Re: [dpdk-dev] [PATCH 2/2] test/memzone: handle previously allocated memzones

[Radoslaw Biernacki]

Thanks, looks OK for me.

Reviewed-by: Radoslaw Biernacki 
adoslaw.bierna...@linaro.com>

On 26 January 2018 at 18:40, Anatoly Burakov 
wrote:

> Currently, memzone autotest expects there to be no memzones
> present by the time the test is run. Some hardware drivers
> will allocate memzones for internal use during initialization,
> resulting in tests failing due to unexpected memzones being
> allocated before the test was run.
>
> Fix this by making callback increment a counter instead. This
> also doubles as a test for correct operation of memzone_walk().
>
> Fixes: 71330483a193 ("test/memzone: fix memory leak")
> Cc: radoslaw.bierna...@linaro.org
> Cc: sta...@dpdk.org
>
> Signed-off-by: Anatoly Burakov 
> ---
>  test/test/test_memzone.c | 10 +++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/test/test/test_memzone.c b/test/test/test_memzone.c
> index 00d340f..5428b35 100644
> --- a/test/test/test_memzone.c
> +++ b/test/test/test_memzone.c
> @@ -953,16 +953,19 @@ test_memzone_basic(void)
> return 0;
>  }
>
> -static int memzone_walk_called;
> +static int memzone_walk_cnt;
>  static void memzone_walk_clb(const struct rte_memzone *mz __rte_unused,
>  void *arg __rte_unused)
>  {
> -   memzone_walk_called = 1;
> +   memzone_walk_cnt++;
>  }
>
>  static int
>  test_memzone(void)
>  {
> +   /* take note of how many memzones were allocated before running */
> +   int memzone_cnt = rte_eal_get_configuration()->
> mem_config->memzone_cnt;
> +
> printf("test basic memzone API\n");
> if (test_memzone_basic() < 0)
> return -1;
> @@ -1000,8 +1003,9 @@ test_memzone(void)
> return -1;
>
> printf("check memzone cleanup\n");
> +   memzone_walk_cnt = 0;
> rte_memzone_walk(memzone_walk_clb, NULL);
> -   if (memzone_walk_called) {
> +   if (memzone_walk_cnt != memzone_cnt) {
> printf("there are some memzones left after test\n");
> rte_memzone_dump(stdout);
> return -1;
> --
> 2.7.4
>

Re: [dpdk-dev] [PATCH 1/2] test/memzone: add test for memzone count in eal mem config

[Radoslaw Biernacki]

Looks OK.

Following note is aside from the patch.
Might be beneficial (in some rare cases) to add bailout recovery with
goto's in test_memzone_basic()
Just in case one of the rte_memzone_reserve() we should not make return -1,
but instead a goto to below section where we call rte_memzone_free().
This way we would be able to free only the allocated memzones and prevent
leaking out those memzones to other tests.

Reviewed-by: Radoslaw Biernacki 
adoslaw.bierna...@linaro.com>

On 26 January 2018 at 18:40, Anatoly Burakov 
wrote:

> Ensure that memzone count in eal mem config is incremented and
> decremented whenever memzones are allocated and freed.
>
> Signed-off-by: Anatoly Burakov 
> ---
>  test/test/test_memzone.c | 20 
>  1 file changed, 20 insertions(+)
>
> diff --git a/test/test/test_memzone.c b/test/test/test_memzone.c
> index f6c9b56..00d340f 100644
> --- a/test/test/test_memzone.c
> +++ b/test/test/test_memzone.c
> @@ -841,6 +841,9 @@ test_memzone_basic(void)
> const struct rte_memzone *memzone3;
> const struct rte_memzone *memzone4;
> const struct rte_memzone *mz;
> +   int memzone_cnt_after, memzone_cnt_expected;
> +   int memzone_cnt_before =
> +   rte_eal_get_configuration()->
> mem_config->memzone_cnt;
>
> memzone1 = rte_memzone_reserve("testzone1", 100,
> SOCKET_ID_ANY, 0);
> @@ -858,6 +861,18 @@ test_memzone_basic(void)
> if (memzone1 == NULL || memzone2 == NULL || memzone4 == NULL)
> return -1;
>
> +   /* check how many memzones we are expecting */
> +   memzone_cnt_expected = memzone_cnt_before +
> +   (memzone1 != NULL) + (memzone2 != NULL) +
> +   (memzone3 != NULL) + (memzone4 != NULL);
> +
> +   memzone_cnt_after =
> +   rte_eal_get_configuration()->
> mem_config->memzone_cnt;
> +
> +   if (memzone_cnt_after != memzone_cnt_expected)
> +   return -1;
> +
> +
> rte_memzone_dump(stdout);
>
> /* check cache-line alignments */
> @@ -930,6 +945,11 @@ test_memzone_basic(void)
> return -1;
> }
>
> +   memzone_cnt_after =
> +   rte_eal_get_configuration()->
> mem_config->memzone_cnt;
> +   if (memzone_cnt_after != memzone_cnt_before)
> +   return -1;
> +
> return 0;
>  }
>
> --
> 2.7.4
>

Re: [dpdk-dev] [PATCH v2 2/4] net/mlx4: spawn rdma-core dependency plug-in

[Marcelo Ricardo Leitner]

On Fri, Jan 26, 2018 at 03:19:00PM +0100, Adrien Mazarguil wrote:
...
> +static int
> +mlx4_glue_init(void)
> +{
> + char file[] = "/tmp/" MLX4_DRIVER_NAME "_XX";
> + int fd = mkstemp(file);
...
> + while (off != mlx4_glue_lib_size) {
> + ssize_t ret;
> +
> + ret = write(fd, (const uint8_t *)mlx4_glue_lib + off,
> + mlx4_glue_lib_size - off);
> + if (ret == -1) {
> + if (errno != EINTR) {
> + rte_errno = errno;
> + goto glue_error;
> + }
> + ret = 0;
> + }
> + off += ret;
> + }
> + close(fd);
> + fd = -1;
> + handle = dlopen(file, RTLD_LAZY);
> + unlink(file);

This is a potential security issue. There are no guarantees that the
file dlopen() will open is the file that was just written above. It
could have been changed by something else in between.

  Marcelo

Re: [dpdk-dev] [PATCH] app/testpmd: do not enable Rx offloads by default

[Shahaf Shuler]

Friday, January 26, 2018 10:36 AM, Thomas Monjalon:
> 26/01/2018 09:06, Lu, Wenzhuo:
> > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > > 26/01/2018 08:30, Lu, Wenzhuo:
> > > > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > > > > 25/01/2018 02:11, Lu, Wenzhuo:
> > > > > Why these offload flags are silently enabled?
> > > >
> > > > I don't think it's silently. It's a global configuration. In this
> > > > case, testpmd is
> > > the user, it does request it explicitly. If it's not so explicit,
> > > maybe we can consider moving all the configuration to a specific
> configure file.
> > > > Talking about why it's enabled by default. Hard to figure out the 
> > > > history.
> > > To my opinion, the original designer wants to simulate the common case.
> > >
> > > Please do not justify a design mistake by history.
> > >
> > > This is a test tool, so we don't care about the common case.
> > > A test tool should not try to guess the best configuration.
> > > Only the user should decide the configuration to apply, and the
> > > default should be empty, as the API is.
> >
> > I see the divergence. You think testpmd is a tool, so it should be white
> paper as the tool users may suppose it is.
> > I think testpmd is an example to let the APP developers know how to use
> DPDK. So any pre-configuration is acceptable.

If so, then what this patch shows that it is not such a good example to 
application writers. 
Application should not set offloads that are not supported by the PMD. 
Pre-configuration is OK, but adjustment is a must according to the PMD caps. 
I think that many PMD have a wrong behavior due to it. They are forced to 
silently drop un-supported offload configuration (to enable testpmd to run on 
top of them).
This will end up, for example, in application which expects to have the VLAN 
tag in the mbuf (because the configuration passed successfully), while in fact 
it will be always in the packet headers. 

I didn't change it on my series because I wanted to convert the application "as 
is" from the old to the new offloads API,  however I do think a change is 
needed here.

 
> 
> Ah ah ah :)
> No
> Examples are in examples/ directory.

[dpdk-dev] [PATCH 1/4] service: move finalize to internal

[Harry van Haaren]

This commit moves the rte_service_finalize() function
to be in the component header, and marks it as @internal.
The function is only called internally by rte_eal_finalize().

Signed-off-by: Harry van Haaren 

---

Cc: tho...@monjalon.net
Cc: vipin.vargh...@intel.com
---
 lib/librte_eal/common/include/rte_service.h   | 11 ---
 lib/librte_eal/common/include/rte_service_component.h | 11 +++
 2 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/lib/librte_eal/common/include/rte_service.h 
b/lib/librte_eal/common/include/rte_service.h
index 5e3e3a6..02b1512 100644
--- a/lib/librte_eal/common/include/rte_service.h
+++ b/lib/librte_eal/common/include/rte_service.h
@@ -429,17 +429,6 @@ int32_t rte_service_attr_get(uint32_t id, uint32_t attr_id,
  */
 int32_t rte_service_attr_reset_all(uint32_t id);
 
-/**
- * @warning
- * @b EXPERIMENTAL: this API may change without prior notice
- *
- * Free up the memory that has been initialized. This routine
- * is to be invoked prior to process termination.
- *
- * @retval None
- */
-void rte_service_finalize(void);
-
 #ifdef __cplusplus
 }
 #endif
diff --git a/lib/librte_eal/common/include/rte_service_component.h 
b/lib/librte_eal/common/include/rte_service_component.h
index 849c71d..f881ac0 100644
--- a/lib/librte_eal/common/include/rte_service_component.h
+++ b/lib/librte_eal/common/include/rte_service_component.h
@@ -139,4 +139,15 @@ int32_t rte_service_component_runstate_set(uint32_t id, 
uint32_t runstate);
  */
 int32_t rte_service_init(void);
 
+/**
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
+ *
+ * @internal Free up the memory that has been initialized.
+ * This routine is to be invoked prior to process termination.
+ *
+ * @retval None
+ */
+void rte_service_finalize(void);
+
 #endif /* _RTE_SERVICE_PRIVATE_H_ */
-- 
2.7.4

[dpdk-dev] [PATCH 2/4] eal: add function to release internal resources

[Harry van Haaren]

This commit adds a new function rte_eal_finalize().
The function serves as a hook to allow DPDK to release
internal resources (e.g.: hugepage allocations).

This function allows DPDK to become more like an ordinary
library, where the library context itself can be initialized
and finalized by the application.

The rte_exit() and rte_panic() functions must be considered,
particularly if they should call finalize() to cleanup any
resources or not. This patch adds the cleanup to rte_exit(),
but does not clean up on rte_panic(). The reason to not clean
up on panicing is that the developer may wish to inspect the
exact internal state of EAL.

Signed-off-by: Harry van Haaren 

---

Cc: tho...@monjalon.net
Cc: vipin.vargh...@intel.com
---
 doc/guides/prog_guide/env_abstraction_layer.rst |  8 
 doc/guides/rel_notes/release_18_02.rst  |  9 +
 lib/librte_eal/bsdapp/eal/Makefile  |  1 +
 lib/librte_eal/bsdapp/eal/eal_debug.c   |  5 +
 lib/librte_eal/common/include/rte_eal.h | 15 +++
 lib/librte_eal/linuxapp/eal/Makefile|  1 +
 lib/librte_eal/linuxapp/eal/eal_debug.c |  5 +
 lib/librte_eal/rte_eal_version.map  |  1 +
 8 files changed, 45 insertions(+)

diff --git a/doc/guides/prog_guide/env_abstraction_layer.rst 
b/doc/guides/prog_guide/env_abstraction_layer.rst
index 34d871c..f020041 100644
--- a/doc/guides/prog_guide/env_abstraction_layer.rst
+++ b/doc/guides/prog_guide/env_abstraction_layer.rst
@@ -99,6 +99,14 @@ It consist of calls to the pthread library (more 
specifically, pthread_self(), p
 The creation and initialization functions for these objects are not 
multi-thread safe.
 However, once initialized, the objects themselves can safely be used in 
multiple threads simultaneously.
 
+Finalizing and Cleanup
+~~
+
+During the initialization of EAL resources such as hugepage backed memory can 
be
+allocated by core components.  The memory allocated during ``rte_eal_init()``
+can be released by calling the ``rte_eal_finalize()`` function. Refer to the
+API documentation for details.
+
 Multi-process Support
 ~
 
diff --git a/doc/guides/rel_notes/release_18_02.rst 
b/doc/guides/rel_notes/release_18_02.rst
index 00b3224..5c7410e 100644
--- a/doc/guides/rel_notes/release_18_02.rst
+++ b/doc/guides/rel_notes/release_18_02.rst
@@ -41,6 +41,15 @@ New Features
  Also, make sure to start the actual text at the margin.
  =
 
+* **Add function to allow releasing internal EAL resources on exit**
+
+  During ``rte_eal_init()`` EAL allocates memory from hugepages to enable its
+  core libraries to perform their tasks. The ``rte_eal_finalize()`` function
+  releases these resources, ensuring that no hugepage memory is leaked. It is
+  expected that all DPDK applications call ``rte_eal_finalize()`` before
+  exiting. Not calling this function could result in leaking hugepages, leading
+  to failure during initialization of secondary processes.
+
 * **Added the ixgbe ethernet driver to support RSS with flow API.**
 
   Rte_flow actually defined to include RSS, but till now, RSS is out of
diff --git a/lib/librte_eal/bsdapp/eal/Makefile 
b/lib/librte_eal/bsdapp/eal/Makefile
index c694076..7480f98 100644
--- a/lib/librte_eal/bsdapp/eal/Makefile
+++ b/lib/librte_eal/bsdapp/eal/Makefile
@@ -34,6 +34,7 @@ SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_interrupts.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_alarm.c
 
 # from common dir
+SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_common.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_common_lcore.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_common_timer.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_common_memzone.c
diff --git a/lib/librte_eal/bsdapp/eal/eal_debug.c 
b/lib/librte_eal/bsdapp/eal/eal_debug.c
index b0ae2b7..1a17ce3 100644
--- a/lib/librte_eal/bsdapp/eal/eal_debug.c
+++ b/lib/librte_eal/bsdapp/eal/eal_debug.c
@@ -14,6 +14,7 @@
 #include 
 #include 
 #include 
+#include 
 
 #define BACKTRACE_SIZE 256
 
@@ -79,6 +80,10 @@ rte_exit(int exit_code, const char *format, ...)
va_end(ap);
 
 #ifndef RTE_EAL_ALWAYS_PANIC_ON_ERROR
+   int ret = rte_eal_finalize();
+   if (ret)
+   RTE_LOG(CRIT, EAL,
+   "EAL could not release all resources, code %d\n", ret);
exit(exit_code);
 #else
rte_dump_stack();
diff --git a/lib/librte_eal/common/include/rte_eal.h 
b/lib/librte_eal/common/include/rte_eal.h
index 2aba2c8..b2f5869 100644
--- a/lib/librte_eal/common/include/rte_eal.h
+++ b/lib/librte_eal/common/include/rte_eal.h
@@ -170,6 +170,21 @@ int rte_eal_iopl_init(void);
 int rte_eal_init(int argc, char **argv);
 
 /**
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
+ * Finalize the Environment Abstraction Layer (EAL)
+ *
+ * This function must be called to release any internal resources that EAL has
+ * 

[dpdk-dev] [PATCH 3/4] app/pdump: call eal finalize before exit

[Harry van Haaren]

This patch adds a call to the newly introduced finalize()
function just before quitting the pdump app.

Adding this function call before quitting from a secondary processes
is important, as otherwise it will leak hugepage memory. For a secondary
process that is run multiple times, this could cause hugepage memory
to become depleted and stop a secondary process from starting.

Signed-off-by: Harry van Haaren 

---

Cc: tho...@monjalon.net
Cc: vipin.vargh...@intel.com
---
 app/pdump/main.c | 5 +
 1 file changed, 5 insertions(+)

diff --git a/app/pdump/main.c b/app/pdump/main.c
index 0f70c75..6c58cda 100644
--- a/app/pdump/main.c
+++ b/app/pdump/main.c
@@ -882,5 +882,10 @@ main(int argc, char **argv)
/* dump debug stats */
print_pdump_stats();
 
+   ret = rte_eal_finalize();
+   if (ret)
+   printf("Error from rte_eal_finalize(), %d\n", ret);
+
+
return 0;
 }
-- 
2.7.4

[dpdk-dev] [PATCH 4/4] app/proc_info: call eal finalize before exit

[Harry van Haaren]

This patch adds a call to the newly introduced finalize()
function just before quitting the app.

Adding this function call before quitting from a secondary processes
is important, as otherwise it will leak hugepage memory. For a secondary
process that is run multiple times, this could cause hugepage memory
to become depleted and stop a secondary process from starting.

Signed-off-by: Harry van Haaren 

---

Cc: tho...@monjalon.net
Cc: vipin.vargh...@intel.com
---
 app/proc_info/main.c | 4 
 1 file changed, 4 insertions(+)

diff --git a/app/proc_info/main.c b/app/proc_info/main.c
index 94d53f5..44efbc4 100644
--- a/app/proc_info/main.c
+++ b/app/proc_info/main.c
@@ -660,5 +660,9 @@ main(int argc, char **argv)
if (enable_metrics)
metrics_display(RTE_METRICS_GLOBAL);
 
+   ret = rte_eal_finalize();
+   if (ret)
+   printf("Error from rte_eal_finalize(), %d\n", ret);
+
return 0;
 }
-- 
2.7.4

[dpdk-dev] [PATCH 00/11] net/qede: update PMD version to 2.7.0.1

[Rasesh Mody]

This patch set includes some bug fixes and minor enhancements for QEDE PMD.
It also includes NIC guide updates for BNX2X and QEDE PMDs.

Please apply.

Thanks!
-Rasesh

Harish Patil (4):
  net/qede/base: fix VF LRO tunnel params configuration
  net/qede: initialize VF tunnel as enabled on start
  net/qede: fix to check if tunnel L3 header is valid
  net/qede: fix tunnel header size in Tx BD configuration

Rasesh Mody (6):
  net/qede: remove DEBUG INFO config option
  net/qede: fix MTU set and max Rx pkt len usage
  net/qede: fix clearing of queue stats
  doc: update qede guide
  doc: update bnx2x guide
  net/qede: update PMD version

Shahed Shaikh (1):
  net/qede: add check for null return value

 config/common_base|1 -
 doc/guides/nics/bnx2x.rst |2 +-
 doc/guides/nics/qede.rst  |   26 ++---
 drivers/net/qede/base/ecore_vf.c  |6 ++
 drivers/net/qede/base/ecore_vfpf_if.h |2 +
 drivers/net/qede/qede_ethdev.c|  155 +--
 drivers/net/qede/qede_ethdev.h|2 +-
 drivers/net/qede/qede_logs.h  |4 -
 drivers/net/qede/qede_rxtx.c  |  189 +
 drivers/net/qede/qede_rxtx.h  |2 +-
 10 files changed, 265 insertions(+), 124 deletions(-)

-- 
1.7.10.3

[dpdk-dev] [PATCH 01/11] net/qede/base: fix VF LRO tunnel params configuration

[Rasesh Mody]

From: Harish Patil 

Add missing LRO tunnel configuration parameters for VF over sriov channel.

Fixes: ec94dbc57362 ("qede: add base driver")
Cc: sta...@dpdk.org

Signed-off-by: Harish Patil 
---
 drivers/net/qede/base/ecore_vf.c  |6 ++
 drivers/net/qede/base/ecore_vfpf_if.h |2 ++
 2 files changed, 8 insertions(+)

diff --git a/drivers/net/qede/base/ecore_vf.c b/drivers/net/qede/base/ecore_vf.c
index 25109db..e0f2dd5 100644
--- a/drivers/net/qede/base/ecore_vf.c
+++ b/drivers/net/qede/base/ecore_vf.c
@@ -1385,6 +1385,12 @@ enum _ecore_status_t
if (sge_tpa_params->tpa_gro_consistent_flg)
p_sge_tpa_tlv->sge_tpa_flags |=
VFPF_TPA_GRO_CONSIST_FLAG;
+   if (sge_tpa_params->tpa_ipv4_tunn_en_flg)
+   p_sge_tpa_tlv->sge_tpa_flags |=
+   VFPF_TPA_TUNN_IPV4_EN_FLAG;
+   if (sge_tpa_params->tpa_ipv6_tunn_en_flg)
+   p_sge_tpa_tlv->sge_tpa_flags |=
+   VFPF_TPA_TUNN_IPV6_EN_FLAG;
 
p_sge_tpa_tlv->tpa_max_aggs_num =
sge_tpa_params->tpa_max_aggs_num;
diff --git a/drivers/net/qede/base/ecore_vfpf_if.h 
b/drivers/net/qede/base/ecore_vfpf_if.h
index 3ccc766..ecb0064 100644
--- a/drivers/net/qede/base/ecore_vfpf_if.h
+++ b/drivers/net/qede/base/ecore_vfpf_if.h
@@ -424,6 +424,8 @@ struct vfpf_vport_update_sge_tpa_tlv {
#define VFPF_TPA_PKT_SPLIT_FLAG  (1 << 2)
#define VFPF_TPA_HDR_DATA_SPLIT_FLAG (1 << 3)
#define VFPF_TPA_GRO_CONSIST_FLAG(1 << 4)
+   #define VFPF_TPA_TUNN_IPV4_EN_FLAG   (1 << 5)
+   #define VFPF_TPA_TUNN_IPV6_EN_FLAG   (1 << 6)
 
u8  update_sge_tpa_flags;
#define VFPF_UPDATE_SGE_DEPRECATED_FLAG(1 << 0)
-- 
1.7.10.3

[dpdk-dev] [PATCH 02/11] net/qede: initialize VF tunnel as enabled on start

[Rasesh Mody]

From: Harish Patil 

By default, the PF driver enables tunnel offload for its child VF.
So mark tunnel offloads as enabled in the VF driver to reflect the
actual state.

Fixes: 52d94b57e1c7 ("net/qede: add slowpath support for VXLAN tunneling")
Fixes: d378cefab84e ("net/qede: add support for GENEVE tunneling offload")

Cc: sta...@dpdk.org

Signed-off-by: Harish Patil 
---
 drivers/net/qede/qede_ethdev.c |   15 ++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/drivers/net/qede/qede_ethdev.c b/drivers/net/qede/qede_ethdev.c
index 8d82661..8a69d20 100644
--- a/drivers/net/qede/qede_ethdev.c
+++ b/drivers/net/qede/qede_ethdev.c
@@ -3040,9 +3040,22 @@ static int qede_common_dev_init(struct rte_eth_dev 
*eth_dev, bool is_vf)
SLIST_INIT(&adapter->uc_list_head);
adapter->mtu = ETHER_MTU;
adapter->new_mtu = ETHER_MTU;
-   if (!is_vf)
+   if (!is_vf) {
if (qede_start_vport(adapter, adapter->mtu))
return -1;
+   } else {
+   /* VF tunnel offloads is enabled by default in PF driver */
+   adapter->vxlan.enable = true;
+   adapter->vxlan.num_filters = 0;
+   adapter->vxlan.filter_type = ETH_TUNNEL_FILTER_IMAC |
+ETH_TUNNEL_FILTER_IVLAN;
+   adapter->vxlan.udp_port = QEDE_VXLAN_DEF_PORT;
+   adapter->geneve.enable = true;
+   adapter->vxlan.num_filters = 0;
+   adapter->vxlan.filter_type = ETH_TUNNEL_FILTER_IMAC |
+ETH_TUNNEL_FILTER_IVLAN;
+   adapter->vxlan.udp_port = QEDE_GENEVE_DEF_PORT;
+   }
 
DP_INFO(edev, "MAC address : %02x:%02x:%02x:%02x:%02x:%02x\n",
adapter->primary_mac.addr_bytes[0],
-- 
1.7.10.3

[dpdk-dev] [PATCH 03/11] net/qede: fix to check if tunnel L3 header is valid

[Rasesh Mody]

From: Harish Patil 

- Add a check to verify tunnel IP header checksum is valid and mark MBUF
flag as appropriate.
- Bit of refactoring so that inner frame handling for tunneled packets can
be made common as regular (non-tunneled) packets.
- make qede_tunn_exist() as inline.
- remove RTE_PTYPE_L2_ETHER as default L2 pkt_type.

Fixes: 3d4bb4411683 ("net/qede: add fastpath support for VXLAN tunneling")

Cc: sta...@dpdk.org

Signed-off-by: Harish Patil 
---
 drivers/net/qede/qede_rxtx.c |  143 +++---
 1 file changed, 91 insertions(+), 52 deletions(-)

diff --git a/drivers/net/qede/qede_rxtx.c b/drivers/net/qede/qede_rxtx.c
index 184f0e1..2af1bfc 100644
--- a/drivers/net/qede/qede_rxtx.c
+++ b/drivers/net/qede/qede_rxtx.c
@@ -812,12 +812,18 @@ void qede_stop_queues(struct rte_eth_dev *eth_dev)
}
 }
 
-static bool qede_tunn_exist(uint16_t flag)
+static inline bool qede_tunn_exist(uint16_t flag)
 {
return !!((PARSING_AND_ERR_FLAGS_TUNNELEXIST_MASK <<
PARSING_AND_ERR_FLAGS_TUNNELEXIST_SHIFT) & flag);
 }
 
+static inline uint8_t qede_check_tunn_csum_l3(uint16_t flag)
+{
+   return !!((PARSING_AND_ERR_FLAGS_TUNNELIPHDRERROR_MASK <<
+   PARSING_AND_ERR_FLAGS_TUNNELIPHDRERROR_SHIFT) & flag);
+}
+
 /*
  * qede_check_tunn_csum_l4:
  * Returns:
@@ -844,33 +850,51 @@ static inline uint8_t qede_check_notunn_csum_l4(uint16_t 
flag)
return 0;
 }
 
-/* Returns outer L3 and L4 packet_type for tunneled packets */
+/* Returns outer L2, L3 and L4 packet_type for tunneled packets */
 static inline uint32_t qede_rx_cqe_to_pkt_type_outer(struct rte_mbuf *m)
 {
uint32_t packet_type = RTE_PTYPE_UNKNOWN;
struct ether_hdr *eth_hdr;
struct ipv4_hdr *ipv4_hdr;
struct ipv6_hdr *ipv6_hdr;
+   struct vlan_hdr *vlan_hdr;
+   uint16_t ethertype;
+   bool vlan_tagged = 0;
+   uint16_t len;
 
eth_hdr = rte_pktmbuf_mtod(m, struct ether_hdr *);
-   if (eth_hdr->ether_type == rte_cpu_to_be_16(ETHER_TYPE_IPv4)) {
+   len = sizeof(struct ether_hdr);
+   ethertype = rte_cpu_to_be_16(eth_hdr->ether_type);
+
+/* Note: Valid only if VLAN stripping is disabled */
+   if (ethertype == ETHER_TYPE_VLAN) {
+   vlan_tagged = 1;
+   vlan_hdr = (struct vlan_hdr *)(eth_hdr + 1);
+   len += sizeof(struct vlan_hdr);
+   ethertype = rte_cpu_to_be_16(vlan_hdr->eth_proto);
+   }
+
+   if (ethertype == ETHER_TYPE_IPv4) {
packet_type |= RTE_PTYPE_L3_IPV4;
-   ipv4_hdr = rte_pktmbuf_mtod_offset(m, struct ipv4_hdr *,
-  sizeof(struct ether_hdr));
+   ipv4_hdr = rte_pktmbuf_mtod_offset(m, struct ipv4_hdr *, len);
if (ipv4_hdr->next_proto_id == IPPROTO_TCP)
packet_type |= RTE_PTYPE_L4_TCP;
else if (ipv4_hdr->next_proto_id == IPPROTO_UDP)
packet_type |= RTE_PTYPE_L4_UDP;
-   } else if (eth_hdr->ether_type == rte_cpu_to_be_16(ETHER_TYPE_IPv6)) {
+   } else if (ethertype == ETHER_TYPE_IPv6) {
packet_type |= RTE_PTYPE_L3_IPV6;
-   ipv6_hdr = rte_pktmbuf_mtod_offset(m, struct ipv6_hdr *,
-  sizeof(struct ether_hdr));
+   ipv6_hdr = rte_pktmbuf_mtod_offset(m, struct ipv6_hdr *, len);
if (ipv6_hdr->proto == IPPROTO_TCP)
packet_type |= RTE_PTYPE_L4_TCP;
else if (ipv6_hdr->proto == IPPROTO_UDP)
packet_type |= RTE_PTYPE_L4_UDP;
}
 
+   if (vlan_tagged)
+   packet_type |= RTE_PTYPE_L2_ETHER_VLAN;
+   else
+   packet_type |= RTE_PTYPE_L2_ETHER;
+
return packet_type;
 }
 
@@ -1163,17 +1187,17 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
[QEDE_PKT_TYPE_TUNN_GRE] = RTE_PTYPE_TUNNEL_GRE,
[QEDE_PKT_TYPE_TUNN_VXLAN] = RTE_PTYPE_TUNNEL_VXLAN,
[QEDE_PKT_TYPE_TUNN_L2_TENID_NOEXIST_GENEVE] =
-   RTE_PTYPE_TUNNEL_GENEVE | RTE_PTYPE_L2_ETHER,
+   RTE_PTYPE_TUNNEL_GENEVE,
[QEDE_PKT_TYPE_TUNN_L2_TENID_NOEXIST_GRE] =
-   RTE_PTYPE_TUNNEL_GRE | RTE_PTYPE_L2_ETHER,
+   RTE_PTYPE_TUNNEL_GRE,
[QEDE_PKT_TYPE_TUNN_L2_TENID_NOEXIST_VXLAN] =
-   RTE_PTYPE_TUNNEL_VXLAN | RTE_PTYPE_L2_ETHER,
+   RTE_PTYPE_TUNNEL_VXLAN,
[QEDE_PKT_TYPE_TUNN_L2_TENID_EXIST_GENEVE] =
-   RTE_PTYPE_TUNNEL_GENEVE | RTE_PTYPE_L2_ETHER,
+   RTE_PTYPE_TUNNEL_GENEVE,
[QEDE_PKT_TYPE_TUNN_L2_TENID_EXIST_GRE] =
-   RTE_PTYPE_TUNNEL_GRE | RTE_P

[dpdk-dev] [PATCH 07/11] net/qede: fix clearing of queue stats

[Rasesh Mody]

Add support to clear the per queue statistics thereby clearing xstats
counters.
Fixes: 7634c5f91569 ("net/qede: add queue statistics")
Cc: sta...@dpdk.org

Signed-off-by: Rasesh Mody 
---
 drivers/net/qede/qede_ethdev.c |   58 
 1 file changed, 58 insertions(+)

diff --git a/drivers/net/qede/qede_ethdev.c b/drivers/net/qede/qede_ethdev.c
index 895a0da..cab5059 100644
--- a/drivers/net/qede/qede_ethdev.c
+++ b/drivers/net/qede/qede_ethdev.c
@@ -387,6 +387,60 @@ static void qede_print_adapter_info(struct qede_dev *qdev)
DP_INFO(edev, "*\n");
 }
 
+static void qede_reset_queue_stats(struct qede_dev *qdev, bool xstats)
+{
+   struct ecore_dev *edev = QEDE_INIT_EDEV(qdev);
+   unsigned int i = 0, j = 0, qid;
+   unsigned int rxq_stat_cntrs, txq_stat_cntrs;
+   struct qede_tx_queue *txq;
+
+   DP_VERBOSE(edev, ECORE_MSG_DEBUG, "Clearing queue stats\n");
+
+   rxq_stat_cntrs = RTE_MIN(QEDE_RSS_COUNT(qdev),
+  RTE_ETHDEV_QUEUE_STAT_CNTRS);
+   txq_stat_cntrs = RTE_MIN(QEDE_TSS_COUNT(qdev),
+  RTE_ETHDEV_QUEUE_STAT_CNTRS);
+
+   for_each_rss(qid) {
+   OSAL_MEMSET(((char *)(qdev->fp_array[qid].rxq)) +
+offsetof(struct qede_rx_queue, rcv_pkts), 0,
+   sizeof(uint64_t));
+   OSAL_MEMSET(((char *)(qdev->fp_array[qid].rxq)) +
+offsetof(struct qede_rx_queue, rx_hw_errors), 0,
+   sizeof(uint64_t));
+   OSAL_MEMSET(((char *)(qdev->fp_array[qid].rxq)) +
+offsetof(struct qede_rx_queue, rx_alloc_errors), 0,
+   sizeof(uint64_t));
+
+   if (xstats)
+   for (j = 0; j < RTE_DIM(qede_rxq_xstats_strings); j++)
+   OSAL_MEMSETchar *)
+ (qdev->fp_array[qid].rxq)) +
+qede_rxq_xstats_strings[j].offset),
+   0,
+   sizeof(uint64_t));
+
+   i++;
+   if (i == rxq_stat_cntrs)
+   break;
+   }
+
+   i = 0;
+
+   for_each_tss(qid) {
+   txq = qdev->fp_array[qid].txq;
+
+   OSAL_MEMSET((uint64_t *)(uintptr_t)
+   (((uint64_t)(uintptr_t)(txq)) +
+offsetof(struct qede_tx_queue, xmit_pkts)), 0,
+   sizeof(uint64_t));
+
+   i++;
+   if (i == txq_stat_cntrs)
+   break;
+   }
+}
+
 static int
 qede_start_vport(struct qede_dev *qdev, uint16_t mtu)
 {
@@ -412,6 +466,8 @@ static void qede_print_adapter_info(struct qede_dev *qdev)
}
}
ecore_reset_vport_stats(edev);
+   if (IS_PF(edev))
+   qede_reset_queue_stats(qdev, true);
DP_INFO(edev, "VPORT started with MTU = %u\n", mtu);
 
return 0;
@@ -1885,6 +1941,7 @@ static void qede_dev_close(struct rte_eth_dev *eth_dev)
struct ecore_dev *edev = &qdev->edev;
 
ecore_reset_vport_stats(edev);
+   qede_reset_queue_stats(qdev, true);
 }
 
 int qede_dev_set_link_state(struct rte_eth_dev *eth_dev, bool link_up)
@@ -1920,6 +1977,7 @@ static void qede_reset_stats(struct rte_eth_dev *eth_dev)
struct ecore_dev *edev = &qdev->edev;
 
ecore_reset_vport_stats(edev);
+   qede_reset_queue_stats(qdev, false);
 }
 
 static void qede_allmulticast_enable(struct rte_eth_dev *eth_dev)
-- 
1.7.10.3

[dpdk-dev] [PATCH 06/11] net/qede: fix MTU set and max Rx pkt len usage

[Rasesh Mody]

This patch fixes issues related to MTU set and max_rx_pkt_len usage.
 - Adjust MTU during device configuration when jumbo is enabled

 - In qede_set_mtu():
   Return not supported for VF as currently we do not support it.

   Cache new mtu value in mtu_new for proper update.

   Add check for RXQ allocation before calculating RX buffer size
   if not allocated defer RX buffer size calculation till RXQ setup.

   Add check for before performing device start/stop.

 - Use max_rx_pkt_len appropriately

 - Change QEDE_ETH_OVERHEAD macro to adjust driver specifics

Fixes: 4c4bdadfa9e7 ("net/qede: refactoring multi-queue implementation")
Fixes: 9a6d30ae6d46 ("net/qede: refactoring vport handling code")
Fixes: 1ef4c3a5c1f7 ("net/qede: prevent crash while changing MTU dynamically")
Cc: sta...@dpdk.org

Signed-off-by: Rasesh Mody 
---
 drivers/net/qede/qede_ethdev.c |   63 +++-
 drivers/net/qede/qede_rxtx.c   |6 ++--
 drivers/net/qede/qede_rxtx.h   |2 +-
 3 files changed, 47 insertions(+), 24 deletions(-)

diff --git a/drivers/net/qede/qede_ethdev.c b/drivers/net/qede/qede_ethdev.c
index 323e8ed..895a0da 100644
--- a/drivers/net/qede/qede_ethdev.c
+++ b/drivers/net/qede/qede_ethdev.c
@@ -1414,18 +1414,24 @@ static int qede_dev_configure(struct rte_eth_dev 
*eth_dev)
return -ENOMEM;
}
 
+   /* If jumbo enabled adjust MTU */
+   if (eth_dev->data->dev_conf.rxmode.jumbo_frame)
+   eth_dev->data->mtu =
+   eth_dev->data->dev_conf.rxmode.max_rx_pkt_len -
+   ETHER_HDR_LEN - ETHER_CRC_LEN;
+
/* VF's MTU has to be set using vport-start where as
 * PF's MTU can be updated via vport-update.
 */
if (IS_VF(edev)) {
-   if (qede_start_vport(qdev, rxmode->max_rx_pkt_len))
+   if (qede_start_vport(qdev, eth_dev->data->mtu))
return -1;
} else {
-   if (qede_update_mtu(eth_dev, rxmode->max_rx_pkt_len))
+   if (qede_update_mtu(eth_dev, eth_dev->data->mtu))
return -1;
}
 
-   qdev->mtu = rxmode->max_rx_pkt_len;
+   qdev->mtu = eth_dev->data->mtu;
qdev->new_mtu = qdev->mtu;
 
/* Enable VLAN offloads by default */
@@ -2306,16 +2312,23 @@ static int qede_set_mtu(struct rte_eth_dev *dev, 
uint16_t mtu)
struct ecore_dev *edev = QEDE_INIT_EDEV(qdev);
struct rte_eth_dev_info dev_info = {0};
struct qede_fastpath *fp;
+   uint32_t max_rx_pkt_len;
uint32_t frame_size;
uint16_t rx_buf_size;
uint16_t bufsz;
+   bool restart = false;
int i;
 
PMD_INIT_FUNC_TRACE(edev);
+   if (IS_VF(edev))
+   return -ENOTSUP;
qede_dev_info_get(dev, &dev_info);
-   frame_size = mtu + QEDE_ETH_OVERHEAD;
+   max_rx_pkt_len = mtu + ETHER_HDR_LEN + ETHER_CRC_LEN;
+   frame_size = max_rx_pkt_len + QEDE_ETH_OVERHEAD;
if ((mtu < ETHER_MIN_MTU) || (frame_size > dev_info.max_rx_pktlen)) {
-   DP_ERR(edev, "MTU %u out of range\n", mtu);
+   DP_ERR(edev, "MTU %u out of range, %u is maximum allowable\n",
+  mtu, dev_info.max_rx_pktlen - ETHER_HDR_LEN -
+   ETHER_CRC_LEN - QEDE_ETH_OVERHEAD);
return -EINVAL;
}
if (!dev->data->scattered_rx &&
@@ -2329,29 +2342,39 @@ static int qede_set_mtu(struct rte_eth_dev *dev, 
uint16_t mtu)
 */
dev->rx_pkt_burst = qede_rxtx_pkts_dummy;
dev->tx_pkt_burst = qede_rxtx_pkts_dummy;
-   qede_dev_stop(dev);
+   if (dev->data->dev_started) {
+   dev->data->dev_started = 0;
+   qede_dev_stop(dev);
+   restart = true;
+   }
rte_delay_ms(1000);
-   qdev->mtu = mtu;
+   qdev->new_mtu = mtu;
/* Fix up RX buf size for all queues of the port */
for_each_rss(i) {
fp = &qdev->fp_array[i];
-   bufsz = (uint16_t)rte_pktmbuf_data_room_size(
-   fp->rxq->mb_pool) - RTE_PKTMBUF_HEADROOM;
-   if (dev->data->scattered_rx)
-   rx_buf_size = bufsz + QEDE_ETH_OVERHEAD;
-   else
-   rx_buf_size = mtu + QEDE_ETH_OVERHEAD;
-   rx_buf_size = QEDE_CEIL_TO_CACHE_LINE_SIZE(rx_buf_size);
-   fp->rxq->rx_buf_size = rx_buf_size;
-   DP_INFO(edev, "buf_size adjusted to %u\n", rx_buf_size);
-   }
-   qede_dev_start(dev);
-   if (frame_size > ETHER_MAX_LEN)
+   if (fp->rxq != NULL) {
+   bufsz = (uint16_t)rte_pktmbuf_data_room_size(
+   fp->rxq->mb_pool) - RTE_PKTMBUF_HEADROOM;
+   if (dev->data->scattered_rx)
+   rx_buf_size = bufsz + ETHER_HDR_LEN +
+  

[dpdk-dev] [PATCH 04/11] net/qede: fix tunnel header size in Tx BD configuration

[Rasesh Mody]

From: Harish Patil 

- Fix incorrect header size. In the tunnel case, the outer L2/L3 lengths
should be included to calculate tunnel header_size.
- In TSO case, skip manipulating TX BD1 and TX BD2 data buffer fields
since those fields are already updated with header and payload lengths
respectively.
- Update TX BD debug data collection.

Fixes: 3d4bb4411683 ("net/qede: add fastpath support for VXLAN tunneling")

Cc: sta...@dpdk.org

Signed-off-by: Harish Patil 
---
 drivers/net/qede/qede_rxtx.c |   40 +---
 1 file changed, 25 insertions(+), 15 deletions(-)

diff --git a/drivers/net/qede/qede_rxtx.c b/drivers/net/qede/qede_rxtx.c
index 2af1bfc..df248cf 100644
--- a/drivers/net/qede/qede_rxtx.c
+++ b/drivers/net/qede/qede_rxtx.c
@@ -1588,7 +1588,8 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
 /* Populate scatter gather buffer descriptor fields */
 static inline uint16_t
 qede_encode_sg_bd(struct qede_tx_queue *p_txq, struct rte_mbuf *m_seg,
- struct eth_tx_2nd_bd **bd2, struct eth_tx_3rd_bd **bd3)
+ struct eth_tx_2nd_bd **bd2, struct eth_tx_3rd_bd **bd3,
+ uint16_t start_seg)
 {
struct qede_tx_queue *txq = p_txq;
struct eth_tx_bd *tx_bd = NULL;
@@ -1597,7 +1598,7 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
 
/* Check for scattered buffers */
while (m_seg) {
-   if (nb_segs == 0) {
+   if (start_seg == 0) {
if (!*bd2) {
*bd2 = (struct eth_tx_2nd_bd *)
ecore_chain_produce(&txq->tx_pbl);
@@ -1607,7 +1608,7 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
mapping = rte_mbuf_data_iova(m_seg);
QEDE_BD_SET_ADDR_LEN(*bd2, mapping, m_seg->data_len);
PMD_TX_LOG(DEBUG, txq, "BD2 len %04x", m_seg->data_len);
-   } else if (nb_segs == 1) {
+   } else if (start_seg == 1) {
if (!*bd3) {
*bd3 = (struct eth_tx_3rd_bd *)
ecore_chain_produce(&txq->tx_pbl);
@@ -1645,20 +1646,24 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
 
if (bd1)
PMD_TX_LOG(INFO, txq,
-  "BD1: nbytes=%u nbds=%u bd_flags=%04x bf=%04x",
-  rte_cpu_to_le_16(bd1->nbytes), bd1->data.nbds,
-  bd1->data.bd_flags.bitfields,
-  rte_cpu_to_le_16(bd1->data.bitfields));
+  "BD1: nbytes=0x%04x nbds=0x%04x bd_flags=0x%04x bf=0x%04x",
+  rte_cpu_to_le_16(bd1->nbytes), bd1->data.nbds,
+  bd1->data.bd_flags.bitfields,
+  rte_cpu_to_le_16(bd1->data.bitfields));
if (bd2)
PMD_TX_LOG(INFO, txq,
-  "BD2: nbytes=%u bf=%04x\n",
-  rte_cpu_to_le_16(bd2->nbytes), bd2->data.bitfields1);
+  "BD2: nbytes=0x%04x bf1=0x%04x bf2=0x%04x tunn_ip=0x%04x\n",
+  rte_cpu_to_le_16(bd2->nbytes), bd2->data.bitfields1,
+  bd2->data.bitfields2, bd2->data.tunn_ip_size);
if (bd3)
PMD_TX_LOG(INFO, txq,
-  "BD3: nbytes=%u bf=%04x mss=%u\n",
-  rte_cpu_to_le_16(bd3->nbytes),
-  rte_cpu_to_le_16(bd3->data.bitfields),
-  rte_cpu_to_le_16(bd3->data.lso_mss));
+  "BD3: nbytes=0x%04x bf=0x%04x MSS=0x%04x "
+  "tunn_l4_hdr_start_offset_w=0x%04x tunn_hdr_size=0x%04x\n",
+  rte_cpu_to_le_16(bd3->nbytes),
+  rte_cpu_to_le_16(bd3->data.bitfields),
+  rte_cpu_to_le_16(bd3->data.lso_mss),
+  bd3->data.tunn_l4_hdr_start_offset_w,
+  bd3->data.tunn_hdr_size_w);
 
rte_get_tx_ol_flag_list(tx_ol_flags, ol_buf, sizeof(ol_buf));
PMD_TX_LOG(INFO, txq, "TX offloads = %s\n", ol_buf);
@@ -1938,6 +1943,10 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
 * and BD2 onwards for data.
 */
hdr_size = mbuf->l2_len + mbuf->l3_len + mbuf->l4_len;
+   if (tunn_flg)
+   hdr_size += mbuf->outer_l2_len +
+   mbuf->outer_l3_len;
+
bd1_bd_flags_bf |= 1 << ETH_TX_1ST_BD_FLAGS_LSO_SHIFT;
bd1_bd_flags_bf |=
1 << ETH_TX_1ST_BD_FLAGS_IP_CSUM_SHIFT;
@@ -2054,9 +2063,11 @@ static inline uint32_t 
qede_rx_cqe_to_tunn_pkt_type(uint16_t flags)
 
/* Handle fragmented MBUF */
m_seg = mbuf->ne

[dpdk-dev] [PATCH 05/11] net/qede: remove DEBUG INFO config option

[Rasesh Mody]

With dynamic logging, we no longer need DEBUG INFO config option.

Signed-off-by: Rasesh Mody 
---
 config/common_base |1 -
 doc/guides/nics/qede.rst   |   10 +-
 drivers/net/qede/qede_ethdev.c |8 
 drivers/net/qede/qede_logs.h   |4 
 4 files changed, 1 insertion(+), 22 deletions(-)

diff --git a/config/common_base b/config/common_base
index 170a389..d851161 100644
--- a/config/common_base
+++ b/config/common_base
@@ -317,7 +317,6 @@ CONFIG_RTE_LIBRTE_NFP_DEBUG_RX=n
 # QLogic 10G/25G/40G/50G/100G PMD
 #
 CONFIG_RTE_LIBRTE_QEDE_PMD=y
-CONFIG_RTE_LIBRTE_QEDE_DEBUG_INFO=n
 CONFIG_RTE_LIBRTE_QEDE_DEBUG_TX=n
 CONFIG_RTE_LIBRTE_QEDE_DEBUG_RX=n
 #Provides abs path/name of the firmware file.
diff --git a/doc/guides/nics/qede.rst b/doc/guides/nics/qede.rst
index b958a5b..43b0f3f 100644
--- a/doc/guides/nics/qede.rst
+++ b/doc/guides/nics/qede.rst
@@ -87,14 +87,6 @@ enabling debugging options may affect system performance.
 
   Toggle compilation of QEDE PMD driver.
 
-- ``CONFIG_RTE_LIBRTE_QEDE_DEBUG_INFO`` (default **n**)
-
-  Toggle display of generic debugging messages.
-
-- ``CONFIG_RTE_LIBRTE_QEDE_DEBUG_DRIVER`` (default **n**)
-
-  Toggle display of ecore related messages.
-
 - ``CONFIG_RTE_LIBRTE_QEDE_DEBUG_TX`` (default **n**)
 
   Toggle display of transmit fast path run-time messages.
@@ -199,7 +191,7 @@ This section provides instructions to configure SR-IOV with 
Linux OS.
 
 
 #. Running testpmd
-   (Enable QEDE_DEBUG_INFO=y to view informational messages):
+   (Supply ``--log-level="pmd.qede.driver",7`` to view informational messages):
 
Refer to the document
:ref:`compiling and testing a PMD for a NIC ` to run
diff --git a/drivers/net/qede/qede_ethdev.c b/drivers/net/qede/qede_ethdev.c
index 8a69d20..323e8ed 100644
--- a/drivers/net/qede/qede_ethdev.c
+++ b/drivers/net/qede/qede_ethdev.c
@@ -357,7 +357,6 @@ static void qede_interrupt_action(struct ecore_hwfn *p_hwfn)
qdev->ops = qed_ops;
 }
 
-#ifdef RTE_LIBRTE_QEDE_DEBUG_INFO
 static void qede_print_adapter_info(struct qede_dev *qdev)
 {
struct ecore_dev *edev = &qdev->edev;
@@ -387,7 +386,6 @@ static void qede_print_adapter_info(struct qede_dev *qdev)
DP_INFO(edev, " Firmware file : %s\n", fw_file);
DP_INFO(edev, "*\n");
 }
-#endif
 
 static int
 qede_start_vport(struct qede_dev *qdev, uint16_t mtu)
@@ -1149,9 +1147,7 @@ static void qede_prandom_bytes(uint32_t *buff)
 int qede_config_rss(struct rte_eth_dev *eth_dev)
 {
struct qede_dev *qdev = QEDE_INIT_QDEV(eth_dev);
-#ifdef RTE_LIBRTE_QEDE_DEBUG_INFO
struct ecore_dev *edev = QEDE_INIT_EDEV(qdev);
-#endif
uint32_t def_rss_key[ECORE_RSS_KEY_SIZE];
struct rte_eth_rss_reta_entry64 reta_conf[2];
struct rte_eth_rss_conf rss_conf;
@@ -1295,9 +1291,7 @@ static int qede_args_check(const char *key, const char 
*val, void *opaque)
int ret = 0;
struct rte_eth_dev *eth_dev = opaque;
struct qede_dev *qdev = QEDE_INIT_QDEV(eth_dev);
-#ifdef RTE_LIBRTE_QEDE_DEBUG_INFO
struct ecore_dev *edev = QEDE_INIT_EDEV(qdev);
-#endif
 
errno = 0;
tmp = strtoul(val, NULL, 0);
@@ -3027,9 +3021,7 @@ static int qede_common_dev_init(struct rte_eth_dev 
*eth_dev, bool is_vf)
eth_dev->dev_ops = (is_vf) ? &qede_eth_vf_dev_ops : &qede_eth_dev_ops;
 
if (do_once) {
-#ifdef RTE_LIBRTE_QEDE_DEBUG_INFO
qede_print_adapter_info(adapter);
-#endif
do_once = false;
}
 
diff --git a/drivers/net/qede/qede_logs.h b/drivers/net/qede/qede_logs.h
index b7ecf1c..159315e 100644
--- a/drivers/net/qede/qede_logs.h
+++ b/drivers/net/qede/qede_logs.h
@@ -34,16 +34,12 @@
##__VA_ARGS__); \
 } while (0)
 
-#ifdef RTE_LIBRTE_QEDE_DEBUG_INFO
 #define DP_INFO(p_dev, fmt, ...) \
rte_log(RTE_LOG_INFO, qede_logtype_driver, \
"[%s:%d(%s)]" fmt, \
__func__, __LINE__, \
(p_dev)->name ? (p_dev)->name : "", \
##__VA_ARGS__)
-#else
-#define DP_INFO(p_dev, fmt, ...) do { } while (0)
-#endif
 
 #define DP_VERBOSE(p_dev, module, fmt, ...)\
do {\
-- 
1.7.10.3

[dpdk-dev] [PATCH 08/11] doc: update qede guide

[Rasesh Mody]

Signed-off-by: Rasesh Mody 
---
 doc/guides/nics/qede.rst |   16 +---
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/doc/guides/nics/qede.rst b/doc/guides/nics/qede.rst
index 43b0f3f..0ca5084 100644
--- a/doc/guides/nics/qede.rst
+++ b/doc/guides/nics/qede.rst
@@ -35,14 +35,20 @@ Supported Features
 - N-tuple filter and flow director (limited support)
 - NPAR (NIC Partitioning)
 - SR-IOV VF
-- VXLAN tunneling offload
-- MPLSoUDP Tx tunnel offload
+- VXLAN Tunneling offload
+- GENEVE Tunneling offload
+- MPLSoUDP Tx Tunneling offload
 
 Non-supported Features
 --
 
 - SR-IOV PF
-- GENEVE and NVGRE Tunneling offloads
+- GRE and NVGRE Tunneling offloads
+
+Co-existence considerations
+---
+- QLogic FastLinQ QL4 CNAs can have both NIC and Storage personalities. 
However, coexistence with storage protocol drivers (qedi and qedf) is not 
supported on the same adapter. So storage personality has to be disabled on 
that adapter when used in DPDK applications.
+- For SR-IOV case, qede PMD will be used to bind to SR-IOV VF device and Linux 
native kernel driver (qede) will be attached to SR-IOV PF.
 
 Supported QLogic Adapters
 -
@@ -95,10 +101,6 @@ enabling debugging options may affect system performance.
 
   Toggle display of receive fast path run-time messages.
 
-- ``CONFIG_RTE_LIBRTE_QEDE_VF_TX_SWITCH`` (default **"y"**)
-
-  A knob to control per-VF Tx switching feature.
-
 - ``CONFIG_RTE_LIBRTE_QEDE_FW`` (default **""**)
 
   Gives absolute path of firmware file.
-- 
1.7.10.3

[dpdk-dev] [PATCH 09/11] doc: update bnx2x guide

[Rasesh Mody]

Signed-off-by: Rasesh Mody 
---
 doc/guides/nics/bnx2x.rst |2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/doc/guides/nics/bnx2x.rst b/doc/guides/nics/bnx2x.rst
index b1fec8a..31f146a 100644
--- a/doc/guides/nics/bnx2x.rst
+++ b/doc/guides/nics/bnx2x.rst
@@ -84,7 +84,7 @@ Prerequisites
 
 - Requires firmware version **7.2.51.0**. It is included in most of the
   standard Linux distros. If it is not available visit
-  `QLogic Driver Download Center `_
+  `linux-firmware git repository 
`_
   to get the required firmware.
 
 Pre-Installation Configuration
-- 
1.7.10.3

[dpdk-dev] [PATCH 11/11] net/qede: update PMD version

[Rasesh Mody]

Signed-off-by: Rasesh Mody 
---
 drivers/net/qede/qede_ethdev.h |2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/qede/qede_ethdev.h b/drivers/net/qede/qede_ethdev.h
index cb95973..c7bb415 100644
--- a/drivers/net/qede/qede_ethdev.h
+++ b/drivers/net/qede/qede_ethdev.h
@@ -45,7 +45,7 @@
 /* Driver versions */
 #define QEDE_PMD_VER_PREFIX"QEDE PMD"
 #define QEDE_PMD_VERSION_MAJOR 2
-#define QEDE_PMD_VERSION_MINOR 6
+#define QEDE_PMD_VERSION_MINOR 7
 #define QEDE_PMD_VERSION_REVISION   0
 #define QEDE_PMD_VERSION_PATCH 1
 
-- 
1.7.10.3

[dpdk-dev] [PATCH 10/11] net/qede: add check for null return value

[Rasesh Mody]

From: Shahed Shaikh 

Test the return value of ecore_ptt_acquire for NULL.

Fixes: d378cefab84e ("net/qede: add support for GENEVE tunneling offload")
Coverity issue: 257049

Signed-off-by: Shahed Shaikh 
---
 drivers/net/qede/qede_ethdev.c |   11 ++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/drivers/net/qede/qede_ethdev.c b/drivers/net/qede/qede_ethdev.c
index cab5059..c3cea6d 100644
--- a/drivers/net/qede/qede_ethdev.c
+++ b/drivers/net/qede/qede_ethdev.c
@@ -681,7 +681,16 @@ static void qede_set_ucast_cmn_params(struct 
ecore_filter_ucast *ucast)
 
for_each_hwfn(edev, i) {
p_hwfn = &edev->hwfns[i];
-   p_ptt = IS_PF(edev) ? ecore_ptt_acquire(p_hwfn) : NULL;
+   if (IS_PF(edev)) {
+   p_ptt = ecore_ptt_acquire(p_hwfn);
+   if (!p_ptt) {
+   DP_ERR(p_hwfn, "Can't acquire PTT\n");
+   return -EAGAIN;
+   }
+   } else {
+   p_ptt = NULL;
+   }
+
rc = ecore_sp_pf_update_tunn_cfg(p_hwfn, p_ptt,
tunn_info, ECORE_SPQ_MODE_CB, NULL);
if (IS_PF(edev))
-- 
1.7.10.3

Re: [dpdk-dev] [PATCH v8 1/3] eal: introduce integer divide through reciprocal

[Thomas Monjalon]

26/01/2018 06:04, Pavan Nikhilesh:
> In some use cases of integer division, denominator remains constant and
> numerator varies. It is possible to optimize division for such specific
> scenarios.
> 
> The librte_sched uses rte_reciprocal to optimize division so, moving it to
> eal/common would allow other libraries and applications to use it.
> 
> Signed-off-by: Pavan Nikhilesh 

Series applied, thanks

Re: [dpdk-dev] [dpdk-stable] [PATCH v1] net/mlx4: fix single port configuration

[Shahaf Shuler]

Friday, January 26, 2018 6:44 PM,  Adrien Mazarguil:
> Hi Ophir,
> 
> On Tue, Jan 23, 2018 at 11:32:47PM +, Ophir Munk wrote:
> > The number of mlx4 present ports is calculated as follows:
> > conf.ports.present |= (UINT64_C(1) << device_attr.phys_port_cnt) - 1;
> >
> > That is - all ones sequence (due to -1 subtraction) When retrieving
> > the number of ports, 1 must be added in order to obtain the correct
> > number of ports to the power of 2, as follows:
> > uint32_t ports = rte_log2_u32(conf->ports.present + 1); If 1 was not
> > added, in the case of one port, the number of ports would be falsely
> > calculated as 0.
> >
> > Fixes: 82642799 ("net/mlx4: check max number of ports dynamically")
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: Ophir Munk 
> 
> I was too busy to reply earlier, thanks for taking care of this issue.
> 
> Acked-by: Adrien Mazarguil 

Applied to next-net-mlx, thanks. 

> 
> --
> Adrien Mazarguil
> 6WIND

Re: [dpdk-dev] [PATCH v4 0/9] introduce coherent I/O memory barriers

[Thomas Monjalon]

25/01/2018 22:02, Yongseok Koh:
> This patchset is to introduce coherent I/O memory barriers, which could be 
> more
> efficient for coherent memory between I/O device and CPU, especially for 
> ARMv8.
> 
> v4:
> * rename barriers to "coherent I/O memory barrier".
> * Make groups for various barriers in Doxygen doc.
> 
> v3:
> * add more detailed comments about the new memory barriers.
> 
> v2:
> * introduce DMA memory barriers.
> 
> Yongseok Koh (9):
>   eal: add Doxygen grouping for memory barriers
>   eal: introduce coherent I/O memory barriers
>   eal/x86: define coherent I/O memory barriers
>   eal/ppc64: define coherent I/O memory barriers
>   eal/armv7: define coherent I/O memory barriers
>   eal/arm64: define coherent I/O memory barriers
>   net/mlx5: remove unnecessary memory barrier
>   net/mlx5: replace I/O memory barrier with coherent version
>   net/mlx5: fix synchronization on polling Rx completions

Applied, thanks