addr(struct virtio_user_dev *dev, struct
vhost_vring_addr *addr)
+{
+ int ret, fd;
+ unsigned int index = addr->index;
+
+ /* Convert from queue index to queue-pair & offset */
+ fd = dev->vhostfds[addr->index / 2];
+ addr->index %= 2;
+
+ ret = vhost_kernel_ioctl(fd, VHOST_SET_VRING_ADDR, addr);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to set vring address");
+ return -1;
+ }
+
+ /* restore index back to queue index */
+ addr->index = index;
+
+ return 0;
+}
+
static uint64_t vhost_req_user_to_kernel[] = {
[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
- [VHOST_USER_SET_VRING_ADDR] = VHOST_SET_VRING_ADDR,
};
static int
@@ -308,20 +329,12 @@ vhost_kernel_send_request(struct virtio_user_dev *dev,
unsigned int i;
uint64_t req_kernel;
int vhostfd;
- unsigned int queue_sel;
PMD_DRV_LOG(INFO, "%s", vhost_msg_strings[req]);
req_kernel = vhost_req_user_to_kernel[req];
switch (req_kernel) {
- case VHOST_SET_VRING_ADDR:
- queue_sel = *(unsigned int *)arg;
- vhostfd = dev->vhostfds[queue_sel / 2];
- *(unsigned int *)arg = queue_sel % 2;
- PMD_DRV_LOG(DEBUG, "vhostfd=%d, index=%u",
- vhostfd, *(unsigned int *)arg);
- break;
default:
vhostfd = -1;
}
@@ -473,6 +486,7 @@ struct virtio_user_backend_ops virtio_ops_kernel = {
.get_vring_base = vhost_kernel_get_vring_base,
.set_vring_call = vhost_kernel_set_vring_call,
.set_vring_kick = vhost_kernel_set_vring_kick,
+ .set_vring_addr = vhost_kernel_set_vring_addr,
.send_request = vhost_kernel_send_request,
.enable_qp = vhost_kernel_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index ccb40484f4..9ff49d2b1d 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -538,11 +538,32 @@ vhost_user_set_vring_kick(struct virtio_user_dev *dev,
struct vhost_vring_file *
return vhost_user_set_vring_file(dev, VHOST_USER_SET_VRING_KICK, file);
}
+
+static int
+vhost_user_set_vring_addr(struct virtio_user_dev *dev, struct vhost_vring_addr
*addr)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = VHOST_USER_SET_VRING_ADDR,
+ .flags = VHOST_USER_VERSION,
+ .size = sizeof(*addr),
+ .payload.addr = *addr,
+ };
+
+ ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to send vring addresses");
+ return -1;
+ }
+
+ return 0;
+}
+
+
static struct vhost_user_msg m;
const char * const vhost_msg_strings[] = {
[VHOST_USER_RESET_OWNER] = "VHOST_RESET_OWNER",
- [VHOST_USER_SET_VRING_ADDR] = "VHOST_SET_VRING_ADDR",
[VHOST_USER_SET_STATUS] = "VHOST_SET_STATUS",
[VHOST_USER_GET_STATUS] = "VHOST_GET_STATUS",
};
@@ -610,11 +631,6 @@ vhost_user_sock(struct virtio_user_dev *dev,
[dpdk-dev] [PATCH v4 33/44] net/virtio: add Virtio-user status ops
This patch introduces new callbacks to
get and set the device status.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 2 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 14 ++
drivers/net/virtio/virtio_user/vhost_user.c | 121 +-
drivers/net/virtio/virtio_user/vhost_vdpa.c | 16 ++-
.../net/virtio/virtio_user/virtio_user_dev.c | 42 ++
5 files changed, 129 insertions(+), 66 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 19040dfdc2..c896f0836d 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -121,6 +121,8 @@ struct virtio_user_backend_ops {
int (*set_vring_call)(struct virtio_user_dev *dev, struct
vhost_vring_file *file);
int (*set_vring_kick)(struct virtio_user_dev *dev, struct
vhost_vring_file *file);
int (*set_vring_addr)(struct virtio_user_dev *dev, struct
vhost_vring_addr *addr);
+ int (*get_status)(struct virtio_user_dev *dev, uint8_t *status);
+ int (*set_status)(struct virtio_user_dev *dev, uint8_t status);
int (*send_request)(struct virtio_user_dev *dev,
enum vhost_user_request req,
void *arg);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 48eeb77aec..4de4b669d2 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -316,6 +316,18 @@ vhost_kernel_set_vring_addr(struct virtio_user_dev *dev,
struct vhost_vring_addr
return 0;
}
+static int
+vhost_kernel_get_status(struct virtio_user_dev *dev __rte_unused, uint8_t
*status __rte_unused)
+{
+ return -ENOTSUP;
+}
+
+static int
+vhost_kernel_set_status(struct virtio_user_dev *dev __rte_unused, uint8_t
status __rte_unused)
+{
+ return -ENOTSUP;
+}
+
static uint64_t vhost_req_user_to_kernel[] = {
[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
};
@@ -487,6 +499,8 @@ struct virtio_user_backend_ops virtio_ops_kernel = {
.set_vring_call = vhost_kernel_set_vring_call,
.set_vring_kick = vhost_kernel_set_vring_kick,
.set_vring_addr = vhost_kernel_set_vring_addr,
+ .get_status = vhost_kernel_get_status,
+ .set_status = vhost_kernel_set_status,
.send_request = vhost_kernel_send_request,
.enable_qp = vhost_kernel_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 9ff49d2b1d..91ecb50342 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -559,13 +559,100 @@ vhost_user_set_vring_addr(struct virtio_user_dev *dev,
struct vhost_vring_addr *
return 0;
}
+static int
+vhost_user_get_status(struct virtio_user_dev *dev, uint8_t *status)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = VHOST_USER_GET_STATUS,
+ .flags = VHOST_USER_VERSION,
+ };
+
+ /*
+* If features have not been negotiated, we don't know if the backend
+* supports protocol features
+*/
+ if (!(dev->status & VIRTIO_CONFIG_STATUS_FEATURES_OK))
+ return -ENOTSUP;
+
+ /* Status protocol feature requires protocol features support */
+ if (!(dev->device_features & (1ULL << VHOST_USER_F_PROTOCOL_FEATURES)))
+ return -ENOTSUP;
+
+ if (!(dev->protocol_features & (1ULL << VHOST_USER_PROTOCOL_F_STATUS)))
+ return -ENOTSUP;
+
+ ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to send request");
+ goto err;
+ }
+
+ ret = vhost_user_read(dev->vhostfd, &msg);
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to recv request");
+ goto err;
+ }
+
+ if (msg.request != VHOST_USER_GET_STATUS) {
+ PMD_DRV_LOG(ERR, "Unexpected request type (%d)", msg.request);
+ goto err;
+ }
+
+ if (msg.size != sizeof(msg.payload.u64)) {
+ PMD_DRV_LOG(ERR, "Unexpected payload size (%u)", msg.size);
+ goto err;
+ }
+
+ *status = (uint8_t)msg.payload.u64;
+
+ return 0;
+err:
+ PMD_DRV_LOG(ERR, "Failed to get device status");
+ return -1;
+}
+
+static int
+vhost_user_set_status(struct virtio_user_dev *dev, uint8_t status)
+{
+ int ret;
+ struct vhost_user_msg msg = {
+ .request = VHOST_USER_SET_STATUS,
+ .flags = VHOST_USER_VERSION,
+ .size = sizeof(msg.payload.u64),
+ .payload.u64 = status,
+ };
+
+ /*
+* If features have not been negotiated, we don't know if the backend
+* supports protocol features
+
[dpdk-dev] [PATCH v4 35/44] net/virtio: improve Virtio-user errors handling
This patch adds error logs and propagates errors reported
by the backend. It also adds the device path in error logs
to make it easier to distinguish which device is facing the
issue.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
.../net/virtio/virtio_user/virtio_user_dev.c | 154 --
1 file changed, 104 insertions(+), 50 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index 33a25f4684..95204ea955 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -37,10 +37,15 @@ virtio_user_create_queue(struct virtio_user_dev *dev,
uint32_t queue_sel)
* pair.
*/
struct vhost_vring_file file;
+ int ret;
file.index = queue_sel;
file.fd = dev->callfds[queue_sel];
- dev->ops->set_vring_call(dev, &file);
+ ret = dev->ops->set_vring_call(dev, &file);
+ if (ret < 0) {
+ PMD_INIT_LOG(ERR, "(%s) Failed to create queue %u\n",
dev->path, queue_sel);
+ return -1;
+ }
return 0;
}
@@ -48,6 +53,7 @@ virtio_user_create_queue(struct virtio_user_dev *dev,
uint32_t queue_sel)
static int
virtio_user_kick_queue(struct virtio_user_dev *dev, uint32_t queue_sel)
{
+ int ret;
struct vhost_vring_file file;
struct vhost_vring_state state;
struct vring *vring = &dev->vrings[queue_sel];
@@ -73,15 +79,21 @@ virtio_user_kick_queue(struct virtio_user_dev *dev,
uint32_t queue_sel)
state.index = queue_sel;
state.num = vring->num;
- dev->ops->set_vring_num(dev, &state);
+ ret = dev->ops->set_vring_num(dev, &state);
+ if (ret < 0)
+ goto err;
state.index = queue_sel;
state.num = 0; /* no reservation */
if (dev->features & (1ULL << VIRTIO_F_RING_PACKED))
state.num |= (1 << 15);
- dev->ops->set_vring_base(dev, &state);
+ ret = dev->ops->set_vring_base(dev, &state);
+ if (ret < 0)
+ goto err;
- dev->ops->set_vring_addr(dev, &addr);
+ ret = dev->ops->set_vring_addr(dev, &addr);
+ if (ret < 0)
+ goto err;
/* Of all per virtqueue MSGs, make sure VHOST_USER_SET_VRING_KICK comes
* lastly because vhost depends on this msg to judge if
@@ -89,9 +101,15 @@ virtio_user_kick_queue(struct virtio_user_dev *dev,
uint32_t queue_sel)
*/
file.index = queue_sel;
file.fd = dev->kickfds[queue_sel];
- dev->ops->set_vring_kick(dev, &file);
+ ret = dev->ops->set_vring_kick(dev, &file);
+ if (ret < 0)
+ goto err;
return 0;
+err:
+ PMD_INIT_LOG(ERR, "(%s) Failed to kick queue %u\n", dev->path,
queue_sel);
+
+ return -1;
}
static int
@@ -103,14 +121,14 @@ virtio_user_queue_setup(struct virtio_user_dev *dev,
for (i = 0; i < dev->max_queue_pairs; ++i) {
queue_sel = 2 * i + VTNET_SQ_RQ_QUEUE_IDX;
if (fn(dev, queue_sel) < 0) {
- PMD_DRV_LOG(INFO, "setup rx vq fails: %u", i);
+ PMD_DRV_LOG(ERR, "(%s) setup rx vq %u failed",
dev->path, i);
return -1;
}
}
for (i = 0; i < dev->max_queue_pairs; ++i) {
queue_sel = 2 * i + VTNET_SQ_TQ_QUEUE_IDX;
if (fn(dev, queue_sel) < 0) {
- PMD_DRV_LOG(INFO, "setup tx vq fails: %u", i);
+ PMD_DRV_LOG(INFO, "(%s) setup tx vq %u failed",
dev->path, i);
return -1;
}
}
@@ -144,7 +162,7 @@ virtio_user_dev_set_features(struct virtio_user_dev *dev)
ret = dev->ops->set_features(dev, features);
if (ret < 0)
goto error;
- PMD_DRV_LOG(INFO, "set features: %" PRIx64, features);
+ PMD_DRV_LOG(INFO, "(%s) set features: 0x%" PRIx64, dev->path, features);
error:
pthread_mutex_unlock(&dev->mutex);
@@ -172,9 +190,10 @@ virtio_user_start_device(struct virtio_user_dev *dev)
rte_mcfg_mem_read_lock();
pthread_mutex_lock(&dev->mutex);
+ /* Vhost-user client not connected yet, will start later */
if (dev->backend_type == VIRTIO_USER_BACKEND_VHOST_USER &&
dev->vhostfd < 0)
- goto error;
+ goto out;
/* Step 2: share memory regions */
ret = dev->ops->set_memory_table(dev);
@@ -182,15 +201,19 @@ virtio_user_start_device(struct virtio_user_dev *dev)
goto error;
/* Step 3: kick queues */
- if (virtio_user_queue_setup(dev, virtio_user_kick_queue) < 0)
+ ret = virtio_user_queue_setup(dev, virtio_user_kick_queue);
+ if (ret < 0)
goto error;
/* Step 4: enable queues
* we enable the 1st queue pair by default.
[dpdk-dev] [PATCH v4 36/44] net/virtio: move Vhost-user requests to Vhost-user backend
Now that we have a proper isolation of the backends,
we can move Vhost-user requests declaration to the
Vhost-user backend file.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h | 25 -
drivers/net/virtio/virtio_user/vhost_user.c | 25 +
2 files changed, 25 insertions(+), 25 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 6294b8afee..2aa6b2cb70 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -63,31 +63,6 @@ struct vhost_vring_addr {
#define VHOST_USER_PROTOCOL_F_STATUS 16
#endif
-enum vhost_user_request {
- VHOST_USER_NONE = 0,
- VHOST_USER_GET_FEATURES = 1,
- VHOST_USER_SET_FEATURES = 2,
- VHOST_USER_SET_OWNER = 3,
- VHOST_USER_RESET_OWNER = 4,
- VHOST_USER_SET_MEM_TABLE = 5,
- VHOST_USER_SET_LOG_BASE = 6,
- VHOST_USER_SET_LOG_FD = 7,
- VHOST_USER_SET_VRING_NUM = 8,
- VHOST_USER_SET_VRING_ADDR = 9,
- VHOST_USER_SET_VRING_BASE = 10,
- VHOST_USER_GET_VRING_BASE = 11,
- VHOST_USER_SET_VRING_KICK = 12,
- VHOST_USER_SET_VRING_CALL = 13,
- VHOST_USER_SET_VRING_ERR = 14,
- VHOST_USER_GET_PROTOCOL_FEATURES = 15,
- VHOST_USER_SET_PROTOCOL_FEATURES = 16,
- VHOST_USER_GET_QUEUE_NUM = 17,
- VHOST_USER_SET_VRING_ENABLE = 18,
- VHOST_USER_SET_STATUS = 39,
- VHOST_USER_GET_STATUS = 40,
- VHOST_USER_MAX
-};
-
#ifndef VHOST_BACKEND_F_IOTLB_MSG_V2
#define VHOST_BACKEND_F_IOTLB_MSG_V2 1
#endif
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 165cf820b4..ba02fdcb65 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -27,6 +27,31 @@ struct vhost_memory {
struct vhost_memory_region regions[VHOST_MEMORY_MAX_NREGIONS];
};
+enum vhost_user_request {
+ VHOST_USER_NONE = 0,
+ VHOST_USER_GET_FEATURES = 1,
+ VHOST_USER_SET_FEATURES = 2,
+ VHOST_USER_SET_OWNER = 3,
+ VHOST_USER_RESET_OWNER = 4,
+ VHOST_USER_SET_MEM_TABLE = 5,
+ VHOST_USER_SET_LOG_BASE = 6,
+ VHOST_USER_SET_LOG_FD = 7,
+ VHOST_USER_SET_VRING_NUM = 8,
+ VHOST_USER_SET_VRING_ADDR = 9,
+ VHOST_USER_SET_VRING_BASE = 10,
+ VHOST_USER_GET_VRING_BASE = 11,
+ VHOST_USER_SET_VRING_KICK = 12,
+ VHOST_USER_SET_VRING_CALL = 13,
+ VHOST_USER_SET_VRING_ERR = 14,
+ VHOST_USER_GET_PROTOCOL_FEATURES = 15,
+ VHOST_USER_SET_PROTOCOL_FEATURES = 16,
+ VHOST_USER_GET_QUEUE_NUM = 17,
+ VHOST_USER_SET_VRING_ENABLE = 18,
+ VHOST_USER_SET_STATUS = 39,
+ VHOST_USER_GET_STATUS = 40,
+ VHOST_USER_MAX
+};
+
struct vhost_user_msg {
enum vhost_user_request request;
--
2.29.2
[dpdk-dev] [PATCH v4 34/44] net/virtio: remove useless request ops
Now that all the ops have been implemented, we
can remove the send_request ops for all backends.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 15 +--
drivers/net/virtio/virtio_user/vhost_kernel.c | 43
drivers/net/virtio/virtio_user/vhost_user.c | 102 --
drivers/net/virtio/virtio_user/vhost_vdpa.c | 49 -
4 files changed, 3 insertions(+), 206 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index c896f0836d..6294b8afee 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -96,8 +96,6 @@ enum vhost_user_request {
#define VHOST_BACKEND_F_IOTLB_BATCH 2
#endif
-extern const char * const vhost_msg_strings[VHOST_USER_MAX];
-
struct vhost_memory_region {
uint64_t guest_phys_addr;
uint64_t memory_size; /* bytes */
@@ -123,16 +121,9 @@ struct virtio_user_backend_ops {
int (*set_vring_addr)(struct virtio_user_dev *dev, struct
vhost_vring_addr *addr);
int (*get_status)(struct virtio_user_dev *dev, uint8_t *status);
int (*set_status)(struct virtio_user_dev *dev, uint8_t status);
- int (*send_request)(struct virtio_user_dev *dev,
- enum vhost_user_request req,
- void *arg);
- int (*enable_qp)(struct virtio_user_dev *dev,
-uint16_t pair_idx,
-int enable);
- int (*dma_map)(struct virtio_user_dev *dev, void *addr,
- uint64_t iova, size_t len);
- int (*dma_unmap)(struct virtio_user_dev *dev, void *addr,
- uint64_t iova, size_t len);
+ int (*enable_qp)(struct virtio_user_dev *dev, uint16_t pair_idx, int
enable);
+ int (*dma_map)(struct virtio_user_dev *dev, void *addr, uint64_t iova,
size_t len);
+ int (*dma_unmap)(struct virtio_user_dev *dev, void *addr, uint64_t
iova, size_t len);
};
extern struct virtio_user_backend_ops virtio_ops_user;
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 4de4b669d2..16065c8d11 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -328,48 +328,6 @@ vhost_kernel_set_status(struct virtio_user_dev *dev
__rte_unused, uint8_t status
return -ENOTSUP;
}
-static uint64_t vhost_req_user_to_kernel[] = {
- [VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
-};
-
-static int
-vhost_kernel_send_request(struct virtio_user_dev *dev,
- enum vhost_user_request req,
- void *arg)
-{
- int ret = -1;
- unsigned int i;
- uint64_t req_kernel;
- int vhostfd;
-
- PMD_DRV_LOG(INFO, "%s", vhost_msg_strings[req]);
-
- req_kernel = vhost_req_user_to_kernel[req];
-
- switch (req_kernel) {
- default:
- vhostfd = -1;
- }
- if (vhostfd == -1) {
- for (i = 0; i < dev->max_queue_pairs; ++i) {
- if (dev->vhostfds[i] < 0)
- continue;
-
- ret = ioctl(dev->vhostfds[i], req_kernel, arg);
- if (ret < 0)
- break;
- }
- } else {
- ret = ioctl(vhostfd, req_kernel, arg);
- }
-
- if (ret < 0)
- PMD_DRV_LOG(ERR, "%s failed: %s",
- vhost_msg_strings[req], strerror(errno));
-
- return ret;
-}
-
/**
* Set up environment to talk with a vhost kernel backend.
*
@@ -501,6 +459,5 @@ struct virtio_user_backend_ops virtio_ops_kernel = {
.set_vring_addr = vhost_kernel_set_vring_addr,
.get_status = vhost_kernel_get_status,
.set_status = vhost_kernel_set_status,
- .send_request = vhost_kernel_send_request,
.enable_qp = vhost_kernel_enable_queue_pair
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 91ecb50342..165cf820b4 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -649,107 +649,6 @@ vhost_user_set_status(struct virtio_user_dev *dev,
uint8_t status)
return vhost_user_check_reply_ack(dev, &msg);
}
-static struct vhost_user_msg m;
-
-const char * const vhost_msg_strings[] = {
- [VHOST_USER_RESET_OWNER] = "VHOST_RESET_OWNER",
-};
-
-static int
-vhost_user_sock(struct virtio_user_dev *dev,
- enum vhost_user_request req,
- void *arg)
-{
- struct vhost_user_msg msg;
- struct vhost_vring_file *file = 0;
- int need_reply = 0;
- int fds[VHOST_MEMORY_MAX_NREGIONS];
- int fd_num = 0;
- int vhostfd = dev->vhostfd;
-
- RTE_SET_USED(m);
-
- PMD_DRV_LOG(INFO, "%s", vhost_ms
[dpdk-dev] [PATCH v4 37/44] net/virtio: make server mode blocking
This patch makes the Vhost-user backend server mode
blocking at init, waiting for the client connection.
The goal is to make the driver more reliable, as without
waiting for client connection, the Virtio driver has to
assume the Vhost-user backend will support all the
features it has advertized, which could lead to undefined
behaviour.
For example, without this patch, if the user enables packed
ring Virtio feature but the backend does not support it,
the ring initialized by the driver will not be compatible
with the backend.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost_user.c | 9 +-
.../net/virtio/virtio_user/virtio_user_dev.c | 121 +++---
drivers/net/virtio/virtio_user_ethdev.c | 5 -
3 files changed, 55 insertions(+), 80 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index ba02fdcb65..90fed6fe97 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -692,6 +692,14 @@ virtio_user_start_server(struct virtio_user_dev *dev,
struct sockaddr_un *un)
if (ret < 0)
return -1;
+ PMD_DRV_LOG(NOTICE, "(%s) waiting for client connection...", dev->path);
+ dev->vhostfd = accept(fd, NULL, NULL);
+ if (dev->vhostfd < 0) {
+ PMD_DRV_LOG(ERR, "Failed to accept initial client connection
(%s)",
+ strerror(errno));
+ return -1;
+ }
+
flag = fcntl(fd, F_GETFL);
if (fcntl(fd, F_SETFL, flag | O_NONBLOCK) < 0) {
PMD_DRV_LOG(ERR, "fcntl failed, %s", strerror(errno));
@@ -736,7 +744,6 @@ vhost_user_setup(struct virtio_user_dev *dev)
close(fd);
return -1;
}
- dev->vhostfd = -1;
} else {
if (connect(fd, (struct sockaddr *)&un, sizeof(un)) < 0) {
PMD_DRV_LOG(ERR, "connect error, %s", strerror(errno));
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index 95204ea955..c2a41fe3a0 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -144,10 +144,6 @@ virtio_user_dev_set_features(struct virtio_user_dev *dev)
pthread_mutex_lock(&dev->mutex);
- if (dev->backend_type == VIRTIO_USER_BACKEND_VHOST_USER &&
- dev->vhostfd < 0)
- goto error;
-
/* Step 0: tell vhost to create queues */
if (virtio_user_queue_setup(dev, virtio_user_create_queue) < 0)
goto error;
@@ -190,11 +186,6 @@ virtio_user_start_device(struct virtio_user_dev *dev)
rte_mcfg_mem_read_lock();
pthread_mutex_lock(&dev->mutex);
- /* Vhost-user client not connected yet, will start later */
- if (dev->backend_type == VIRTIO_USER_BACKEND_VHOST_USER &&
- dev->vhostfd < 0)
- goto out;
-
/* Step 2: share memory regions */
ret = dev->ops->set_memory_table(dev);
if (ret < 0)
@@ -213,7 +204,7 @@ virtio_user_start_device(struct virtio_user_dev *dev)
goto error;
dev->started = true;
-out:
+
pthread_mutex_unlock(&dev->mutex);
rte_mcfg_mem_read_unlock();
@@ -422,36 +413,36 @@ virtio_user_dev_setup(struct virtio_user_dev *dev)
PMD_DRV_LOG(ERR, "Server mode only supports
vhost-user!");
return -1;
}
+ }
+
+ if (dev->backend_type == VIRTIO_USER_BACKEND_VHOST_USER) {
dev->ops = &virtio_ops_user;
- } else {
- if (dev->backend_type == VIRTIO_USER_BACKEND_VHOST_USER) {
- dev->ops = &virtio_ops_user;
- } else if (dev->backend_type ==
- VIRTIO_USER_BACKEND_VHOST_KERNEL) {
- dev->ops = &virtio_ops_kernel;
-
- dev->vhostfds = malloc(dev->max_queue_pairs *
- sizeof(int));
- dev->tapfds = malloc(dev->max_queue_pairs *
-sizeof(int));
- if (!dev->vhostfds || !dev->tapfds) {
- PMD_INIT_LOG(ERR, "(%s) Failed to allocate
FDs", dev->path);
- return -1;
- }
-
- for (q = 0; q < dev->max_queue_pairs; ++q) {
- dev->vhostfds[q] = -1;
- dev->tapfds[q] = -1;
- }
- } else if (dev->backend_type ==
- VIRTIO_USER_BACKEND_VHOST_VDPA) {
- dev->ops = &virtio_ops_vdpa;
- } else {
-
[dpdk-dev] [PATCH v4 39/44] net/virtio: introduce backend data
The goal of this patch is to introduce backend-specific
data in order to better isolate what is backend-specific
from what is generic to Virtio-user.
For now, only Vhost-user protocol features are moved to
Vhost-user backend data.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 1 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 7 +++
drivers/net/virtio/virtio_user/vhost_user.c | 62 +++
drivers/net/virtio/virtio_user/vhost_vdpa.c | 7 +++
.../net/virtio/virtio_user/virtio_user_dev.c | 2 +
.../net/virtio/virtio_user/virtio_user_dev.h | 2 +
6 files changed, 68 insertions(+), 13 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index dfa2735a20..fc4f059c02 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -65,6 +65,7 @@ struct virtio_user_dev;
struct virtio_user_backend_ops {
int (*setup)(struct virtio_user_dev *dev);
+ int (*destroy)(struct virtio_user_dev *dev);
int (*get_backend_features)(uint64_t *features);
int (*set_owner)(struct virtio_user_dev *dev);
int (*get_features)(struct virtio_user_dev *dev, uint64_t *features);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 2047d41a50..26a71ad07b 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -357,6 +357,12 @@ vhost_kernel_setup(struct virtio_user_dev *dev)
return 0;
}
+static int
+vhost_kernel_destroy(struct virtio_user_dev *dev __rte_unused)
+{
+ return 0;
+}
+
static int
vhost_kernel_set_backend(int vhostfd, int tapfd)
{
@@ -455,6 +461,7 @@ vhost_kernel_get_backend_features(uint64_t *features)
struct virtio_user_backend_ops virtio_ops_kernel = {
.setup = vhost_kernel_setup,
+ .destroy = vhost_kernel_destroy,
.get_backend_features = vhost_kernel_get_backend_features,
.set_owner = vhost_kernel_set_owner,
.get_features = vhost_kernel_get_features,
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index e3bc5a9b03..26b8c9e9c9 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -17,6 +17,9 @@
#include "vhost.h"
#include "virtio_user_dev.h"
+struct vhost_user_data {
+ uint64_t protocol_features;
+};
#ifndef VHOST_USER_F_PROTOCOL_FEATURES
#define VHOST_USER_F_PROTOCOL_FEATURES 30
@@ -287,6 +290,7 @@ static int
vhost_user_get_features(struct virtio_user_dev *dev, uint64_t *features)
{
int ret;
+ struct vhost_user_data *data = dev->backend_data;
struct vhost_user_msg msg = {
.request = VHOST_USER_GET_FEATURES,
.flags = VHOST_USER_VERSION,
@@ -316,17 +320,17 @@ vhost_user_get_features(struct virtio_user_dev *dev,
uint64_t *features)
return 0;
/* Negotiate protocol features */
- ret = vhost_user_get_protocol_features(dev, &dev->protocol_features);
+ ret = vhost_user_get_protocol_features(dev, &data->protocol_features);
if (ret < 0)
goto err;
- dev->protocol_features &= VHOST_USER_SUPPORTED_PROTOCOL_FEATURES;
+ data->protocol_features &= VHOST_USER_SUPPORTED_PROTOCOL_FEATURES;
- ret = vhost_user_set_protocol_features(dev, dev->protocol_features);
+ ret = vhost_user_set_protocol_features(dev, data->protocol_features);
if (ret < 0)
goto err;
- if (!(dev->protocol_features & (1ULL << VHOST_USER_PROTOCOL_F_MQ)))
+ if (!(data->protocol_features & (1ULL << VHOST_USER_PROTOCOL_F_MQ)))
dev->unsupported_features |= (1ull << VIRTIO_NET_F_MQ);
return 0;
@@ -444,12 +448,13 @@ vhost_user_set_memory_table(struct virtio_user_dev *dev)
struct walk_arg wa;
int fds[VHOST_MEMORY_MAX_NREGIONS];
int ret, fd_num;
+ struct vhost_user_data *data = dev->backend_data;
struct vhost_user_msg msg = {
.request = VHOST_USER_SET_MEM_TABLE,
.flags = VHOST_USER_VERSION,
};
- if (dev->protocol_features & (1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK))
+ if (data->protocol_features & (1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK))
msg.flags |= VHOST_USER_NEED_REPLY_MASK;
wa.region_nr = 0;
@@ -628,6 +633,7 @@ static int
vhost_user_get_status(struct virtio_user_dev *dev, uint8_t *status)
{
int ret;
+ struct vhost_user_data *data = dev->backend_data;
struct vhost_user_msg msg = {
.request = VHOST_USER_GET_STATUS,
.flags = VHOST_USER_VERSION,
@@ -644,7 +650,7 @@ vhost_user_get_status(struct virtio_user_dev *dev, uint8_t
*status)
if (!(dev->device_features & (1ULL << VHOST_USER
[dpdk-dev] [PATCH v4 38/44] net/virtio: move protocol features to Vhost-user
Since only protocol features are specific to Vhost-user
backend, this patch moves all related code to Vhost-user
file.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 20 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 9 ++
drivers/net/virtio/virtio_user/vhost_user.c | 83 +++
drivers/net/virtio/virtio_user/vhost_vdpa.c | 34 +++-
.../net/virtio/virtio_user/virtio_user_dev.c | 47 ++-
drivers/net/virtio/virtio_user_ethdev.c | 17
6 files changed, 113 insertions(+), 97 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index 2aa6b2cb70..dfa2735a20 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -46,23 +46,6 @@ struct vhost_vring_addr {
uint64_t log_guest_addr;
};
-#ifndef VHOST_USER_F_PROTOCOL_FEATURES
-#define VHOST_USER_F_PROTOCOL_FEATURES 30
-#endif
-
-/** Protocol features. */
-#ifndef VHOST_USER_PROTOCOL_F_MQ
-#define VHOST_USER_PROTOCOL_F_MQ 0
-#endif
-
-#ifndef VHOST_USER_PROTOCOL_F_REPLY_ACK
-#define VHOST_USER_PROTOCOL_F_REPLY_ACK 3
-#endif
-
-#ifndef VHOST_USER_PROTOCOL_F_STATUS
-#define VHOST_USER_PROTOCOL_F_STATUS 16
-#endif
-
#ifndef VHOST_BACKEND_F_IOTLB_MSG_V2
#define VHOST_BACKEND_F_IOTLB_MSG_V2 1
#endif
@@ -82,11 +65,10 @@ struct virtio_user_dev;
struct virtio_user_backend_ops {
int (*setup)(struct virtio_user_dev *dev);
+ int (*get_backend_features)(uint64_t *features);
int (*set_owner)(struct virtio_user_dev *dev);
int (*get_features)(struct virtio_user_dev *dev, uint64_t *features);
int (*set_features)(struct virtio_user_dev *dev, uint64_t features);
- int (*get_protocol_features)(struct virtio_user_dev *dev, uint64_t
*features);
- int (*set_protocol_features)(struct virtio_user_dev *dev, uint64_t
features);
int (*set_memory_table)(struct virtio_user_dev *dev);
int (*set_vring_num)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
int (*set_vring_base)(struct virtio_user_dev *dev, struct
vhost_vring_state *state);
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 16065c8d11..2047d41a50 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -445,8 +445,17 @@ vhost_kernel_enable_queue_pair(struct virtio_user_dev *dev,
return 0;
}
+static int
+vhost_kernel_get_backend_features(uint64_t *features)
+{
+ *features = 0;
+
+ return 0;
+}
+
struct virtio_user_backend_ops virtio_ops_kernel = {
.setup = vhost_kernel_setup,
+ .get_backend_features = vhost_kernel_get_backend_features,
.set_owner = vhost_kernel_set_owner,
.get_features = vhost_kernel_get_features,
.set_features = vhost_kernel_set_features,
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 90fed6fe97..e3bc5a9b03 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -17,6 +17,29 @@
#include "vhost.h"
#include "virtio_user_dev.h"
+
+#ifndef VHOST_USER_F_PROTOCOL_FEATURES
+#define VHOST_USER_F_PROTOCOL_FEATURES 30
+#endif
+
+/** Protocol features. */
+#ifndef VHOST_USER_PROTOCOL_F_MQ
+#define VHOST_USER_PROTOCOL_F_MQ 0
+#endif
+
+#ifndef VHOST_USER_PROTOCOL_F_REPLY_ACK
+#define VHOST_USER_PROTOCOL_F_REPLY_ACK 3
+#endif
+
+#ifndef VHOST_USER_PROTOCOL_F_STATUS
+#define VHOST_USER_PROTOCOL_F_STATUS 16
+#endif
+
+#define VHOST_USER_SUPPORTED_PROTOCOL_FEATURES \
+ (1ULL << VHOST_USER_PROTOCOL_F_MQ | \
+1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK | \
+1ULL << VHOST_USER_PROTOCOL_F_STATUS)
+
/* The version of the protocol we support */
#define VHOST_USER_VERSION0x1
@@ -205,11 +228,11 @@ vhost_user_set_owner(struct virtio_user_dev *dev)
}
static int
-vhost_user_get_features(struct virtio_user_dev *dev, uint64_t *features)
+vhost_user_get_protocol_features(struct virtio_user_dev *dev, uint64_t
*features)
{
int ret;
struct vhost_user_msg msg = {
- .request = VHOST_USER_GET_FEATURES,
+ .request = VHOST_USER_GET_PROTOCOL_FEATURES,
.flags = VHOST_USER_VERSION,
};
@@ -221,7 +244,7 @@ vhost_user_get_features(struct virtio_user_dev *dev,
uint64_t *features)
if (ret < 0)
goto err;
- if (msg.request != VHOST_USER_GET_FEATURES) {
+ if (msg.request != VHOST_USER_GET_PROTOCOL_FEATURES) {
PMD_DRV_LOG(ERR, "Unexpected request type (%d)", msg.request);
goto err;
}
@@ -235,27 +258,25 @@ vhost_user_get_features(struct virtio_user_dev *dev,
uint64_t *features)
return 0;
err:
- PMD_DRV_LOG(ERR, "Failed to get backend feat
[dpdk-dev] [PATCH v4 41/44] net/virtio: move Vhost-kernel data to its backend
As done earlier for Vhost-user, this patch moves the
Vhost-Kernel specific data to its backend file.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost_kernel.c | 106 +++---
.../net/virtio/virtio_user/virtio_user_dev.c | 43 ++-
.../net/virtio/virtio_user/virtio_user_dev.h | 7 +-
3 files changed, 98 insertions(+), 58 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 0d05e7d339..768db55a6c 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -14,6 +14,11 @@
#include "virtio_user_dev.h"
#include "vhost_kernel_tap.h"
+struct vhost_kernel_data {
+ int *vhostfds;
+ int *tapfds;
+};
+
struct vhost_memory_kernel {
uint32_t nregions;
uint32_t padding;
@@ -96,7 +101,9 @@ vhost_kernel_ioctl(int fd, uint64_t request, void *arg)
static int
vhost_kernel_set_owner(struct virtio_user_dev *dev)
{
- return vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_OWNER, NULL);
+ struct vhost_kernel_data *data = dev->backend_data;
+
+ return vhost_kernel_ioctl(data->vhostfds[0], VHOST_SET_OWNER, NULL);
}
static int
@@ -104,8 +111,9 @@ vhost_kernel_get_features(struct virtio_user_dev *dev,
uint64_t *features)
{
int ret;
unsigned int tap_features;
+ struct vhost_kernel_data *data = dev->backend_data;
- ret = vhost_kernel_ioctl(dev->vhostfds[0], VHOST_GET_FEATURES,
features);
+ ret = vhost_kernel_ioctl(data->vhostfds[0], VHOST_GET_FEATURES,
features);
if (ret < 0) {
PMD_DRV_LOG(ERR, "Failed to get features");
return -1;
@@ -138,6 +146,8 @@ vhost_kernel_get_features(struct virtio_user_dev *dev,
uint64_t *features)
static int
vhost_kernel_set_features(struct virtio_user_dev *dev, uint64_t features)
{
+ struct vhost_kernel_data *data = dev->backend_data;
+
/* We don't need memory protection here */
features &= ~(1ULL << VIRTIO_F_IOMMU_PLATFORM);
/* VHOST kernel does not know about below flags */
@@ -145,7 +155,7 @@ vhost_kernel_set_features(struct virtio_user_dev *dev,
uint64_t features)
features &= ~VHOST_KERNEL_HOST_OFFLOADS_MASK;
features &= ~(1ULL << VIRTIO_NET_F_MQ);
- return vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_FEATURES,
&features);
+ return vhost_kernel_ioctl(data->vhostfds[0], VHOST_SET_FEATURES,
&features);
}
static int
@@ -185,6 +195,7 @@ add_memseg_list(const struct rte_memseg_list *msl, void
*arg)
static int
vhost_kernel_set_memory_table(struct virtio_user_dev *dev)
{
+ struct vhost_kernel_data *data = dev->backend_data;
struct vhost_memory_kernel *vm;
int ret;
@@ -205,7 +216,7 @@ vhost_kernel_set_memory_table(struct virtio_user_dev *dev)
if (ret < 0)
goto err_free;
- ret = vhost_kernel_ioctl(dev->vhostfds[0], VHOST_SET_MEM_TABLE, vm);
+ ret = vhost_kernel_ioctl(data->vhostfds[0], VHOST_SET_MEM_TABLE, vm);
if (ret < 0)
goto err_free;
@@ -224,9 +235,10 @@ vhost_kernel_set_vring(struct virtio_user_dev *dev,
uint64_t req, struct vhost_v
{
int ret, fd;
unsigned int index = state->index;
+ struct vhost_kernel_data *data = dev->backend_data;
/* Convert from queue index to queue-pair & offset */
- fd = dev->vhostfds[state->index / 2];
+ fd = data->vhostfds[state->index / 2];
state->index %= 2;
ret = vhost_kernel_ioctl(fd, req, state);
@@ -265,9 +277,10 @@ vhost_kernel_set_vring_file(struct virtio_user_dev *dev,
uint64_t req,
{
int ret, fd;
unsigned int index = file->index;
+ struct vhost_kernel_data *data = dev->backend_data;
/* Convert from queue index to queue-pair & offset */
- fd = dev->vhostfds[file->index / 2];
+ fd = data->vhostfds[file->index / 2];
file->index %= 2;
ret = vhost_kernel_ioctl(fd, req, file);
@@ -299,9 +312,10 @@ vhost_kernel_set_vring_addr(struct virtio_user_dev *dev,
struct vhost_vring_addr
{
int ret, fd;
unsigned int index = addr->index;
+ struct vhost_kernel_data *data = dev->backend_data;
/* Convert from queue index to queue-pair & offset */
- fd = dev->vhostfds[addr->index / 2];
+ fd = data->vhostfds[addr->index / 2];
addr->index %= 2;
ret = vhost_kernel_ioctl(fd, VHOST_SET_VRING_ADDR, addr);
@@ -339,27 +353,82 @@ static int
vhost_kernel_setup(struct virtio_user_dev *dev)
{
int vhostfd;
- uint32_t i;
+ uint32_t q, i;
+ struct vhost_kernel_data *data;
+
+ data = malloc(sizeof(*data));
+ if (!data) {
+ PMD_INIT_LOG(ERR, "(%s) Failed to allocate Vhost-kernel data",
dev->path);
+ return -1;
+ }
+
+ data->vhostfds = malloc(dev->max_q
[dpdk-dev] [PATCH v4 40/44] net/virtio: move Vhost-user specifics to its backend
This patch moves all the Vhost-user backend specific
logic like Vhost FD, listen FD and interrupt handling
to the vhost-user backend implementation.
In order to achieve that, new ops are created to update
the link status, disconnect and reconnect the server,
and fetch the link state interrupt FD.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost.h| 4 +
drivers/net/virtio/virtio_user/vhost_kernel.c | 18 +-
drivers/net/virtio/virtio_user/vhost_user.c | 176 ++---
drivers/net/virtio/virtio_user/vhost_vdpa.c | 16 ++
.../net/virtio/virtio_user/virtio_user_dev.c | 181 +++---
.../net/virtio/virtio_user/virtio_user_dev.h | 9 +-
drivers/net/virtio/virtio_user_ethdev.c | 179 +
7 files changed, 350 insertions(+), 233 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost.h
b/drivers/net/virtio/virtio_user/vhost.h
index fc4f059c02..c49e88036d 100644
--- a/drivers/net/virtio/virtio_user/vhost.h
+++ b/drivers/net/virtio/virtio_user/vhost.h
@@ -82,6 +82,10 @@ struct virtio_user_backend_ops {
int (*enable_qp)(struct virtio_user_dev *dev, uint16_t pair_idx, int
enable);
int (*dma_map)(struct virtio_user_dev *dev, void *addr, uint64_t iova,
size_t len);
int (*dma_unmap)(struct virtio_user_dev *dev, void *addr, uint64_t
iova, size_t len);
+ int (*update_link_state)(struct virtio_user_dev *dev);
+ int (*server_disconnect)(struct virtio_user_dev *dev);
+ int (*server_reconnect)(struct virtio_user_dev *dev);
+ int (*get_intr_fd)(struct virtio_user_dev *dev);
};
extern struct virtio_user_backend_ops virtio_ops_user;
diff --git a/drivers/net/virtio/virtio_user/vhost_kernel.c
b/drivers/net/virtio/virtio_user/vhost_kernel.c
index 26a71ad07b..0d05e7d339 100644
--- a/drivers/net/virtio/virtio_user/vhost_kernel.c
+++ b/drivers/net/virtio/virtio_user/vhost_kernel.c
@@ -459,6 +459,20 @@ vhost_kernel_get_backend_features(uint64_t *features)
return 0;
}
+static int
+vhost_kernel_update_link_state(struct virtio_user_dev *dev __rte_unused)
+{
+ /* Nothing to update (Maybe get TAP interface link state?) */
+ return 0;
+}
+
+static int
+vhost_kernel_get_intr_fd(struct virtio_user_dev *dev __rte_unused)
+{
+ /* No link state interrupt with Vhost-kernel */
+ return -1;
+}
+
struct virtio_user_backend_ops virtio_ops_kernel = {
.setup = vhost_kernel_setup,
.destroy = vhost_kernel_destroy,
@@ -475,5 +489,7 @@ struct virtio_user_backend_ops virtio_ops_kernel = {
.set_vring_addr = vhost_kernel_set_vring_addr,
.get_status = vhost_kernel_get_status,
.set_status = vhost_kernel_set_status,
- .enable_qp = vhost_kernel_enable_queue_pair
+ .enable_qp = vhost_kernel_enable_queue_pair,
+ .update_link_state = vhost_kernel_update_link_state,
+ .get_intr_fd = vhost_kernel_get_intr_fd,
};
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 26b8c9e9c9..088aae3aa3 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -11,6 +11,7 @@
#include
#include
+#include
#include
#include
@@ -18,6 +19,8 @@
#include "virtio_user_dev.h"
struct vhost_user_data {
+ int vhostfd;
+ int listenfd;
uint64_t protocol_features;
};
@@ -182,13 +185,14 @@ vhost_user_read(int fd, struct vhost_user_msg *msg)
static int
vhost_user_check_reply_ack(struct virtio_user_dev *dev, struct vhost_user_msg
*msg)
{
+ struct vhost_user_data *data = dev->backend_data;
enum vhost_user_request req = msg->request;
int ret;
if (!(msg->flags & VHOST_USER_NEED_REPLY_MASK))
return 0;
- ret = vhost_user_read(dev->vhostfd, msg);
+ ret = vhost_user_read(data->vhostfd, msg);
if (ret < 0) {
PMD_DRV_LOG(ERR, "Failed to read reply-ack");
return -1;
@@ -216,12 +220,13 @@ static int
vhost_user_set_owner(struct virtio_user_dev *dev)
{
int ret;
+ struct vhost_user_data *data = dev->backend_data;
struct vhost_user_msg msg = {
.request = VHOST_USER_SET_OWNER,
.flags = VHOST_USER_VERSION,
};
- ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
+ ret = vhost_user_write(data->vhostfd, &msg, NULL, 0);
if (ret < 0) {
PMD_DRV_LOG(ERR, "Failed to set owner");
return -1;
@@ -234,16 +239,17 @@ static int
vhost_user_get_protocol_features(struct virtio_user_dev *dev, uint64_t
*features)
{
int ret;
+ struct vhost_user_data *data = dev->backend_data;
struct vhost_user_msg msg = {
.request = VHOST_USER_GET_PROTOCOL_FEATURES,
.flags = VHOST_USER_VERSION,
};
- ret = vhost_user_write(dev->vhostfd, &msg, NULL, 0);
[dpdk-dev] [PATCH v4 42/44] net/virtio: move Vhost-vDPA data to its backend
As done earlier for Vhost-user and Vhost-kernel, this
patch moves the Vhost-vDPA specific data to its backend
file.
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost_vdpa.c | 120 +-
.../net/virtio/virtio_user/virtio_user_dev.h | 5 -
2 files changed, 89 insertions(+), 36 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost_vdpa.c
b/drivers/net/virtio/virtio_user/vhost_vdpa.c
index 5e3778b682..e2d6d3504d 100644
--- a/drivers/net/virtio/virtio_user/vhost_vdpa.c
+++ b/drivers/net/virtio/virtio_user/vhost_vdpa.c
@@ -13,6 +13,11 @@
#include "vhost.h"
#include "virtio_user_dev.h"
+struct vhost_vdpa_data {
+ int vhostfd;
+ uint64_t protocol_features;
+};
+
#define VHOST_VDPA_SUPPORTED_BACKEND_FEATURES \
(1ULL << VHOST_BACKEND_F_IOTLB_MSG_V2 | \
1ULL << VHOST_BACKEND_F_IOTLB_BATCH)
@@ -88,27 +93,34 @@ vhost_vdpa_ioctl(int fd, uint64_t request, void *arg)
static int
vhost_vdpa_set_owner(struct virtio_user_dev *dev)
{
- return vhost_vdpa_ioctl(dev->vhostfd, VHOST_SET_OWNER, NULL);
+ struct vhost_vdpa_data *data = dev->backend_data;
+
+ return vhost_vdpa_ioctl(data->vhostfd, VHOST_SET_OWNER, NULL);
}
static int
vhost_vdpa_get_protocol_features(struct virtio_user_dev *dev, uint64_t
*features)
{
- return vhost_vdpa_ioctl(dev->vhostfd, VHOST_GET_BACKEND_FEATURES,
features);
+ struct vhost_vdpa_data *data = dev->backend_data;
+
+ return vhost_vdpa_ioctl(data->vhostfd, VHOST_GET_BACKEND_FEATURES,
features);
}
static int
vhost_vdpa_set_protocol_features(struct virtio_user_dev *dev, uint64_t
features)
{
- return vhost_vdpa_ioctl(dev->vhostfd, VHOST_SET_BACKEND_FEATURES,
&features);
+ struct vhost_vdpa_data *data = dev->backend_data;
+
+ return vhost_vdpa_ioctl(data->vhostfd, VHOST_SET_BACKEND_FEATURES,
&features);
}
static int
vhost_vdpa_get_features(struct virtio_user_dev *dev, uint64_t *features)
{
+ struct vhost_vdpa_data *data = dev->backend_data;
int ret;
- ret = vhost_vdpa_ioctl(dev->vhostfd, VHOST_GET_FEATURES, features);
+ ret = vhost_vdpa_ioctl(data->vhostfd, VHOST_GET_FEATURES, features);
if (ret) {
PMD_DRV_LOG(ERR, "Failed to get features");
return -1;
@@ -118,15 +130,15 @@ vhost_vdpa_get_features(struct virtio_user_dev *dev,
uint64_t *features)
*features &= ~(1ULL << VIRTIO_NET_F_MQ);
/* Negotiated vDPA backend features */
- ret = vhost_vdpa_get_protocol_features(dev, &dev->protocol_features);
+ ret = vhost_vdpa_get_protocol_features(dev, &data->protocol_features);
if (ret < 0) {
PMD_DRV_LOG(ERR, "Failed to get backend features");
return -1;
}
- dev->protocol_features &= VHOST_VDPA_SUPPORTED_BACKEND_FEATURES;
+ data->protocol_features &= VHOST_VDPA_SUPPORTED_BACKEND_FEATURES;
- ret = vhost_vdpa_set_protocol_features(dev, dev->protocol_features);
+ ret = vhost_vdpa_set_protocol_features(dev, data->protocol_features);
if (ret < 0) {
PMD_DRV_LOG(ERR, "Failed to set backend features");
return -1;
@@ -138,21 +150,24 @@ vhost_vdpa_get_features(struct virtio_user_dev *dev,
uint64_t *features)
static int
vhost_vdpa_set_features(struct virtio_user_dev *dev, uint64_t features)
{
+ struct vhost_vdpa_data *data = dev->backend_data;
+
/* WORKAROUND */
features |= 1ULL << VIRTIO_F_IOMMU_PLATFORM;
- return vhost_vdpa_ioctl(dev->vhostfd, VHOST_SET_FEATURES, &features);
+ return vhost_vdpa_ioctl(data->vhostfd, VHOST_SET_FEATURES, &features);
}
static int
vhost_vdpa_iotlb_batch_begin(struct virtio_user_dev *dev)
{
+ struct vhost_vdpa_data *data = dev->backend_data;
struct vhost_msg msg = {};
- if (!(dev->protocol_features & (1ULL << VHOST_BACKEND_F_IOTLB_BATCH)))
+ if (!(data->protocol_features & (1ULL << VHOST_BACKEND_F_IOTLB_BATCH)))
return 0;
- if (!(dev->protocol_features & (1ULL << VHOST_BACKEND_F_IOTLB_MSG_V2)))
{
+ if (!(data->protocol_features & (1ULL <<
VHOST_BACKEND_F_IOTLB_MSG_V2))) {
PMD_DRV_LOG(ERR, "IOTLB_MSG_V2 not supported by the backend.");
return -1;
}
@@ -160,7 +175,7 @@ vhost_vdpa_iotlb_batch_begin(struct virtio_user_dev *dev)
msg.type = VHOST_IOTLB_MSG_V2;
msg.iotlb.type = VHOST_IOTLB_BATCH_BEGIN;
- if (write(dev->vhostfd, &msg, sizeof(msg)) != sizeof(msg)) {
+ if (write(data->vhostfd, &msg, sizeof(msg)) != sizeof(msg)) {
PMD_DRV_LOG(ERR, "Failed to send IOTLB batch begin (%s)",
strerror(errno));
return -1;
@@ -172,12 +187,13 @@ vhost_vdpa_iotlb_batch_begin(struct virtio_user_dev *dev)
static int
vhost_vdpa_iotlb_batch_end(struct virtio_user_dev *de
[dpdk-dev] [PATCH v4 43/44] net/virtio: improve Vhost-user error logging
This patch improves error logging in vhost_user_read,
especially printing errno when recv() fails.
Suggested-by: Adrian Moreno
Signed-off-by: Maxime Coquelin
Reviewed-by: Chenbo Xia
---
drivers/net/virtio/virtio_user/vhost_user.c | 31 -
1 file changed, 18 insertions(+), 13 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/vhost_user.c
b/drivers/net/virtio/virtio_user/vhost_user.c
index 088aae3aa3..ec2c53c8fb 100644
--- a/drivers/net/virtio/virtio_user/vhost_user.c
+++ b/drivers/net/virtio/virtio_user/vhost_user.c
@@ -148,38 +148,43 @@ vhost_user_read(int fd, struct vhost_user_msg *msg)
int ret, sz_hdr = VHOST_USER_HDR_SIZE, sz_payload;
ret = recv(fd, (void *)msg, sz_hdr, 0);
- if (ret < sz_hdr) {
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to recv msg header: %s",
strerror(errno));
+ return -1;
+ } else if (ret < sz_hdr) {
PMD_DRV_LOG(ERR, "Failed to recv msg hdr: %d instead of %d.",
ret, sz_hdr);
- goto fail;
+ return -1;
}
/* validate msg flags */
if (msg->flags != (valid_flags)) {
- PMD_DRV_LOG(ERR, "Failed to recv msg: flags %x instead of %x.",
+ PMD_DRV_LOG(ERR, "Failed to recv msg: flags 0x%x instead of
0x%x.",
msg->flags, valid_flags);
- goto fail;
+ return -1;
}
sz_payload = msg->size;
- if ((size_t)sz_payload > sizeof(msg->payload))
- goto fail;
+ if ((size_t)sz_payload > sizeof(msg->payload)) {
+ PMD_DRV_LOG(ERR, "Payload size overflow, header says %d but max
%zu\n",
+ sz_payload, sizeof(msg->payload));
+ return -1;
+ }
if (sz_payload) {
ret = recv(fd, (void *)((char *)msg + sz_hdr), sz_payload, 0);
- if (ret < sz_payload) {
- PMD_DRV_LOG(ERR,
- "Failed to recv msg payload: %d instead of %d.",
+ if (ret < 0) {
+ PMD_DRV_LOG(ERR, "Failed to recv msg payload: %s",
strerror(errno));
+ return -1;
+ } else if (ret < sz_payload) {
+ PMD_DRV_LOG(ERR, "Failed to recv msg payload: %d
instead of %u.",
ret, msg->size);
- goto fail;
+ return -1;
}
}
return 0;
-
-fail:
- return -1;
}
static int
--
2.29.2
[dpdk-dev] [PATCH v4 44/44] net/virtio: handle Virtio-user setup failure properly
This patch fixes virtio_user_dev_setup() error path,
by cleaning all resources it allocates. It introduces
virtio_user_dev_uninit_notify() that cleans all open
FDs. It implies assigning all FDs to -1 at init time.
With these changes done, virtio_user_dev_init_notify()
can be simplified.
Suggested-by: Adrian Moreno
Signed-off-by: Maxime Coquelin
---
.../net/virtio/virtio_user/virtio_user_dev.c | 70 +--
1 file changed, 47 insertions(+), 23 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index a1e32158bb..2998473622 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -283,13 +283,7 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
int callfd;
int kickfd;
- for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
- if (i >= dev->max_queue_pairs * 2) {
- dev->kickfds[i] = -1;
- dev->callfds[i] = -1;
- continue;
- }
-
+ for (i = 0; i < dev->max_queue_pairs * 2; i++) {
/* May use invalid flag, but some backend uses kickfd and
* callfd as criteria to judge if dev is alive. so finally we
* use real event_fd.
@@ -297,28 +291,49 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
callfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
if (callfd < 0) {
PMD_DRV_LOG(ERR, "(%s) callfd error, %s", dev->path,
strerror(errno));
- break;
+ goto err;
}
kickfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
if (kickfd < 0) {
close(callfd);
PMD_DRV_LOG(ERR, "(%s) kickfd error, %s", dev->path,
strerror(errno));
- break;
+ goto err;
}
dev->callfds[i] = callfd;
dev->kickfds[i] = kickfd;
}
- if (i < VIRTIO_MAX_VIRTQUEUES) {
- for (j = 0; j < i; ++j) {
- close(dev->callfds[j]);
+ return 0;
+err:
+ for (j = 0; j < i; j++) {
+ if (dev->kickfds[j] >= 0) {
close(dev->kickfds[j]);
+ dev->kickfds[j] = -1;
+ }
+ if (dev->callfds[j] >= 0) {
+ close(dev->callfds[j]);
+ dev->callfds[j] = -1;
}
-
- return -1;
}
- return 0;
+ return -1;
+}
+
+static void
+virtio_user_dev_uninit_notify(struct virtio_user_dev *dev)
+{
+ uint32_t i;
+
+ for (i = 0; i < dev->max_queue_pairs; ++i) {
+ if (dev->kickfds[i] >= 0) {
+ close(dev->kickfds[i]);
+ dev->kickfds[i] = -1;
+ }
+ if (dev->callfds[i] >= 0) {
+ close(dev->callfds[i]);
+ dev->callfds[i] = -1;
+ }
+ }
}
static int
@@ -427,15 +442,22 @@ virtio_user_dev_setup(struct virtio_user_dev *dev)
if (virtio_user_dev_init_notify(dev) < 0) {
PMD_INIT_LOG(ERR, "(%s) Failed to init notifiers\n", dev->path);
- return -1;
+ goto destroy;
}
if (virtio_user_fill_intr_handle(dev) < 0) {
PMD_INIT_LOG(ERR, "(%s) Failed to init interrupt handler\n",
dev->path);
- return -1;
+ goto uninit;
}
return 0;
+
+uninit:
+ virtio_user_dev_uninit_notify(dev);
+destroy:
+ dev->ops->destroy(dev);
+
+ return -1;
}
/* Use below macro to filter features from vhost backend */
@@ -466,9 +488,16 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
*path, int queues,
enum virtio_user_backend_type backend_type)
{
uint64_t backend_features;
+ int i;
pthread_mutex_init(&dev->mutex, NULL);
strlcpy(dev->path, path, PATH_MAX);
+
+ for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; i++) {
+ dev->kickfds[i] = -1;
+ dev->callfds[i] = -1;
+ }
+
dev->started = 0;
dev->max_queue_pairs = queues;
dev->queue_pairs = 1; /* mq disabled by default */
@@ -565,16 +594,11 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
*path, int queues,
void
virtio_user_dev_uninit(struct virtio_user_dev *dev)
{
- uint32_t i;
-
virtio_user_stop_device(dev);
rte_mem_event_callback_unregister(VIRTIO_USER_MEM_EVENT_CLB_NAME, dev);
- for (i = 0; i < dev->max_queue_pairs * 2; ++i) {
- close(dev->callfds[i]);
- close(dev->kickfds[i]);
- }
+ virtio_user_dev_uninit_notify(dev);
free(dev->ifname);
--
2.
Re: [dpdk-dev] [dpdk-stable] [PATCH] vdpa/mlx5: fix configuration mutex cleanup
On 1/21/21 9:13 PM, Matan Azrad wrote:
>
>
> From: Maxime Coquelin
>> On 1/14/21 4:23 PM, Matan Azrad wrote:
>>>
>>>
>>> From: Maxime Coquelin
On 1/14/21 2:09 PM, Matan Azrad wrote:
>
>
> From: Maxime Coquelin
>> Hi Matan,
>>
>> On 1/14/21 12:49 PM, Matan Azrad wrote:
>>> Hi Maxime and David
>>>
>>> Thank you for Review.
>>>
>>> From: David Marchand
On Fri, Jan 8, 2021 at 9:48 AM David Marchand
wrote:
>> I wonder if it would be possible and cleaner to disable
>> cancellation on the thread while the mutex is held?
>>>
>>> Yes, we can cause thread to return by some global variable sync.
>>> It is the same logic.
>>
>> No, that was not my suggestion. My suggestion is to block the
>> thread cancellation while in the critical section, using
>> pthread_setcancelstate().
>
> Yes, Generally it is better to let the thread control his
> cancellation, either
cancel itself or enabling\disabling cancellations.
>
> I don't see a reason to wait for the thread in current logic - the
> critical section
is not important to be completed here.
The reason I see is there are quite a few things done in this
critical section. And if tomorrow someone add new things in it, he
may not know the thread can be cancelled at any time, which could cause
>> hard to debug issues.
>>>
>>> As I said, here it is not needed, this thread designed just to cause guest
>> notifications.
>>>
>>> The optional future developer mistake can be done also outside the critical
>> section in in any other place - we cannot protect it.
>>>
>>> The design choice is to close the thread fast.
>>
>> But why is it so urgent that it cannot been stopped cleanly?
>> I don't think it would add seconds delay by doing it in a clean way.
>
> We have system calls there per queue.
> No need this optional delay just because of mutex cleaning.
OK, up to you...
And what about the timer lock?
>
>
>> Thanks,
>> Maxime
>>
> We just want to close the thread and to clean the mutex.
>
> +1
IEEE Std 1003.1-2001/Cor 2-2004, item XBD/TC2/D6/26 is applied,
adding pthread_t to the list of types that are not required to be
arithmetic types, thus allowing pthread_t to be defined as a structure.
It would be better to leave pthread_t alone and not interpret it:
if (priv->timer_tid) {
pthread_cancel(priv->timer_tid);
pthread_join(priv->timer_tid, &status); }
priv->timer_tid = 0;
>>>
>>>
>>> I'm not sure why you think it is better in this specific case.
>>> The cancellation will close the thread in faster way, no need to
>>> wait for the
>> thread to close itself.
>>>
>>>
--
David Marchand
>>>
>
>>>
>
Re: [dpdk-dev] [PATCH v5 3/3] doc: add clang to aarch64 cross build guide
21/01/2021 16:51, Juraj Linkeš: > Reorganize and update the aarch64 cross guide with clang cross > compilation. Update the GNU toolchain version which clang also uses. > Reorganize into common part, GNU part and clang part. > > Signed-off-by: Juraj Linkeš > Acked-by: Ruifeng Wang I fix some spacing, quoting and avoid "you" form, while applying.
Re: [dpdk-dev] [PATCH v3 0/2] add mbuf fast free offload support
From: Viacheslav Ovsiienko > This patch adds support of the mbuf fast free offload to the transmit > datapath. > This offload allows to free the mbufs on transmit completion in the most > efficient way. It requires the all mbufs were allocated from the same pool, > have the reference counter value as 1, and have no any externally attached > buffers. > > The patchset is split in two parts, the first one is overall send loop > optimization > and can be ported back to stable release, and also is the preparation step > before introducing the fast free offload. The second part provides the code > for > the feature. > > Signed-off-by: Viacheslav Ovsiienko > > --- > v1: > https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fpatches.d > pdk.org%2Fpatch%2F85482%2F&data=04%7C01%7Cmatan%40nvidia.com > %7Cd873758f2c354a1af0aa08d8bef8ddd9%7C43083d15727340c1b7db39efd9c > cc17a%7C0%7C0%7C637469323804842896%7CUnknown%7CTWFpbGZsb3d8eyJ > WIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C > 1000&sdata=BqgJn08GepCw1gpCfG7rKmqZcp1ht02HG%2Bi%2FlEXKjjc%3 > D&reserved=0 > v2: release notes, PMD features and documentation update > v3: typo fixes, documentation update > > Viacheslav Ovsiienko (2): > net/mlx5: optimize inline mbuf freeing > net/mlx5: add mbuf fast free offload support > > doc/guides/nics/features/mlx5.ini | 1 + > doc/guides/nics/mlx5.rst | 12 ++ > doc/guides/rel_notes/release_21_02.rst | 2 +- > drivers/net/mlx5/mlx5_rxtx.c | 67 +--- > -- > drivers/net/mlx5/mlx5_rxtx.h | 2 + > drivers/net/mlx5/mlx5_txq.c| 6 +++ > 6 files changed, 80 insertions(+), 10 deletions(-) > > -- > 1.8.3.1 Series-acked-by: Matan Azrad
Re: [dpdk-dev] [PATCH v5 3/3] PCI: don't use vfio ioctl call to access PIO resource
On 1/22/21 8:25 AM, 谢华伟(此时此刻） wrote: > > On 2021/1/21 23:38, Maxime Coquelin wrote: >>> Do you mean we apply or abandon patch 3? I am both OK. The first >>> priority to me is to enable MMIO bar support. >> OK, so yes, I think we should abandon patch 2 and patch 3. >> For patch 1, it looks valid to me, but I'll let Ferruh decide. >> >> For your device, if my understanding is correct, what we need to do is >> to support MMIO for legacy devices. Correct? > yes. >> If so, the change should be in virtio_pci.c. In vtpci_init(), after >> modern detection has failed, we should check the the BAR is PIO or MMIO >> based on the flag. the result can be saved in struct virtio_pci_dev. >> >> >> We would introduce new wrappers like vtpci_legacy_read, >> vtpci_legacy_write that would either call rte_pci_ioport_read, >> rte_pci_ioport_read in case of PIO, or rte_read32, rte_write32 in case >> of MMIO. > > There are two choices. > > 1, apply patch 2. > > IO/MMIO port are mapped and accessed using the same API. Kernel is > doing in the same way like the following. > > io_addr = pci_iomap > > get PIO directly or ioremap > > iowrite16/32(val, io_addr + offset) > > I think applying patch 2 is a correct choice. It is a fix. Driver had > better not know if bar is PIO or MMIO. ioport in ioport_xx API means > IO, not PIO. > > Btw, it only affects virtio PMD, not that intrusive. > > 2, virtio specific change to enable MMIO support. > > Comparing with choice 1, i feels it is not that clean and pretty. OK, that makes sense. I am OK with keeping patch 2, but would like Ferruh's ACK. Could you please post v6? Thanks, Maxime >> >> It is not too late for this release, as the change will not be that >> intrusive. But if you prepare such patch, please base it on top of my >> virtio rework series; To make it easier to you, I added it to the dpdk- >> next-virtio tree: >> https://git.dpdk.org/next/dpdk-next-virtio/log/?h=virtio_pmd_rework_v2 >> >> Thanks, >> Maxime >> >
Re: [dpdk-dev] [dpdk-stable] [PATCH] vdpa/mlx5: fix configuration mutex cleanup
From: Maxime Coquelin
> > From: Maxime Coquelin
> >> On 1/14/21 4:23 PM, Matan Azrad wrote:
> >>>
> >>>
> >>> From: Maxime Coquelin
> On 1/14/21 2:09 PM, Matan Azrad wrote:
> >
> >
> > From: Maxime Coquelin
> >> Hi Matan,
> >>
> >> On 1/14/21 12:49 PM, Matan Azrad wrote:
> >>> Hi Maxime and David
> >>>
> >>> Thank you for Review.
> >>>
> >>> From: David Marchand
> On Fri, Jan 8, 2021 at 9:48 AM David Marchand
> wrote:
> >> I wonder if it would be possible and cleaner to disable
> >> cancellation on the thread while the mutex is held?
> >>>
> >>> Yes, we can cause thread to return by some global variable sync.
> >>> It is the same logic.
> >>
> >> No, that was not my suggestion. My suggestion is to block the
> >> thread cancellation while in the critical section, using
> >> pthread_setcancelstate().
> >
> > Yes, Generally it is better to let the thread control his
> > cancellation, either
> cancel itself or enabling\disabling cancellations.
> >
> > I don't see a reason to wait for the thread in current logic - the
> > critical section
> is not important to be completed here.
>
> The reason I see is there are quite a few things done in this
> critical section. And if tomorrow someone add new things in it, he
> may not know the thread can be cancelled at any time, which could
> cause
> >> hard to debug issues.
> >>>
> >>> As I said, here it is not needed, this thread designed just to cause
> >>> guest
> >> notifications.
> >>>
> >>> The optional future developer mistake can be done also outside the
> >>> critical
> >> section in in any other place - we cannot protect it.
> >>>
> >>> The design choice is to close the thread fast.
> >>
> >> But why is it so urgent that it cannot been stopped cleanly?
> >> I don't think it would add seconds delay by doing it in a clean way.
> >
> > We have system calls there per queue.
> > No need this optional delay just because of mutex cleaning.
>
> OK, up to you...
>
> And what about the timer lock?
Existing code initiates it before reusing...
Thanks.
>
> >
> >
> >> Thanks,
> >> Maxime
> >>
> > We just want to close the thread and to clean the mutex.
> >
> > +1
>
> IEEE Std 1003.1-2001/Cor 2-2004, item XBD/TC2/D6/26 is applied,
> adding pthread_t to the list of types that are not required to
> be arithmetic types, thus allowing pthread_t to be defined as a
> structure.
>
> It would be better to leave pthread_t alone and not interpret it:
>
> if (priv->timer_tid) {
> pthread_cancel(priv->timer_tid);
> pthread_join(priv->timer_tid, &status); }
> priv->timer_tid = 0;
> >>>
> >>>
> >>> I'm not sure why you think it is better in this specific case.
> >>> The cancellation will close the thread in faster way, no need to
> >>> wait for the
> >> thread to close itself.
> >>>
> >>>
>
> --
> David Marchand
> >>>
> >
> >>>
> >
Re: [dpdk-dev] [PATCH v1 0/2] fix bugs of app eventdev
On Fri, Jan 22, 2021 at 10:49 AM Feifei Wang wrote: > > Fix bugs of app eventdev > > Feifei Wang (2): > app/eventdev: adjust event count order for pipeline test > app/eventdev: remove redundant enqueue in burst Tx Acked-by: Jerin Jacob Applied to dpdk-next-net-eventdev/for-main. Thanks > > app/test-eventdev/test_pipeline_queue.c | 17 - > 1 file changed, 8 insertions(+), 9 deletions(-) > > -- > 2.25.1 >
Re: [dpdk-dev] [PATCH v5] app/testpmd: fix setting maximum packet length
On 1/26/2021 3:45 AM, Lance Richardson wrote:
On Mon, Jan 25, 2021 at 7:44 PM Ferruh Yigit wrote:
+ if (rx_offloads != port->dev_conf.rxmode.offloads) {
+ uint16_t qid;
+
+ port->dev_conf.rxmode.offloads = rx_offloads;
+
+ /* Apply JUMBO_FRAME offload configuration to Rx queue(s) */
+ for (qid = 0; qid < port->dev_info.nb_rx_queues; qid++) {
+ if (on)
+ port->rx_conf[qid].offloads |=
DEV_RX_OFFLOAD_JUMBO_FRAME;
+ else
+ port->rx_conf[qid].offloads &=
~DEV_RX_OFFLOAD_JUMBO_FRAME;
+ }
Is it correct to set per-queue offloads that aren't advertised by the PMD
as supported in rx_queue_offload_capa?
'port->rx_conf[]' is testpmd struct, and 'port->dev_conf.rxmode.offloads' values
are reflected to 'port->rx_conf[].offloads' for all queues.
We should set the offload in 'port->rx_conf[].offloads' if it is set in
'port->dev_conf.rxmode.offloads'.
If a port has capability for 'JUMBO_FRAME', 'port->rx_conf[].offloads' can have
it. And the port level capability is already checked above.
I'm still not 100% clear about the per-queue offload question.
With this patch, and jumbo max packet size configured (on the command
line in this case), I see:
testpmd> show port 0 rx_offload configuration
Rx Offloading Configuration of port 0 :
Port : JUMBO_FRAME
Queue[ 0] : JUMBO_FRAME
testpmd> show port 0 rx_offload capabilities
Rx Offloading Capabilities of port 0 :
Per Queue :
Per Port : VLAN_STRIP IPV4_CKSUM UDP_CKSUM TCP_CKSUM TCP_LRO
OUTER_IPV4_CKSUM VLAN_FILTER VLAN_EXTEND JUMBO_FRAME SCATTER TIMESTAMP
KEEP_CRC OUTER_UDP_CKSUM RSS_HASH
The port level offload is applied to all queues on the port, testpmd config
structure reflects this logic in implementation.
If Rx offload X is set for a port, it is set for all Rx queue offloads, this is
not new behavior and not related to this patch.
In the ethdev, lets assume X & Y are port level offloads,
after X, Y are set via 'rte_eth_dev_configure()'
if user calls 'rte_eth_rx_queue_setup()' with X & Y offload, this is a valid
call and API will return success, since those offloads already enabled in port
level means they are enabled for all queues.
Because of above ethdev behavior, testpmd keeps all enabled port level offload
in the queue level offload too, and display them as enabled offloads for the queue.
To request a queue specific offload, it is added to specific queue's config
before calling queue setup. Lets say that queue specific offload is Z, after
setup testpmd config struct will show that specific queue has X, Y & Z offloads.
I hope it is more clear now.
Yet if I configure a jumbo MTU starting with standard max packet size,
jumbo is only enabled at the port level:
testpmd> port config mtu 0 9000
testpmd> port start all
testpmd> show port 0 rx_offload configuration
Rx Offloading Configuration of port 0 :
Port : JUMBO_FRAME
Queue[ 0] :
It still seems odd for a per-queue offload to be enabled on a PMD that
doesn't support per-queue receive offloads.
"port config mtu" should take queue offloads into account, it looks wrong right
now.
Re: [dpdk-dev] [PATCH v3 0/4] add checking of header includes
On Mon, Jan 25, 2021 at 04:51:19PM +0100, David Marchand wrote: > On Mon, Jan 25, 2021 at 3:11 PM Bruce Richardson > wrote: > > > > As a general principle, each header file should include any other > > headers it needs to provide data type definitions or macros. For > > example, any header using the uintX_t types in structures or function > > prototypes should include "stdint.h" to provide those type definitions. > > > > In practice, while many, but not all, headers in DPDK did include all > > necessary headers, it was never actually checked that each header could > > be included in a C file and compiled without having any compiler errors > > about missing definitions. The script "check-includes.sh" could be used > > for this job, but it was not called out in the documentation, so many > > contributors may not have been aware of it's existance. It also was > > difficult to run from a source-code directory, as the script did not > > automatically allow finding of headers from one DPDK library directory > > to another [this was probably based on running it on a build created by > > the "make" build system, where all headers were in a single directory]. > > To attempt to have a build-system integrated replacement, this patchset > > adds a "chkincs" app in the buildtools directory to verify this on an > > ongoing basis. > > > > This chkincs app does nothing when run, and is not installed as part of > > a DPDK "ninja install", it's for build-time checking only. Its source > > code consists of one C file per public DPDK header, where that C file > > contains nothing except an include for that header. Therefore, if any > > header is added to the lib folder which fails to compile when included > > alone, the build of chkincs will fail with a suitable error message. > > Since this compile checking is not needed on most builds of DPDK, the > > building of chkincs is disabled by default, but can be enabled by the > > "test_includes" meson option. To catch errors with patch submissions, > > the final patch of this series enables it for a single build in > > test-meson-builds script. > > > > Future work could involve doing similar checks on headers for C++ > > compatibility, which was something done by the check-includes.sh script > > but which is missing here.. > > > > V3: > > * Shrunk patchset as most header fixes already applied > > * Moved chkincs from "apps" to the "buildtools" directory, which is a > > better location for something not for installation for end-user use. > > * Added patch to drop check-includes script. > > > > V2: > > * Add maintainers file entry for new app > > * Drop patch for c11 ring header > > * Use build variable "headers_no_chkincs" for tracking exceptions > > > > Bruce Richardson (4): > > eal: add missing include to mcslock > > build: separate out headers for include checking > > buildtools/chkincs: add app to verify header includes > > devtools: remove check-includes script > > > > MAINTAINERS | 5 +- > > buildtools/chkincs/gen_c_file_for_header.py | 12 + > > buildtools/chkincs/main.c| 4 + > > buildtools/chkincs/meson.build | 40 +++ > > devtools/check-includes.sh | 259 --- > > devtools/test-meson-builds.sh| 2 +- > > doc/guides/contributing/coding_style.rst | 12 + > > lib/librte_eal/include/generic/rte_mcslock.h | 1 + > > lib/librte_eal/include/meson.build | 2 +- > > lib/librte_eal/x86/include/meson.build | 14 +- > > lib/librte_ethdev/meson.build| 4 +- > > lib/librte_hash/meson.build | 4 +- > > lib/librte_ipsec/meson.build | 3 +- > > lib/librte_lpm/meson.build | 2 +- > > lib/librte_regexdev/meson.build | 2 +- > > lib/librte_ring/meson.build | 4 +- > > lib/librte_stack/meson.build | 4 +- > > lib/librte_table/meson.build | 7 +- > > lib/meson.build | 3 + > > meson.build | 6 + > > meson_options.txt| 2 + > > 21 files changed, 112 insertions(+), 280 deletions(-) > > create mode 100755 buildtools/chkincs/gen_c_file_for_header.py > > create mode 100644 buildtools/chkincs/main.c > > create mode 100644 buildtools/chkincs/meson.build > > delete mode 100755 devtools/check-includes.sh > > - clang is not happy when enabling the check: > $ meson configure $HOME/builds/build-clang-static -Dcheck_includes=true > $ devtools/test-meson-builds.sh > ... > [362/464] Compiling C object > buildtools/chkincs/chkincs.p/meson-generated_rte_ethdev_vdev.c.o > FAILED: buildtools/chkincs/chkincs.p/meson-generated_rte_ethdev_vdev.c.o > clang -Ibuildtools/chkincs/chkincs.p -Ibuildtools/chkincs > -I../../dpdk/buildtools/chkincs -Idrivers/bus/pci > -I../../dpdk/drivers/bus/pci -Idrivers/bus/vdev > -I
Re: [dpdk-dev] [PATCH v2 3/3] ring: refactor ring library
> For legacy modes, rename ring_generic/c11 to ring_generic/c11_pvt. > Furthermore, add new file ring_elem_pvt.h which includes ring_do_eq/deq > and ring element copy/delete APIs. > > For other modes, rename xx_c11_mem to xx_elem_pvt. Move all private APIs > into these new header files. > > Finally, the external APIs and internal APIs will be separated from each > other. This can remind users not to use internal APIs and make ring > library easier to maintain. > > Suggested-by: Honnappa Nagarahalli > Signed-off-by: Feifei Wang > Reviewed-by: Honnappa Nagarahalli > Reviewed-by: Ruifeng Wang > --- Acked-by: Konstantin Ananyev > -- > 2.25.1
Re: [dpdk-dev] [PATCH v5 0/3] aarch64 clang cross compilation
> Juraj Linkeš (3): > build: add aarch64 clang to meson cross-compile > ci: add aarch64 clang cross-compilation builds > doc: add clang to aarch64 cross build guide Applied with doc fixups, thanks
Re: [dpdk-dev] [PATCH v4 02/44] bus/vdev: add driver IOVA VA mode requirement
> -Original Message-
> From: Maxime Coquelin
> Sent: Tuesday, January 26, 2021 6:16 PM
> To: dev@dpdk.org; Xia, Chenbo ; olivier.m...@6wind.com;
> amore...@redhat.com; david.march...@redhat.com
> Cc: Maxime Coquelin
> Subject: [PATCH v4 02/44] bus/vdev: add driver IOVA VA mode requirement
>
> This patch adds driver flag in vdev bus driver so that
> vdev drivers can require VA IOVA mode to be used, which
> for example the case of Virtio-user PMD.
>
> The patch implements the .get_iommu_class() callback, that
> is called before devices probing to determine the IOVA mode
> to be used, and adds a check right before the device is
> probed to ensure compatible IOVA mode has been selected.
>
> It also adds a ABI exception rule to accommodate with an
> update on the driver registration API
>
> Signed-off-by: Maxime Coquelin
> Signed-off-by: David Marchand
> ---
> devtools/libabigail.abignore| 2 ++
> drivers/bus/vdev/rte_bus_vdev.h | 4
> drivers/bus/vdev/vdev.c | 29 +
> 3 files changed, 35 insertions(+)
>
> diff --git a/devtools/libabigail.abignore b/devtools/libabigail.abignore
> index 1dc84fa74b..170304c876 100644
> --- a/devtools/libabigail.abignore
> +++ b/devtools/libabigail.abignore
> @@ -11,6 +11,8 @@
> ; Explicit ignore for driver-only ABI
> [suppress_type]
> name = eth_dev_ops
> +[suppress_function]
> +name_regexp = rte_vdev_(|un)register
>
> ; Ignore fields inserted in cacheline boundary of rte_cryptodev
> [suppress_type]
> diff --git a/drivers/bus/vdev/rte_bus_vdev.h b/drivers/bus/vdev/rte_bus_vdev.h
> index f99a41f825..fc315d10fa 100644
> --- a/drivers/bus/vdev/rte_bus_vdev.h
> +++ b/drivers/bus/vdev/rte_bus_vdev.h
> @@ -113,8 +113,12 @@ struct rte_vdev_driver {
> rte_vdev_remove_t *remove; /**< Virtual device remove function. */
> rte_vdev_dma_map_t *dma_map; /**< Virtual device DMA map function.
> */
> rte_vdev_dma_unmap_t *dma_unmap; /**< Virtual device DMA unmap function.
> */
> + uint32_t drv_flags; /**< Flags RTE_VDEV_DRV_*. */
> };
>
> +/** Device driver needs IOVA as VA and cannot work with IOVA as PA */
> +#define RTE_VDEV_DRV_NEED_IOVA_AS_VA 0x0001
> +
> /**
> * Register a virtual device driver.
> *
> diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
> index acfd78828f..9a673347ae 100644
> --- a/drivers/bus/vdev/vdev.c
> +++ b/drivers/bus/vdev/vdev.c
> @@ -189,6 +189,7 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
> {
> const char *name;
> struct rte_vdev_driver *driver;
> + enum rte_iova_mode iova_mode;
> int ret;
>
> if (rte_dev_is_probed(&dev->device))
> @@ -199,6 +200,14 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
>
> if (vdev_parse(name, &driver))
> return -1;
> +
> + iova_mode = rte_eal_iova_mode();
> + if ((driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA) && (iova_mode ==
> RTE_IOVA_PA)) {
> + VDEV_LOG(ERR, "%s requires VA IOVA mode but current mode is PA,
> not initializing",
> + name);
> + return -1;
> + }
> +
> ret = driver->probe(dev);
> if (ret == 0)
> dev->device.driver = &driver->driver;
> @@ -594,6 +603,25 @@ vdev_unplug(struct rte_device *dev)
> return rte_vdev_uninit(dev->name);
> }
>
> +static enum rte_iova_mode
> +vdev_get_iommu_class(void)
> +{
> + const char *name;
> + struct rte_vdev_device *dev;
> + struct rte_vdev_driver *driver;
> +
> + TAILQ_FOREACH(dev, &vdev_device_list, next) {
> + name = rte_vdev_device_name(dev);
> + if (vdev_parse(name, &driver))
> + continue;
> +
> + if (driver->drv_flags & RTE_VDEV_DRV_NEED_IOVA_AS_VA)
> + return RTE_IOVA_VA;
> + }
> +
> + return RTE_IOVA_DC;
> +}
> +
> static struct rte_bus rte_vdev_bus = {
> .scan = vdev_scan,
> .probe = vdev_probe,
> @@ -603,6 +631,7 @@ static struct rte_bus rte_vdev_bus = {
> .parse = vdev_parse,
> .dma_map = vdev_dma_map,
> .dma_unmap = vdev_dma_unmap,
> + .get_iommu_class = vdev_get_iommu_class,
> .dev_iterate = rte_vdev_dev_iterate,
> };
>
> --
> 2.29.2
Reviewed-by: Chenbo Xia
Re: [dpdk-dev] [PATCH v1] devtools: update abi ignore for cryptodev
20/01/2021 15:25, Ray Kinsella:
> Update the ignore entry for crytodev to use named fields instead of
> bit positions.
>
> Fixes: 1c3ffb9559
>
> Signed-off-by: Ray Kinsella
> ---
> --- a/devtools/libabigail.abignore
> +++ b/devtools/libabigail.abignore
> @@ -15,4 +15,4 @@
> ; Ignore fields inserted in cacheline boundary of rte_cryptodev
> [suppress_type]
> name = rte_cryptodev
> -has_data_member_inserted_between = {0, 1023}
> +has_data_member_inserted_between = {offset_after(attached), end}
Adding a bit more explanations in the commit message:
It is allowing changes between the last field (attached) in ABI 21.0,
and the end of the padded struct in ABI 21.
Fixes: 1c3ffb95595e ("cryptodev: add enqueue and dequeue callbacks")
Acked-by: Thomas Monjalon
Applied, thanks.
Re: [dpdk-dev] [PATCH v4 44/44] net/virtio: handle Virtio-user setup failure properly
Hi Maxime,
> -Original Message-
> From: Maxime Coquelin
> Sent: Tuesday, January 26, 2021 6:17 PM
> To: dev@dpdk.org; Xia, Chenbo ; olivier.m...@6wind.com;
> amore...@redhat.com; david.march...@redhat.com
> Cc: Maxime Coquelin
> Subject: [PATCH v4 44/44] net/virtio: handle Virtio-user setup failure
> properly
>
> This patch fixes virtio_user_dev_setup() error path,
> by cleaning all resources it allocates. It introduces
> virtio_user_dev_uninit_notify() that cleans all open
> FDs. It implies assigning all FDs to -1 at init time.
>
> With these changes done, virtio_user_dev_init_notify()
> can be simplified.
>
> Suggested-by: Adrian Moreno
> Signed-off-by: Maxime Coquelin
> ---
> .../net/virtio/virtio_user/virtio_user_dev.c | 70 +--
> 1 file changed, 47 insertions(+), 23 deletions(-)
>
> diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
> b/drivers/net/virtio/virtio_user/virtio_user_dev.c
> index a1e32158bb..2998473622 100644
> --- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
> +++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
> @@ -283,13 +283,7 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
> int callfd;
> int kickfd;
>
> - for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
> - if (i >= dev->max_queue_pairs * 2) {
> - dev->kickfds[i] = -1;
> - dev->callfds[i] = -1;
> - continue;
> - }
> -
> + for (i = 0; i < dev->max_queue_pairs * 2; i++) {
> /* May use invalid flag, but some backend uses kickfd and
>* callfd as criteria to judge if dev is alive. so finally we
>* use real event_fd.
> @@ -297,28 +291,49 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
> callfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
> if (callfd < 0) {
> PMD_DRV_LOG(ERR, "(%s) callfd error, %s", dev->path,
> strerror(errno));
> - break;
> + goto err;
> }
> kickfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
> if (kickfd < 0) {
> close(callfd);
> PMD_DRV_LOG(ERR, "(%s) kickfd error, %s", dev->path,
> strerror(errno));
> - break;
> + goto err;
> }
> dev->callfds[i] = callfd;
> dev->kickfds[i] = kickfd;
> }
>
> - if (i < VIRTIO_MAX_VIRTQUEUES) {
> - for (j = 0; j < i; ++j) {
> - close(dev->callfds[j]);
> + return 0;
> +err:
> + for (j = 0; j < i; j++) {
> + if (dev->kickfds[j] >= 0) {
> close(dev->kickfds[j]);
> + dev->kickfds[j] = -1;
> + }
> + if (dev->callfds[j] >= 0) {
> + close(dev->callfds[j]);
> + dev->callfds[j] = -1;
> }
> -
> - return -1;
> }
>
> - return 0;
> + return -1;
> +}
> +
> +static void
> +virtio_user_dev_uninit_notify(struct virtio_user_dev *dev)
> +{
> + uint32_t i;
> +
> + for (i = 0; i < dev->max_queue_pairs; ++i) {
Should be 'dev->max_queue_pairs * 2'?
I believe you will fix this in the final version before applying them
in the tree.
So with above fixed:
Reviewed-by: Chenbo Xia
> + if (dev->kickfds[i] >= 0) {
> + close(dev->kickfds[i]);
> + dev->kickfds[i] = -1;
> + }
> + if (dev->callfds[i] >= 0) {
> + close(dev->callfds[i]);
> + dev->callfds[i] = -1;
> + }
> + }
> }
>
> static int
> @@ -427,15 +442,22 @@ virtio_user_dev_setup(struct virtio_user_dev *dev)
>
> if (virtio_user_dev_init_notify(dev) < 0) {
> PMD_INIT_LOG(ERR, "(%s) Failed to init notifiers\n", dev->path);
> - return -1;
> + goto destroy;
> }
>
> if (virtio_user_fill_intr_handle(dev) < 0) {
> PMD_INIT_LOG(ERR, "(%s) Failed to init interrupt handler\n",
> dev-
> >path);
> - return -1;
> + goto uninit;
> }
>
> return 0;
> +
> +uninit:
> + virtio_user_dev_uninit_notify(dev);
> +destroy:
> + dev->ops->destroy(dev);
> +
> + return -1;
> }
>
> /* Use below macro to filter features from vhost backend */
> @@ -466,9 +488,16 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
> *path, int queues,
>enum virtio_user_backend_type backend_type)
> {
> uint64_t backend_features;
> + int i;
>
> pthread_mutex_init(&dev->mutex, NULL);
> strlcpy(dev->path, path, PATH_MAX);
> +
> + for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; i++) {
> + dev->kickfds[i] = -1;
> + dev->callfds[i] = -1;
> + }
> +
> dev->started = 0;
> dev->max_queue_pairs =
Re: [dpdk-dev] [PATCH v5 3/3] PCI: don't use vfio ioctl call to access PIO resource
On 2021/1/22 15:25, chris wrote: On 2021/1/21 23:38, Maxime Coquelin wrote: Do you mean we apply or abandon patch 3? I am both OK. The first priority to me is to enable MMIO bar support. OK, so yes, I think we should abandon patch 2 and patch 3. For patch 1, it looks valid to me, but I'll let Ferruh decide. For your device, if my understanding is correct, what we need to do is to support MMIO for legacy devices. Correct? yes. If so, the change should be in virtio_pci.c. In vtpci_init(), after modern detection has failed, we should check the the BAR is PIO or MMIO based on the flag. the result can be saved in struct virtio_pci_dev. We would introduce new wrappers like vtpci_legacy_read, vtpci_legacy_write that would either call rte_pci_ioport_read, rte_pci_ioport_read in case of PIO, or rte_read32, rte_write32 in case of MMIO. There are two choices. 1, apply patch 2. IO/MMIO port are mapped and accessed using the same API. Kernel is doing in the same way like the following. io_addr = pci_iomap get PIO directly or ioremap iowrite16/32(val, io_addr + offset) I think applying patch 2 is a correct choice. It is a fix. Driver had better not know if bar is PIO or MMIO. ioport in ioport_xx API means IO, not PIO. Btw, it only affects virtio PMD, not that intrusive. 2, virtio specific change to enable MMIO support. Comparing with choice 1, i feels it is not that clean and pretty. It is not too late for this release, as the change will not be that intrusive. But if you prepare such patch, please base it on top of my virtio rework series; To make it easier to you, I added it to the dpdk- next-virtio tree: https://git.dpdk.org/next/dpdk-next-virtio/log/?h=virtio_pmd_rework_v2 Hi Maxime: Decision on patch 2? I still think current patch 2 is cleaner. Thanks, huawei Maxime
Re: [dpdk-dev] [PATCH v5 3/3] PCI: don't use vfio ioctl call to access PIO resource
On 1/26/21 1:30 PM, 谢华伟(此时此刻） wrote: > > On 2021/1/22 15:25, chris wrote: >> >> On 2021/1/21 23:38, Maxime Coquelin wrote: Do you mean we apply or abandon patch 3? I am both OK. The first priority to me is to enable MMIO bar support. >>> OK, so yes, I think we should abandon patch 2 and patch 3. >>> For patch 1, it looks valid to me, but I'll let Ferruh decide. >>> >>> For your device, if my understanding is correct, what we need to do is >>> to support MMIO for legacy devices. Correct? >> yes. >>> If so, the change should be in virtio_pci.c. In vtpci_init(), after >>> modern detection has failed, we should check the the BAR is PIO or MMIO >>> based on the flag. the result can be saved in struct virtio_pci_dev. >>> >>> >>> We would introduce new wrappers like vtpci_legacy_read, >>> vtpci_legacy_write that would either call rte_pci_ioport_read, >>> rte_pci_ioport_read in case of PIO, or rte_read32, rte_write32 in case >>> of MMIO. >> >> There are two choices. >> >> 1, apply patch 2. >> >> IO/MMIO port are mapped and accessed using the same API. Kernel is >> doing in the same way like the following. >> >> io_addr = pci_iomap >> >> get PIO directly or ioremap >> >> iowrite16/32(val, io_addr + offset) >> >> I think applying patch 2 is a correct choice. It is a fix. Driver had >> better not know if bar is PIO or MMIO. ioport in ioport_xx API means >> IO, not PIO. >> >> Btw, it only affects virtio PMD, not that intrusive. >> >> 2, virtio specific change to enable MMIO support. >> >> Comparing with choice 1, i feels it is not that clean and pretty. >> >>> >>> It is not too late for this release, as the change will not be that >>> intrusive. But if you prepare such patch, please base it on top of my >>> virtio rework series; To make it easier to you, I added it to the dpdk- >>> next-virtio tree: >>> https://git.dpdk.org/next/dpdk-next-virtio/log/?h=virtio_pmd_rework_v2 >>> > Hi Maxime: > > Decision on patch 2? > > I still think current patch 2 is cleaner. Hi, I actually replied one hour ago: " OK, that makes sense. I am OK with keeping patch 2, but would like Ferruh's ACK. Could you please post v6? " Thanks, Maxime > Thanks, huawei > > >>> Maxime >>> >
Re: [dpdk-dev] [PATCH v4 02/44] bus/vdev: add driver IOVA VA mode requirement
On Tue, Jan 26, 2021 at 11:16 AM Maxime Coquelin wrote: > > This patch adds driver flag in vdev bus driver so that > vdev drivers can require VA IOVA mode to be used, which > for example the case of Virtio-user PMD. > > The patch implements the .get_iommu_class() callback, that > is called before devices probing to determine the IOVA mode > to be used, and adds a check right before the device is > probed to ensure compatible IOVA mode has been selected. > > It also adds a ABI exception rule to accommodate with an > update on the driver registration API > > Signed-off-by: Maxime Coquelin > Signed-off-by: David Marchand > --- > devtools/libabigail.abignore| 2 ++ > drivers/bus/vdev/rte_bus_vdev.h | 4 > drivers/bus/vdev/vdev.c | 29 + > 3 files changed, 35 insertions(+) > > diff --git a/devtools/libabigail.abignore b/devtools/libabigail.abignore > index 1dc84fa74b..170304c876 100644 > --- a/devtools/libabigail.abignore > +++ b/devtools/libabigail.abignore > @@ -11,6 +11,8 @@ > ; Explicit ignore for driver-only ABI > [suppress_type] > name = eth_dev_ops > +[suppress_function] > +name_regexp = rte_vdev_(|un)register > > ; Ignore fields inserted in cacheline boundary of rte_cryptodev > [suppress_type] Ray, Are you okay with this exception? Thanks. -- David Marchand
[dpdk-dev] [PATCH 2/2] net/mlx5: use global default miss for E-Switch sampling
In E-Switch steering domain there was dedicated default miss action
created for every sampling flow.
The patch replaces this one with the global default miss action.
Cc: sta...@dpdk.org
Signed-off-by: Jiawei Wang
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/mlx5_flow.h| 1 -
drivers/net/mlx5/mlx5_flow_dv.c | 23 +--
2 files changed, 5 insertions(+), 19 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow.h b/drivers/net/mlx5/mlx5_flow.h
index 2178a04..e9e3397 100644
--- a/drivers/net/mlx5/mlx5_flow.h
+++ b/drivers/net/mlx5/mlx5_flow.h
@@ -598,7 +598,6 @@ struct mlx5_flow_dv_sample_resource {
uint32_t ratio; /** Sample Ratio */
uint64_t set_action; /** Restore reg_c0 value */
void *normal_path_tbl; /** Flow Table pointer */
- void *default_miss; /** default_miss dr_action. */
struct mlx5_flow_sub_actions_idx sample_idx;
/**< Action index resources. */
struct mlx5_flow_sub_actions_list sample_act;
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 27d711d..5845b98 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -9149,22 +9149,18 @@ struct mlx5_cache_entry *
"for sample");
goto error;
}
- int ret;
-
cache_resource->normal_path_tbl = tbl;
if (resource->ft_type == MLX5DV_FLOW_TABLE_TYPE_FDB) {
- ret = mlx5_flow_os_create_flow_action_default_miss
- (&cache_resource->default_miss);
- if (ret) {
+ if (!sh->default_miss_action) {
rte_flow_error_set(error, ENOMEM,
RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
NULL,
- "cannot create default miss "
- "action");
+ "default miss action was not "
+ "created");
goto error;
}
sample_dv_actions[resource->sample_act.actions_num++] =
- cache_resource->default_miss;
+ sh->default_miss_action;
}
/* Create a DR sample action */
sampler_attr.sample_ratio = cache_resource->ratio;
@@ -9184,11 +9180,7 @@ struct mlx5_cache_entry *
cache_resource->dev = dev;
return &cache_resource->entry;
error:
- if (cache_resource->ft_type == MLX5DV_FLOW_TABLE_TYPE_FDB &&
- cache_resource->default_miss)
- claim_zero(mlx5_flow_os_destroy_flow_action
- (cache_resource->default_miss));
- else
+ if (cache_resource->ft_type != MLX5DV_FLOW_TABLE_TYPE_FDB)
flow_dv_sample_sub_actions_release(dev,
&cache_resource->sample_idx);
if (cache_resource->normal_path_tbl)
@@ -11591,11 +11583,6 @@ struct mlx5_cache_entry *
if (cache_resource->verbs_action)
claim_zero(mlx5_flow_os_destroy_flow_action
(cache_resource->verbs_action));
- if (cache_resource->ft_type == MLX5DV_FLOW_TABLE_TYPE_FDB) {
- if (cache_resource->default_miss)
- claim_zero(mlx5_flow_os_destroy_flow_action
- (cache_resource->default_miss));
- }
if (cache_resource->normal_path_tbl)
flow_dv_tbl_resource_release(MLX5_SH(dev),
cache_resource->normal_path_tbl);
--
1.8.3.1
[dpdk-dev] [PATCH 0/2] net/mlx5: use global default miss for E-Switch sampling
This patchset fixes the E-Switch sampling flow creation issue, and also uses the global default miss for each E-Switch sampling flow. Jiawei Wang (2): net/mlx5: fix the E-Switch sample action creation failure net/mlx5: use global default miss for E-Switch sampling drivers/net/mlx5/mlx5_flow.h| 1 - drivers/net/mlx5/mlx5_flow_dv.c | 23 +-- 2 files changed, 5 insertions(+), 19 deletions(-) -- 1.8.3.1
[dpdk-dev] [PATCH 1/2] net/mlx5: fix the E-Switch sample action creation failure
This patch fixes the incorrect checking for the return value
of default miss action creation.
Fixes: 14020ad53d4e ("net/mlx5: wrap default miss flow action per OS")
Cc: sta...@dpdk.org
Signed-off-by: Jiawei Wang
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/mlx5_flow_dv.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 1a0c0be..27d711d 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -9155,7 +9155,7 @@ struct mlx5_cache_entry *
if (resource->ft_type == MLX5DV_FLOW_TABLE_TYPE_FDB) {
ret = mlx5_flow_os_create_flow_action_default_miss
(&cache_resource->default_miss);
- if (!ret) {
+ if (ret) {
rte_flow_error_set(error, ENOMEM,
RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
NULL,
--
1.8.3.1
Re: [dpdk-dev] [PATCH v4 44/44] net/virtio: handle Virtio-user setup failure properly
On 1/26/21 1:02 PM, Xia, Chenbo wrote:
> Hi Maxime,
>
>> -Original Message-
>> From: Maxime Coquelin
>> Sent: Tuesday, January 26, 2021 6:17 PM
>> To: dev@dpdk.org; Xia, Chenbo ; olivier.m...@6wind.com;
>> amore...@redhat.com; david.march...@redhat.com
>> Cc: Maxime Coquelin
>> Subject: [PATCH v4 44/44] net/virtio: handle Virtio-user setup failure
>> properly
>>
>> This patch fixes virtio_user_dev_setup() error path,
>> by cleaning all resources it allocates. It introduces
>> virtio_user_dev_uninit_notify() that cleans all open
>> FDs. It implies assigning all FDs to -1 at init time.
>>
>> With these changes done, virtio_user_dev_init_notify()
>> can be simplified.
>>
>> Suggested-by: Adrian Moreno
>> Signed-off-by: Maxime Coquelin
>> ---
>> .../net/virtio/virtio_user/virtio_user_dev.c | 70 +--
>> 1 file changed, 47 insertions(+), 23 deletions(-)
>>
>> diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> index a1e32158bb..2998473622 100644
>> --- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> +++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> @@ -283,13 +283,7 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
>> int callfd;
>> int kickfd;
>>
>> -for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
>> -if (i >= dev->max_queue_pairs * 2) {
>> -dev->kickfds[i] = -1;
>> -dev->callfds[i] = -1;
>> -continue;
>> -}
>> -
>> +for (i = 0; i < dev->max_queue_pairs * 2; i++) {
>> /* May use invalid flag, but some backend uses kickfd and
>> * callfd as criteria to judge if dev is alive. so finally we
>> * use real event_fd.
>> @@ -297,28 +291,49 @@ virtio_user_dev_init_notify(struct virtio_user_dev
>> *dev)
>> callfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
>> if (callfd < 0) {
>> PMD_DRV_LOG(ERR, "(%s) callfd error, %s", dev->path,
>> strerror(errno));
>> -break;
>> +goto err;
>> }
>> kickfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
>> if (kickfd < 0) {
>> close(callfd);
>> PMD_DRV_LOG(ERR, "(%s) kickfd error, %s", dev->path,
>> strerror(errno));
>> -break;
>> +goto err;
>> }
>> dev->callfds[i] = callfd;
>> dev->kickfds[i] = kickfd;
>> }
>>
>> -if (i < VIRTIO_MAX_VIRTQUEUES) {
>> -for (j = 0; j < i; ++j) {
>> -close(dev->callfds[j]);
>> +return 0;
>> +err:
>> +for (j = 0; j < i; j++) {
>> +if (dev->kickfds[j] >= 0) {
>> close(dev->kickfds[j]);
>> +dev->kickfds[j] = -1;
>> +}
>> +if (dev->callfds[j] >= 0) {
>> +close(dev->callfds[j]);
>> +dev->callfds[j] = -1;
>> }
>> -
>> -return -1;
>> }
>>
>> -return 0;
>> +return -1;
>> +}
>> +
>> +static void
>> +virtio_user_dev_uninit_notify(struct virtio_user_dev *dev)
>> +{
>> +uint32_t i;
>> +
>> +for (i = 0; i < dev->max_queue_pairs; ++i) {
>
> Should be 'dev->max_queue_pairs * 2'?
Oops, good catch!
> I believe you will fix this in the final version before applying them
> in the tree.
>
> So with above fixed:
>
> Reviewed-by: Chenbo Xia
Thanks, I will do the change while applying.
Thanks,
Maxime
>> +if (dev->kickfds[i] >= 0) {
>> +close(dev->kickfds[i]);
>> +dev->kickfds[i] = -1;
>> +}
>> +if (dev->callfds[i] >= 0) {
>> +close(dev->callfds[i]);
>> +dev->callfds[i] = -1;
>> +}
>> +}
>> }
>>
>> static int
>> @@ -427,15 +442,22 @@ virtio_user_dev_setup(struct virtio_user_dev *dev)
>>
>> if (virtio_user_dev_init_notify(dev) < 0) {
>> PMD_INIT_LOG(ERR, "(%s) Failed to init notifiers\n", dev->path);
>> -return -1;
>> +goto destroy;
>> }
>>
>> if (virtio_user_fill_intr_handle(dev) < 0) {
>> PMD_INIT_LOG(ERR, "(%s) Failed to init interrupt handler\n",
>> dev-
>>> path);
>> -return -1;
>> +goto uninit;
>> }
>>
>> return 0;
>> +
>> +uninit:
>> +virtio_user_dev_uninit_notify(dev);
>> +destroy:
>> +dev->ops->destroy(dev);
>> +
>> +return -1;
>> }
>>
>> /* Use below macro to filter features from vhost backend */
>> @@ -466,9 +488,16 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
>> *path, int queues,
>> enum virtio_user_backend_type backend_type)
>> {
>> uint64_t backend_features;
>> +int i;
>>
>> pthread_mutex_init(&dev->mutex, NULL);
>> strlcpy(dev->path, path, PATH_MAX);
>> +
>
Re: [dpdk-dev] [dpdk-stable] [PATCH] vdpa/mlx5: fix configuration mutex cleanup
On 1/26/21 11:45 AM, Matan Azrad wrote:
>
>
> From: Maxime Coquelin
>>> From: Maxime Coquelin
On 1/14/21 4:23 PM, Matan Azrad wrote:
>
>
> From: Maxime Coquelin
>> On 1/14/21 2:09 PM, Matan Azrad wrote:
>>>
>>>
>>> From: Maxime Coquelin
Hi Matan,
On 1/14/21 12:49 PM, Matan Azrad wrote:
> Hi Maxime and David
>
> Thank you for Review.
>
> From: David Marchand
>> On Fri, Jan 8, 2021 at 9:48 AM David Marchand
>> wrote:
I wonder if it would be possible and cleaner to disable
cancellation on the thread while the mutex is held?
>
> Yes, we can cause thread to return by some global variable sync.
> It is the same logic.
No, that was not my suggestion. My suggestion is to block the
thread cancellation while in the critical section, using
pthread_setcancelstate().
>>>
>>> Yes, Generally it is better to let the thread control his
>>> cancellation, either
>> cancel itself or enabling\disabling cancellations.
>>>
>>> I don't see a reason to wait for the thread in current logic - the
>>> critical section
>> is not important to be completed here.
>>
>> The reason I see is there are quite a few things done in this
>> critical section. And if tomorrow someone add new things in it, he
>> may not know the thread can be cancelled at any time, which could
>> cause
hard to debug issues.
>
> As I said, here it is not needed, this thread designed just to cause
> guest
notifications.
>
> The optional future developer mistake can be done also outside the
> critical
section in in any other place - we cannot protect it.
>
> The design choice is to close the thread fast.
But why is it so urgent that it cannot been stopped cleanly?
I don't think it would add seconds delay by doing it in a clean way.
>>>
>>> We have system calls there per queue.
>>> No need this optional delay just because of mutex cleaning.
>>
>> OK, up to you...
>>
>> And what about the timer lock?
>
> Existing code initiates it before reusing...
Ok, so why not applying same logic for both mutexes?
> Thanks.
>
>>
>>>
>>>
Thanks,
Maxime
>>> We just want to close the thread and to clean the mutex.
>>>
>>> +1
>>
>> IEEE Std 1003.1-2001/Cor 2-2004, item XBD/TC2/D6/26 is applied,
>> adding pthread_t to the list of types that are not required to
>> be arithmetic types, thus allowing pthread_t to be defined as a
>> structure.
>>
>> It would be better to leave pthread_t alone and not interpret it:
>>
>> if (priv->timer_tid) {
>> pthread_cancel(priv->timer_tid);
>> pthread_join(priv->timer_tid, &status); }
>> priv->timer_tid = 0;
>
>
> I'm not sure why you think it is better in this specific case.
> The cancellation will close the thread in faster way, no need to
> wait for the
thread to close itself.
>
>
>>
>> --
>> David Marchand
>
>>>
>
>>>
>
[dpdk-dev] [PATCH v3 1/1] app/procinfo: fix security context info
From: Hemant Agrawal
We need to differentiate between crypto and ethernet security
context as they belong to different devices.
Fixes: d82d6ac64338 ("app/procinfo: add crypto security context info")
Cc: sta...@dpdk.org
Signed-off-by: Hemant Agrawal
Signed-off-by: Thomas Monjalon
---
v3: include stdbool.h and use true/false instead of 1/0
---
app/proc-info/main.c | 14 ++
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/app/proc-info/main.c b/app/proc-info/main.c
index 44249dd2cb..b9587f7ded 100644
--- a/app/proc-info/main.c
+++ b/app/proc-info/main.c
@@ -5,6 +5,7 @@
#include
#include
#include
+#include
#include
#include
#include
@@ -645,11 +646,16 @@ metrics_display(int port_id)
}
static void
-show_security_context(uint16_t portid)
+show_security_context(uint16_t portid, bool inline_offload)
{
- void *p_ctx = rte_eth_dev_get_sec_ctx(portid);
+ void *p_ctx;
const struct rte_security_capability *s_cap;
+ if (inline_offload)
+ p_ctx = rte_eth_dev_get_sec_ctx(portid);
+ else
+ p_ctx = rte_cryptodev_get_sec_ctx(portid);
+
if (p_ctx == NULL)
return;
@@ -856,7 +862,7 @@ show_port(void)
}
#ifdef RTE_LIB_SECURITY
- show_security_context(i);
+ show_security_context(i, true);
#endif
}
}
@@ -1220,7 +1226,7 @@ show_crypto(void)
}
#ifdef RTE_LIB_SECURITY
- show_security_context(i);
+ show_security_context(i, false);
#endif
}
}
--
2.30.0
Re: [dpdk-dev] [PATCH v3 1/1] app/procinfo: fix security context info
26/01/2021 14:06, Thomas Monjalon:
> From: Hemant Agrawal
>
> We need to differentiate between crypto and ethernet security
> context as they belong to different devices.
>
> Fixes: d82d6ac64338 ("app/procinfo: add crypto security context info")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Hemant Agrawal
> Signed-off-by: Thomas Monjalon
> ---
> v3: include stdbool.h and use true/false instead of 1/0
Applied, thanks
Re: [dpdk-dev] [PATCH v4 02/44] bus/vdev: add driver IOVA VA mode requirement
On 26/01/2021 12:50, David Marchand wrote: > On Tue, Jan 26, 2021 at 11:16 AM Maxime Coquelin > wrote: >> >> This patch adds driver flag in vdev bus driver so that >> vdev drivers can require VA IOVA mode to be used, which >> for example the case of Virtio-user PMD. >> >> The patch implements the .get_iommu_class() callback, that >> is called before devices probing to determine the IOVA mode >> to be used, and adds a check right before the device is >> probed to ensure compatible IOVA mode has been selected. >> >> It also adds a ABI exception rule to accommodate with an >> update on the driver registration API >> >> Signed-off-by: Maxime Coquelin >> Signed-off-by: David Marchand >> --- >> devtools/libabigail.abignore| 2 ++ >> drivers/bus/vdev/rte_bus_vdev.h | 4 >> drivers/bus/vdev/vdev.c | 29 + >> 3 files changed, 35 insertions(+) >> >> diff --git a/devtools/libabigail.abignore b/devtools/libabigail.abignore >> index 1dc84fa74b..170304c876 100644 >> --- a/devtools/libabigail.abignore >> +++ b/devtools/libabigail.abignore >> @@ -11,6 +11,8 @@ >> ; Explicit ignore for driver-only ABI >> [suppress_type] >> name = eth_dev_ops >> +[suppress_function] >> +name_regexp = rte_vdev_(|un)register >> >> ; Ignore fields inserted in cacheline boundary of rte_cryptodev >> [suppress_type] > > Ray, > Are you okay with this exception? Ask a perhaps silly question, shouldn't rte_vdev_register & rte_vdev_unregister have been INTERNAL in any case? > Thanks. > Ray K
Re: [dpdk-dev] [EXT] [PATCH v4 3/3] examples/eventdev: move eth stop to the end
On Mon, Jan 25, 2021 at 11:22 PM Pavan Nikhilesh Bhagavatula
wrote:
>
> >Move eth stop code from "signal_handler" function to the end of
> >"main"
> >function. There are two reasons for this:
> >
> >First, this improves code maintenance and makes code look simple and
> >clear. Based on this change, after receiving the interrupt signal,
> >"fdata->done" is set as 1. Then the main thread will wait all worker
> >lcores to jump out of the loop. Finally, the main thread will stop and
> >then close eth dev port.
> >
> >Second, for older version, the main thread first stops eth dev port and
> >then waits the end of worker lcore. This may cause errors because it
> >may
> >stop the eth dev port which worker lcores are using. This moving
> >change
> >can fix this by waiting all worker threads to exit and then stop the
> >eth dev port.
> >
> >In the meanwhile, remove wmb in signal_handler.
> >
> >This is because when the main lcore receive the stop signal, it stores 1
> >into fdata->done. And then the worker lcores load "fdata->done" and
> >jump
> >out of the loop to stop running. Nothing should be stored after
> >updating
> >fdata->done, so the wmb is unnecessary.
> >
> >Fixes: 085edac2ca38 ("examples/eventdev_pipeline: support Tx
> >adapter")
> >Cc: pbhagavat...@marvell.com
> >Cc: sta...@dpdk.org
> >
> >Suggested-by: Ruifeng Wang
> >Signed-off-by: Feifei Wang
> >Reviewed-by: Ruifeng Wang
> >Reviewed-by: Honnappa Nagarahalli
> >
> >Acked-by: Harry van Haaren
>
> Acked-by: Pavan Nikhilesh
Series applied to dpdk-next-eventdev/for-main. Thanks.
>
> >---
> > examples/eventdev_pipeline/main.c | 16
> > 1 file changed, 4 insertions(+), 12 deletions(-)
> >
> >diff --git a/examples/eventdev_pipeline/main.c
> >b/examples/eventdev_pipeline/main.c
> >index 3526d4d3d..4621e8a89 100644
> >--- a/examples/eventdev_pipeline/main.c
> >+++ b/examples/eventdev_pipeline/main.c
> >@@ -311,7 +311,6 @@ static void
> > signal_handler(int signum)
> > {
> > static uint8_t once;
> >- uint16_t portid;
> >
> > if (fdata->done)
> > rte_exit(1, "Exiting on signal %d\n", signum);
> >@@ -322,17 +321,6 @@ signal_handler(int signum)
> > rte_event_dev_dump(0, stdout);
> > once = 1;
> > fdata->done = 1;
> >- rte_smp_wmb();
> >-
> >- RTE_ETH_FOREACH_DEV(portid) {
> >- rte_event_eth_rx_adapter_stop(portid);
> >- rte_event_eth_tx_adapter_stop(portid);
> >- if (rte_eth_dev_stop(portid) < 0)
> >- printf("Failed to stop port %u", portid);
> >- }
> >-
> >- rte_eal_mp_wait_lcore();
> >-
> > }
> > if (signum == SIGTSTP)
> > rte_event_dev_dump(0, stdout);
> >@@ -483,6 +471,10 @@ main(int argc, char **argv)
> > }
> >
> > RTE_ETH_FOREACH_DEV(portid) {
> >+ rte_event_eth_rx_adapter_stop(portid);
> >+ rte_event_eth_tx_adapter_stop(portid);
> >+ if (rte_eth_dev_stop(portid) < 0)
> >+ printf("Failed to stop port %u", portid);
> > rte_eth_dev_close(portid);
> > }
> >
> >--
> >2.25.1
>
Re: [dpdk-dev] [PATCH v3 0/4] add checking of header includes
On Tue, Jan 26, 2021 at 12:15 PM Bruce Richardson wrote: > > On Mon, Jan 25, 2021 at 04:51:19PM +0100, David Marchand wrote: > > On Mon, Jan 25, 2021 at 3:11 PM Bruce Richardson > > wrote: > > > > > > As a general principle, each header file should include any other > > > headers it needs to provide data type definitions or macros. For > > > example, any header using the uintX_t types in structures or function > > > prototypes should include "stdint.h" to provide those type definitions. > > > > > > In practice, while many, but not all, headers in DPDK did include all > > > necessary headers, it was never actually checked that each header could > > > be included in a C file and compiled without having any compiler errors > > > about missing definitions. The script "check-includes.sh" could be used > > > for this job, but it was not called out in the documentation, so many > > > contributors may not have been aware of it's existance. It also was > > > difficult to run from a source-code directory, as the script did not > > > automatically allow finding of headers from one DPDK library directory > > > to another [this was probably based on running it on a build created by > > > the "make" build system, where all headers were in a single directory]. > > > To attempt to have a build-system integrated replacement, this patchset > > > adds a "chkincs" app in the buildtools directory to verify this on an > > > ongoing basis. > > > > > > This chkincs app does nothing when run, and is not installed as part of > > > a DPDK "ninja install", it's for build-time checking only. Its source > > > code consists of one C file per public DPDK header, where that C file > > > contains nothing except an include for that header. Therefore, if any > > > header is added to the lib folder which fails to compile when included > > > alone, the build of chkincs will fail with a suitable error message. > > > Since this compile checking is not needed on most builds of DPDK, the > > > building of chkincs is disabled by default, but can be enabled by the > > > "test_includes" meson option. To catch errors with patch submissions, > > > the final patch of this series enables it for a single build in > > > test-meson-builds script. > > > > > > Future work could involve doing similar checks on headers for C++ > > > compatibility, which was something done by the check-includes.sh script > > > but which is missing here.. > > > > > > V3: > > > * Shrunk patchset as most header fixes already applied > > > * Moved chkincs from "apps" to the "buildtools" directory, which is a > > > better location for something not for installation for end-user use. > > > * Added patch to drop check-includes script. > > > > > > V2: > > > * Add maintainers file entry for new app > > > * Drop patch for c11 ring header > > > * Use build variable "headers_no_chkincs" for tracking exceptions > > > > > > Bruce Richardson (4): > > > eal: add missing include to mcslock > > > build: separate out headers for include checking > > > buildtools/chkincs: add app to verify header includes > > > devtools: remove check-includes script > > > > > > MAINTAINERS | 5 +- > > > buildtools/chkincs/gen_c_file_for_header.py | 12 + > > > buildtools/chkincs/main.c| 4 + > > > buildtools/chkincs/meson.build | 40 +++ > > > devtools/check-includes.sh | 259 --- > > > devtools/test-meson-builds.sh| 2 +- > > > doc/guides/contributing/coding_style.rst | 12 + > > > lib/librte_eal/include/generic/rte_mcslock.h | 1 + > > > lib/librte_eal/include/meson.build | 2 +- > > > lib/librte_eal/x86/include/meson.build | 14 +- > > > lib/librte_ethdev/meson.build| 4 +- > > > lib/librte_hash/meson.build | 4 +- > > > lib/librte_ipsec/meson.build | 3 +- > > > lib/librte_lpm/meson.build | 2 +- > > > lib/librte_regexdev/meson.build | 2 +- > > > lib/librte_ring/meson.build | 4 +- > > > lib/librte_stack/meson.build | 4 +- > > > lib/librte_table/meson.build | 7 +- > > > lib/meson.build | 3 + > > > meson.build | 6 + > > > meson_options.txt| 2 + > > > 21 files changed, 112 insertions(+), 280 deletions(-) > > > create mode 100755 buildtools/chkincs/gen_c_file_for_header.py > > > create mode 100644 buildtools/chkincs/main.c > > > create mode 100644 buildtools/chkincs/meson.build > > > delete mode 100755 devtools/check-includes.sh > > > > - clang is not happy when enabling the check: > > $ meson configure $HOME/builds/build-clang-static -Dcheck_includes=true > > $ devtools/test-meson-builds.sh > > ... > > [362/464] Compiling C object > > buildtools/chkincs/chkincs.p/meson-generated_rte_ethdev_vdev.c.o > >
[dpdk-dev] [PATCH v4 0/7] add checking of header includes
As a general principle, each header file should include any other headers it needs to provide data type definitions or macros. For example, any header using the uintX_t types in structures or function prototypes should include "stdint.h" to provide those type definitions. In practice, while many, but not all, headers in DPDK did include all necessary headers, it was never actually checked that each header could be included in a C file and compiled without having any compiler errors about missing definitions. The script "check-includes.sh" could be used for this job, but it was not called out in the documentation, so many contributors may not have been aware of it's existance. It also was difficult to run from a source-code directory, as the script did not automatically allow finding of headers from one DPDK library directory to another [this was probably based on running it on a build created by the "make" build system, where all headers were in a single directory]. To attempt to have a build-system integrated replacement, this patchset adds a "chkincs" app in the buildtools directory to verify this on an ongoing basis. This chkincs app does nothing when run, and is not installed as part of a DPDK "ninja install", it's for build-time checking only. Its source code consists of one C file per public DPDK header, where that C file contains nothing except an include for that header. Therefore, if any header is added to the lib folder which fails to compile when included alone, the build of chkincs will fail with a suitable error message. Since this compile checking is not needed on most builds of DPDK, the building of chkincs is disabled by default, but can be enabled by the "test_includes" meson option. To catch errors with patch submissions, the final patch of this series enables it for a single build in test-meson-builds script. Future work could involve doing similar checks on headers for C++ compatibility, which was something done by the check-includes.sh script but which is missing here. V4: * Fixed build errors with clang and arm builds * Added support for running chkincs as part of github actions CI V3: * Shrunk patchset as most header fixes already applied * Moved chkincs from "apps" to the "buildtools" directory, which is a better location for something not for installation for end-user use. * Added patch to drop check-includes script. V2: * Add maintainers file entry for new app * Drop patch for c11 ring header * Use build variable "headers_no_chkincs" for tracking exceptions Bruce Richardson (7): eal: add missing include to mcslock eal: fix error attribute use for clang rib: fix missing header include build: separate out headers for include checking buildtools/chkincs: add app to verify header includes devtools: remove check-includes script ci: add checking of includes to CI builds .ci/linux-build.sh | 1 + MAINTAINERS | 5 +- buildtools/chkincs/gen_c_file_for_header.py | 12 + buildtools/chkincs/main.c| 4 + buildtools/chkincs/meson.build | 41 +++ devtools/check-includes.sh | 259 --- devtools/test-meson-builds.sh| 2 +- doc/guides/contributing/coding_style.rst | 12 + lib/librte_eal/include/generic/rte_mcslock.h | 1 + lib/librte_eal/include/meson.build | 2 +- lib/librte_eal/include/rte_compat.h | 8 +- lib/librte_eal/x86/include/meson.build | 14 +- lib/librte_ethdev/meson.build| 4 +- lib/librte_hash/meson.build | 4 +- lib/librte_ipsec/meson.build | 3 +- lib/librte_lpm/meson.build | 2 +- lib/librte_regexdev/meson.build | 2 +- lib/librte_rib/rte_rib6.h| 1 + lib/librte_ring/meson.build | 4 +- lib/librte_stack/meson.build | 4 +- lib/librte_table/meson.build | 7 +- lib/meson.build | 3 + meson.build | 6 + meson_options.txt| 2 + 24 files changed, 122 insertions(+), 281 deletions(-) create mode 100755 buildtools/chkincs/gen_c_file_for_header.py create mode 100644 buildtools/chkincs/main.c create mode 100644 buildtools/chkincs/meson.build delete mode 100755 devtools/check-includes.sh -- 2.27.0
[dpdk-dev] [PATCH v4 1/7] eal: add missing include to mcslock
Include 'rte_branch_prediction.h' to get the likely/unlikely macro
definitions.
Fixes: 2173fb61 ("mcslock: add MCS queued lock implementation")
Cc: sta...@dpdk.org
Signed-off-by: Bruce Richardson
---
lib/librte_eal/include/generic/rte_mcslock.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/librte_eal/include/generic/rte_mcslock.h
b/lib/librte_eal/include/generic/rte_mcslock.h
index d370bef17a..9f323bd2a2 100644
--- a/lib/librte_eal/include/generic/rte_mcslock.h
+++ b/lib/librte_eal/include/generic/rte_mcslock.h
@@ -22,6 +22,7 @@
#include
#include
#include
+#include
/**
* The rte_mcslock_t type.
--
2.27.0
[dpdk-dev] [PATCH v4 2/7] eal: fix error attribute use for clang
Clang does not have an "error" attribute for functions, so for marking
internal functions we need to check for the error attribute, and provide
a fallback if it is not present. For clang, we can use "diagnose_if"
attribute, similarly checking for its presence before use.
Fixes: fba5af82adc8 ("eal: add internal ABI tag definition")
Cc: haiyue.w...@intel.com
Signed-off-by: Bruce Richardson
---
lib/librte_eal/include/rte_compat.h | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/lib/librte_eal/include/rte_compat.h
b/lib/librte_eal/include/rte_compat.h
index 4cd8f68d68..c30f072aa3 100644
--- a/lib/librte_eal/include/rte_compat.h
+++ b/lib/librte_eal/include/rte_compat.h
@@ -19,12 +19,18 @@ __attribute__((section(".text.experimental")))
#endif
-#ifndef ALLOW_INTERNAL_API
+#if !defined ALLOW_INTERNAL_API && __has_attribute(error) /* For GCC */
#define __rte_internal \
__attribute__((error("Symbol is not public ABI"), \
section(".text.internal")))
+#elif !defined ALLOW_INTERNAL_API && __has_attribute(diagnose_if) /* For clang
*/
+
+#define __rte_internal \
+__attribute__((diagnose_if(1, "Symbol is not public ABI", "error"), \
+section(".text.internal")))
+
#else
#define __rte_internal \
--
2.27.0
[dpdk-dev] [PATCH v4 3/7] rib: fix missing header include
The rte_rib6 header was using RTE_MIN macro from rte_common.h but not
including the header file.
Fixes: f7e861e21c46 ("rib: support IPv6")
Cc: vladimir.medved...@intel.com
Signed-off-by: Bruce Richardson
---
lib/librte_rib/rte_rib6.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/librte_rib/rte_rib6.h b/lib/librte_rib/rte_rib6.h
index b5e10569b9..dbd52928a2 100644
--- a/lib/librte_rib/rte_rib6.h
+++ b/lib/librte_rib/rte_rib6.h
@@ -20,6 +20,7 @@
#include
#include
+#include
#ifdef __cplusplus
extern "C" {
--
2.27.0
[dpdk-dev] [PATCH v4 4/7] build: separate out headers for include checking
For some libraries, there may be some header files which are not for direct
inclusion, but rather are to be included via other header files. To allow
later checking of these files for missing includes, we separate out the
indirect include files from the direct ones.
Signed-off-by: Bruce Richardson
---
doc/guides/contributing/coding_style.rst | 12
lib/librte_eal/include/meson.build | 2 +-
lib/librte_eal/x86/include/meson.build | 14 +-
lib/librte_ethdev/meson.build| 4 ++--
lib/librte_hash/meson.build | 4 ++--
lib/librte_ipsec/meson.build | 3 ++-
lib/librte_lpm/meson.build | 2 +-
lib/librte_regexdev/meson.build | 2 +-
lib/librte_ring/meson.build | 4 +++-
lib/librte_stack/meson.build | 4 +++-
lib/librte_table/meson.build | 7 +++
lib/meson.build | 3 +++
meson.build | 1 +
meson_options.txt| 2 ++
14 files changed, 45 insertions(+), 19 deletions(-)
diff --git a/doc/guides/contributing/coding_style.rst
b/doc/guides/contributing/coding_style.rst
index bb3f3efcbc..dba4145228 100644
--- a/doc/guides/contributing/coding_style.rst
+++ b/doc/guides/contributing/coding_style.rst
@@ -891,6 +891,18 @@ headers
installed to $PREFIX/include when ``ninja install`` is run. As with
source files, these should be specified using the meson ``files()``
function.
+ When ``check_includes`` build option is set to ``true``, each header
file
+ has additional checks performed on it, for example to ensure that it is
+ not missing any include statements for dependent headers.
+ For header files which are public, but only included indirectly in
+ applications, these checks can be skipped by using the
``headers_no_chkincs``
+ variable rather than ``headers``.
+
+headers_no_chkincs
+ **Default Value = []**.
+ As with ``headers`` option above, except that the files are not checked
+ for all needed include files as part of a DPDK build when
+ ``check_includes`` is set to ``true``.
includes:
**Default Value = []**.
diff --git a/lib/librte_eal/include/meson.build
b/lib/librte_eal/include/meson.build
index 0dea342e1d..449740e510 100644
--- a/lib/librte_eal/include/meson.build
+++ b/lib/librte_eal/include/meson.build
@@ -16,7 +16,6 @@ headers += files(
'rte_dev.h',
'rte_devargs.h',
'rte_eal.h',
- 'rte_eal_interrupts.h',
'rte_eal_memconfig.h',
'rte_eal_trace.h',
'rte_errno.h',
@@ -49,6 +48,7 @@ headers += files(
'rte_version.h',
'rte_vfio.h',
)
+headers_no_chkincs += files('rte_eal_interrupts.h')
# special case install the generic headers, since they go in a subdir
generic_headers = files(
diff --git a/lib/librte_eal/x86/include/meson.build
b/lib/librte_eal/x86/include/meson.build
index 549cc21a42..835ea22947 100644
--- a/lib/librte_eal/x86/include/meson.build
+++ b/lib/librte_eal/x86/include/meson.build
@@ -2,11 +2,7 @@
# Copyright(c) 2017 Intel Corporation
arch_headers = files(
- 'rte_atomic_32.h',
- 'rte_atomic_64.h',
'rte_atomic.h',
- 'rte_byteorder_32.h',
- 'rte_byteorder_64.h',
'rte_byteorder.h',
'rte_cpuflags.h',
'rte_cycles.h',
@@ -22,4 +18,12 @@ arch_headers = files(
'rte_ticketlock.h',
'rte_vect.h',
)
-install_headers(arch_headers, subdir: get_option('include_subdir_arch'))
+arch_headers_no_chkincs = files(
+ 'rte_atomic_32.h',
+ 'rte_atomic_64.h',
+ 'rte_byteorder_32.h',
+ 'rte_byteorder_64.h',
+)
+install_headers(arch_headers + arch_headers_no_chkincs,
+ subdir: get_option('include_subdir_arch'))
+dpdk_chkinc_headers += arch_headers
diff --git a/lib/librte_ethdev/meson.build b/lib/librte_ethdev/meson.build
index e4b610246f..ab84869ea8 100644
--- a/lib/librte_ethdev/meson.build
+++ b/lib/librte_ethdev/meson.build
@@ -12,12 +12,10 @@ sources = files('ethdev_private.c',
headers = files('rte_ethdev.h',
'rte_ethdev_driver.h',
- 'rte_ethdev_core.h',
'rte_ethdev_pci.h',
'rte_ethdev_trace.h',
'rte_ethdev_trace_fp.h',
'rte_ethdev_vdev.h',
- 'rte_eth_ctrl.h',
'rte_dev_info.h',
'rte_flow.h',
'rte_flow_driver.h',
@@ -25,5 +23,7 @@ headers = files('rte_ethdev.h',
'rte_mtr_driver.h',
'rte_tm.h',
'rte_tm_driver.h')
+headers_no_chkincs += files('rte_eth_ctrl.h',
+ 'rte_ethdev_core.h')
deps += ['net', 'kvargs', 'meter', 'telemetry']
diff --git a/lib/librte_hash/meson.build b/lib/librte_hash/meson.build
index 0977a63fd2..b3ebc8b078 100644
--- a/lib/librte_hash/meson.build
+++ b/lib/librte_hash/meson.build
@@ -1,12 +1,12 @@
# SPDX-License-Identifier: BSD-3-Clause
# Copyright(c) 2017 Intel Corporation
-headers = files('rte_
[dpdk-dev] [PATCH v4 5/7] buildtools/chkincs: add app to verify header includes
To verify that all DPDK headers are ok for inclusion directly in a C file,
and are not missing any other pre-requisite headers, we can auto-generate
for each header an empty C file that includes that header. Compiling these
files will throw errors if any header has unmet dependencies.
To ensure ongoing compliance, we enable this build test as part of the
default x86 build in "test-meson-builds.sh".
Signed-off-by: Bruce Richardson
---
MAINTAINERS | 4 ++
buildtools/chkincs/gen_c_file_for_header.py | 12 ++
buildtools/chkincs/main.c | 4 ++
buildtools/chkincs/meson.build | 41 +
devtools/test-meson-builds.sh | 2 +-
meson.build | 5 +++
6 files changed, 67 insertions(+), 1 deletion(-)
create mode 100755 buildtools/chkincs/gen_c_file_for_header.py
create mode 100644 buildtools/chkincs/main.c
create mode 100644 buildtools/chkincs/meson.build
diff --git a/MAINTAINERS b/MAINTAINERS
index 1a12916f56..d4f9ebe46d 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1562,6 +1562,10 @@ F: app/test/test_resource.c
F: app/test/virtual_pmd.c
F: app/test/virtual_pmd.h
+Header build sanity checking
+M: Bruce Richardson
+F: buildtools/chkincs/
+
Sample packet helper functions for unit test
M: Reshma Pattan
F: app/test/sample_packet_forward.c
diff --git a/buildtools/chkincs/gen_c_file_for_header.py
b/buildtools/chkincs/gen_c_file_for_header.py
new file mode 100755
index 00..ed46948aea
--- /dev/null
+++ b/buildtools/chkincs/gen_c_file_for_header.py
@@ -0,0 +1,12 @@
+#! /usr/bin/env python3
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2021 Intel Corporation
+
+from sys import argv
+from os.path import abspath
+
+(h_file, c_file) = argv[1:]
+
+contents = '#include "' + abspath(h_file) + '"'
+with open(c_file, 'w') as cf:
+cf.write(contents)
diff --git a/buildtools/chkincs/main.c b/buildtools/chkincs/main.c
new file mode 100644
index 00..d25bb8852a
--- /dev/null
+++ b/buildtools/chkincs/main.c
@@ -0,0 +1,4 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2021 Intel Corporation
+ */
+int main(void) { return 0; }
diff --git a/buildtools/chkincs/meson.build b/buildtools/chkincs/meson.build
new file mode 100644
index 00..97f9ad13b3
--- /dev/null
+++ b/buildtools/chkincs/meson.build
@@ -0,0 +1,41 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2021 Intel Corporation
+
+if not get_option('check_includes')
+ build = false
+ subdir_done()
+endif
+
+if is_windows
+ # for windows, the shebang line in the script won't work.
+ error('option "check_includes" is not supported on windows')
+endif
+
+gen_c_file_for_header = find_program('gen_c_file_for_header.py')
+gen_c_files = generator(gen_c_file_for_header,
+ output: '@BASENAME@.c',
+ arguments: ['@INPUT@', '@OUTPUT@'])
+
+cflags = machine_args
+cflags += '-Wno-unused-function' # needed if we include generic headers
+cflags += '-DALLOW_EXPERIMENTAL_API'
+cflags += '-DALLOW_INTERNAL_API'
+
+# some ethdev headers depend on bus headers
+includes = include_directories('../../drivers/bus/pci',
+ '../../drivers/bus/vdev')
+
+sources = files('main.c')
+sources += gen_c_files.process(dpdk_chkinc_headers)
+
+deps = []
+foreach l:enabled_libs
+ deps += get_variable('static_rte_' + l)
+endforeach
+
+executable('chkincs', sources,
+ c_args: cflags,
+ include_directories: includes,
+ dependencies: deps,
+ link_whole: dpdk_static_libraries + dpdk_drivers,
+ install: false)
diff --git a/devtools/test-meson-builds.sh b/devtools/test-meson-builds.sh
index efa91e0e40..07b5e6aeca 100755
--- a/devtools/test-meson-builds.sh
+++ b/devtools/test-meson-builds.sh
@@ -227,7 +227,7 @@ default_machine='nehalem'
if ! check_cc_flags "-march=$default_machine" ; then
default_machine='corei7'
fi
-build build-x86-default cc skipABI \
+build build-x86-default cc skipABI -Dcheck_includes=true\
-Dlibdir=lib -Dmachine=$default_machine $use_shared
# 32-bit with default compiler
diff --git a/meson.build b/meson.build
index e6e34d0a98..fcc4d4c900 100644
--- a/meson.build
+++ b/meson.build
@@ -68,6 +68,11 @@ if get_option('enable_kmods')
subdir('kernel')
endif
+# check header includes if requested
+if get_option('check_includes')
+ subdir('buildtools/chkincs')
+endif
+
# write the build config
build_cfg = 'rte_build_config.h'
configure_file(output: build_cfg,
--
2.27.0
[dpdk-dev] [PATCH v4 6/7] devtools: remove check-includes script
The check-includes script allowed checking header files in a given
directory to ensure that each header compiled alone without requiring
any other header inclusions.
With header checking now being done by the chkincs app in the build
system this script can be removed.
Signed-off-by: Bruce Richardson
---
MAINTAINERS| 1 -
devtools/check-includes.sh | 259 -
2 files changed, 260 deletions(-)
delete mode 100755 devtools/check-includes.sh
diff --git a/MAINTAINERS b/MAINTAINERS
index d4f9ebe46d..8498b402db 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -81,7 +81,6 @@ F: devtools/check-dup-includes.sh
F: devtools/check-maintainers.sh
F: devtools/check-forbidden-tokens.awk
F: devtools/check-git-log.sh
-F: devtools/check-includes.sh
F: devtools/check-spdx-tag.sh
F: devtools/check-symbol-maps.sh
F: devtools/checkpatches.sh
diff --git a/devtools/check-includes.sh b/devtools/check-includes.sh
deleted file mode 100755
index 940cf0eb2a..00
--- a/devtools/check-includes.sh
+++ /dev/null
@@ -1,259 +0,0 @@
-#!/bin/sh -e
-# SPDX-License-Identifier: BSD-3-Clause
-# Copyright 2016 6WIND S.A.
-
-# This script checks that header files in a given directory do not miss
-# dependencies when included on their own, do not conflict and accept being
-# compiled with the strictest possible flags.
-#
-# Files are looked up in the directory provided as the first argument,
-# otherwise build/include by default.
-#
-# Recognized environment variables:
-#
-# VERBOSE=1 is the same as -v.
-#
-# QUIET=1 is the same as -q.
-#
-# SUMMARY=1 is the same as -s.
-#
-# CC, CPPFLAGS, CFLAGS, CXX, CXXFLAGS are taken into account.
-#
-# PEDANTIC_CFLAGS, PEDANTIC_CXXFLAGS and PEDANTIC_CPPFLAGS provide strict
-# C/C++ compilation flags.
-#
-# IGNORE contains a list of globbing patterns matching files (relative to the
-# include directory) to avoid. It is set by default to known DPDK headers
-# which must not be included on their own.
-#
-# IGNORE_CXX provides additional files for C++.
-
-while getopts hqvs arg; do
- case $arg in
- h)
- cat < /dev/null
-
-[ "$VERBOSE" = 1 ] &&
-output ()
-{
- local CCV
- local CXXV
-
- shift
- CCV=$CC
- CXXV=$CXX
- CC="echo $CC" CXX="echo $CXX" "$@"
- CC=$CCV
- CXX=$CXXV
-
- "$@"
-} ||
-output ()
-{
-
- printf ' %s\n' "$1"
- shift
- "$@"
-}
-
-trap 'rm -f "$temp_cc" "$temp_cxx"' EXIT
-
-compile_cc ()
-{
- ${CC} -I"$include_dir" \
- ${PEDANTIC_CPPFLAGS} ${CPPFLAGS} \
- ${PEDANTIC_CFLAGS} ${CFLAGS} \
- -c -o /dev/null "${temp_cc}"
-}
-
-compile_cxx ()
-{
- ${CXX} -I"$include_dir" \
- ${PEDANTIC_CPPFLAGS} ${CPPFLAGS} \
- ${PEDANTIC_CXXFLAGS} ${CXXFLAGS} \
- -c -o /dev/null "${temp_cxx}"
-}
-
-ignore ()
-{
- file="$1"
- shift
- while [ $# -ne 0 ]; do
- case "$file" in
- $1)
- return 0
- ;;
- esac
- shift
- done
- return 1
-}
-
-# Check C/C++ compilation for each header file.
-
-while read -r path
-do
- file=${path#$include_dir}
- file=${file##/}
- if ignore "$file" $IGNORE; then
- output "SKIP $file" :
- continue
- fi
- if printf "\
-#include <%s>
-
-int main(void)
-{
- return 0;
-}
-" "$file" > "$temp_cc" &&
- output "CC $file" compile_cc
- then
- pass_cc="$pass_cc $file"
- else
- failures_cc=$(($failures_cc + 1))
- fi
- if ignore "$file" $IGNORE_CXX; then
- output "SKIP CXX $file" :
- continue
- fi
- if printf "\
-#include <%s>
-
-int main()
-{
-}
-" "$file" > "$temp_cxx" &&
- output "CXX $file" compile_cxx
- then
- pass_cxx="$pass_cxx $file"
- else
- failures_cxx=$(($failures_cxx + 1))
- fi
-done < "$temp_cc" &&
-for file in $pass_cc; do
- printf "\
-#include <%s>
-" "$file" >> $temp_cc
-done
-if printf "\
-int main(void)
-{
- return 0;
-}
-" >> "$temp_cc" &&
- output "CC (all includes that did not fail)" compile_cc
-then
- :
-else
- failures_cc=$(($failures_cc + 1))
-fi
-
-# Check C++ compilation with all includes.
-
-: > "$temp_cxx" &&
-for file in $pass_cxx; do
- printf "\
-#include <%s>
-" "$file" >> $temp_cxx
-done
-if printf "\
-int main()
-{
-}
-" >> "$temp_cxx" &&
- output "CXX (all includes that did not fail)" compile_cxx
-then
- :
-else
- failures_cxx=$(($failures_cxx + 1))
-fi
-
-# Report results.
-
-if [ "$SUMMARY" = 1 ]; then
- printf "\
-Summary:
- %u failure(s) for C using '%s'.
- %u failure(s) for C++ using '%s'.
-" $failures_cc "$CC" $failures_cxx "$CXX" 1>&2
-fi
-
-# Exit with nonzero status if there are failures.
-
-[ $failures_c
[dpdk-dev] [PATCH v4 7/7] ci: add checking of includes to CI builds
For CI builds, turn on the checking of includes. Signed-off-by: Bruce Richardson --- For simplicity, this patch just enables the header includes checks globally, rather than just limiting it to certain builds. Since the C files generated are all just a single #include line , the extra compilation time is fairly short compared to the overall existing build time in the CI. --- .ci/linux-build.sh | 1 + 1 file changed, 1 insertion(+) diff --git a/.ci/linux-build.sh b/.ci/linux-build.sh index afa3689a09..fdbeb5a616 100755 --- a/.ci/linux-build.sh +++ b/.ci/linux-build.sh @@ -57,6 +57,7 @@ fi OPTS="$OPTS -Dmachine=default" OPTS="$OPTS --default-library=$DEF_LIB" OPTS="$OPTS --buildtype=debugoptimized" +OPTS="$OPTS -Dcheck_includes=true" meson build --werror $OPTS ninja -C build -- 2.27.0
Re: [dpdk-dev] [PATCH v4 3/7] rib: fix missing header include
On 26/01/2021 14:18, Bruce Richardson wrote:
The rte_rib6 header was using RTE_MIN macro from rte_common.h but not
including the header file.
Fixes: f7e861e21c46 ("rib: support IPv6")
Cc: vladimir.medved...@intel.com
Signed-off-by: Bruce Richardson
---
lib/librte_rib/rte_rib6.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/librte_rib/rte_rib6.h b/lib/librte_rib/rte_rib6.h
index b5e10569b9..dbd52928a2 100644
--- a/lib/librte_rib/rte_rib6.h
+++ b/lib/librte_rib/rte_rib6.h
@@ -20,6 +20,7 @@
#include
#include
+#include
#ifdef __cplusplus
extern "C" {
Acked-by: Vladimir Medvedkin
--
Regards,
Vladimir
Re: [dpdk-dev] [PATCH v3 0/4] add checking of header includes
On Tue, Jan 26, 2021 at 03:04:25PM +0100, David Marchand wrote: > On Tue, Jan 26, 2021 at 12:15 PM Bruce Richardson > wrote: > > > > On Mon, Jan 25, 2021 at 04:51:19PM +0100, David Marchand wrote: > > > On Mon, Jan 25, 2021 at 3:11 PM Bruce Richardson > > > wrote: > > > > > > > > As a general principle, each header file should include any other > > > > headers it needs to provide data type definitions or macros. For > > > > example, any header using the uintX_t types in structures or function > > > > prototypes should include "stdint.h" to provide those type definitions. > > > > > > > > In practice, while many, but not all, headers in DPDK did include all > > > > necessary headers, it was never actually checked that each header could > > > > be included in a C file and compiled without having any compiler errors > > > > about missing definitions. The script "check-includes.sh" could be used > > > > for this job, but it was not called out in the documentation, so many > > > > contributors may not have been aware of it's existance. It also was > > > > difficult to run from a source-code directory, as the script did not > > > > automatically allow finding of headers from one DPDK library directory > > > > to another [this was probably based on running it on a build created by > > > > the "make" build system, where all headers were in a single directory]. > > > > To attempt to have a build-system integrated replacement, this patchset > > > > adds a "chkincs" app in the buildtools directory to verify this on an > > > > ongoing basis. > > > > > > > > This chkincs app does nothing when run, and is not installed as part of > > > > a DPDK "ninja install", it's for build-time checking only. Its source > > > > code consists of one C file per public DPDK header, where that C file > > > > contains nothing except an include for that header. Therefore, if any > > > > header is added to the lib folder which fails to compile when included > > > > alone, the build of chkincs will fail with a suitable error message. > > > > Since this compile checking is not needed on most builds of DPDK, the > > > > building of chkincs is disabled by default, but can be enabled by the > > > > "test_includes" meson option. To catch errors with patch submissions, > > > > the final patch of this series enables it for a single build in > > > > test-meson-builds script. > > > > > > > > Future work could involve doing similar checks on headers for C++ > > > > compatibility, which was something done by the check-includes.sh script > > > > but which is missing here.. > > > > > > > > V3: > > > > * Shrunk patchset as most header fixes already applied > > > > * Moved chkincs from "apps" to the "buildtools" directory, which is a > > > > better location for something not for installation for end-user use. > > > > * Added patch to drop check-includes script. > > > > > > > > V2: > > > > * Add maintainers file entry for new app > > > > * Drop patch for c11 ring header > > > > * Use build variable "headers_no_chkincs" for tracking exceptions > > > > > > > > Bruce Richardson (4): > > > > eal: add missing include to mcslock > > > > build: separate out headers for include checking > > > > buildtools/chkincs: add app to verify header includes > > > > devtools: remove check-includes script > > > > > > > > MAINTAINERS | 5 +- > > > > buildtools/chkincs/gen_c_file_for_header.py | 12 + > > > > buildtools/chkincs/main.c| 4 + > > > > buildtools/chkincs/meson.build | 40 +++ > > > > devtools/check-includes.sh | 259 --- > > > > devtools/test-meson-builds.sh| 2 +- > > > > doc/guides/contributing/coding_style.rst | 12 + > > > > lib/librte_eal/include/generic/rte_mcslock.h | 1 + > > > > lib/librte_eal/include/meson.build | 2 +- > > > > lib/librte_eal/x86/include/meson.build | 14 +- > > > > lib/librte_ethdev/meson.build| 4 +- > > > > lib/librte_hash/meson.build | 4 +- > > > > lib/librte_ipsec/meson.build | 3 +- > > > > lib/librte_lpm/meson.build | 2 +- > > > > lib/librte_regexdev/meson.build | 2 +- > > > > lib/librte_ring/meson.build | 4 +- > > > > lib/librte_stack/meson.build | 4 +- > > > > lib/librte_table/meson.build | 7 +- > > > > lib/meson.build | 3 + > > > > meson.build | 6 + > > > > meson_options.txt| 2 + > > > > 21 files changed, 112 insertions(+), 280 deletions(-) > > > > create mode 100755 buildtools/chkincs/gen_c_file_for_header.py > > > > create mode 100644 buildtools/chkincs/main.c > > > > create mode 100644 buildtools/chkincs/meson.build > > > > delete mode 100755 devtools/check-includes.sh > > > > > > - clang is not happy when enabling
