Re: [dpdk-dev] [PATCH v2] config/ppc: ignore gcc 11 psabi warnings

[David Marchand]

On Fri, Sep 3, 2021 at 1:53 AM David Christensen  
wrote:
>
> Suppress the gcc warning "note: the layout of aggregates containing
> vectors with 4-byte alignment has changed in GCC 5" on POWER systems
> by setting "-Wno-psabi".  Warning was originally added to gcc in
> commit https://gcc.gnu.org/git/gitweb.cgi?p=gcc.git;h=9832651 to warn
> of the vector alignment changes introduced in GCC 5.  Older gcc
> versions forced vector alignment to 16 bytes due to requirements for
> POWER 6 and earlier CPUs, but these restrictions don't apply to CPUs
> supported by DPDK.
>
> Bugzilla ID: 739

I guess the intent was to have this backported.
So added explicit Cc: sta...@dpdk.org

>
> Signed-off-by: David Christensen 
> ---
> v2:
> - update copyright year
> - rebase for 21.11-rc0
> ---
>  config/ppc/meson.build | 7 +++
>  1 file changed, 7 insertions(+)
>
> diff --git a/config/ppc/meson.build b/config/ppc/meson.build
> index adf49e1f42..5354db4e0a 100644
> --- a/config/ppc/meson.build
> +++ b/config/ppc/meson.build
> @@ -1,5 +1,6 @@
>  # SPDX-License-Identifier: BSD-3-Clause
>  # Copyright(c) 2018 Luca Boccassi 
> +# Copyright(c) 2021 IBM Corporation
>
>  if not dpdk_conf.get('RTE_ARCH_64')
>  error('Only 64-bit compiles are supported for this platform type')
> @@ -17,6 +18,12 @@ if not power9_supported
>  dpdk_conf.set('RTE_MACHINE','power8')
>  endif
>
> +# Suppress the gcc warning "note: the layout of aggregates containing
> +# vectors with 4-byte alignment has changed in GCC 5".
> +if cc.get_id() == 'gcc' and cc.version().version_compare('>=10.0') and 
> cc.version().version_compare('<12.0') and cc.has_argument('-Wno-psabi')

Wrapped this to next line.


> +add_project_arguments('-Wno-psabi', language: 'c')
> +endif
> +
>  # Certain POWER9 systems can scale as high as 1536 LCORES, but setting such a
>  # high value can waste memory, cause timeouts in time limited autotests, and 
> is
>  # unlikely to be used in many production situations.  Similarly, keeping the
> --
> 2.27.0
>

Applied, thanks.


-- 
David Marchand

Re: [dpdk-dev] [PATCH] eal/ppc: ignore gcc 10 stringop-overflow warnings

[David Marchand]

On Fri, Sep 3, 2021 at 12:15 AM David Christensen
 wrote:
>
> Suppress gcc warning "warning: writing 16 bytes into a region of
> size 0" for users of the POWER rte_memcpy() function.  Existing
> rte_memcpy() code takes different code paths based on the actual
> size of the move so the warning is already addressed. See also
> commit b5b3ea803e47 ("eal/x86: ignore gcc 10 stringop-overflow warnings")
>

Cc: sta...@dpdk.org

> Signed-off-by: David Christensen 

Applied, thanks.


-- 
David Marchand

Re: [dpdk-dev] [EXT] Re: [PATCH] RFC: ethdev: add reassembly offload

[Andrew Rybchenko]

On 9/13/21 9:56 AM, Xu, Rosen wrote:
> Hi,
> 
>> -Original Message-
>> From: Anoob Joseph 
>> Sent: Wednesday, September 08, 2021 18:30
>> To: Yigit, Ferruh ; Xu, Rosen ;
>> Andrew Rybchenko 
>> Cc: Nicolau, Radu ; Doherty, Declan
>> ; hemant.agra...@nxp.com;
>> ma...@nvidia.com; Ananyev, Konstantin ;
>> tho...@monjalon.net; Ankur Dwivedi ;
>> andrew.rybche...@oktetlabs.ru; Akhil Goyal ;
>> dev@dpdk.org
>> Subject: RE: [EXT] Re: [PATCH] RFC: ethdev: add reassembly offload
>>
>> Hi Ferruh, Rosen, Andrew,
>>
>> Please see inline.
>>
>> Thanks,
>> Anoob
>>
>>> Subject: [EXT] Re: [PATCH] RFC: ethdev: add reassembly offload
>>>
>>> External Email
>>>
>>> --
>>> On 8/23/2021 11:02 AM, Akhil Goyal wrote:
 Reassembly is a costly operation if it is done in software, however,
 if it is offloaded to HW, it can considerably save application cycles.
 The operation becomes even more costlier if IP fragmants are
 encrypted.

 To resolve above two issues, a new offload
>>> DEV_RX_OFFLOAD_REASSEMBLY
 is introduced in ethdev for devices which can attempt reassembly of
 packets in hardware.
 rte_eth_dev_info is added with the reassembly capabilities which a
 device can support.
 Now, if IP fragments are encrypted, reassembly can also be attempted
 while doing inline IPsec processing.
 This is controlled by a flag in rte_security_ipsec_sa_options to
 enable reassembly of encrypted IP fragments in the inline path.

 The resulting reassembled packet would be a typical segmented mbuf
 in case of success.

 And if reassembly of fragments is failed or is incomplete (if
 fragments do not come before the reass_timeout), the mbuf is updated
 with an ol_flag PKT_RX_REASSEMBLY_INCOMPLETE and mbuf is returned
>>> as
 is. Now application may decide the fate of the packet to wait more
 for fragments to come or drop.

 Signed-off-by: Akhil Goyal 
 ---
  lib/ethdev/rte_ethdev.c |  1 +
  lib/ethdev/rte_ethdev.h | 18 +-
  lib/mbuf/rte_mbuf_core.h|  3 ++-
  lib/security/rte_security.h | 10 ++
  4 files changed, 30 insertions(+), 2 deletions(-)

 diff --git a/lib/ethdev/rte_ethdev.c b/lib/ethdev/rte_ethdev.c index
 9d95cd11e1..1ab3a093cf 100644
 --- a/lib/ethdev/rte_ethdev.c
 +++ b/lib/ethdev/rte_ethdev.c
 @@ -119,6 +119,7 @@ static const struct {
RTE_RX_OFFLOAD_BIT2STR(VLAN_FILTER),
RTE_RX_OFFLOAD_BIT2STR(VLAN_EXTEND),
RTE_RX_OFFLOAD_BIT2STR(JUMBO_FRAME),
 +  RTE_RX_OFFLOAD_BIT2STR(REASSEMBLY),
RTE_RX_OFFLOAD_BIT2STR(SCATTER),
RTE_RX_OFFLOAD_BIT2STR(TIMESTAMP),
RTE_RX_OFFLOAD_BIT2STR(SECURITY),
 diff --git a/lib/ethdev/rte_ethdev.h b/lib/ethdev/rte_ethdev.h index
 d2b27c351f..e89a4dc1eb 100644
 --- a/lib/ethdev/rte_ethdev.h
 +++ b/lib/ethdev/rte_ethdev.h
 @@ -1360,6 +1360,7 @@ struct rte_eth_conf {
  #define DEV_RX_OFFLOAD_VLAN_FILTER0x0200
  #define DEV_RX_OFFLOAD_VLAN_EXTEND0x0400
  #define DEV_RX_OFFLOAD_JUMBO_FRAME0x0800
 +#define DEV_RX_OFFLOAD_REASSEMBLY 0x1000
>>>
>>> previous '0x1000' was 'DEV_RX_OFFLOAD_CRC_STRIP', it has been
>> long
>>> that offload has been removed, but not sure if it cause any problem to
>>> re- use it.
>>>
  #define DEV_RX_OFFLOAD_SCATTER0x2000
  /**
   * Timestamp is set by the driver in
>>> RTE_MBUF_DYNFIELD_TIMESTAMP_NAME
 @@ -1477,6 +1478,20 @@ struct rte_eth_dev_portconf {
   */
  #define RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID
>>> (UINT16_MAX)

 +/**
 + * Reassembly capabilities that a device can support.
 + * The device which can support reassembly offload should set
 + * DEV_RX_OFFLOAD_REASSEMBLY
 + */
 +struct rte_eth_reass_capa {
 +  /** Maximum time in ns that a fragment can wait for further
>>> fragments */
 +  uint64_t reass_timeout;
 +  /** Maximum number of fragments that device can reassemble */
 +  uint16_t max_frags;
 +  /** Reserved for future capabilities */
 +  uint16_t reserved[3];
 +};
 +
>>>
>>> I wonder if there is any other hardware around supports reassembly
>>> offload, it would be good to get more feedback on the capabilities list.
>>>
  /**
   * Ethernet device associated switch information
   */
 @@ -1582,8 +1597,9 @@ struct rte_eth_dev_info {
 * embedded managed interconnect/switch.
 */
struct rte_eth_switch_info switch_info;
 +  /* Reassembly capabilities of a device for reassembly offload */
 +  struct rte_eth_reass_capa reass_capa;

 -  uint64_t reserved_64s[2]; /**< Reserved for future fields */
>>>
>>> Reserved fields were added to be able to update the struct without
>>> breaking the ABI, so that a critical change doesn't have to wait un

Re: [dpdk-dev] [PATCH] checkpatches: suppress warning about strlcpy

[Bruce Richardson]

On Fri, Sep 10, 2021 at 10:54:43AM -0700, Stephen Hemminger wrote:
> Latest versions of Linux kernel checkpatch now complain about
> uses of strlcpy (kernel perfers strcspy).
> 
> WARNING:STRLCPY: Prefer strscpy over strlcpy
> 
> But DPDK does not have strcspy so suppress the warning.
   ^^^
typo :-)
> 
> Signed-off-by: Stephen Hemminger 

We actually do have an rte_strscpy function, but on the other hand I
believe that strlcpy should very much also be acceptable.

Acked-by: Bruce Richardson 
> ---
>  devtools/checkpatches.sh | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/devtools/checkpatches.sh b/devtools/checkpatches.sh
> index c314d83a29d3..057208c8e1ea 100755
> --- a/devtools/checkpatches.sh
> +++ b/devtools/checkpatches.sh
> @@ -33,7 +33,7 @@ VOLATILE,PREFER_PACKED,PREFER_ALIGNED,PREFER_PRINTF,\
>  PREFER_KERNEL_TYPES,PREFER_FALLTHROUGH,BIT_MACRO,CONST_STRUCT,\
>  SPLIT_STRING,LONG_LINE_STRING,C99_COMMENT_TOLERANCE,\
>  LINE_SPACING,PARENTHESIS_ALIGNMENT,NETWORKING_BLOCK_COMMENT_STYLE,\
> -NEW_TYPEDEFS,COMPARISON_TO_NULL"
> +NEW_TYPEDEFS,COMPARISON_TO_NULL,STRLCPY"
>  options="$options $DPDK_CHECKPATCH_OPTIONS"
>  
>  print_usage () {
> -- 
> 2.30.2
> 

[dpdk-dev] [PATCH] eal: promote rte_mcfg_get_single_file_segment to stable ABI

[Jakub Grajciar -X (jgrajcia - PANTHEON TECH SRO at Cisco)]

Signed-off-by: Jakub Grajciar jgraj...@cisco.com
---
 doc/guides/rel_notes/release_21_11.rst | 3 +++
 lib/eal/include/rte_eal_memconfig.h| 4 
 lib/eal/version.map| 4 +---
 3 files changed, 4 insertions(+), 7 deletions(-)

diff --git a/doc/guides/rel_notes/release_21_11.rst 
b/doc/guides/rel_notes/release_21_11.rst
index 675b573834..1611562f6a 100644
--- a/doc/guides/rel_notes/release_21_11.rst
+++ b/doc/guides/rel_notes/release_21_11.rst
@@ -107,6 +107,9 @@ ABI Changes
Also, make sure to start the actual text at the margin.
===

+* The experimental function ``rte_mcfg_get_single_file_segments`` has been
+  promoted to stable ABI.
+

 Known Issues
 
diff --git a/lib/eal/include/rte_eal_memconfig.h 
b/lib/eal/include/rte_eal_memconfig.h
index dede2ee324..d61023d80a 100644
--- a/lib/eal/include/rte_eal_memconfig.h
+++ b/lib/eal/include/rte_eal_memconfig.h
@@ -112,13 +112,9 @@ void
 rte_mcfg_timer_unlock(void);

 /**
- * @warning
- * @b EXPERIMENTAL: this API may change without prior notice
- *
  * If true, pages are put in single files (per memseg list),
  * as opposed to creating a file per page.
  */
-__rte_experimental
 bool
 rte_mcfg_get_single_file_segments(void);

diff --git a/lib/eal/version.map b/lib/eal/version.map
index beeb986adc..0cb757a1cf 100644
--- a/lib/eal/version.map
+++ b/lib/eal/version.map
@@ -121,6 +121,7 @@ DPDK_22 {
rte_malloc_socket;
rte_malloc_validate;
rte_malloc_virt2iova;
+   rte_mcfg_get_single_file_segments;
rte_mcfg_mem_read_lock;
rte_mcfg_mem_read_unlock;
rte_mcfg_mem_write_lock;
@@ -328,9 +329,6 @@ EXPERIMENTAL {
rte_mcfg_timer_unlock;
rte_rand_max; # WINDOWS_NO_EXPORT

-   # added in 19.11
-   rte_mcfg_get_single_file_segments;
-
# added in 20.02
rte_thread_is_intr;

--
2.25.1

[dpdk-dev] Questions about rte_eth_link_speed_to_str API

[Min Hu (Connor)]

Hi all,
I have questions about rte_eth_link_speed_to_str API.
The API converts link speed to string for display, But it only
supports the following speeds, like that:
case ETH_SPEED_NUM_NONE: return "None";
case ETH_SPEED_NUM_10M:  return "10 Mbps";
case ETH_SPEED_NUM_100M: return "100 Mbps";
case ETH_SPEED_NUM_1G:   return "1 Gbps";
case ETH_SPEED_NUM_2_5G: return "2.5 Gbps";
case ETH_SPEED_NUM_5G:   return "5 Gbps";
case ETH_SPEED_NUM_10G:  return "10 Gbps";
case ETH_SPEED_NUM_20G:  return "20 Gbps";
case ETH_SPEED_NUM_25G:  return "25 Gbps";
case ETH_SPEED_NUM_40G:  return "40 Gbps";
case ETH_SPEED_NUM_50G:  return "50 Gbps";
case ETH_SPEED_NUM_56G:  return "56 Gbps";
case ETH_SPEED_NUM_100G: return "100 Gbps";
case ETH_SPEED_NUM_200G: return "200 Gbps";
case ETH_SPEED_NUM_UNKNOWN: return "Unknown";
default: return "Invalid";

In some cases, like bonding, for example, three slaves which
link speed are 10Gbps, so link speed of bonding port will be
30Gbps, but it shows "Invalid".

Is this reasonable？ any comments will be welcome.

Best wishes

[dpdk-dev] [PATCH v2] net/iavf: fix QFI fields of GTPU UL and DL for FDIR

[Lingyu Liu]

Fix QFI fields matching of GTPU UL/DL for FDIR.

Fixes: 78e8a87f6324 ("net/iavf: fix GTPU UL and DL support for flow director")
Cc: sta...@dpdk.org
Signed-off-by: Junfeng Guo 
Signed-off-by: Lingyu Liu 
---
V2:
 * Fix coding style issue
---
 drivers/net/iavf/iavf_fdir.c | 12 +++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/net/iavf/iavf_fdir.c b/drivers/net/iavf/iavf_fdir.c
index 32b06044f2..560589a496 100644
--- a/drivers/net/iavf/iavf_fdir.c
+++ b/drivers/net/iavf/iavf_fdir.c
@@ -1171,7 +1171,17 @@ iavf_fdir_parse_pattern(__rte_unused struct iavf_adapter 
*ad,
if (gtp_psc_spec && gtp_psc_mask) {
if (gtp_psc_mask->qfi == UINT8_MAX) {
input_set |= IAVF_INSET_GTPU_QFI;
-   VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr, 
GTPU_EH, QFI);
+   if (gtp_psc_spec->pdu_type ==
+   
IAVF_GTPU_EH_UPLINK)
+   
VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr,
+   
 GTPU_UP, QFI);
+   else if (gtp_psc_spec->pdu_type ==
+   
IAVF_GTPU_EH_DWLINK)
+   
VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr,
+   
 GTPU_DWN, QFI);
+   else
+   
VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr,
+   
 GTPU_EH, QFI);
}
 
rte_memcpy(hdr->buffer, gtp_psc_spec,
-- 
2.25.1

Re: [dpdk-dev] [PATCH 2/3] app/testpmd: fix RSS hash type update

[Maxime Coquelin]

Hi Nélio,

On 9/10/21 4:16 PM, Nélio Laranjeiro wrote:
> On Fri, Sep 10, 2021 at 01:06:53PM +0300, Andrew Rybchenko wrote:
>> On 9/10/21 12:57 PM, Maxime Coquelin wrote:
>>>
>>>
>>> On 9/10/21 11:51 AM, Andrew Rybchenko wrote:
 On 9/10/21 12:17 PM, Maxime Coquelin wrote:
> port_rss_hash_key_update() initializes rss_conf with the
> RSS hash type and key provided by the user, but it calls
> rte_eth_dev_rss_hash_conf_get() before calling
> rte_eth_dev_rss_hash_update(), which overides the parsed
> config with current NIC's config.
>
> While the RSS key value is set again after, this is not
> the case of the key length and the type of hash.
>
> There is no need to read the RSS config from the NIC, let's
> just try to set the user defined one.
>
> Fixes: 8205e241b2b0 ("app/testpmd: add missing type to RSS hash commands")
> Cc: sta...@dpdk.org
> Cc: nelio.laranje...@6wind.com
>
> Signed-off-by: Maxime Coquelin 
> ---
>  app/test-pmd/config.c | 8 ++--
>  1 file changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c
> index 31d8ba1b91..451bda53b1 100644
> --- a/app/test-pmd/config.c
> +++ b/app/test-pmd/config.c
> @@ -2853,18 +2853,14 @@ port_rss_hash_key_update(portid_t port_id, char 
> rss_type[], uint8_t *hash_key,
>   int diag;
>   unsigned int i;
>  
> - rss_conf.rss_key = NULL;
> + rss_conf.rss_key = hash_key;
>   rss_conf.rss_key_len = hash_key_len;
>   rss_conf.rss_hf = 0;
>   for (i = 0; rss_type_table[i].str; i++) {
>   if (!strcmp(rss_type_table[i].str, rss_type))
>   rss_conf.rss_hf = rss_type_table[i].rss_type;
>   }
> - diag = rte_eth_dev_rss_hash_conf_get(port_id, &rss_conf);
> - if (diag == 0) {
> - rss_conf.rss_key = hash_key;
> - diag = rte_eth_dev_rss_hash_update(port_id, &rss_conf);
> - }
> + diag = rte_eth_dev_rss_hash_update(port_id, &rss_conf);

 I'm not 100% sure, but I'd say the intent above could be
 to update key only as the function name says. I.e. keep
 rss_hf as is. That could be the reason to get first.
> 
> True,
> 
>>> I think that was the intial purpose of the command, but patch
>>> 8205e241b2b0 added setting the hash type as mandatory. There are
>>> no other command to configure the hash type from testpmd AFAICT.
> 
> Also for the same initial purpose, some NIC have an hash key per
> protocol, by default it uses the same key for all of them but it can be
> configured individually making for example key0 for all protocols expect
> IPv4 which uses key1.

Thanks for the info, I have looked at most drivers but didn't found one
that support this feature, could you give some pointer?

Given how the drivers implément the callback, do you agree with the fix,
or do you have something else in mind?

Thanks,
Maxime

>>> Also, even without 8205e241b2b0, the function was broken because the
>>> key length was overiden.
>>
>> I see, many thanks for explanations.
> 

Re: [dpdk-dev] [PATCH RFC 0/1] vhost-add-DMADEV-support-for-async-datapath

[Pai G, Sunil]

Just as FYI, 
An alternate approach is up on the OVS mailing list : 
http://patchwork.ozlabs.org/project/openvswitch/list/?series=261277 

Re: [dpdk-dev] Questions about rte_eth_link_speed_to_str API

[Thomas Monjalon]

13/09/2021 10:45, Min Hu (Connor):
> Hi all,
>   I have questions about rte_eth_link_speed_to_str API.
>   The API converts link speed to string for display, But it only
> supports the following speeds, like that:
>   case ETH_SPEED_NUM_NONE: return "None";
>   case ETH_SPEED_NUM_10M:  return "10 Mbps";
>   case ETH_SPEED_NUM_100M: return "100 Mbps";
>   case ETH_SPEED_NUM_1G:   return "1 Gbps";
>   case ETH_SPEED_NUM_2_5G: return "2.5 Gbps";
>   case ETH_SPEED_NUM_5G:   return "5 Gbps";
>   case ETH_SPEED_NUM_10G:  return "10 Gbps";
>   case ETH_SPEED_NUM_20G:  return "20 Gbps";
>   case ETH_SPEED_NUM_25G:  return "25 Gbps";
>   case ETH_SPEED_NUM_40G:  return "40 Gbps";
>   case ETH_SPEED_NUM_50G:  return "50 Gbps";
>   case ETH_SPEED_NUM_56G:  return "56 Gbps";
>   case ETH_SPEED_NUM_100G: return "100 Gbps";
>   case ETH_SPEED_NUM_200G: return "200 Gbps";
>   case ETH_SPEED_NUM_UNKNOWN: return "Unknown";
>   default: return "Invalid";
> 
>   In some cases, like bonding, for example, three slaves which
>   link speed are 10Gbps, so link speed of bonding port will be
>   30Gbps, but it shows "Invalid".
> 
>   Is this reasonable？ any comments will be welcome.

Is it meaningful to print combined slaves speed?
If yes, we can do better then this fixed switch/case logic,
it shouldn't be too hard given it is a standard uint32_t value.

Re: [dpdk-dev] [PATCH v1] usertools/telemetry: add non-interactive mode

[Bruce Richardson]

On Thu, Sep 09, 2021 at 04:56:25PM +0100, David Hunt wrote:
> Add non-interactive mode to dpdk-telemetry.py so that a query string
> can be supplied on the command line, and script dumps out data and
> exits. Handing for calling from scripts.
> 
> Signed-off-by: David Hunt 
> ---
Hi Dave,

I'm not sure I like the use of "-q" for adding a query mode - it's more a
shortcut parameter for a "quiet" mode. If I may, I'd suggest an alternative
approach here might be to improve support for piping the input commands to
the script instead so that you can do e.g. 

"echo /ethdev/stats,0 | dpdk-telemetry.py"

and have that work well in a script.

I'll do up a patchset for improving that and upstream it for feedback.

/Bruce

[dpdk-dev] [PATCH 0/3] improvements for telemetry script

[Bruce Richardson]

Patch 1 fixes errors reported by flake8 in the telemetry python script.
Inspired by the work by Dave Hunt [1] the final two patches look to adjust the
script so that it works nicer when commands come from an input pipe rather than
from an interactive terminal.

Without this set:
  $ echo "/eal/params" | ./usertools/dpdk-telemetry.py
  Connecting to /var/run/dpdk/rte/dpdk_telemetry.v2
  {"version": "DPDK 21.11.0-rc0", "pid": 130033, "max_output_len": 16384}
  Connected to application: "dpdk-test"
  --> {"/eal/params": ["./build/app/test/dpdk-test", "-c", "F", "--no-huge"]}
  --> Traceback (most recent call last):
File "/home/bruce/dpdk.org/./usertools/dpdk-telemetry.py", line 109, in 

  handle_socket(os.path.join(rdir, 
'dpdk_telemetry.{}'.format(TELEMETRY_VERSION)))
File "/home/bruce/dpdk.org/./usertools/dpdk-telemetry.py", line 78, in 
handle_socket
  text = input('--> ').strip()
  EOFError: EOF when reading a line

With this patchset:
  $ echo "/eal/params" | ./usertools/dpdk-telemetry.py
  {"/eal/params": ["./build/app/test/dpdk-test", "-c", "F", "--no-huge"]}


[1] 
http://patches.dpdk.org/project/dpdk/patch/20210909155625.24581-1-david.h...@intel.com/

Bruce Richardson (3):
  usertools/dpdk-telemetry: fix flake8 errors
  usertools/dpdk_telemetry: fix handling EOF for input pipe
  usertools/dpdk-telemetry: silence prompts for input pipes

 usertools/dpdk-telemetry.py | 38 ++---
 1 file changed, 23 insertions(+), 15 deletions(-)

--
2.30.2

[dpdk-dev] [PATCH 1/3] usertools/dpdk-telemetry: fix flake8 errors

[Bruce Richardson]

Fix style errors reported by flake8.

Fixes: 6a2967c112a3 ("usertools: add new telemetry script")
Fixes: 2d9a697e41ca ("usertools: add file-prefix option for telemetry")
Cc: sta...@dpdk.org

Signed-off-by: Bruce Richardson 
---
 usertools/dpdk-telemetry.py | 9 -
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/usertools/dpdk-telemetry.py b/usertools/dpdk-telemetry.py
index e04aa04702..bdc617db18 100755
--- a/usertools/dpdk-telemetry.py
+++ b/usertools/dpdk-telemetry.py
@@ -9,7 +9,6 @@
 
 import socket
 import os
-import glob
 import json
 import errno
 import readline
@@ -102,8 +101,8 @@ def get_dpdk_runtime_dir(fp):
 readline.set_completer_delims(readline.get_completer_delims().replace('/', ''))
 
 parser = argparse.ArgumentParser()
-parser.add_argument('-f', '--file-prefix', \
-help='Provide file-prefix for DPDK runtime directory', default='rte')
+parser.add_argument('-f', '--file-prefix', default='rte',
+help='Provide file-prefix for DPDK runtime directory')
 args = parser.parse_args()
-rdir = get_dpdk_runtime_dir(args.file_prefix)
-handle_socket(os.path.join(rdir, 
'dpdk_telemetry.{}'.format(TELEMETRY_VERSION)))
+rd = get_dpdk_runtime_dir(args.file_prefix)
+handle_socket(os.path.join(rd, 'dpdk_telemetry.{}'.format(TELEMETRY_VERSION)))
-- 
2.30.2

[dpdk-dev] [PATCH 2/3] usertools/dpdk_telemetry: fix handling EOF for input pipe

[Bruce Richardson]

To allow the script to take queries from input pipes e.g. "echo
/ethdev/stats,0 | dpdk-telemetry.py", we need to handle the case of EOF
correctly without crashing with an exception. Do this by using a
try-except block around the input handling.

Fixes: 6a2967c112a3 ("usertools: add new telemetry script")
Cc: sta...@dpdk.org

Signed-off-by: Bruce Richardson 
---
 usertools/dpdk-telemetry.py | 16 ++--
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/usertools/dpdk-telemetry.py b/usertools/dpdk-telemetry.py
index bdc617db18..7ebbb64fce 100755
--- a/usertools/dpdk-telemetry.py
+++ b/usertools/dpdk-telemetry.py
@@ -69,13 +69,17 @@ def handle_socket(path):
 CMDS = read_socket(sock, output_buf_len, False)["/"]
 
 # interactive prompt
-text = input('--> ').strip()
-while text != "quit":
-if text.startswith('/'):
-sock.send(text.encode())
-read_socket(sock, output_buf_len)
+try:
 text = input('--> ').strip()
-sock.close()
+while text != "quit":
+if text.startswith('/'):
+sock.send(text.encode())
+read_socket(sock, output_buf_len)
+text = input('--> ').strip()
+except EOFError:
+pass
+finally:
+sock.close()
 
 
 def readline_complete(text, state):
-- 
2.30.2

[dpdk-dev] [PATCH 3/3] usertools/dpdk-telemetry: silence prompts for input pipes

[Bruce Richardson]

When the input to the script is coming from a device which is not a TTY
then we become less verbose and skip the prompts and helpful messages
about what is happening.

Signed-off-by: Bruce Richardson 
---
 usertools/dpdk-telemetry.py | 15 ++-
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/usertools/dpdk-telemetry.py b/usertools/dpdk-telemetry.py
index 7ebbb64fce..2974a64732 100755
--- a/usertools/dpdk-telemetry.py
+++ b/usertools/dpdk-telemetry.py
@@ -9,6 +9,7 @@
 
 import socket
 import os
+import sys
 import json
 import errno
 import readline
@@ -49,19 +50,23 @@ def get_app_name(pid):
 
 def handle_socket(path):
 """ Connect to socket and handle user input """
+prompt = ''  # this evaluates to false in conditions
 sock = socket.socket(socket.AF_UNIX, socket.SOCK_SEQPACKET)
 global CMDS
-print("Connecting to " + path)
+
+if os.isatty(sys.stdin.fileno()):
+prompt = '--> '
+print("Connecting to " + path)
 try:
 sock.connect(path)
 except OSError:
 print("Error connecting to " + path)
 sock.close()
 return
-json_reply = read_socket(sock, 1024)
+json_reply = read_socket(sock, 1024, prompt)
 output_buf_len = json_reply["max_output_len"]
 app_name = get_app_name(json_reply["pid"])
-if app_name:
+if app_name and prompt:
 print('Connected to application: "%s"' % app_name)
 
 # get list of commands for readline completion
@@ -70,12 +75,12 @@ def handle_socket(path):
 
 # interactive prompt
 try:
-text = input('--> ').strip()
+text = input(prompt).strip()
 while text != "quit":
 if text.startswith('/'):
 sock.send(text.encode())
 read_socket(sock, output_buf_len)
-text = input('--> ').strip()
+text = input(prompt).strip()
 except EOFError:
 pass
 finally:
-- 
2.30.2

Re: [dpdk-dev] [PATCH v1] usertools/telemetry: add non-interactive mode

[Bruce Richardson]

On Mon, Sep 13, 2021 at 11:43:25AM +0100, Bruce Richardson wrote:
> On Thu, Sep 09, 2021 at 04:56:25PM +0100, David Hunt wrote:
> > Add non-interactive mode to dpdk-telemetry.py so that a query string
> > can be supplied on the command line, and script dumps out data and
> > exits. Handing for calling from scripts.
> > 
> > Signed-off-by: David Hunt 
> > ---
> Hi Dave,
> 
> I'm not sure I like the use of "-q" for adding a query mode - it's more a
> shortcut parameter for a "quiet" mode. If I may, I'd suggest an alternative
> approach here might be to improve support for piping the input commands to
> the script instead so that you can do e.g. 
> 
> "echo /ethdev/stats,0 | dpdk-telemetry.py"
> 
> and have that work well in a script.
> 
> I'll do up a patchset for improving that and upstream it for feedback.
> 

Now at: http://patches.dpdk.org/project/dpdk/list/?series=18867

/Bruce

[dpdk-dev] [PATCH] eal/freebsd: lock memory device to prevent conflicts

[Bruce Richardson]

Only a single DPDK process on the system can be using the /dev/contigmem
mappings at a time, but this was never explicitly enforced, e.g. when
using --in-memory flag on two processes. To prevent possible conflict
issues, we lock the dev node when it's in use, preventing other DPDK
processes from starting up and causing problems for us.

Fixes: 764bf26873b9 ("add FreeBSD support")
Cc: sta...@dpdk.org

Signed-off-by: Bruce Richardson 
---
 lib/eal/freebsd/eal_hugepage_info.c | 4 
 1 file changed, 4 insertions(+)

diff --git a/lib/eal/freebsd/eal_hugepage_info.c 
b/lib/eal/freebsd/eal_hugepage_info.c
index 408f054f7a..4a8d87c23e 100644
--- a/lib/eal/freebsd/eal_hugepage_info.c
+++ b/lib/eal/freebsd/eal_hugepage_info.c
@@ -90,6 +90,10 @@ eal_hugepage_info_init(void)
RTE_LOG(ERR, EAL, "could not open "CONTIGMEM_DEV"\n");
return -1;
}
+   if (flock(fd, LOCK_EX) < 0) {
+   RTE_LOG(ERR, EAL, "could not lock memory. Is another DPDK 
process running?\n");
+   return -1;
+   }
 
if (buffer_size >= 1<<30)
RTE_LOG(INFO, EAL, "Contigmem driver has %d buffers, each of 
size %dGB\n",
-- 
2.30.2

[dpdk-dev] [PATCH v5] ethdev: fix representor port ID search by name

[Andrew Rybchenko]

From: Viacheslav Galaktionov 

Getting a list of representors from a representor does not make sense.
Instead, a parent device should be used.

To this end, extend the rte_eth_dev_data structure to include the port ID
of the backing device for representors.

Signed-off-by: Viacheslav Galaktionov 
Signed-off-by: Andrew Rybchenko 
Acked-by: Haiyue Wang 
Acked-by: Beilei Xing 
---
The new field is added into the hole in rte_eth_dev_data structure.
The patch does not change ABI, but extra care is required since ABI
check is disabled for the structure because of the libabigail bug [1].
It should not be a problem anyway since 21.11 is a ABI breaking release.

Potentially it is bad for out-of-tree drivers which implement
representors but do not fill in a new parert_port_id field in
rte_eth_dev_data structure. Get ID by name will not work.

mlx5 changes should be reviwed by maintainers very carefully, since
we are not sure if we patch it correctly.

[1] https://sourceware.org/bugzilla/show_bug.cgi?id=28060

v5:
- try to improve name: backer_port_id instead of parent_port_id
- init new field to RTE_MAX_ETHPORTS on allocation to avoid
  zero port usage by default

v4:
- apply mlx5 review notes: remove fallback from generic ethdev
  code and add fallback to mlx5 code to handle legacy usecase

v3:
- fix mlx5 build breakage

v2:
- fix mlx5 review notes
- try device port ID first before parent in order to address
  backward compatibility issue

 drivers/net/bnxt/bnxt_reps.c |  1 +
 drivers/net/enic/enic_vf_representor.c   |  1 +
 drivers/net/i40e/i40e_vf_representor.c   |  1 +
 drivers/net/ice/ice_dcf_vf_representor.c |  1 +
 drivers/net/ixgbe/ixgbe_vf_representor.c |  1 +
 drivers/net/mlx5/linux/mlx5_os.c | 13 +
 drivers/net/mlx5/windows/mlx5_os.c   | 13 +
 lib/ethdev/ethdev_driver.h   |  6 +++---
 lib/ethdev/rte_class_eth.c   |  2 +-
 lib/ethdev/rte_ethdev.c  |  9 +
 lib/ethdev/rte_ethdev_core.h |  6 ++
 11 files changed, 46 insertions(+), 8 deletions(-)

diff --git a/drivers/net/bnxt/bnxt_reps.c b/drivers/net/bnxt/bnxt_reps.c
index bdbad53b7d..0d50c0f1da 100644
--- a/drivers/net/bnxt/bnxt_reps.c
+++ b/drivers/net/bnxt/bnxt_reps.c
@@ -187,6 +187,7 @@ int bnxt_representor_init(struct rte_eth_dev *eth_dev, void 
*params)
eth_dev->data->dev_flags |= RTE_ETH_DEV_REPRESENTOR |
RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS;
eth_dev->data->representor_id = rep_params->vf_id;
+   eth_dev->data->backer_port_id = rep_params->parent_dev->data->port_id;
 
rte_eth_random_addr(vf_rep_bp->dflt_mac_addr);
memcpy(vf_rep_bp->mac_addr, vf_rep_bp->dflt_mac_addr,
diff --git a/drivers/net/enic/enic_vf_representor.c 
b/drivers/net/enic/enic_vf_representor.c
index 79dd6e5640..fedb09ecd6 100644
--- a/drivers/net/enic/enic_vf_representor.c
+++ b/drivers/net/enic/enic_vf_representor.c
@@ -662,6 +662,7 @@ int enic_vf_representor_init(struct rte_eth_dev *eth_dev, 
void *init_params)
eth_dev->data->dev_flags |= RTE_ETH_DEV_REPRESENTOR |
RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS;
eth_dev->data->representor_id = vf->vf_id;
+   eth_dev->data->backer_port_id = pf->port_id;
eth_dev->data->mac_addrs = rte_zmalloc("enic_mac_addr_vf",
sizeof(struct rte_ether_addr) *
ENIC_UNICAST_PERFECT_FILTERS, 0);
diff --git a/drivers/net/i40e/i40e_vf_representor.c 
b/drivers/net/i40e/i40e_vf_representor.c
index 0481b55381..d65b821a01 100644
--- a/drivers/net/i40e/i40e_vf_representor.c
+++ b/drivers/net/i40e/i40e_vf_representor.c
@@ -514,6 +514,7 @@ i40e_vf_representor_init(struct rte_eth_dev *ethdev, void 
*init_params)
ethdev->data->dev_flags |= RTE_ETH_DEV_REPRESENTOR |
RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS;
ethdev->data->representor_id = representor->vf_id;
+   ethdev->data->backer_port_id = pf->dev_data->port_id;
 
/* Setting the number queues allocated to the VF */
ethdev->data->nb_rx_queues = vf->vsi->nb_qps;
diff --git a/drivers/net/ice/ice_dcf_vf_representor.c 
b/drivers/net/ice/ice_dcf_vf_representor.c
index 970461f3e9..e51d0aa6b9 100644
--- a/drivers/net/ice/ice_dcf_vf_representor.c
+++ b/drivers/net/ice/ice_dcf_vf_representor.c
@@ -418,6 +418,7 @@ ice_dcf_vf_repr_init(struct rte_eth_dev *vf_rep_eth_dev, 
void *init_param)
 
vf_rep_eth_dev->data->dev_flags |= RTE_ETH_DEV_REPRESENTOR;
vf_rep_eth_dev->data->representor_id = repr->vf_id;
+   vf_rep_eth_dev->data->backer_port_id = repr->dcf_eth_dev->data->port_id;
 
vf_rep_eth_dev->data->mac_addrs = &repr->mac_addr;
 
diff --git a/drivers/net/ixgbe/ixgbe_vf_representor.c 
b/drivers/net/ixgbe/ixgbe_vf_representor.c
index d5b636a194..9fa75984fb 100644
--- a/drivers/net/ixgbe/ixgbe_vf_representor.c
+++ b/drivers/net/ixg

Re: [dpdk-dev] [PATCH 1/8] bus/pci: add new memory resource access APIs

[Kinsella, Ray]

On 10/09/2021 03:23, Chenbo Xia wrote:
> Some applications wants to access PCI memory resource. Currently
> applications use struct rte_pci_device to access it. Since the
> structure will be made internal later, this patch adds two APIs
> for memory resource access.
> 
> Signed-off-by: Chenbo Xia 
> ---
>  doc/guides/rel_notes/release_21_11.rst |  5 ++
>  drivers/bus/pci/pci_common.c   | 78 ++
>  drivers/bus/pci/rte_bus_pci.h  | 36 
>  drivers/bus/pci/version.map|  4 ++
>  4 files changed, 123 insertions(+)
> 
Acked-by: Ray Kinsella 

Re: [dpdk-dev] [PATCH 8/8] bus/pci: remove ABIs in PCI bus

[Kinsella, Ray]

On 10/09/2021 03:24, Chenbo Xia wrote:
> As announced in the deprecation note, most of ABIs in PCI bus are
> removed in this patch. Only the function rte_pci_dump is still ABI
> and experimental APIs are kept for future promotion.
> 
> This patch creates a new file named pci_driver.h and moves most of
> the content in original rte_bus_pci.h to it. After that, pci_driver.h
> is considered the interface for drivers and rte_bus_pci.h for
> applications. pci_driver.h is defined as driver_sdk_headers so that
> out-of-tree drivers can use it.
> 
> Then this patch replaces the including of rte_bus_pci.h with pci_driver.h
> in all related drivers.
> 
> Signed-off-by: Chenbo Xia 
> ---
>  app/test/virtual_pmd.c|   2 +-
>  doc/guides/rel_notes/release_21_11.rst|   2 +
>  drivers/baseband/acc100/rte_acc100_pmd.c  |   2 +-
>  .../fpga_5gnr_fec/rte_fpga_5gnr_fec.c |   2 +-
>  drivers/baseband/fpga_lte_fec/fpga_lte_fec.c  |   2 +-
>  drivers/bus/pci/bsd/pci.c |   1 -
>  drivers/bus/pci/linux/pci.c   |   1 -
>  drivers/bus/pci/linux/pci_uio.c   |   1 -
>  drivers/bus/pci/linux/pci_vfio.c  |   1 -
>  drivers/bus/pci/meson.build   |   4 +
>  drivers/bus/pci/pci_common_uio.c  |   1 -
>  drivers/bus/pci/pci_driver.h  | 402 ++
>  drivers/bus/pci/pci_params.c  |   1 -
>  drivers/bus/pci/private.h |   3 +-
>  drivers/bus/pci/rte_bus_pci.h | 375 +---
>  drivers/bus/pci/version.map   |  32 +-
>  drivers/common/cnxk/roc_platform.h|   2 +-
>  drivers/common/mlx5/linux/mlx5_common_verbs.c |   2 +-
>  drivers/common/mlx5/mlx5_common_pci.c |   2 +-
>  drivers/common/octeontx2/otx2_dev.h   |   2 +-
>  drivers/common/octeontx2/otx2_sec_idev.c  |   2 +-
>  drivers/common/qat/qat_device.h   |   2 +-
>  drivers/common/qat/qat_qp.c   |   2 +-
>  drivers/common/sfc_efx/sfc_efx.h  |   2 +-
>  drivers/compress/mlx5/mlx5_compress.c |   2 +-
>  drivers/compress/octeontx/otx_zip.h   |   2 +-
>  drivers/compress/qat/qat_comp.c   |   2 +-
>  drivers/crypto/ccp/ccp_dev.h  |   2 +-
>  drivers/crypto/ccp/ccp_pci.h  |   2 +-
>  drivers/crypto/ccp/rte_ccp_pmd.c  |   2 +-
>  drivers/crypto/cnxk/cn10k_cryptodev.c |   2 +-
>  drivers/crypto/cnxk/cn9k_cryptodev.c  |   2 +-
>  drivers/crypto/mlx5/mlx5_crypto.c |   2 +-
>  drivers/crypto/nitrox/nitrox_device.h |   2 +-
>  drivers/crypto/octeontx/otx_cryptodev.c   |   2 +-
>  drivers/crypto/octeontx/otx_cryptodev_ops.c   |   2 +-
>  drivers/crypto/octeontx2/otx2_cryptodev.c |   2 +-
>  drivers/crypto/qat/qat_sym.c  |   2 +-
>  drivers/crypto/qat/qat_sym_pmd.c  |   2 +-
>  drivers/crypto/virtio/virtio_cryptodev.c  |   2 +-
>  drivers/crypto/virtio/virtio_pci.h|   2 +-
>  drivers/event/dlb2/pf/dlb2_main.h |   2 +-
>  drivers/event/dlb2/pf/dlb2_pf.c   |   2 +-
>  drivers/event/octeontx/ssovf_probe.c  |   2 +-
>  drivers/event/octeontx/timvf_probe.c  |   2 +-
>  drivers/event/octeontx2/otx2_evdev.c  |   2 +-
>  drivers/mempool/cnxk/cnxk_mempool.c   |   2 +-
>  drivers/mempool/octeontx/octeontx_fpavf.c |   2 +-
>  drivers/mempool/octeontx2/otx2_mempool.c  |   2 +-
>  drivers/mempool/octeontx2/otx2_mempool.h  |   2 +-
>  drivers/mempool/octeontx2/otx2_mempool_irq.c  |   2 +-
>  drivers/meson.build   |   4 +
>  drivers/net/ark/ark_ethdev.c  |   2 +-
>  drivers/net/avp/avp_ethdev.c  |   2 +-
>  drivers/net/bnx2x/bnx2x.h |   2 +-
>  drivers/net/bnxt/bnxt.h   |   2 +-
>  drivers/net/bonding/rte_eth_bond_args.c   |   2 +-
>  drivers/net/cxgbe/base/adapter.h  |   2 +-
>  drivers/net/cxgbe/cxgbe_ethdev.c  |   2 +-
>  drivers/net/e1000/em_ethdev.c |   2 +-
>  drivers/net/e1000/em_rxtx.c   |   2 +-
>  drivers/net/e1000/igb_ethdev.c|   2 +-
>  drivers/net/e1000/igb_pf.c|   2 +-
>  drivers/net/ena/ena_ethdev.h  |   2 +-
>  drivers/net/enic/base/vnic_dev.h  |   2 +-
>  drivers/net/enic/enic_ethdev.c|   2 +-
>  drivers/net/enic/enic_main.c  |   2 +-
>  drivers/net/enic/enic_vf_representor.c|   2 +-
>  drivers/net/hinic/base/hinic_pmd_hwdev.c  |   2 +-
>  drivers/net/hinic/base/hinic_pmd_hwif.c   |   2 +-
>  drivers/net/hinic/base/hinic_pmd_nicio.c  |   2 +-
>  drivers/net/hinic/hinic_pmd_ethdev.c  |   2 +-
>  drivers/net/hns3/hns3_ethdev.c|   2 +-
>  drivers/net/hns3/hns3_rxtx.c  |   2 

[dpdk-dev] [PATCH] examples: ipv4, udp and tcp checksum offload warning

[usamanadeem321]

Closes gracefully if IPV4 checksum offload is not available.
Gives warning if UDP or TCP checksum offloads are not available.

Signed-off-by: usamanadeem321 
---
 examples/l3fwd/main.c | 26 +-
 1 file changed, 25 insertions(+), 1 deletion(-)

diff --git a/examples/l3fwd/main.c b/examples/l3fwd/main.c
index 00ac267af1..81e605700e 100644
--- a/examples/l3fwd/main.c
+++ b/examples/l3fwd/main.c
@@ -123,7 +123,7 @@ static struct rte_eth_conf port_conf = {
.mq_mode = ETH_MQ_RX_RSS,
.max_rx_pkt_len = RTE_ETHER_MAX_LEN,
.split_hdr_size = 0,
-   .offloads = DEV_RX_OFFLOAD_CHECKSUM,
+
},
.rx_adv_conf = {
.rss_conf = {
@@ -1039,6 +1039,30 @@ l3fwd_poll_resource_setup(void)
local_port_conf.txmode.offloads |=
DEV_TX_OFFLOAD_MBUF_FAST_FREE;
 
+   if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_IPV4_CKSUM)
+   local_port_conf.rxmode.offloads |=
+DEV_RX_OFFLOAD_IPV4_CKSUM;
+   else {
+   rte_exit(EXIT_FAILURE,
+   "IPV4 Checksum offload not available. (port %u) 
",
+   portid);
+   }
+
+   if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_UDP_CKSUM)
+   local_port_conf.rxmode.offloads |=
+   DEV_RX_OFFLOAD_UDP_CKSUM;
+
+   else
+   printf("WARNING: UDP Checksum offload not 
available.\n");
+
+   if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_CKSUM)
+   local_port_conf.rxmode.offloads |=
+   DEV_RX_OFFLOAD_TCP_CKSUM;
+
+   else
+   printf("WARNING: TCP Checksum offload not 
available.\n");
+
+
local_port_conf.rx_adv_conf.rss_conf.rss_hf &=
dev_info.flow_type_rss_offloads;
 
-- 
2.25.1

Re: [dpdk-dev] [RFC 1/1] net/mlx5: set txq affinity in round-robin

[Kinsella, Ray]

On 10/09/2021 05:23, Rongwei Liu wrote:
> Previously, we set txq affinity to 0 and let firmware
> to perform round-robin when bonding. Firmware uses a
> global counter to assign txq affinity to different
> physical ports accord to remainder after division.
> 
> There are three dis-advantages:
> 1. The global counter is shared between kernel and dpdk.
> 2. After restarting pmd or port, the previous counter value
> is reused, so the new affinity is unpredictable.
> 3. There is no way to get what affinity is set by firmware.
> 
> In this update, we will create several TISs up to the
> number of bonding ports and bind each TIS to one PF port.
> 
> For each port, it will start to pick up TIS using its port
> index. Upper layer application can quickly calculate each txq's
> affinity without querying.
> 
> At DPDK layer, when creating txq with 2 bonding ports, the
> affinity is set like:
> port 0: 1-->2-->1-->2
> port 1: 2-->1-->2-->1
> port 2: 1-->2-->1-->2
> 
> Note: Only applicable to Devx api.
> 
> Signed-off-by: Jiawei Wang 
> Signed-off-by: Rongwei Liu 
> ---
>  drivers/common/mlx5/mlx5_devx_cmds.c | 38 ++
>  drivers/common/mlx5/mlx5_devx_cmds.h | 12 +++
>  drivers/common/mlx5/mlx5_prm.h   | 26 +++
>  drivers/common/mlx5/version.map  |  1 +
>  drivers/net/mlx5/linux/mlx5_os.c | 37 ++
>  drivers/net/mlx5/mlx5.c  | 17 +++---
>  drivers/net/mlx5/mlx5.h  |  8 -
>  drivers/net/mlx5/mlx5_devx.c | 47 ++--
>  drivers/net/mlx5/mlx5_tx.h   |  1 +
>  drivers/net/mlx5/mlx5_txpp.c |  4 +--
>  10 files changed, 174 insertions(+), 17 deletions(-)
> 
Acked-by: Ray Kinsella 

Re: [dpdk-dev] [PATCH] eal/freebsd: lock memory device to prevent conflicts

[Burakov, Anatoly]

On 13-Sep-21 12:06 PM, Bruce Richardson wrote:

Only a single DPDK process on the system can be using the /dev/contigmem
mappings at a time, but this was never explicitly enforced, e.g. when
using --in-memory flag on two processes. To prevent possible conflict
issues, we lock the dev node when it's in use, preventing other DPDK
processes from starting up and causing problems for us.

Fixes: 764bf26873b9 ("add FreeBSD support")
Cc: sta...@dpdk.org

Signed-off-by: Bruce Richardson 
---
  lib/eal/freebsd/eal_hugepage_info.c | 4 
  1 file changed, 4 insertions(+)

diff --git a/lib/eal/freebsd/eal_hugepage_info.c 
b/lib/eal/freebsd/eal_hugepage_info.c
index 408f054f7a..4a8d87c23e 100644
--- a/lib/eal/freebsd/eal_hugepage_info.c
+++ b/lib/eal/freebsd/eal_hugepage_info.c
@@ -90,6 +90,10 @@ eal_hugepage_info_init(void)
RTE_LOG(ERR, EAL, "could not open "CONTIGMEM_DEV"\n");
return -1;
}
+   if (flock(fd, LOCK_EX) < 0) {
+   RTE_LOG(ERR, EAL, "could not lock memory. Is another DPDK process 
running?\n");
+   return -1;
+   }
  
  	if (buffer_size >= 1<<30)

RTE_LOG(INFO, EAL, "Contigmem driver has %d buffers, each of size 
%dGB\n",



This only gets triggered when regular init path is chosen, i.e. 
--no-huge still works. I'm a bit uneasy with --in-memory mode pretending 
to work on FreeBSD and Windows, but that's a separate problem :) As far 
as the patch goes, the problem it addresses does get fixed.


Reviewed-by: Anatoly Burakov 

--
Thanks,
Anatoly

Re: [dpdk-dev] [PATCH] eal/freebsd: lock memory device to prevent conflicts

[Bruce Richardson]

On Mon, Sep 13, 2021 at 02:14:55PM +0100, Burakov, Anatoly wrote:
> On 13-Sep-21 12:06 PM, Bruce Richardson wrote:
> > Only a single DPDK process on the system can be using the /dev/contigmem
> > mappings at a time, but this was never explicitly enforced, e.g. when
> > using --in-memory flag on two processes. To prevent possible conflict
> > issues, we lock the dev node when it's in use, preventing other DPDK
> > processes from starting up and causing problems for us.
> > 
> > Fixes: 764bf26873b9 ("add FreeBSD support")
> > Cc: sta...@dpdk.org
> > 
> > Signed-off-by: Bruce Richardson 
> > ---
> >   lib/eal/freebsd/eal_hugepage_info.c | 4 
> >   1 file changed, 4 insertions(+)
> > 
> > diff --git a/lib/eal/freebsd/eal_hugepage_info.c 
> > b/lib/eal/freebsd/eal_hugepage_info.c
> > index 408f054f7a..4a8d87c23e 100644
> > --- a/lib/eal/freebsd/eal_hugepage_info.c
> > +++ b/lib/eal/freebsd/eal_hugepage_info.c
> > @@ -90,6 +90,10 @@ eal_hugepage_info_init(void)
> > RTE_LOG(ERR, EAL, "could not open "CONTIGMEM_DEV"\n");
> > return -1;
> > }
> > +   if (flock(fd, LOCK_EX) < 0) {
> > +   RTE_LOG(ERR, EAL, "could not lock memory. Is another DPDK 
> > process running?\n");
> > +   return -1;
> > +   }
> > if (buffer_size >= 1<<30)
> > RTE_LOG(INFO, EAL, "Contigmem driver has %d buffers, each of 
> > size %dGB\n",
> > 
> 
> This only gets triggered when regular init path is chosen, i.e. --no-huge
> still works.

Yes, but that is ok, I think, since no-huge doesn't use these resources or
suffer from this problem. On the other hand, except for running unit tests,
no-huge mode is pretty useless on FreeBSD as we don't have any
vfio-equivalent support, so all HW access has to use physical addresses
which can only be got using contigmem.

> I'm a bit uneasy with --in-memory mode pretending to work on
> FreeBSD and Windows, but that's a separate problem :)

Yes, it is, though one that does belong is the same area as this one. The
"fix" is probably to just print a warning when --in-memory is used,
informing the user that the flag is ignored and then continue.
Alternatively we can error out, but I think the warn+continue is better,
myself.

> As far as the patch
> goes, the problem it addresses does get fixed.
> 
> Reviewed-by: Anatoly Burakov 
> 
Thanks.

/Bruce

[dpdk-dev] [PATCH v2] net/af_packet: fix ignoring full ring on tx

[Tudor Cornea]

The poll call can return POLLERR which is ignored, or it can return
POLLOUT, even if there are no free frames in the mmap-ed area.

We can account for both of these cases by re-checking if the next
frame is empty before writing into it.

Signed-off-by: Mihai Pogonaru 
Signed-off-by: Tudor Cornea 
---
 drivers/net/af_packet/rte_eth_af_packet.c | 19 +++
 1 file changed, 19 insertions(+)

diff --git a/drivers/net/af_packet/rte_eth_af_packet.c 
b/drivers/net/af_packet/rte_eth_af_packet.c
index b73b211..087c196 100644
--- a/drivers/net/af_packet/rte_eth_af_packet.c
+++ b/drivers/net/af_packet/rte_eth_af_packet.c
@@ -216,6 +216,25 @@ eth_af_packet_tx(void *queue, struct rte_mbuf **bufs, 
uint16_t nb_pkts)
(poll(&pfd, 1, -1) < 0))
break;
 
+   /*
+* Poll can return POLLERR if the interface is down
+*
+* It will almost always return POLLOUT, even if there
+* are no extra buffers available
+*
+* This happens, because packet_poll() calls datagram_poll()
+* which checks the space left in the socket buffer and,
+* in the case of packet_mmap, the default socket buffer length
+* doesn't match the requested size for the tx_ring.
+* As such, there is almost always space left in socket buffer,
+* which doesn't seem to be correlated to the requested size
+* for the tx_ring in packet_mmap.
+*
+* This results in poll() returning POLLOUT.
+*/
+   if (ppd->tp_status != TP_STATUS_AVAILABLE)
+   break;
+
/* copy the tx frame data */
pbuf = (uint8_t *) ppd + TPACKET2_HDRLEN -
sizeof(struct sockaddr_ll);
-- 
2.7.4

Re: [dpdk-dev] [PATCH] eal/freebsd: lock memory device to prevent conflicts

[Dmitry Kozlyuk]

2021-09-13 14:14 (UTC+0100), Burakov, Anatoly:
> [...]
> I'm a bit uneasy with --in-memory mode pretending 
> to work on FreeBSD and Windows, but that's a separate problem :)

On Windows, --in-memory does not pretend to work, just the opposite,
it is enabled implicitly as it's the only working mode.

[dpdk-dev] [PATCH v2] eal/freebsd: lock memory device to prevent conflicts

[Bruce Richardson]

Only a single DPDK process on the system can be using the /dev/contigmem
mappings at a time, but this was never explicitly enforced, e.g. when
using --in-memory flag on two processes. To prevent possible conflict
issues, we lock the dev node when it's in use, preventing other DPDK
processes from starting up and causing problems for us.

Fixes: 764bf26873b9 ("add FreeBSD support")
Cc: sta...@dpdk.org

Signed-off-by: Bruce Richardson 
Reviewed-by: Anatoly Burakov 
---

V2: Adding missing LOCK_NB flag to make sure the process doesn't sit
waiting for the lock to be released

 lib/eal/freebsd/eal_hugepage_info.c | 4 
 1 file changed, 4 insertions(+)

diff --git a/lib/eal/freebsd/eal_hugepage_info.c 
b/lib/eal/freebsd/eal_hugepage_info.c
index 408f054f7a..9dbe375bd3 100644
--- a/lib/eal/freebsd/eal_hugepage_info.c
+++ b/lib/eal/freebsd/eal_hugepage_info.c
@@ -90,6 +90,10 @@ eal_hugepage_info_init(void)
RTE_LOG(ERR, EAL, "could not open "CONTIGMEM_DEV"\n");
return -1;
}
+   if (flock(fd, LOCK_EX | LOCK_NB) < 0) {
+   RTE_LOG(ERR, EAL, "could not lock memory. Is another DPDK 
process running?\n");
+   return -1;
+   }

if (buffer_size >= 1<<30)
RTE_LOG(INFO, EAL, "Contigmem driver has %d buffers, each of 
size %dGB\n",
--
2.30.2

[dpdk-dev] [PATCH] net/af_packet: remove timestamp from packet status

[Tudor Cornea]

We should eliminate the timestamp status from the packet
status. This should only matter if timestamping is enabled
on the socket, but we might hit a kernel bug, which is fixed
in newer releases.

For interfaces of type 'veth', the sent skb is forwarded
to the peer and back into the network stack which timestamps
it on the RX path if timestamping is enabled globally
(which happens if any socket enables timestamping).

When the skb is destructed, tpacket_destruct_skb() is called
and it calls __packet_set_timestamp() which doesn't check
the flags on the socket and returns the timestamp if it is
set in the skb (and for veth it is, as mentioned above).

See the following kernel commit for reference [1]:

net: packetmmap: fix only tx timestamp on request

The packetmmap tx ring should only return timestamps if requested
via setsockopt PACKET_TIMESTAMP, as documented. This allows
compatibility with non-timestamp aware user-space code which checks
tp_status == TP_STATUS_AVAILABLE; not expecting additional timestamp
flags to be set in tp_status.

[1] https://www.spinics.net/lists/kernel/msg3959391.html

Signed-off-by: Mihai Pogonaru 
Signed-off-by: Tudor Cornea 
---
 drivers/net/af_packet/rte_eth_af_packet.c | 22 --
 1 file changed, 20 insertions(+), 2 deletions(-)

diff --git a/drivers/net/af_packet/rte_eth_af_packet.c 
b/drivers/net/af_packet/rte_eth_af_packet.c
index b73b211..a6638a2 100644
--- a/drivers/net/af_packet/rte_eth_af_packet.c
+++ b/drivers/net/af_packet/rte_eth_af_packet.c
@@ -17,6 +17,7 @@
 #include 
 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
@@ -167,6 +168,23 @@ eth_af_packet_rx(void *queue, struct rte_mbuf **bufs, 
uint16_t nb_pkts)
return num_rx;
 }
 
+static inline bool tx_ring_status_unavailable(uint32_t tp_status)
+{
+#if KERNEL_VERSION(5, 10, 0) > LINUX_VERSION_CODE
+   /*
+* We eliminate the timestamp status from the packet status.
+* This should only matter if timestamping is enabled on the socket,
+* but there is a bug in the kernel which is fixed in newer releases.
+*
+* See the following kernel commit for reference:
+* commit 171c3b151118a2fe0fc1e2a9d1b5a1570cfe82d2
+* net: packetmmap: fix only tx timestamp on request
+*/
+   tp_status &= ~(TP_STATUS_TS_SOFTWARE | TP_STATUS_TS_RAW_HARDWARE);
+#endif
+   return tp_status != TP_STATUS_AVAILABLE;
+}
+
 /*
  * Callback to handle sending packets through a real NIC.
  */
@@ -212,8 +230,8 @@ eth_af_packet_tx(void *queue, struct rte_mbuf **bufs, 
uint16_t nb_pkts)
}
 
/* point at the next incoming frame */
-   if ((ppd->tp_status != TP_STATUS_AVAILABLE) &&
-   (poll(&pfd, 1, -1) < 0))
+   if (tx_ring_status_unavailable(ppd->tp_status) &&
+   poll(&pfd, 1, -1) < 0)
break;
 
/* copy the tx frame data */
-- 
2.7.4

Re: [dpdk-dev] [PATCH 0/8] cryptodev: hide internal strutures

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 0/8] cryptodev: hide internal strutures
> 
> Structures rte_cryptodev and rte_cryptodev_data are not
> supposed to be directly used by the application. These
> are made public as they are used by inline datapath
> public APIs.
> This patchset, creates a new rte_cryptodev_core.h file
> which helps in defining a data structure to hold datapath
> APIs in a flat array based on the device identifier which
> is filled by the PMD.
> The patchset updates all crypto PMDs to use the new
> framework and hence the internal structs are made hidden.
> 
> Similar series for ethdev and eventdev are also floated on ML.
> http://patches.dpdk.org/project/dpdk/list/?series=18382
> http://patches.dpdk.org/project/dpdk/list/?series=18422
> 
> 
> Akhil Goyal (8):
>   cryptodev: separate out internal structures
>   cryptodev: move inline APIs into separate structure
>   cryptodev: add helper functions for new datapath interface
>   cryptodev: use new API for datapath functions
>   drivers/crypto: use new framework for datapath
>   crypto/scheduler: rename enq-deq functions
>   crypto/scheduler: update for new datapath framework
>   cryptodev: move device specific structures
> 
>  drivers/crypto/aesni_gcm/aesni_gcm_pmd.c  |  10 +-
>  drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c|  11 +-
>  drivers/crypto/armv8/rte_armv8_pmd.c  |  11 +-
>  drivers/crypto/bcmfs/bcmfs_sym_pmd.c  |  11 +-
>  drivers/crypto/caam_jr/caam_jr.c  |  11 +-
>  drivers/crypto/ccp/ccp_dev.c  |   1 +
>  drivers/crypto/ccp/rte_ccp_pmd.c  |  11 +-
>  drivers/crypto/cnxk/cn10k_cryptodev_ops.c |   8 +-
>  drivers/crypto/cnxk/cn10k_cryptodev_ops.h |   3 +
>  drivers/crypto/cnxk/cn10k_ipsec.c |   1 +
>  drivers/crypto/cnxk/cn9k_cryptodev_ops.c  |   9 +-
>  drivers/crypto/cnxk/cn9k_cryptodev_ops.h  |   3 +
>  .../crypto/cnxk/cnxk_cryptodev_capabilities.c |   1 +
>  drivers/crypto/cnxk/cnxk_cryptodev_sec.c  |   1 +
>  drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c   |  12 +-
>  drivers/crypto/dpaa_sec/dpaa_sec.c|  11 +-
>  drivers/crypto/kasumi/rte_kasumi_pmd.c|  11 +-
>  drivers/crypto/mlx5/mlx5_crypto.c |  11 +-
>  drivers/crypto/mvsam/rte_mrvl_pmd.c   |  11 +-
>  drivers/crypto/nitrox/nitrox_sym.c|  11 +-
>  drivers/crypto/nitrox/nitrox_sym_reqmgr.c |   1 +
>  drivers/crypto/null/null_crypto_pmd.c |  11 +-
>  .../crypto/octeontx/otx_cryptodev_hw_access.c |   1 +
>  drivers/crypto/octeontx/otx_cryptodev_ops.c   |  16 +-
>  drivers/crypto/octeontx/otx_cryptodev_ops.h   |   5 +
>  .../crypto/octeontx2/otx2_cryptodev_mbox.c|   1 +
>  drivers/crypto/octeontx2/otx2_cryptodev_ops.c |  11 +-
>  drivers/crypto/openssl/rte_openssl_pmd.c  |  11 +-
>  drivers/crypto/qat/qat_asym_pmd.c |  11 +-
>  drivers/crypto/qat/qat_sym_pmd.c  |  10 +-
>  drivers/crypto/scheduler/scheduler_failover.c |  35 +-
>  .../crypto/scheduler/scheduler_multicore.c|  32 +-
>  .../scheduler/scheduler_pkt_size_distr.c  |  34 +-
>  .../crypto/scheduler/scheduler_roundrobin.c   |  34 +-
>  drivers/crypto/snow3g/rte_snow3g_pmd.c|  11 +-
>  drivers/crypto/virtio/virtio_cryptodev.c  |  10 +-
>  drivers/crypto/virtio/virtio_cryptodev.h  |   2 +
>  drivers/crypto/virtio/virtio_rxtx.c   |   2 +
>  drivers/crypto/zuc/rte_zuc_pmd.c  |  11 +-
>  lib/cryptodev/cryptodev_pmd.c |  33 ++
>  lib/cryptodev/cryptodev_pmd.h | 323 +-
>  lib/cryptodev/meson.build |   4 +-
>  lib/cryptodev/rte_cryptodev.c |  43 ++-
>  lib/cryptodev/rte_cryptodev.h | 308 +
>  lib/cryptodev/rte_cryptodev_core.h|  43 +++
>  lib/cryptodev/version.map |   8 +
>  46 files changed, 854 insertions(+), 316 deletions(-)
>  create mode 100644 lib/cryptodev/rte_cryptodev_core.h
> 
> --
> 2.25.1

Series-tested-by: Rebecca Troy 
Series-acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 1/8] cryptodev: separate out internal structures

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 1/8] cryptodev: separate out internal structures
> 
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 2/8] cryptodev: move inline APIs into separate structure

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 2/8] cryptodev: move inline APIs into separate structure
> 
> Move fastpath inline function pointers from rte_cryptodev into a
> separate structure accessed via a flat array.
> The intension is to make rte_cryptodev and related structures private
> to avoid future API/ABI breakages.
> 
> Signed-off-by: Akhil Goyal 
> ---
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 3/8] cryptodev: add helper functions for new datapath interface

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 3/8] cryptodev: add helper functions for new datapath
> interface
> 
> Add helper functions and macros to help drivers to
> transition to new datapath interface.
> 
> Signed-off-by: Akhil Goyal 
> ---

Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 4/8] cryptodev: use new API for datapath functions

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 4/8] cryptodev: use new API for datapath functions
> 
> The datapath inline APIs (rte_cryptodev_enqueue_burst/
> rte_cryptodev_dequeue_burst) are updated to use the new
> rte_crytodev_api->enqueue_burst/rte_cryptodev_api->dequeue_burst
> APIs based on the dev_id
> 
> Signed-off-by: Akhil Goyal 
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 5/8] drivers/crypto: use new framework for datapath

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 5/8] drivers/crypto: use new framework for datapath
> 
> All crypto drivers are updated to use the new API
> for all enqueue and dequeue paths.
> 
> Signed-off-by: Akhil Goyal 
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 6/8] crypto/scheduler: rename enq-deq functions

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 6/8] crypto/scheduler: rename enq-deq functions
> 
> scheduler PMD has 4 variants, which uses same
> name for all the enqueue and dequeue functions.
> This causes multiple definitions of same function
> with the new framework of datapath APIs.
> Hence the function names are updated to specify the
> the variant it is for.
> 
> Signed-off-by: Akhil Goyal 
> ---
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 7/8] crypto/scheduler: update for new datapath framework

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 7/8] crypto/scheduler: update for new datapath framework
> 
> PMD is updated to use the new API for all enqueue
> and dequeue paths.
> 
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

Re: [dpdk-dev] [PATCH 8/8] cryptodev: move device specific structures

[Zhang, Roy Fan]

> -Original Message-
> From: Akhil Goyal 
> Sent: Sunday, August 29, 2021 1:52 PM
> To: dev@dpdk.org
> Cc: ano...@marvell.com; Nicolau, Radu ; Doherty,
> Declan ; hemant.agra...@nxp.com;
> ma...@nvidia.com; Ananyev, Konstantin ;
> tho...@monjalon.net; Zhang, Roy Fan ;
> asoma...@amd.com; ruifeng.w...@arm.com;
> ajit.khapa...@broadcom.com; De Lara Guarch, Pablo
> ; Trahe, Fiona ;
> adwiv...@marvell.com; michae...@marvell.com;
> rnagadhee...@marvell.com; jianjay.z...@huawei.com; jer...@marvell.com;
> Akhil Goyal 
> Subject: [PATCH 8/8] cryptodev: move device specific structures
> 
> The device specific structures - rte_cryptodev
> and rte_cryptodev_data are moved to cryptodev_pmd.h
> to hide it from the applications.
> 
> Signed-off-by: Akhil Goyal 
> ---
Tested-by: Rebecca Troy 
Acked-by: Fan Zhang 

[dpdk-dev] [PATCH] eal/freebsd: ignore in-memory option

[Bruce Richardson]

The in-memory option is not supported on FreeBSD so print a warning and
ignore the flag when it is specified for BSD apps. The lack of support
is due to the different way in which memory is managed on FreeBSD using
the contigmem driver rather than via a hugetlbfs filesystem.

Fixes: 14de8734c401 ("eal: add --in-memory option")
Cc: anatoly.bura...@intel.com

Signed-off-by: Bruce Richardson 
---
 lib/eal/freebsd/eal.c | 4 
 1 file changed, 4 insertions(+)

diff --git a/lib/eal/freebsd/eal.c b/lib/eal/freebsd/eal.c
index 6cee5ae369..fb734012a4 100644
--- a/lib/eal/freebsd/eal.c
+++ b/lib/eal/freebsd/eal.c
@@ -718,6 +718,10 @@ rte_eal_init(int argc, char **argv)
 
/* FreeBSD always uses legacy memory model */
internal_conf->legacy_mem = true;
+   if (internal_conf->in_memory) {
+   RTE_LOG(WARNING, EAL, "Warning: ignoring unsupported flag, 
'%s'\n", OPT_IN_MEMORY);
+   internal_conf->in_memory = false;
+   }
 
if (eal_plugins_init() < 0) {
rte_eal_init_alert("Cannot init plugins");
-- 
2.30.2

Re: [dpdk-dev] [PATCH] eal/freebsd: lock memory device to prevent conflicts

[Burakov, Anatoly]

On 13-Sep-21 2:36 PM, Bruce Richardson wrote:

On Mon, Sep 13, 2021 at 02:14:55PM +0100, Burakov, Anatoly wrote:

On 13-Sep-21 12:06 PM, Bruce Richardson wrote:

Only a single DPDK process on the system can be using the /dev/contigmem
mappings at a time, but this was never explicitly enforced, e.g. when
using --in-memory flag on two processes. To prevent possible conflict
issues, we lock the dev node when it's in use, preventing other DPDK
processes from starting up and causing problems for us.

Fixes: 764bf26873b9 ("add FreeBSD support")
Cc: sta...@dpdk.org

Signed-off-by: Bruce Richardson 
---
   lib/eal/freebsd/eal_hugepage_info.c | 4 
   1 file changed, 4 insertions(+)

diff --git a/lib/eal/freebsd/eal_hugepage_info.c 
b/lib/eal/freebsd/eal_hugepage_info.c
index 408f054f7a..4a8d87c23e 100644
--- a/lib/eal/freebsd/eal_hugepage_info.c
+++ b/lib/eal/freebsd/eal_hugepage_info.c
@@ -90,6 +90,10 @@ eal_hugepage_info_init(void)
RTE_LOG(ERR, EAL, "could not open "CONTIGMEM_DEV"\n");
return -1;
}
+   if (flock(fd, LOCK_EX) < 0) {
+   RTE_LOG(ERR, EAL, "could not lock memory. Is another DPDK process 
running?\n");
+   return -1;
+   }
if (buffer_size >= 1<<30)
RTE_LOG(INFO, EAL, "Contigmem driver has %d buffers, each of size 
%dGB\n",



This only gets triggered when regular init path is chosen, i.e. --no-huge
still works.


Yes, but that is ok, I think, since no-huge doesn't use these resources or
suffer from this problem. On the other hand, except for running unit tests,
no-huge mode is pretty useless on FreeBSD as we don't have any
vfio-equivalent support, so all HW access has to use physical addresses
which can only be got using contigmem.


What i meant to say was, i've checked this against '--no-huge' which 
*should* still work with this patch, and it does :) So, the phrasing was 
unfortunate, but we agree!





I'm a bit uneasy with --in-memory mode pretending to work on
FreeBSD and Windows, but that's a separate problem :)


Yes, it is, though one that does belong is the same area as this one. The
"fix" is probably to just print a warning when --in-memory is used,
informing the user that the flag is ignored and then continue.
Alternatively we can error out, but I think the warn+continue is better,
myself.


I think erroring out is better. The feature is intended to work a 
certain way, so if we can't guarantee that it does, we can't pretend it 
is "supported" or "is working". But again, irrelevant to this patch :)





As far as the patch
goes, the problem it addresses does get fixed.

Reviewed-by: Anatoly Burakov 


Thanks.

/Bruce




--
Thanks,
Anatoly

Re: [dpdk-dev] [PATCH] Enable AddressSanitizer feature on DPDK

[Stephen Hemminger]

On Mon, 13 Sep 2021 05:27:12 +
"Peng, ZhihongX"  wrote:

> > -Original Message-
> > From: Stephen Hemminger 
> > Sent: Friday, September 10, 2021 10:48 AM
> > To: Peng, ZhihongX 
> > Cc: Burakov, Anatoly ; Ananyev, Konstantin
> > ; dev@dpdk.org; Lin, Xueqin
> > 
> > Subject: Re: [PATCH] Enable AddressSanitizer feature on DPDK
> > 
> > On Fri, 10 Sep 2021 02:01:47 +
> > zhihongx.p...@intel.com wrote:
> >   
> > >
> > > +if get_option('b_sanitize').startswith('address')
> > > + cflags += '-DRTE_MALLOC_ASAN'
> > > +endif
> > > +  
> > 
> > This looks great, but can we make it just do-the-right-thing and get rid of 
> > the
> > nerd knobs (i.e no meson configure).
> > 
> > The address sanitizer already has a way to detect if enabled.
> > 
> > GCC uses:
> > __SANITIZE_ADDRESS__
> > 
> > Clang uses:
> > #if defined(__has_feature)
> > #  if __has_feature(address_sanitizer)  
> 
> Tried this method you said. It can run successfully. Because gcc and clang 
> have different
> Methods for determining whether to turn on the asan function, so if you judge 
> the two
> methods in the code, it feels not simple to judge in meson.

There is already compiler specific #ifdef's why not do this contained in one 
header file?

The point is DPDK is trying to get away from having configuration settings if 
at all
possible. Configuration creates dependency nightmares and also leaves many code 
paths
as never tested.

Re: [dpdk-dev] [PATCH] net/af_packet: remove timestamp from packet status

[Stephen Hemminger]

On Mon, 13 Sep 2021 17:09:11 +0300
Tudor Cornea  wrote:

> +static inline bool tx_ring_status_unavailable(uint32_t tp_status)
> +{
> +#if KERNEL_VERSION(5, 10, 0) > LINUX_VERSION_CODE

No, having kernel dependent userspace in DPDK is not good practice.

Distribution vendors don't number their kernels the same as upstream.
RHEL for example, keeps same version over life or release but backports
many fixes.

Also, the system DPDK runs on is often not the system DPDK is built
on.

Re: [dpdk-dev] [PATCH] examples: ipv4, udp and tcp checksum offload warning

[Stephen Hemminger]

On Mon, 13 Sep 2021 17:09:51 +0500
usamanadeem321  wrote:

>  
> + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_IPV4_CKSUM)
> + local_port_conf.rxmode.offloads |=
> +  DEV_RX_OFFLOAD_IPV4_CKSUM;
> + else {
> + rte_exit(EXIT_FAILURE,
> + "IPV4 Checksum offload not available. (port %u) 
> ",
> + portid);
> + }
> +

Why not just do it in software if not available. The IPv4 checksum
is so cheap many operating systems just always do it in software.

Re: [dpdk-dev] [PATCH] Enable AddressSanitizer feature on DPDK

[Bruce Richardson]

On Mon, Sep 13, 2021 at 08:05:58AM -0700, Stephen Hemminger wrote:
> On Mon, 13 Sep 2021 05:27:12 +
> "Peng, ZhihongX"  wrote:
> 
> > > -Original Message-
> > > From: Stephen Hemminger 
> > > Sent: Friday, September 10, 2021 10:48 AM
> > > To: Peng, ZhihongX 
> > > Cc: Burakov, Anatoly ; Ananyev, Konstantin
> > > ; dev@dpdk.org; Lin, Xueqin
> > > 
> > > Subject: Re: [PATCH] Enable AddressSanitizer feature on DPDK
> > > 
> > > On Fri, 10 Sep 2021 02:01:47 +
> > > zhihongx.p...@intel.com wrote:
> > >   
> > > >
> > > > +if get_option('b_sanitize').startswith('address')
> > > > +   cflags += '-DRTE_MALLOC_ASAN'
> > > > +endif
> > > > +  
> > > 
> > > This looks great, but can we make it just do-the-right-thing and get rid 
> > > of the
> > > nerd knobs (i.e no meson configure).
> > > 

There are no new meson options being added here. Turning on/off address
sanitizing is a built-in meson option that is there already.

> > > The address sanitizer already has a way to detect if enabled.
> > > 
> > > GCC uses:
> > > __SANITIZE_ADDRESS__
> > > 
> > > Clang uses:
> > > #if defined(__has_feature)
> > > #  if __has_feature(address_sanitizer)  
> > 
> > Tried this method you said. It can run successfully. Because gcc and clang 
> > have different
> > Methods for determining whether to turn on the asan function, so if you 
> > judge the two
> > methods in the code, it feels not simple to judge in meson.
> 
> There is already compiler specific #ifdef's why not do this contained in one 
> header file?
> 
> The point is DPDK is trying to get away from having configuration settings if 
> at all
> possible. Configuration creates dependency nightmares and also leaves many 
> code paths
> as never tested.

Not sure I follow your point here. We need some macro to easily tell if we
are running with address sanitization enabled or not, so as to avoid having
the multi-compiler detection rules all over the place. The only question is
where it's better to have this in a header file or a meson.build file.
Given your objection and the fact that the meson.build code above looks a
little awkward, I'd suggest putting the conditional checks in malloc_elem.h.

Is something like the following what you had in mind?

  #ifdef __SANITIZE_ADDRESS__
  #define RTE_MALLOC_ASAN
  #elif defined(__has_feature) && __has_feature(address_sanitizer)
  #define RTE_MALLOC_ASAN
  #endif

/Bruce

Re: [dpdk-dev] [DPDK] net/virtio: fix check scatter on all Rx queues

[Maxime Coquelin]

Hi Zhihong,

On 8/4/21 10:31 AM, zhihongx.p...@intel.com wrote:
> From: Zhihong Peng 
> 
> This patch fixes the wrong way to obtain virtqueue.
> The end of virtqueue cannot be judged based on whether
> the array is NULL.

My understanding is that it is causing issue because it is confusing the
control queue with a Rx queue? I so, please be more specific on the
issue it is fixing in the commit message.

> Fixes: 4e8169eb0d2d (net/virtio: fix Rx scatter offload)
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Zhihong Peng 
> ---
>  drivers/net/virtio/virtio_ethdev.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/virtio/virtio_ethdev.c 
> b/drivers/net/virtio/virtio_ethdev.c
> index e58085a2c9..f2d19dc9d6 100644
> --- a/drivers/net/virtio/virtio_ethdev.c
> +++ b/drivers/net/virtio/virtio_ethdev.c
> @@ -873,8 +873,8 @@ virtio_check_scatter_on_all_rx_queues(struct rte_eth_dev 
> *dev,
>   if (hw->vqs == NULL)
>   return true;
>  
> - for (qidx = 0; (vq = hw->vqs[2 * qidx + VTNET_SQ_RQ_QUEUE_IDX]) != NULL;
> -  qidx++) {
> + for (qidx = 0; qidx < hw->max_queue_pairs; qidx++) {
> + vq = hw->vqs[2 * qidx + VTNET_SQ_RQ_QUEUE_IDX];

I agree with the change, but I would add a check to ensure vq is not
NULL to be safe wrt to NULL pointer dereferencing.

>   rxvq = &vq->rxq;
>   if (rxvq->mpool == NULL)
>   continue;
> 

Thanks,
Maxime

Re: [dpdk-dev] [PATCH 1/2] net/virtio: fix Tx cleanup functions to have same signature

[Maxime Coquelin]

On 8/18/21 4:13 PM, Andrew Rybchenko wrote:
> From: Ivan Ilchenko 
> 
> There is a family of cleanup from completed transmits functions.
> Fix packed virtqueues cleanup functions to have the same signature
> as split virtqueues have. This lets all functions of the family to
> match the same callback prototype.
> 
> Fixes: 892dc798fa9 ("net/virtio: implement Tx path for packed queues")
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Ivan Ilchenko 
> Signed-off-by: Andrew Rybchenko 
> ---
>  drivers/net/virtio/virtqueue.h | 11 ++-
>  1 file changed, 6 insertions(+), 5 deletions(-)
> 

Reviewed-by: Maxime Coquelin 

Thanks,
Maxime

Re: [dpdk-dev] [PATCH 2/2] net/virtio: fix Tx completed mbufs leak on device stop

[Maxime Coquelin]

On 8/18/21 4:13 PM, Andrew Rybchenko wrote:
> From: Ivan Ilchenko 
> 
> Free Tx completed mbufs on device stop. Not completed Tx mbufs cannot be
> freed since they are still in use.
> 
> Fixes: c1f86306a02 ("virtio: add new driver")
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Ivan Ilchenko 
> Signed-off-by: Andrew Rybchenko 
> ---
>  drivers/net/virtio/virtio_ethdev.c | 30 ++
>  1 file changed, 30 insertions(+)
> 
> diff --git a/drivers/net/virtio/virtio_ethdev.c 
> b/drivers/net/virtio/virtio_ethdev.c
> index e58085a2c9..ed3fefee7c 100644
> --- a/drivers/net/virtio/virtio_ethdev.c
> +++ b/drivers/net/virtio/virtio_ethdev.c
> @@ -2393,6 +2393,34 @@ static void virtio_dev_free_mbufs(struct rte_eth_dev 
> *dev)
>   PMD_INIT_LOG(DEBUG, "%d mbufs freed", mbuf_num);
>  }
>  
> +static void
> +virtio_tx_completed_cleanup(struct rte_eth_dev *dev)
> +{
> + struct virtio_hw *hw = dev->data->dev_private;
> + struct virtqueue *vq;
> + int qidx;
> + void (*xmit_cleanup)(struct virtqueue *vq, uint16_t nb_used);
> +
> + if (virtio_with_packed_queue(hw)) {
> + if (hw->use_vec_tx)
> + xmit_cleanup = &virtio_xmit_cleanup_inorder_packed;
> + else if (virtio_with_feature(hw, VIRTIO_F_IN_ORDER))
> + xmit_cleanup = &virtio_xmit_cleanup_inorder_packed;
> + else
> + xmit_cleanup = &virtio_xmit_cleanup_normal_packed;
> + } else {
> + if (hw->use_inorder_tx)
> + xmit_cleanup = &virtio_xmit_cleanup_inorder;
> + else
> + xmit_cleanup = &virtio_xmit_cleanup;
> + }
> +
> + for (qidx = 0; qidx < hw->max_queue_pairs; qidx++) {
> + vq = hw->vqs[2 * qidx + VTNET_SQ_TQ_QUEUE_IDX];

Maybe add a check to ensure that vq is non-NULL since it is dereferenced
later without checking.

> + xmit_cleanup(vq, virtqueue_nused(vq));
> + }
> +}
> +
>  /*
>   * Stop device: disable interrupt and mark link down
>   */
> @@ -2411,6 +2439,8 @@ virtio_dev_stop(struct rte_eth_dev *dev)
>   goto out_unlock;
>   hw->started = 0;
>  
> + virtio_tx_completed_cleanup(dev);
> +
>   if (intr_conf->lsc || intr_conf->rxq) {
>   virtio_intr_disable(dev);
>  
> 

Re: [dpdk-dev] DPDK Max Mbuf Allocation

[Ferruh Yigit]

On 9/13/2021 5:56 AM, Kamaraj P wrote:
> Hello All,
> 
> Would like to understand or if there are any guidelines to allocate the max
> no of mbuf per NIC ?
> For example, if i have defined as below:
> #define RX_RING_SIZE 1024
> #define TX_RING_SIZE 1024
> 
> The Maximum RX/TX queues can be defined as 8 per NIC, What would be the max
> no of mbuf can be allocated per NIC ?
> Please share if there is any guildliness or any limitation to increase the
> mbuf ?
> 

Hi Kamaraj,

Max number of the queues and max number of the descriptors per queue depends to
HW and changes form HW to HW.
This information is shared by the PMDs that application needs to take into
account. For example the descriptor limitations are provided by
'rx_desc_lim'/'tx_desc_lim' etc.

After descriptor number is defined, testpmd uses the mbuf count as following,
which can be taken as sample:

nb_mbuf_per_pool = RTE_TEST_RX_DESC_MAX + RTE_TEST_TX_DESC_MAX + MAX_PKT_BURST +
   (nb_lcores * mb_mempool_cache);




> Thanks,
> Kamaraj
> 

Re: [dpdk-dev] [PATCH] Enable AddressSanitizer feature on DPDK

[Stephen Hemminger]

On Mon, 13 Sep 2021 16:22:13 +0100
Bruce Richardson  wrote:

> On Mon, Sep 13, 2021 at 08:05:58AM -0700, Stephen Hemminger wrote:
> > On Mon, 13 Sep 2021 05:27:12 +
> > "Peng, ZhihongX"  wrote:
> >   
> > > > -Original Message-
> > > > From: Stephen Hemminger 
> > > > Sent: Friday, September 10, 2021 10:48 AM
> > > > To: Peng, ZhihongX 
> > > > Cc: Burakov, Anatoly ; Ananyev, Konstantin
> > > > ; dev@dpdk.org; Lin, Xueqin
> > > > 
> > > > Subject: Re: [PATCH] Enable AddressSanitizer feature on DPDK
> > > > 
> > > > On Fri, 10 Sep 2021 02:01:47 +
> > > > zhihongx.p...@intel.com wrote:
> > > > 
> > > > >
> > > > > +if get_option('b_sanitize').startswith('address')
> > > > > + cflags += '-DRTE_MALLOC_ASAN'
> > > > > +endif
> > > > > +
> > > > 
> > > > This looks great, but can we make it just do-the-right-thing and get 
> > > > rid of the
> > > > nerd knobs (i.e no meson configure).
> > > >   
> 
> There are no new meson options being added here. Turning on/off address
> sanitizing is a built-in meson option that is there already.
> 
> > > > The address sanitizer already has a way to detect if enabled.
> > > > 
> > > > GCC uses:
> > > > __SANITIZE_ADDRESS__
> > > > 
> > > > Clang uses:
> > > > #if defined(__has_feature)
> > > > #  if __has_feature(address_sanitizer)
> > > 
> > > Tried this method you said. It can run successfully. Because gcc and 
> > > clang have different
> > > Methods for determining whether to turn on the asan function, so if you 
> > > judge the two
> > > methods in the code, it feels not simple to judge in meson.  
> > 
> > There is already compiler specific #ifdef's why not do this contained in 
> > one header file?
> > 
> > The point is DPDK is trying to get away from having configuration settings 
> > if at all
> > possible. Configuration creates dependency nightmares and also leaves many 
> > code paths
> > as never tested.  
> 
> Not sure I follow your point here. We need some macro to easily tell if we
> are running with address sanitization enabled or not, so as to avoid having
> the multi-compiler detection rules all over the place. The only question is
> where it's better to have this in a header file or a meson.build file.
> Given your objection and the fact that the meson.build code above looks a
> little awkward, I'd suggest putting the conditional checks in malloc_elem.h.

NVM working of meson as commn base seems like good option.

Re: [dpdk-dev] [PATCH] examples/vhost: change the default value of NIC's max queues

[Maxime Coquelin]

On 9/10/21 5:17 AM, Xia, Chenbo wrote:
> Hi Wenwu,
> 
>> -Original Message-
>> From: Ma, WenwuX 
>> Sent: Friday, September 10, 2021 9:52 PM
>> To: dev@dpdk.org
>> Cc: maxime.coque...@redhat.com; Xia, Chenbo ; Jiang,
>> Cheng1 ; Hu, Jiayu ; Wang, Yinan
>> ; Ma, WenwuX 
>> Subject: [PATCH] examples/vhost: change the default value of NIC's max queues
>>
>> vswitch can't launch with 40G FTV due to Device start fails
> 
> Not many people can understand what's FTV. So let's describe it with a driver
> name. Example if it's 'i40e':
> 
> vswitch can't launch with a 40G i40e port...
> 
> And Device -> device
> 
>> if NIC’s max queues > the default number of 128,
>> so, we changed the default value from 128 to 512.
>>
> 
> I'd say it's not cool to still hard-code the MAX_QUEUES so that only 'some' 
> NICs
> can work with the example. The app should have a way to check this kind of 
> info 
> before init/start. But as I would like to see at some point, this example will
> be removed and all our tests go to testpmd. Let's not waste too much effort on
> this example.

+1 on this, I agree with Chenbo it is better to invest time in porting
existing tests to testpmd.

> 
> Besides: it can be a fix. Let's backport it.
> 
> Thanks,
> Chenbo
>  
>> Signed-off-by: Wenwu Ma 
>> ---
>>  examples/vhost/main.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/examples/vhost/main.c b/examples/vhost/main.c
>> index bc3d71c898..36969a4de5 100644
>> --- a/examples/vhost/main.c
>> +++ b/examples/vhost/main.c
>> @@ -29,7 +29,7 @@
>>  #include "main.h"
>>
>>  #ifndef MAX_QUEUES
>> -#define MAX_QUEUES 128
>> +#define MAX_QUEUES 512
>>  #endif
>>
>>  /* the maximum number of external ports supported */
>> --
>> 2.25.1
> 

Re: [dpdk-dev] DPDK Max Mbuf Allocation

[Stephen Hemminger]

On Mon, 13 Sep 2021 16:43:18 +0100
Ferruh Yigit  wrote:

> On 9/13/2021 5:56 AM, Kamaraj P wrote:
> > Hello All,
> > 
> > Would like to understand or if there are any guidelines to allocate the max
> > no of mbuf per NIC ?
> > For example, if i have defined as below:
> > #define RX_RING_SIZE 1024
> > #define TX_RING_SIZE 1024
> > 
> > The Maximum RX/TX queues can be defined as 8 per NIC, What would be the max
> > no of mbuf can be allocated per NIC ?
> > Please share if there is any guildliness or any limitation to increase the
> > mbuf ?
> >   
> 
> Hi Kamaraj,
> 
> Max number of the queues and max number of the descriptors per queue depends 
> to
> HW and changes form HW to HW.
> This information is shared by the PMDs that application needs to take into
> account. For example the descriptor limitations are provided by
> 'rx_desc_lim'/'tx_desc_lim' etc.
> 
> After descriptor number is defined, testpmd uses the mbuf count as following,
> which can be taken as sample:
> 
> nb_mbuf_per_pool = RTE_TEST_RX_DESC_MAX + RTE_TEST_TX_DESC_MAX + 
> MAX_PKT_BURST +
>(nb_lcores * mb_mempool_cache);
> 

It is a a little more complicated since some devices (like bnxt) allocate
multiple mbuf's per packet. Something like

 nb_mbuf_per_pool = MAX_RX_QUEUES * (RTE_TEST_RX_DESC_MAX * MBUF_PER_RX + 
MBUF_PER_Q)
+ MAX_TX_QUEUE * RTE_TEST_TX_DESC_MAX * MBUF_PER_TX
+ nb_lcores * MAX_PKT_BURST
+ nb_lcores * mb_mempool_cache
+ nb_lcores * PKTMBUF_POOL_RESERVED;

Ended up with
   MBUF_PER_RX = 3
   MBUF_PER_Q  = 6
and when using jumbo
   MBUF_PER_TX = MAX_MTU / MBUF_DATA_SIZE = 2

Re: [dpdk-dev] [PATCH] vhost: promote some APIs to stable

[Maxime Coquelin]

On 9/9/21 1:19 PM, Kinsella, Ray wrote:
> 
> 
> On 09/09/2021 03:13, Xia, Chenbo wrote:
>> Hi Kevin,
>>
>>> -Original Message-
>>> From: Kevin Traynor 
>>> Sent: Wednesday, September 8, 2021 8:01 PM
>>> To: Xia, Chenbo ; dev@dpdk.org;
>>> maxime.coque...@redhat.com
>>> Cc: Liu, Changpeng ; m...@ashroe.eu
>>> Subject: Re: [dpdk-dev] [PATCH] vhost: promote some APIs to stable
>>>
>>> On 07/09/2021 03:58, Chenbo Xia wrote:
 As reported by symbol bot, APIs listed in this patch have been
 experimental for more than two years. This patch promotes these
 18 APIs to stable.

>>>
>>> Patch lgtm. One question about a possible follow on below.
>>>
>>> Acked-by: Kevin Traynor 
>>>
 Signed-off-by: Chenbo Xia 
 ---
  lib/vhost/rte_vhost.h| 13 -
  lib/vhost/rte_vhost_crypto.h |  5 -
  lib/vhost/version.map| 36 ++--
  3 files changed, 18 insertions(+), 36 deletions(-)

>>
>> [...]
>>
 diff --git a/lib/vhost/version.map b/lib/vhost/version.map
 index c92a9d4962..8ebde3f694 100644
 --- a/lib/vhost/version.map
 +++ b/lib/vhost/version.map
 @@ -13,6 +13,13 @@ DPDK_22 {
rte_vdpa_reset_stats;
rte_vdpa_unregister_device;
rte_vhost_avail_entries;
 +  rte_vhost_clr_inflight_desc_packed;
 +  rte_vhost_clr_inflight_desc_split;
 +  rte_vhost_crypto_create;
 +  rte_vhost_crypto_fetch_requests;
 +  rte_vhost_crypto_finalize_requests;
 +  rte_vhost_crypto_free;
 +  rte_vhost_crypto_set_zero_copy;
rte_vhost_dequeue_burst;
rte_vhost_driver_attach_vdpa_device;
rte_vhost_driver_callback_register;
 @@ -20,13 +27,17 @@ DPDK_22 {
rte_vhost_driver_disable_features;
rte_vhost_driver_enable_features;
rte_vhost_driver_get_features;
 +  rte_vhost_driver_get_protocol_features;
 +  rte_vhost_driver_get_queue_num;
rte_vhost_driver_get_vdpa_device;
rte_vhost_driver_register;
rte_vhost_driver_set_features;
 +  rte_vhost_driver_set_protocol_features;
rte_vhost_driver_start;
rte_vhost_driver_unregister;
rte_vhost_enable_guest_notification;
rte_vhost_enqueue_burst;
 +  rte_vhost_extern_callback_register;
rte_vhost_get_ifname;
rte_vhost_get_log_base;
rte_vhost_get_mem_table;
 @@ -35,15 +46,22 @@ DPDK_22 {
rte_vhost_get_numa_node;
rte_vhost_get_queue_num;
rte_vhost_get_vdpa_device;
 +  rte_vhost_get_vhost_ring_inflight;
rte_vhost_get_vhost_vring;
rte_vhost_get_vring_base;
 +  rte_vhost_get_vring_base_from_inflight;
rte_vhost_get_vring_num;
>>>
rte_vhost_gpa_to_vva;
>>>
>>> Can this ^^^ be also removed now that rte_vhost_va_from_guest_pa() is
>>> promoted to non-experimental? It is marked as deprecated in API (see
>>> below) but i don't see anything in the deprecation documentation.
>>
>> Good point. I think it can be removed now. But we didn't send the deprecation
>> notice last release. I am not sure if it's ok to remove it this release.
>>
>> @Ray & Maxime,
>>
>> What do you think? I think since this API is unsafe and the safe version is
>> promoted, it makes sense to remove this.
> 
> Strictly speaking there should have been depreciation notice. 
> However if the API has been marked depreciated since 2018 and _is_ unsafe.
> You'd have to imagine that is sufficient to warrant removal at this stage. 
> 
> Thomas, David and Ferruh - any inputs/comments or objections?

I aagree it can be removed in this release. SPDK project was the only
user I'm aware of, and they migrated to the safe variant long time ago.

I propose to apply this patch first, then I will post a patch removing
this deprecated symbol if nobody disagree.

Thanks,
Maxime

>>
>> Thanks,
>> Chenbo
>>
>>>
>>> commit 9553e6e408883b3677e208dc66049bcd7f758529
>>> Author: Maxime Coquelin 
>>> Date:   Wed Mar 14 17:31:25 2018 +0100
>>>
>>> vhost: deprecate unsafe GPA translation API
>>>
>>> This patch marks rte_vhost_gpa_to_vva() as deprecated because
>>> it is unsafe. Application relying on this API should move
>>> to the new rte_vhost_va_from_guest_pa() API, and check
>>> returned length to avoid out-of-bound accesses.
>>>
>>> This issue has been assigned CVE-2018-1059.
>>>
>>> Signed-off-by: Maxime Coquelin 
>>>
>>>
rte_vhost_host_notifier_ctrl;
rte_vhost_log_used_vring;
rte_vhost_log_write;
rte_vhost_rx_queue_count;
 +  rte_vhost_set_inflight_desc_packed;
 +  rte_vhost_set_inflight_desc_split;
 +  rte_vhost_set_last_inflight_io_packed;
 +  rte_vhost_set_last_inflight_io_split;
rte_vhost_set_vring_base;
 +  rte_vhost_va_from_guest_pa;
rte_vhost_vring_call;

local: *;
 @@ -52,25 +70,7 @@ DPDK_22 {
  EXPERIMENTAL {
global:

 -  rte_vhost_driver_get_protocol_features;

[dpdk-dev] [PATCH v2 00/18] fix spelling errors

[Stephen Hemminger]

After finding a few typos in BPF, decided to cast a
wider net and fix lots more places.  These are all contained
to typos in strings or comments.

v2 - more spelling fixes in bpf library

Stephen Hemminger (18):
  node: fix typos
  ipsec: fix spelling errors
  eventdev: fix typo in comment
  power: fix typo in comment
  rcu: fix typo in comment
  bpf: fix spelling in comments
  acl: fix spelling errors
  eal: fix typos in comments
  net: fix spelling error in gtp comment
  bbdev: fix typo in comment
  pipeline: fix spellin error in comment
  hash: fix typo in comment
  rib: fix typo in comment
  sched: fix typo in comment
  vhost: fix typo in comment
  ip_frag: fix typos
  test-pmd: fix spelling errors in comments
  test: fix spelling errors in comments

 app/test-pmd/cmdline.c|  2 +-
 app/test-pmd/config.c |  4 ++--
 app/test-pmd/icmpecho.c   |  2 +-
 app/test/test_atomic.c|  2 +-
 app/test/test_barrier.c   |  2 +-
 app/test/test_bpf.c   |  2 +-
 app/test/test_compressdev.c   |  2 +-
 app/test/test_func_reentrancy.c   |  6 +++---
 app/test/test_ipsec.c |  2 +-
 app/test/test_link_bonding.c  |  2 +-
 app/test/test_lpm.c   |  6 +++---
 app/test/test_lpm6.c  | 14 +++---
 app/test/test_malloc.c|  4 ++--
 app/test/test_mbuf.c  |  2 +-
 app/test/test_mempool.c   |  2 +-
 app/test/test_mempool_perf.c  |  2 +-
 app/test/test_meter.c |  6 +++---
 app/test/test_pmd_perf.c  |  2 +-
 app/test/test_timer.c |  2 +-
 lib/acl/acl.h |  4 ++--
 lib/acl/acl_bld.c |  2 +-
 lib/acl/acl_run_avx2.h|  2 +-
 lib/bbdev/rte_bbdev.c |  2 +-
 lib/bpf/bpf_jit_x86.c |  2 +-
 lib/bpf/bpf_load_elf.c|  2 +-
 lib/bpf/bpf_pkt.c |  2 +-
 lib/bpf/bpf_validate.c|  8 
 lib/eal/include/rte_function_versioning.h |  2 +-
 lib/eal/windows/include/fnmatch.h |  2 +-
 lib/eventdev/rte_event_timer_adapter.c|  2 +-
 lib/hash/rte_thash.c  |  2 +-
 lib/ip_frag/rte_ipv4_reassembly.c |  2 +-
 lib/ipsec/esp_inb.c   |  4 ++--
 lib/ipsec/esp_outb.c  |  2 +-
 lib/ipsec/ipsec_sad.c |  2 +-
 lib/ipsec/sa.c|  2 +-
 lib/ipsec/sa.h|  2 +-
 lib/net/rte_gtp.h |  6 +++---
 lib/node/ethdev_rx_priv.h |  2 +-
 lib/node/ethdev_tx_priv.h |  2 +-
 lib/node/ip4_rewrite_priv.h   |  2 +-
 lib/pipeline/rte_swx_pipeline.h   |  2 +-
 lib/power/power_acpi_cpufreq.c|  2 +-
 lib/rcu/rte_rcu_qsbr.h|  2 +-
 lib/rib/rte_rib6.c|  2 +-
 lib/sched/rte_sched.c |  2 +-
 lib/vhost/rte_vhost.h |  4 ++--
 47 files changed, 69 insertions(+), 69 deletions(-)

-- 
2.30.2

[dpdk-dev] [PATCH v2 01/18] node: fix typos

[Stephen Hemminger]

Spelling errors in comments including doxygen

Signed-off-by: Stephen Hemminger 
---
 lib/node/ethdev_rx_priv.h   | 2 +-
 lib/node/ethdev_tx_priv.h   | 2 +-
 lib/node/ip4_rewrite_priv.h | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/lib/node/ethdev_rx_priv.h b/lib/node/ethdev_rx_priv.h
index efcd66dd4817..76cf77d60315 100644
--- a/lib/node/ethdev_rx_priv.h
+++ b/lib/node/ethdev_rx_priv.h
@@ -71,7 +71,7 @@ struct ethdev_rx_node_main *ethdev_rx_get_node_data_get(void);
  *
  * Get the Ethernet Rx node.
  *
- * @retrun
+ * @return
  *   Pointer to the Ethernet Rx node.
  */
 struct rte_node_register *ethdev_rx_node_get(void);
diff --git a/lib/node/ethdev_tx_priv.h b/lib/node/ethdev_tx_priv.h
index 586bff44a737..9ced520bd209 100644
--- a/lib/node/ethdev_tx_priv.h
+++ b/lib/node/ethdev_tx_priv.h
@@ -50,7 +50,7 @@ struct ethdev_tx_node_main *ethdev_tx_node_data_get(void);
  *
  * Get the Ethernet Tx node.
  *
- * @retrun
+ * @return
  *   Pointer to the Ethernet Tx node.
  */
 struct rte_node_register *ethdev_tx_node_get(void);
diff --git a/lib/node/ip4_rewrite_priv.h b/lib/node/ip4_rewrite_priv.h
index 80f0abdc94eb..036469746b29 100644
--- a/lib/node/ip4_rewrite_priv.h
+++ b/lib/node/ip4_rewrite_priv.h
@@ -53,7 +53,7 @@ struct ip4_rewrite_node_main {
  *
  * Get the ipv4 rewrite node.
  *
- * @retrun
+ * @return
  *   Pointer to the ipv4 rewrite node.
  */
 struct rte_node_register *ip4_rewrite_node_get(void);
-- 
2.30.2

[dpdk-dev] [PATCH v2 02/18] ipsec: fix spelling errors

[Stephen Hemminger]

Lots of typos in comments found by codespell

Signed-off-by: Stephen Hemminger 
---
 lib/ipsec/esp_inb.c   | 4 ++--
 lib/ipsec/esp_outb.c  | 2 +-
 lib/ipsec/ipsec_sad.c | 2 +-
 lib/ipsec/sa.c| 2 +-
 lib/ipsec/sa.h| 2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/lib/ipsec/esp_inb.c b/lib/ipsec/esp_inb.c
index 2b1df6a032a8..846fc0ea71a2 100644
--- a/lib/ipsec/esp_inb.c
+++ b/lib/ipsec/esp_inb.c
@@ -415,7 +415,7 @@ trs_process_check(struct rte_mbuf *mb, struct rte_mbuf **ml,
 
 /*
  * packet checks for tunnel mode:
- * - same as for trasnport mode
+ * - same as for transport mode
  * - esp tail next proto contains expected for that SA value
  */
 static inline int32_t
@@ -501,7 +501,7 @@ trs_process_step3(struct rte_mbuf *mb)
 static inline void
 tun_process_step3(struct rte_mbuf *mb, uint64_t txof_msk, uint64_t txof_val)
 {
-   /* reset mbuf metatdata: L2/L3 len, packet type */
+   /* reset mbuf metadata: L2/L3 len, packet type */
mb->packet_type = RTE_PTYPE_UNKNOWN;
mb->tx_offload = (mb->tx_offload & txof_msk) | txof_val;
 
diff --git a/lib/ipsec/esp_outb.c b/lib/ipsec/esp_outb.c
index 1e181cf2cee4..0bf3cd6bd4da 100644
--- a/lib/ipsec/esp_outb.c
+++ b/lib/ipsec/esp_outb.c
@@ -525,7 +525,7 @@ cpu_outb_trs_pkt_prepare(const struct rte_ipsec_session *ss,
 
 /*
  * process outbound packets for SA with ESN support,
- * for algorithms that require SQN.hibits to be implictly included
+ * for algorithms that require SQN.hibits to be implicitly included
  * into digest computation.
  * In that case we have to move ICV bytes back to their proper place.
  */
diff --git a/lib/ipsec/ipsec_sad.c b/lib/ipsec/ipsec_sad.c
index 3f9533c80a68..531e1e323cdc 100644
--- a/lib/ipsec/ipsec_sad.c
+++ b/lib/ipsec/ipsec_sad.c
@@ -62,7 +62,7 @@ EAL_REGISTER_TAILQ(rte_ipsec_sad_tailq)
  * Inserts a rule into an appropriate hash table,
  * updates the value for a given SPI in SPI_ONLY hash table
  * reflecting presence of more specific rule type in two LSBs.
- * Updates a counter that reflects the number of rules whith the same SPI.
+ * Updates a counter that reflects the number of rules with the same SPI.
  */
 static inline int
 add_specific(struct rte_ipsec_sad *sad, const void *key,
diff --git a/lib/ipsec/sa.c b/lib/ipsec/sa.c
index e59189d215b3..f49b3ec15d4f 100644
--- a/lib/ipsec/sa.c
+++ b/lib/ipsec/sa.c
@@ -126,7 +126,7 @@ ipsec_sa_size(uint64_t type, uint32_t *wnd_sz, uint32_t 
*nb_bucket)
/*
 * RFC 4303 recommends 64 as minimum window size.
 * there is no point to use ESN mode without SQN window,
-* so make sure we have at least 64 window when ESN is enalbed.
+* so make sure we have at least 64 window when ESN is enabled.
 */
wsz = ((type & RTE_IPSEC_SATP_ESN_MASK) ==
RTE_IPSEC_SATP_ESN_DISABLE) ?
diff --git a/lib/ipsec/sa.h b/lib/ipsec/sa.h
index 1bffe751f586..b8ce4e958119 100644
--- a/lib/ipsec/sa.h
+++ b/lib/ipsec/sa.h
@@ -116,7 +116,7 @@ struct rte_ipsec_sa {
 * In case of SA handled by multiple threads *sqn* cacheline
 * could be shared by multiple cores.
 * To minimise performance impact, we try to locate in a separate
-* place from other frequently accesed data.
+* place from other frequently accessed data.
 */
union {
uint64_t outb;
-- 
2.30.2

[dpdk-dev] [PATCH v2 03/18] eventdev: fix typo in comment

[Stephen Hemminger]

Typo in comment found by codespell

Signed-off-by: Stephen Hemminger 
---
 lib/eventdev/rte_event_timer_adapter.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/eventdev/rte_event_timer_adapter.c 
b/lib/eventdev/rte_event_timer_adapter.c
index ee20b39f4b24..b5c2dc1427c6 100644
--- a/lib/eventdev/rte_event_timer_adapter.c
+++ b/lib/eventdev/rte_event_timer_adapter.c
@@ -493,7 +493,7 @@ event_buffer_flush(struct event_buffer *bufp, uint8_t 
dev_id, uint8_t port_id,
 
RTE_ASSERT(head_idx < EVENT_BUFFER_SZ && tail_idx < EVENT_BUFFER_SZ);
 
-   /* Determine the largest contigous run we can attempt to enqueue to the
+   /* Determine the largest contiguous run we can attempt to enqueue to the
 * event device.
 */
if (head_idx > tail_idx)
-- 
2.30.2

[dpdk-dev] [PATCH v2 04/18] power: fix typo in comment

[Stephen Hemminger]

More spelling errors caught by codespell

Signed-off-by: Stephen Hemminger 
---
 lib/power/power_acpi_cpufreq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/power/power_acpi_cpufreq.c b/lib/power/power_acpi_cpufreq.c
index 1e8aeb84037c..402ed8c99b42 100644
--- a/lib/power/power_acpi_cpufreq.c
+++ b/lib/power/power_acpi_cpufreq.c
@@ -159,7 +159,7 @@ power_get_available_freqs(struct acpi_power_info *pi)
goto out;
}
 
-   /* Store the available frequncies into power context */
+   /* Store the available frequencies into power context */
for (i = 0, pi->nb_freqs = 0; i < count; i++) {
POWER_DEBUG_TRACE("Lcore %u frequency[%d]: %s\n", pi->lcore_id,
i, freqs[i]);
-- 
2.30.2

[dpdk-dev] [PATCH v2 05/18] rcu: fix typo in comment

[Stephen Hemminger]

Another codespell find.

Signed-off-by: Stephen Hemminger 
---
 lib/rcu/rte_rcu_qsbr.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/rcu/rte_rcu_qsbr.h b/lib/rcu/rte_rcu_qsbr.h
index 18811c1cc106..62a420a785fc 100644
--- a/lib/rcu/rte_rcu_qsbr.h
+++ b/lib/rcu/rte_rcu_qsbr.h
@@ -362,7 +362,7 @@ rte_rcu_qsbr_thread_offline(struct rte_rcu_qsbr *v, 
unsigned int thread_id)
 
/* The reader can go offline only after the load of the
 * data structure is completed. i.e. any load of the
-* data strcture can not move after this store.
+* data structure can not move after this store.
 */
 
__atomic_store_n(&v->qsbr_cnt[thread_id].cnt,
-- 
2.30.2

[dpdk-dev] [PATCH v2 06/18] bpf: fix spelling in comments

[Stephen Hemminger]

Found by running codespell on the bpf implementation.

Signed-off-by: Stephen Hemminger 
---
 lib/bpf/bpf_jit_x86.c  | 2 +-
 lib/bpf/bpf_load_elf.c | 2 +-
 lib/bpf/bpf_pkt.c  | 2 +-
 lib/bpf/bpf_validate.c | 8 
 4 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/lib/bpf/bpf_jit_x86.c b/lib/bpf/bpf_jit_x86.c
index aa22ea78a01d..518513376a67 100644
--- a/lib/bpf/bpf_jit_x86.c
+++ b/lib/bpf/bpf_jit_x86.c
@@ -1245,7 +1245,7 @@ emit_epilog(struct bpf_jit_state *st)
uint32_t i;
int32_t spil, ofs;
 
-   /* if we allready have an epilog generate a jump to it */
+   /* if we already have an epilog generate a jump to it */
if (st->exit.num++ != 0) {
emit_abs_jmp(st, st->exit.off);
return;
diff --git a/lib/bpf/bpf_load_elf.c b/lib/bpf/bpf_load_elf.c
index 2b11adeb5eb1..02a5d8ba0d07 100644
--- a/lib/bpf/bpf_load_elf.c
+++ b/lib/bpf/bpf_load_elf.c
@@ -80,7 +80,7 @@ resolve_xsym(const char *sn, size_t ofs, struct ebpf_insn 
*ins, size_t ins_sz,
if (type == RTE_BPF_XTYPE_FUNC) {
 
/* we don't support multiple functions per BPF module,
-* so treat EBPF_PSEUDO_CALL to extrernal function
+* so treat EBPF_PSEUDO_CALL to external function
 * as an ordinary EBPF_CALL.
 */
if (ins[idx].src_reg == EBPF_PSEUDO_CALL) {
diff --git a/lib/bpf/bpf_pkt.c b/lib/bpf/bpf_pkt.c
index 6e8248f0d6e4..701e8e2c62a9 100644
--- a/lib/bpf/bpf_pkt.c
+++ b/lib/bpf/bpf_pkt.c
@@ -169,7 +169,7 @@ bpf_eth_cbh_add(struct bpf_eth_cbh *cbh, uint16_t port, 
uint16_t queue)
 }
 
 /*
- * BPF packet processing routinies.
+ * BPF packet processing routines.
  */
 
 static inline uint32_t
diff --git a/lib/bpf/bpf_validate.c b/lib/bpf/bpf_validate.c
index 7b1291b382e9..eb8016ef196c 100644
--- a/lib/bpf/bpf_validate.c
+++ b/lib/bpf/bpf_validate.c
@@ -1723,7 +1723,7 @@ static const struct bpf_ins_check ins_chk[UINT8_MAX + 1] 
= {
 
 /*
  * make sure that instruction syntax is valid,
- * and it fields don't violate partciular instrcution type restrictions.
+ * and its fields don't violate particular instruction type restrictions.
  */
 static const char *
 check_syntax(const struct ebpf_insn *ins)
@@ -1954,7 +1954,7 @@ log_loop(const struct bpf_verifier *bvf)
  * First pass goes though all instructions in the set, checks that each
  * instruction is a valid one (correct syntax, valid field values, etc.)
  * and constructs control flow graph (CFG).
- * Then deapth-first search is performed over the constructed graph.
+ * Then depth-first search is performed over the constructed graph.
  * Programs with unreachable instructions and/or loops will be rejected.
  */
 static int
@@ -1981,7 +1981,7 @@ validate(struct bpf_verifier *bvf)
 
/*
 * construct CFG, jcc nodes have to outgoing edges,
-* 'exit' nodes - none, all others nodes have exaclty one
+* 'exit' nodes - none, all other nodes have exactly one
 * outgoing edge.
 */
switch (ins->code) {
@@ -2251,7 +2251,7 @@ evaluate(struct bpf_verifier *bvf)
idx = get_node_idx(bvf, node);
op = ins[idx].code;
 
-   /* for jcc node make a copy of evaluatoion state */
+   /* for jcc node make a copy of evaluation state */
if (node->nb_edge > 1)
rc |= save_eval_state(bvf, node);
 
-- 
2.30.2

[dpdk-dev] [PATCH v2 07/18] acl: fix spelling errors

[Stephen Hemminger]

Lots of little typos in comments

Signed-off-by: Stephen Hemminger 
---
 lib/acl/acl.h  | 4 ++--
 lib/acl/acl_bld.c  | 2 +-
 lib/acl/acl_run_avx2.h | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/lib/acl/acl.h b/lib/acl/acl.h
index 4089ab2a0462..f5739a475cf0 100644
--- a/lib/acl/acl.h
+++ b/lib/acl/acl.h
@@ -45,7 +45,7 @@ struct rte_acl_bitset {
  * Each transition is 64 bit value with the following format:
  * | node_type_specific : 32 | node_type : 3 | node_addr : 29 |
  * For all node types except RTE_ACL_NODE_MATCH, node_addr is an index
- * to the start of the node in the transtions array.
+ * to the start of the node in the transitions array.
  * Few different node types are used:
  * RTE_ACL_NODE_MATCH:
  * node_addr value is and index into an array that contains the return value
@@ -66,7 +66,7 @@ struct rte_acl_bitset {
  * RTE_ACL_NODE_SINGLE:
  * always transitions to the same node regardless of the input value.
  * RTE_ACL_NODE_DFA:
- * that node consits of up to 256 transitions.
+ * that node consists of up to 256 transitions.
  * In attempt to conserve space all transitions are divided into 4 consecutive
  * groups, by 64 transitions per group:
  * group64[i] contains transitions[i * 64, .. i * 64 + 63].
diff --git a/lib/acl/acl_bld.c b/lib/acl/acl_bld.c
index da10864cd870..f316d3e875ef 100644
--- a/lib/acl/acl_bld.c
+++ b/lib/acl/acl_bld.c
@@ -1494,7 +1494,7 @@ acl_set_data_indexes(struct rte_acl_ctx *ctx)
 /*
  * Internal routine, performs 'build' phase of trie generation:
  * - setups build context.
- * - analizes given set of rules.
+ * - analyzes given set of rules.
  * - builds internal tree(s).
  */
 static int
diff --git a/lib/acl/acl_run_avx2.h b/lib/acl/acl_run_avx2.h
index d06d2e8782d6..0b8967f22e70 100644
--- a/lib/acl/acl_run_avx2.h
+++ b/lib/acl/acl_run_avx2.h
@@ -125,7 +125,7 @@ acl_process_matches_avx2x8(const struct rte_acl_ctx *ctx,
/* For each transition: put low 32 into tr_lo and high 32 into tr_hi */
ACL_TR_HILO(mm256, __m256, t0, t1, lo, hi);
 
-   /* Keep transitions wth NOMATCH intact. */
+   /* Keep transitions with NOMATCH intact. */
*tr_lo = _mm256_blendv_epi8(*tr_lo, lo, matches);
*tr_hi = _mm256_blendv_epi8(*tr_hi, hi, matches);
 }
-- 
2.30.2

[dpdk-dev] [PATCH v2 08/18] eal: fix typos in comments

[Stephen Hemminger]

Minor spelling errors.

Signed-off-by: Stephen Hemminger 
---
 lib/eal/include/rte_function_versioning.h | 2 +-
 lib/eal/windows/include/fnmatch.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/eal/include/rte_function_versioning.h 
b/lib/eal/include/rte_function_versioning.h
index 746a1e19923e..eb6dd2bc1727 100644
--- a/lib/eal/include/rte_function_versioning.h
+++ b/lib/eal/include/rte_function_versioning.h
@@ -15,7 +15,7 @@
 
 /*
  * Provides backwards compatibility when updating exported functions.
- * When a symol is exported from a library to provide an API, it also provides 
a
+ * When a symbol is exported from a library to provide an API, it also 
provides a
  * calling convention (ABI) that is embodied in its name, return type,
  * arguments, etc.  On occasion that function may need to change to accommodate
  * new functionality, behavior, etc.  When that occurs, it is desirable to
diff --git a/lib/eal/windows/include/fnmatch.h 
b/lib/eal/windows/include/fnmatch.h
index 142753c3568d..c272f65ccdc3 100644
--- a/lib/eal/windows/include/fnmatch.h
+++ b/lib/eal/windows/include/fnmatch.h
@@ -30,7 +30,7 @@ extern "C" {
  * with the given regular expression pattern.
  *
  * @param pattern
- * regular expression notation decribing the pattern to match
+ * regular expression notation describing the pattern to match
  *
  * @param string
  * source string to searcg for the pattern
-- 
2.30.2

[dpdk-dev] [PATCH v2 09/18] net: fix spelling error in gtp comment

[Stephen Hemminger]

More codespell finds.

Signed-off-by: Stephen Hemminger 
---
 lib/net/rte_gtp.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/lib/net/rte_gtp.h b/lib/net/rte_gtp.h
index 6a6f9b238db9..f22dbce47c65 100644
--- a/lib/net/rte_gtp.h
+++ b/lib/net/rte_gtp.h
@@ -64,9 +64,9 @@ struct rte_gtp_hdr_ext_word {
 /** GTP header length */
 #define RTE_ETHER_GTP_HLEN \
(sizeof(struct rte_udp_hdr) + sizeof(struct rte_gtp_hdr))
-/* GTP next protocal type */
-#define RTE_GTP_TYPE_IPV4 0x40 /**< GTP next protocal type IPv4 */
-#define RTE_GTP_TYPE_IPV6 0x60 /**< GTP next protocal type IPv6 */
+/* GTP next protocol type */
+#define RTE_GTP_TYPE_IPV4 0x40 /**< GTP next protocol type IPv4 */
+#define RTE_GTP_TYPE_IPV6 0x60 /**< GTP next protocol type IPv6 */
 /* GTP destination port number */
 #define RTE_GTPC_UDP_PORT 2123 /**< GTP-C UDP destination port */
 #define RTE_GTPU_UDP_PORT 2152 /**< GTP-U UDP destination port */
-- 
2.30.2

[dpdk-dev] [PATCH v2 10/18] bbdev: fix typo in comment

[Stephen Hemminger]

Yet another spelling error found by codespell.

Signed-off-by: Stephen Hemminger 
---
 lib/bbdev/rte_bbdev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/bbdev/rte_bbdev.c b/lib/bbdev/rte_bbdev.c
index fc37236195be..49a0de49d8df 100644
--- a/lib/bbdev/rte_bbdev.c
+++ b/lib/bbdev/rte_bbdev.c
@@ -138,7 +138,7 @@ rte_bbdev_data_alloc(void)
 }
 
 /*
- * Find data alocated for the device or if not found return first unused bbdev
+ * Find data allocated for the device or if not found return first unused bbdev
  * data. If all structures are in use and none is used by the device return
  * NULL.
  */
-- 
2.30.2

[dpdk-dev] [PATCH v2 11/18] pipeline: fix spellin error in comment

[Stephen Hemminger]

Minor typo

Signed-off-by: Stephen Hemminger 
---
 lib/pipeline/rte_swx_pipeline.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/pipeline/rte_swx_pipeline.h b/lib/pipeline/rte_swx_pipeline.h
index cd395ac39dbb..9bfc69b2bb4f 100644
--- a/lib/pipeline/rte_swx_pipeline.h
+++ b/lib/pipeline/rte_swx_pipeline.h
@@ -620,7 +620,7 @@ struct rte_swx_pipeline_selector_params {
const char *group_id_field_name;
 
/** The set of fields used to select (through a hashing scheme) the
-* member within the current group. Inputs into the seletion operation.
+* member within the current group. Inputs into the selection operation.
 * Restriction: All the selector fields must be part of the same struct,
 * i.e. part of the same header or part of the meta-data structure.
 */
-- 
2.30.2

[dpdk-dev] [PATCH v2 12/18] hash: fix typo in comment

[Stephen Hemminger]

Yet another spelling error found by codespell.

Signed-off-by: Stephen Hemminger 
---
 lib/hash/rte_thash.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/hash/rte_thash.c b/lib/hash/rte_thash.c
index d5a95a6e009a..f5e9111f0264 100644
--- a/lib/hash/rte_thash.c
+++ b/lib/hash/rte_thash.c
@@ -26,7 +26,7 @@ EAL_REGISTER_TAILQ(rte_thash_tailq)
 
 /**
  * Table of some irreducible polinomials over GF(2).
- * For lfsr they are reperesented in BE bit order, and
+ * For lfsr they are represented in BE bit order, and
  * x^0 is masked out.
  * For example, poly x^5 + x^2 + 1 will be represented
  * as (101001b & 1b) = 01001b = 0x9
-- 
2.30.2

[dpdk-dev] [PATCH v2 13/18] rib: fix typo in comment

[Stephen Hemminger]

Yet another spelling error found by codespell.

Signed-off-by: Stephen Hemminger 
---
 lib/rib/rte_rib6.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/rib/rte_rib6.c b/lib/rib/rte_rib6.c
index 96424e9c9f4f..4a6f29b0040d 100644
--- a/lib/rib/rte_rib6.c
+++ b/lib/rib/rte_rib6.c
@@ -199,7 +199,7 @@ rte_rib6_lookup_exact(struct rte_rib6 *rib,
 }
 
 /*
- *  Traverses on subtree and retreeves more specific routes
+ *  Traverses on subtree and retrieves more specific routes
  *  for a given in args ip/depth prefix
  *  last = NULL means the first invocation
  */
-- 
2.30.2

[dpdk-dev] [PATCH v2 14/18] sched: fix typo in comment

[Stephen Hemminger]

Yet another spelling error found by codespell.

Signed-off-by: Stephen Hemminger 
---
 lib/sched/rte_sched.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/sched/rte_sched.c b/lib/sched/rte_sched.c
index a858f61f9548..329f4b2e7b48 100644
--- a/lib/sched/rte_sched.c
+++ b/lib/sched/rte_sched.c
@@ -579,7 +579,7 @@ rte_sched_subport_config_qsize(struct rte_sched_subport 
*subport)
 
subport->qsize_add[0] = 0;
 
-   /* Strict prority traffic class */
+   /* Strict priority traffic class */
for (i = 1; i < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; i++)
subport->qsize_add[i] = subport->qsize_add[i-1] + 
subport->qsize[i-1];
 
-- 
2.30.2

[dpdk-dev] [PATCH v2 15/18] vhost: fix typo in comment

[Stephen Hemminger]

Yet another spelling error found by codespell.

Signed-off-by: Stephen Hemminger 
---
 lib/vhost/rte_vhost.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/vhost/rte_vhost.h b/lib/vhost/rte_vhost.h
index 8d875e932297..c36dfc705b04 100644
--- a/lib/vhost/rte_vhost.h
+++ b/lib/vhost/rte_vhost.h
@@ -776,7 +776,7 @@ rte_vhost_get_vhost_ring_inflight(int vid, uint16_t 
vring_idx,
 /**
  * Set split inflight descriptor.
  *
- * This function save descriptors that has been comsumed in available
+ * This function save descriptors that has been consumed in available
  * ring
  *
  * @param vid
@@ -796,7 +796,7 @@ rte_vhost_set_inflight_desc_split(int vid, uint16_t 
vring_idx,
 /**
  * Set packed inflight descriptor and get corresponding inflight entry
  *
- * This function save descriptors that has been comsumed
+ * This function save descriptors that has been consumed
  *
  * @param vid
  *  vhost device ID
-- 
2.30.2

[dpdk-dev] [PATCH v2 16/18] ip_frag: fix typos

[Stephen Hemminger]

Minor typos in comment.

Signed-off-by: Stephen Hemminger 
---
 lib/ip_frag/rte_ipv4_reassembly.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/ip_frag/rte_ipv4_reassembly.c 
b/lib/ip_frag/rte_ipv4_reassembly.c
index 69666c8b8271..4a89a5f5365a 100644
--- a/lib/ip_frag/rte_ipv4_reassembly.c
+++ b/lib/ip_frag/rte_ipv4_reassembly.c
@@ -80,7 +80,7 @@ ipv4_frag_reassemble(struct ip_frag_pkt *fp)
 
 /*
  * Process new mbuf with fragment of IPV4 packet.
- * Incoming mbuf should have it's l2_len/l3_len fields setuped correclty.
+ * Incoming mbuf should have it's l2_len/l3_len fields setup correctly.
  * @param tbl
  *   Table where to lookup/add the fragmented packet.
  * @param mb
-- 
2.30.2

[dpdk-dev] [PATCH v2 17/18] test-pmd: fix spelling errors in comments

[Stephen Hemminger]

More typos

Signed-off-by: Stephen Hemminger 
---
 app/test-pmd/cmdline.c  | 2 +-
 app/test-pmd/config.c   | 4 ++--
 app/test-pmd/icmpecho.c | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 82253bc75110..2628bd64dae6 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -3670,7 +3670,7 @@ parse_item_list(const char *str, const char *item_name, 
unsigned int max_items,
return nb_item;
 
/*
-* Then, check that all values in the list are differents.
+* Then, check that all values in the list are different.
 * No optimization here...
 */
for (i = 0; i < nb_item; i++) {
diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c
index 31d8ba1b913c..d274c6669d50 100644
--- a/app/test-pmd/config.c
+++ b/app/test-pmd/config.c
@@ -2777,7 +2777,7 @@ port_rss_reta_info(portid_t port_id,
 }
 
 /*
- * Displays the RSS hash functions of a port, and, optionaly, the RSS hash
+ * Displays the RSS hash functions of a port, and, optionally, the RSS hash
  * key of the port.
  */
 void
@@ -4932,7 +4932,7 @@ mcast_addr_pool_remove(struct rte_port *port, uint32_t 
addr_idx)
 {
port->mc_addr_nb--;
if (addr_idx == port->mc_addr_nb) {
-   /* No need to recompact the set of multicast addressses. */
+   /* No need to recompact the set of multicast addresses. */
if (port->mc_addr_nb == 0) {
/* free the pool of multicast addresses. */
free(port->mc_addr_pool);
diff --git a/app/test-pmd/icmpecho.c b/app/test-pmd/icmpecho.c
index 8948f28eb5af..ebff5ac9794f 100644
--- a/app/test-pmd/icmpecho.c
+++ b/app/test-pmd/icmpecho.c
@@ -54,7 +54,7 @@ arp_op_name(uint16_t arp_op)
default:
break;
}
-   return "Unkwown ARP op";
+   return "Unknown ARP op";
 }
 
 static const char *
-- 
2.30.2

[dpdk-dev] [PATCH v2 18/18] test: fix spelling errors in comments

[Stephen Hemminger]

Lots of little typos.

Signed-off-by: Stephen Hemminger 
---
 app/test/test_atomic.c  |  2 +-
 app/test/test_barrier.c |  2 +-
 app/test/test_bpf.c |  2 +-
 app/test/test_compressdev.c |  2 +-
 app/test/test_func_reentrancy.c |  6 +++---
 app/test/test_ipsec.c   |  2 +-
 app/test/test_link_bonding.c|  2 +-
 app/test/test_lpm.c |  6 +++---
 app/test/test_lpm6.c| 14 +++---
 app/test/test_malloc.c  |  4 ++--
 app/test/test_mbuf.c|  2 +-
 app/test/test_mempool.c |  2 +-
 app/test/test_mempool_perf.c|  2 +-
 app/test/test_meter.c   |  6 +++---
 app/test/test_pmd_perf.c|  2 +-
 app/test/test_timer.c   |  2 +-
 16 files changed, 29 insertions(+), 29 deletions(-)

diff --git a/app/test/test_atomic.c b/app/test/test_atomic.c
index f10f555af8b4..7dfa10a70102 100644
--- a/app/test/test_atomic.c
+++ b/app/test/test_atomic.c
@@ -88,7 +88,7 @@
  *
  *   - Invoke ``test_atomic_exchange`` on each lcore.  Before doing
  * anything else, the cores wait for a synchronization event.
- * Each core then does the follwoing for N iterations:
+ * Each core then does the following for N iterations:
  *
  *   Generate a new token with a data integrity check
  *   Exchange the new token for previously generated token
diff --git a/app/test/test_barrier.c b/app/test/test_barrier.c
index c27f8a0742f2..8fa93c00329a 100644
--- a/app/test/test_barrier.c
+++ b/app/test/test_barrier.c
@@ -66,7 +66,7 @@ struct plock_test {
 struct lcore_plock_test {
struct plock_test *pt[2]; /* shared, lock-protected data */
uint64_t sum[2];  /* local copy of the shared data */
-   uint64_t iter;/* number of iterations to perfom */
+   uint64_t iter;/* number of iterations to perform */
uint32_t lc;  /* given lcore id */
 };
 
diff --git a/app/test/test_bpf.c b/app/test/test_bpf.c
index 527c06b80708..cf6e022f7a92 100644
--- a/app/test/test_bpf.c
+++ b/app/test/test_bpf.c
@@ -2391,7 +2391,7 @@ static const struct ebpf_insn test_call5_prog[] = {
},
 };
 
-/* String comparision impelementation, return 0 if equal else difference */
+/* String comparison implementation, return 0 if equal else difference */
 static uint32_t
 dummy_func5(const char *s1, const char *s2)
 {
diff --git a/app/test/test_compressdev.c b/app/test/test_compressdev.c
index 0571c17ecb8d..a7d04c02c6dd 100644
--- a/app/test/test_compressdev.c
+++ b/app/test/test_compressdev.c
@@ -2035,7 +2035,7 @@ test_deflate_comp_decomp(const struct interim_data_params 
*int_data,
test_priv_data.all_decomp_data = &all_decomp_data;
test_priv_data.decomp_produced_data_size = &decomp_produced_data_size;
 
-   test_priv_data.num_priv_xforms = 0; /* it's used for deompression only 
*/
+   test_priv_data.num_priv_xforms = 0; /* it's used for decompression only 
*/
 
capa = rte_compressdev_capability_get(0, RTE_COMP_ALGO_DEFLATE);
if (capa == NULL) {
diff --git a/app/test/test_func_reentrancy.c b/app/test/test_func_reentrancy.c
index 231c99a9ebc5..69ff421b1824 100644
--- a/app/test/test_func_reentrancy.c
+++ b/app/test/test_func_reentrancy.c
@@ -242,7 +242,7 @@ hash_create_free(__rte_unused void *arg)
rte_atomic32_inc(&obj_count);
}
 
-   /* create mutiple times simultaneously */
+   /* create multiple times simultaneously */
for (i = 0; i < MAX_ITER_MULTI; i++) {
snprintf(hash_name, sizeof(hash_name), "fr_test_%d_%d", 
lcore_self, i);
hash_params.name = hash_name;
@@ -306,7 +306,7 @@ fbk_create_free(__rte_unused void *arg)
rte_atomic32_inc(&obj_count);
}
 
-   /* create mutiple fbk tables simultaneously */
+   /* create multiple fbk tables simultaneously */
for (i = 0; i < MAX_ITER_MULTI; i++) {
snprintf(fbk_name, sizeof(fbk_name), "fr_test_%d_%d", 
lcore_self, i);
fbk_params.name = fbk_name;
@@ -368,7 +368,7 @@ lpm_create_free(__rte_unused void *arg)
rte_atomic32_inc(&obj_count);
}
 
-   /* create mutiple fbk tables simultaneously */
+   /* create multiple fbk tables simultaneously */
for (i = 0; i < MAX_LPM_ITER_TIMES; i++) {
snprintf(lpm_name, sizeof(lpm_name), "fr_test_%d_%d", 
lcore_self, i);
lpm = rte_lpm_create(lpm_name, SOCKET_ID_ANY, &config);
diff --git a/app/test/test_ipsec.c b/app/test/test_ipsec.c
index c6d6b88d6d9e..923bf9864412 100644
--- a/app/test/test_ipsec.c
+++ b/app/test/test_ipsec.c
@@ -653,7 +653,7 @@ create_crypto_session(struct ipsec_unitest_params *ut,
if (s == NULL)
return -ENOMEM;
 
-   /* initiliaze SA crypto session for device */
+   /* initialize SA crypto session for device */
rc = rte_cryptodev_sym_session_init(dev_id, s,
   

[dpdk-dev] [PATCH V3 01/24] pipeline: move data structures to internal header file

[Cristian Dumitrescu]

Start to consolidate the data structures and inline functions required
by the pipeline instructions into an internal header file.

Signed-off-by: Cristian Dumitrescu 
---
Depends-on: series-18297 ("[V4,1/4] table: add support learner tables")

V3:
-fixed meson.build

 lib/pipeline/rte_swx_pipeline.c  | 1373 +
 lib/pipeline/rte_swx_pipeline_internal.h | 1383 ++
 2 files changed, 1384 insertions(+), 1372 deletions(-)
 create mode 100644 lib/pipeline/rte_swx_pipeline_internal.h

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index f89a134a52..ae9b2056db 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -2,24 +2,11 @@
  * Copyright(c) 2020 Intel Corporation
  */
 #include 
-#include 
 #include 
 #include 
-#include 
-#include 
 #include 
 
-#include 
-#include 
-#include 
-#include 
-#include 
-
-#include 
-#include 
-
-#include "rte_swx_pipeline.h"
-#include "rte_swx_ctl.h"
+#include "rte_swx_pipeline_internal.h"
 
 #define CHECK(condition, err_code) 
\
 do {   
\
@@ -40,22 +27,9 @@ do { 
  \
   RTE_SWX_INSTRUCTION_SIZE),  \
  err_code)
 
-#ifndef TRACE_LEVEL
-#define TRACE_LEVEL 0
-#endif
-
-#if TRACE_LEVEL
-#define TRACE(...) printf(__VA_ARGS__)
-#else
-#define TRACE(...)
-#endif
-
 /*
  * Environment.
  */
-#define ntoh64(x) rte_be_to_cpu_64(x)
-#define hton64(x) rte_cpu_to_be_64(x)
-
 #ifndef RTE_SWX_PIPELINE_HUGE_PAGES_DISABLE
 
 #include 
@@ -103,1351 +77,6 @@ env_free(void *start, size_t size)
 
 #endif
 
-/*
- * Struct.
- */
-struct field {
-   char name[RTE_SWX_NAME_SIZE];
-   uint32_t n_bits;
-   uint32_t offset;
-   int var_size;
-};
-
-struct struct_type {
-   TAILQ_ENTRY(struct_type) node;
-   char name[RTE_SWX_NAME_SIZE];
-   struct field *fields;
-   uint32_t n_fields;
-   uint32_t n_bits;
-   uint32_t n_bits_min;
-   int var_size;
-};
-
-TAILQ_HEAD(struct_type_tailq, struct_type);
-
-/*
- * Input port.
- */
-struct port_in_type {
-   TAILQ_ENTRY(port_in_type) node;
-   char name[RTE_SWX_NAME_SIZE];
-   struct rte_swx_port_in_ops ops;
-};
-
-TAILQ_HEAD(port_in_type_tailq, port_in_type);
-
-struct port_in {
-   TAILQ_ENTRY(port_in) node;
-   struct port_in_type *type;
-   void *obj;
-   uint32_t id;
-};
-
-TAILQ_HEAD(port_in_tailq, port_in);
-
-struct port_in_runtime {
-   rte_swx_port_in_pkt_rx_t pkt_rx;
-   void *obj;
-};
-
-/*
- * Output port.
- */
-struct port_out_type {
-   TAILQ_ENTRY(port_out_type) node;
-   char name[RTE_SWX_NAME_SIZE];
-   struct rte_swx_port_out_ops ops;
-};
-
-TAILQ_HEAD(port_out_type_tailq, port_out_type);
-
-struct port_out {
-   TAILQ_ENTRY(port_out) node;
-   struct port_out_type *type;
-   void *obj;
-   uint32_t id;
-};
-
-TAILQ_HEAD(port_out_tailq, port_out);
-
-struct port_out_runtime {
-   rte_swx_port_out_pkt_tx_t pkt_tx;
-   rte_swx_port_out_flush_t flush;
-   void *obj;
-};
-
-/*
- * Extern object.
- */
-struct extern_type_member_func {
-   TAILQ_ENTRY(extern_type_member_func) node;
-   char name[RTE_SWX_NAME_SIZE];
-   rte_swx_extern_type_member_func_t func;
-   uint32_t id;
-};
-
-TAILQ_HEAD(extern_type_member_func_tailq, extern_type_member_func);
-
-struct extern_type {
-   TAILQ_ENTRY(extern_type) node;
-   char name[RTE_SWX_NAME_SIZE];
-   struct struct_type *mailbox_struct_type;
-   rte_swx_extern_type_constructor_t constructor;
-   rte_swx_extern_type_destructor_t destructor;
-   struct extern_type_member_func_tailq funcs;
-   uint32_t n_funcs;
-};
-
-TAILQ_HEAD(extern_type_tailq, extern_type);
-
-struct extern_obj {
-   TAILQ_ENTRY(extern_obj) node;
-   char name[RTE_SWX_NAME_SIZE];
-   struct extern_type *type;
-   void *obj;
-   uint32_t struct_id;
-   uint32_t id;
-};
-
-TAILQ_HEAD(extern_obj_tailq, extern_obj);
-
-#ifndef RTE_SWX_EXTERN_TYPE_MEMBER_FUNCS_MAX
-#define RTE_SWX_EXTERN_TYPE_MEMBER_FUNCS_MAX 8
-#endif
-
-struct extern_obj_runtime {
-   void *obj;
-   uint8_t *mailbox;
-   rte_swx_extern_type_member_func_t 
funcs[RTE_SWX_EXTERN_TYPE_MEMBER_FUNCS_MAX];
-};
-
-/*
- * Extern function.
- */
-struct extern_func {
-   TAILQ_ENTRY(extern_func) node;
-   char name[RTE_SWX_NAME_SIZE];
-   struct struct_type *mailbox_struct_type;
-   rte_swx_extern_func_t func;
-   uint32_t struct_id;
-   uint32_t id;
-};
-
-TAILQ_HEAD(extern_func_tailq, extern_func);
-
-struct extern_func_runtime {
-   uint8_t *mailbox;
-   rte_swx_extern_func_t func;
-};
-
-/*
- * Header.
- */
-struct header {
-   TAILQ_ENTRY(header) node;
-   char name[RTE_SWX_NAME_SIZE];
-   s

[dpdk-dev] [PATCH V3 02/24] pipeline: move thread inline functions to header file

[Cristian Dumitrescu]

Move the thread inline functions to the internal header file.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 56 --
 lib/pipeline/rte_swx_pipeline_internal.h | 59 
 2 files changed, 59 insertions(+), 56 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index ae9b2056db..7e01453c27 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1492,62 +1492,6 @@ struct_field_parse(struct rte_swx_pipeline *p,
}
 }
 
-static inline void
-pipeline_port_inc(struct rte_swx_pipeline *p)
-{
-   p->port_id = (p->port_id + 1) & (p->n_ports_in - 1);
-}
-
-static inline void
-thread_ip_reset(struct rte_swx_pipeline *p, struct thread *t)
-{
-   t->ip = p->instructions;
-}
-
-static inline void
-thread_ip_set(struct thread *t, struct instruction *ip)
-{
-   t->ip = ip;
-}
-
-static inline void
-thread_ip_action_call(struct rte_swx_pipeline *p,
- struct thread *t,
- uint32_t action_id)
-{
-   t->ret = t->ip + 1;
-   t->ip = p->action_instructions[action_id];
-}
-
-static inline void
-thread_ip_inc(struct rte_swx_pipeline *p);
-
-static inline void
-thread_ip_inc(struct rte_swx_pipeline *p)
-{
-   struct thread *t = &p->threads[p->thread_id];
-
-   t->ip++;
-}
-
-static inline void
-thread_ip_inc_cond(struct thread *t, int cond)
-{
-   t->ip += cond;
-}
-
-static inline void
-thread_yield(struct rte_swx_pipeline *p)
-{
-   p->thread_id = (p->thread_id + 1) & (RTE_SWX_PIPELINE_THREADS_MAX - 1);
-}
-
-static inline void
-thread_yield_cond(struct rte_swx_pipeline *p, int cond)
-{
-   p->thread_id = (p->thread_id + cond) & (RTE_SWX_PIPELINE_THREADS_MAX - 
1);
-}
-
 /*
  * rx.
  */
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 5d80dd8451..682f4c86a0 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -1380,4 +1380,63 @@ struct rte_swx_pipeline {
int numa_node;
 };
 
+/*
+ * Instruction.
+ */
+static inline void
+pipeline_port_inc(struct rte_swx_pipeline *p)
+{
+   p->port_id = (p->port_id + 1) & (p->n_ports_in - 1);
+}
+
+static inline void
+thread_ip_reset(struct rte_swx_pipeline *p, struct thread *t)
+{
+   t->ip = p->instructions;
+}
+
+static inline void
+thread_ip_set(struct thread *t, struct instruction *ip)
+{
+   t->ip = ip;
+}
+
+static inline void
+thread_ip_action_call(struct rte_swx_pipeline *p,
+ struct thread *t,
+ uint32_t action_id)
+{
+   t->ret = t->ip + 1;
+   t->ip = p->action_instructions[action_id];
+}
+
+static inline void
+thread_ip_inc(struct rte_swx_pipeline *p);
+
+static inline void
+thread_ip_inc(struct rte_swx_pipeline *p)
+{
+   struct thread *t = &p->threads[p->thread_id];
+
+   t->ip++;
+}
+
+static inline void
+thread_ip_inc_cond(struct thread *t, int cond)
+{
+   t->ip += cond;
+}
+
+static inline void
+thread_yield(struct rte_swx_pipeline *p)
+{
+   p->thread_id = (p->thread_id + 1) & (RTE_SWX_PIPELINE_THREADS_MAX - 1);
+}
+
+static inline void
+thread_yield_cond(struct rte_swx_pipeline *p, int cond)
+{
+   p->thread_id = (p->thread_id + cond) & (RTE_SWX_PIPELINE_THREADS_MAX - 
1);
+}
+
 #endif
-- 
2.17.1

[dpdk-dev] [PATCH V3 03/24] pipeline: create inline functions for RX instruction

[Cristian Dumitrescu]

Create inline functions for the RX instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 38 --
 lib/pipeline/rte_swx_pipeline_internal.h | 51 
 2 files changed, 51 insertions(+), 38 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 7e01453c27..ad1ecfc640 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1517,44 +1517,6 @@ instr_rx_translate(struct rte_swx_pipeline *p,
return 0;
 }
 
-static inline void
-instr_rx_exec(struct rte_swx_pipeline *p);
-
-static inline void
-instr_rx_exec(struct rte_swx_pipeline *p)
-{
-   struct thread *t = &p->threads[p->thread_id];
-   struct instruction *ip = t->ip;
-   struct port_in_runtime *port = &p->in[p->port_id];
-   struct rte_swx_pkt *pkt = &t->pkt;
-   int pkt_received;
-
-   /* Packet. */
-   pkt_received = port->pkt_rx(port->obj, pkt);
-   t->ptr = &pkt->pkt[pkt->offset];
-   rte_prefetch0(t->ptr);
-
-   TRACE("[Thread %2u] rx %s from port %u\n",
- p->thread_id,
- pkt_received ? "1 pkt" : "0 pkts",
- p->port_id);
-
-   /* Headers. */
-   t->valid_headers = 0;
-   t->n_headers_out = 0;
-
-   /* Meta-data. */
-   METADATA_WRITE(t, ip->io.io.offset, ip->io.io.n_bits, p->port_id);
-
-   /* Tables. */
-   t->table_state = p->table_state;
-
-   /* Thread. */
-   pipeline_port_inc(p);
-   thread_ip_inc_cond(t, pkt_received);
-   thread_yield(p);
-}
-
 /*
  * tx.
  */
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 682f4c86a0..9814b5685a 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -1439,4 +1439,55 @@ thread_yield_cond(struct rte_swx_pipeline *p, int cond)
p->thread_id = (p->thread_id + cond) & (RTE_SWX_PIPELINE_THREADS_MAX - 
1);
 }
 
+/*
+ * rx.
+ */
+static inline int
+__instr_rx_exec(struct rte_swx_pipeline *p, struct thread *t, const struct 
instruction *ip)
+{
+   struct port_in_runtime *port = &p->in[p->port_id];
+   struct rte_swx_pkt *pkt = &t->pkt;
+   int pkt_received;
+
+   /* Packet. */
+   pkt_received = port->pkt_rx(port->obj, pkt);
+   t->ptr = &pkt->pkt[pkt->offset];
+   rte_prefetch0(t->ptr);
+
+   TRACE("[Thread %2u] rx %s from port %u\n",
+ p->thread_id,
+ pkt_received ? "1 pkt" : "0 pkts",
+ p->port_id);
+
+   /* Headers. */
+   t->valid_headers = 0;
+   t->n_headers_out = 0;
+
+   /* Meta-data. */
+   METADATA_WRITE(t, ip->io.io.offset, ip->io.io.n_bits, p->port_id);
+
+   /* Tables. */
+   t->table_state = p->table_state;
+
+   /* Thread. */
+   pipeline_port_inc(p);
+
+   return pkt_received;
+}
+
+static inline void
+instr_rx_exec(struct rte_swx_pipeline *p)
+{
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
+   int pkt_received;
+
+   /* Packet. */
+   pkt_received = __instr_rx_exec(p, t, ip);
+
+   /* Thread. */
+   thread_ip_inc_cond(t, pkt_received);
+   thread_yield(p);
+}
+
 #endif
-- 
2.17.1

[dpdk-dev] [PATCH V3 04/24] pipeline: create inline functions for TX instruction

[Cristian Dumitrescu]

Create inline functions for the TX instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 86 +-
 lib/pipeline/rte_swx_pipeline_internal.h | 90 
 2 files changed, 92 insertions(+), 84 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index ad1ecfc640..bcf796f8c3 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1567,84 +1567,13 @@ instr_drop_translate(struct rte_swx_pipeline *p,
return 0;
 }
 
-static inline void
-emit_handler(struct thread *t)
-{
-   struct header_out_runtime *h0 = &t->headers_out[0];
-   struct header_out_runtime *h1 = &t->headers_out[1];
-   uint32_t offset = 0, i;
-
-   /* No header change or header decapsulation. */
-   if ((t->n_headers_out == 1) &&
-   (h0->ptr + h0->n_bytes == t->ptr)) {
-   TRACE("Emit handler: no header change or header decap.\n");
-
-   t->pkt.offset -= h0->n_bytes;
-   t->pkt.length += h0->n_bytes;
-
-   return;
-   }
-
-   /* Header encapsulation (optionally, with prior header decasulation). */
-   if ((t->n_headers_out == 2) &&
-   (h1->ptr + h1->n_bytes == t->ptr) &&
-   (h0->ptr == h0->ptr0)) {
-   uint32_t offset;
-
-   TRACE("Emit handler: header encapsulation.\n");
-
-   offset = h0->n_bytes + h1->n_bytes;
-   memcpy(t->ptr - offset, h0->ptr, h0->n_bytes);
-   t->pkt.offset -= offset;
-   t->pkt.length += offset;
-
-   return;
-   }
-
-   /* Header insertion. */
-   /* TBD */
-
-   /* Header extraction. */
-   /* TBD */
-
-   /* For any other case. */
-   TRACE("Emit handler: complex case.\n");
-
-   for (i = 0; i < t->n_headers_out; i++) {
-   struct header_out_runtime *h = &t->headers_out[i];
-
-   memcpy(&t->header_out_storage[offset], h->ptr, h->n_bytes);
-   offset += h->n_bytes;
-   }
-
-   if (offset) {
-   memcpy(t->ptr - offset, t->header_out_storage, offset);
-   t->pkt.offset -= offset;
-   t->pkt.length += offset;
-   }
-}
-
-static inline void
-instr_tx_exec(struct rte_swx_pipeline *p);
-
 static inline void
 instr_tx_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint64_t port_id = METADATA_READ(t, ip->io.io.offset, ip->io.io.n_bits);
-   struct port_out_runtime *port = &p->out[port_id];
-   struct rte_swx_pkt *pkt = &t->pkt;
 
-   TRACE("[Thread %2u]: tx 1 pkt to port %u\n",
- p->thread_id,
- (uint32_t)port_id);
-
-   /* Headers. */
-   emit_handler(t);
-
-   /* Packet. */
-   port->pkt_tx(port->obj, pkt);
+   __instr_tx_exec(p, t, ip);
 
/* Thread. */
thread_ip_reset(p, t);
@@ -1656,19 +1585,8 @@ instr_tx_i_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint64_t port_id = ip->io.io.val;
-   struct port_out_runtime *port = &p->out[port_id];
-   struct rte_swx_pkt *pkt = &t->pkt;
-
-   TRACE("[Thread %2u]: tx (i) 1 pkt to port %u\n",
- p->thread_id,
- (uint32_t)port_id);
-
-   /* Headers. */
-   emit_handler(t);
 
-   /* Packet. */
-   port->pkt_tx(port->obj, pkt);
+   __instr_tx_i_exec(p, t, ip);
 
/* Thread. */
thread_ip_reset(p, t);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 9814b5685a..e9fe6632b6 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -1490,4 +1490,94 @@ instr_rx_exec(struct rte_swx_pipeline *p)
thread_yield(p);
 }
 
+/*
+ * tx.
+ */
+static inline void
+emit_handler(struct thread *t)
+{
+   struct header_out_runtime *h0 = &t->headers_out[0];
+   struct header_out_runtime *h1 = &t->headers_out[1];
+   uint32_t offset = 0, i;
+
+   /* No header change or header decapsulation. */
+   if ((t->n_headers_out == 1) &&
+   (h0->ptr + h0->n_bytes == t->ptr)) {
+   TRACE("Emit handler: no header change or header decap.\n");
+
+   t->pkt.offset -= h0->n_bytes;
+   t->pkt.length += h0->n_bytes;
+
+   return;
+   }
+
+   /* Header encapsulation (optionally, with prior header decasulation). */
+   if ((t->n_headers_out == 2) &&
+   (h1->ptr + h1->n_bytes == t->ptr) &&
+   (h0->ptr == h0->ptr0)) {
+   uint32_t offset;
+
+   TRACE("Emit handler: header encapsulation.\n");
+
+   offset = h0->n_bytes + h1->n_bytes;
+   memcpy(t->ptr - offset, h0->ptr, h0->n_bytes);
+   t->

[dpdk-dev] [PATCH V3 06/24] pipeline: create inline functions for emit instruction

[Cristian Dumitrescu]

Create inline functions for the emit instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 162 -
 lib/pipeline/rte_swx_pipeline_internal.h | 170 +++
 2 files changed, 228 insertions(+), 104 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index fd7e31b709..80c5fb94bb 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1810,82 +1810,12 @@ instr_hdr_emit_translate(struct rte_swx_pipeline *p,
 }
 
 static inline void
-__instr_hdr_emit_exec(struct rte_swx_pipeline *p, uint32_t n_emit);
-
-static inline void
-__instr_hdr_emit_exec(struct rte_swx_pipeline *p, uint32_t n_emit)
+instr_hdr_emit_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint64_t valid_headers = t->valid_headers;
-   uint32_t n_headers_out = t->n_headers_out;
-   struct header_out_runtime *ho = &t->headers_out[n_headers_out - 1];
-   uint8_t *ho_ptr = NULL;
-   uint32_t ho_nbytes = 0, first = 1, i;
-
-   for (i = 0; i < n_emit; i++) {
-   uint32_t header_id = ip->io.hdr.header_id[i];
-   uint32_t struct_id = ip->io.hdr.struct_id[i];
-
-   struct header_runtime *hi = &t->headers[header_id];
-   uint8_t *hi_ptr0 = hi->ptr0;
-   uint32_t n_bytes = hi->n_bytes;
-
-   uint8_t *hi_ptr = t->structs[struct_id];
-
-   if (!MASK64_BIT_GET(valid_headers, header_id))
-   continue;
-
-   TRACE("[Thread %2u]: emit header %u\n",
- p->thread_id,
- header_id);
-
-   /* Headers. */
-   if (first) {
-   first = 0;
-
-   if (!t->n_headers_out) {
-   ho = &t->headers_out[0];
-
-   ho->ptr0 = hi_ptr0;
-   ho->ptr = hi_ptr;
-
-   ho_ptr = hi_ptr;
-   ho_nbytes = n_bytes;
-
-   n_headers_out = 1;
-
-   continue;
-   } else {
-   ho_ptr = ho->ptr;
-   ho_nbytes = ho->n_bytes;
-   }
-   }
-
-   if (ho_ptr + ho_nbytes == hi_ptr) {
-   ho_nbytes += n_bytes;
-   } else {
-   ho->n_bytes = ho_nbytes;
-
-   ho++;
-   ho->ptr0 = hi_ptr0;
-   ho->ptr = hi_ptr;
 
-   ho_ptr = hi_ptr;
-   ho_nbytes = n_bytes;
-
-   n_headers_out++;
-   }
-   }
-
-   ho->n_bytes = ho_nbytes;
-   t->n_headers_out = n_headers_out;
-}
-
-static inline void
-instr_hdr_emit_exec(struct rte_swx_pipeline *p)
-{
-   __instr_hdr_emit_exec(p, 1);
+   __instr_hdr_emit_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1894,81 +1824,105 @@ instr_hdr_emit_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_emit_tx_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 2 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
+
+   __instr_hdr_emit_tx_exec(p, t, ip);
 
-   __instr_hdr_emit_exec(p, 1);
-   instr_tx_exec(p);
+   /* Thread. */
+   thread_ip_reset(p, t);
+   instr_rx_exec(p);
 }
 
 static inline void
 instr_hdr_emit2_tx_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 3 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_emit_exec(p, 2);
-   instr_tx_exec(p);
+   __instr_hdr_emit2_tx_exec(p, t, ip);
+
+   /* Thread. */
+   thread_ip_reset(p, t);
+   instr_rx_exec(p);
 }
 
 static inline void
 instr_hdr_emit3_tx_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 4 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
+
+   __instr_hdr_emit3_tx_exec(p, t, ip);
 
-   __instr_hdr_emit_exec(p, 3);
-   instr_tx_exec(p);
+   /* Thread. */
+   thread_ip_reset(p, t);
+   instr_rx_exec(p);
 }
 
 static inline void
 instr_hdr_emit4_tx_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 5 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
+
+   __instr_hdr_emit4_tx_exec(p, t, ip);
 
-   __in

[dpdk-dev] [PATCH V3 05/24] pipeline: create inline functions for extract instruction

[Cristian Dumitrescu]

Create inline functions for the extract instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 130 -
 lib/pipeline/rte_swx_pipeline_internal.h | 178 +++
 2 files changed, 203 insertions(+), 105 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index bcf796f8c3..fd7e31b709 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1665,52 +1665,12 @@ instr_hdr_lookahead_translate(struct rte_swx_pipeline 
*p,
 }
 
 static inline void
-__instr_hdr_extract_exec(struct rte_swx_pipeline *p, uint32_t n_extract);
-
-static inline void
-__instr_hdr_extract_exec(struct rte_swx_pipeline *p, uint32_t n_extract)
+instr_hdr_extract_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint64_t valid_headers = t->valid_headers;
-   uint8_t *ptr = t->ptr;
-   uint32_t offset = t->pkt.offset;
-   uint32_t length = t->pkt.length;
-   uint32_t i;
 
-   for (i = 0; i < n_extract; i++) {
-   uint32_t header_id = ip->io.hdr.header_id[i];
-   uint32_t struct_id = ip->io.hdr.struct_id[i];
-   uint32_t n_bytes = ip->io.hdr.n_bytes[i];
-
-   TRACE("[Thread %2u]: extract header %u (%u bytes)\n",
- p->thread_id,
- header_id,
- n_bytes);
-
-   /* Headers. */
-   t->structs[struct_id] = ptr;
-   valid_headers = MASK64_BIT_SET(valid_headers, header_id);
-
-   /* Packet. */
-   offset += n_bytes;
-   length -= n_bytes;
-   ptr += n_bytes;
-   }
-
-   /* Headers. */
-   t->valid_headers = valid_headers;
-
-   /* Packet. */
-   t->pkt.offset = offset;
-   t->pkt.length = length;
-   t->ptr = ptr;
-}
-
-static inline void
-instr_hdr_extract_exec(struct rte_swx_pipeline *p)
-{
-   __instr_hdr_extract_exec(p, 1);
+   __instr_hdr_extract_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1719,10 +1679,10 @@ instr_hdr_extract_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_extract2_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 2 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_extract_exec(p, 2);
+   __instr_hdr_extract2_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1731,10 +1691,10 @@ instr_hdr_extract2_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_extract3_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 3 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_extract_exec(p, 3);
+   __instr_hdr_extract3_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1743,10 +1703,10 @@ instr_hdr_extract3_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_extract4_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 4 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_extract_exec(p, 4);
+   __instr_hdr_extract4_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1755,10 +1715,10 @@ instr_hdr_extract4_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_extract5_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 5 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_extract_exec(p, 5);
+   __instr_hdr_extract5_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1767,10 +1727,10 @@ instr_hdr_extract5_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_extract6_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 6 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_extract_exec(p, 6);
+   __instr_hdr_extract6_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1779,10 +1739,10 @@ instr_hdr_extract6_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_hdr_extract7_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 7 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_hdr_extract_exec(p, 7

[dpdk-dev] [PATCH V3 07/24] pipeline: create inline functions for validate instruction

[Cristian Dumitrescu]

Create inline functions for the validate and invalidate instructions.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 12 ++---
 lib/pipeline/rte_swx_pipeline_internal.h | 32 
 2 files changed, 34 insertions(+), 10 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 80c5fb94bb..6c6d8e52a5 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1953,12 +1953,8 @@ instr_hdr_validate_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint32_t header_id = ip->valid.header_id;
 
-   TRACE("[Thread %2u] validate header %u\n", p->thread_id, header_id);
-
-   /* Headers. */
-   t->valid_headers = MASK64_BIT_SET(t->valid_headers, header_id);
+   __instr_hdr_validate_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -1992,12 +1988,8 @@ instr_hdr_invalidate_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint32_t header_id = ip->valid.header_id;
-
-   TRACE("[Thread %2u] invalidate header %u\n", p->thread_id, header_id);
 
-   /* Headers. */
-   t->valid_headers = MASK64_BIT_CLR(t->valid_headers, header_id);
+   __instr_hdr_invalidate_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 8b37a9812e..312490f11a 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -1928,4 +1928,36 @@ __instr_hdr_emit8_tx_exec(struct rte_swx_pipeline *p,
__instr_tx_exec(p, t, ip);
 }
 
+/*
+ * validate.
+ */
+static inline void
+__instr_hdr_validate_exec(struct rte_swx_pipeline *p __rte_unused,
+ struct thread *t,
+ const struct instruction *ip)
+{
+   uint32_t header_id = ip->valid.header_id;
+
+   TRACE("[Thread %2u] validate header %u\n", p->thread_id, header_id);
+
+   /* Headers. */
+   t->valid_headers = MASK64_BIT_SET(t->valid_headers, header_id);
+}
+
+/*
+ * invalidate.
+ */
+static inline void
+__instr_hdr_invalidate_exec(struct rte_swx_pipeline *p __rte_unused,
+   struct thread *t,
+   const struct instruction *ip)
+{
+   uint32_t header_id = ip->valid.header_id;
+
+   TRACE("[Thread %2u] invalidate header %u\n", p->thread_id, header_id);
+
+   /* Headers. */
+   t->valid_headers = MASK64_BIT_CLR(t->valid_headers, header_id);
+}
+
 #endif
-- 
2.17.1

[dpdk-dev] [PATCH V3 08/24] pipeline: create inline functions for learn instruction

[Cristian Dumitrescu]

Create inline functions for the learn and forget instructions.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 36 ++--
 lib/pipeline/rte_swx_pipeline_internal.h | 55 
 2 files changed, 58 insertions(+), 33 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 6c6d8e52a5..ca12f34b01 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -2230,27 +2230,8 @@ instr_learn_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint64_t action_id = ip->learn.action_id;
-   uint32_t learner_id = t->learner_id;
-   struct rte_swx_table_state *ts = &t->table_state[p->n_tables +
-   p->n_selectors + learner_id];
-   struct learner_runtime *l = &t->learners[learner_id];
-   struct learner_statistics *stats = &p->learner_stats[learner_id];
-   uint32_t status;
-
-   /* Table. */
-   status = rte_swx_table_learner_add(ts->obj,
-  l->mailbox,
-  t->time,
-  action_id,
-  l->action_data[action_id]);
-
-   TRACE("[Thread %2u] learner %u learn %s\n",
- p->thread_id,
- learner_id,
- status ? "ok" : "error");
 
-   stats->n_pkts_learn[status] += 1;
+   __instr_learn_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2279,20 +2260,9 @@ static inline void
 instr_forget_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
-   uint32_t learner_id = t->learner_id;
-   struct rte_swx_table_state *ts = &t->table_state[p->n_tables +
-   p->n_selectors + learner_id];
-   struct learner_runtime *l = &t->learners[learner_id];
-   struct learner_statistics *stats = &p->learner_stats[learner_id];
-
-   /* Table. */
-   rte_swx_table_learner_delete(ts->obj, l->mailbox);
-
-   TRACE("[Thread %2u] learner %u forget\n",
- p->thread_id,
- learner_id);
+   struct instruction *ip = t->ip;
 
-   stats->n_pkts_forget += 1;
+   __instr_forget_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 312490f11a..24096a23b6 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -1960,4 +1960,59 @@ __instr_hdr_invalidate_exec(struct rte_swx_pipeline *p 
__rte_unused,
t->valid_headers = MASK64_BIT_CLR(t->valid_headers, header_id);
 }
 
+/*
+ * learn.
+ */
+static inline void
+__instr_learn_exec(struct rte_swx_pipeline *p,
+  struct thread *t,
+  const struct instruction *ip)
+{
+   uint64_t action_id = ip->learn.action_id;
+   uint32_t learner_id = t->learner_id;
+   struct rte_swx_table_state *ts = &t->table_state[p->n_tables +
+   p->n_selectors + learner_id];
+   struct learner_runtime *l = &t->learners[learner_id];
+   struct learner_statistics *stats = &p->learner_stats[learner_id];
+   uint32_t status;
+
+   /* Table. */
+   status = rte_swx_table_learner_add(ts->obj,
+  l->mailbox,
+  t->time,
+  action_id,
+  l->action_data[action_id]);
+
+   TRACE("[Thread %2u] learner %u learn %s\n",
+ p->thread_id,
+ learner_id,
+ status ? "ok" : "error");
+
+   stats->n_pkts_learn[status] += 1;
+}
+
+/*
+ * forget.
+ */
+static inline void
+__instr_forget_exec(struct rte_swx_pipeline *p,
+   struct thread *t,
+   const struct instruction *ip __rte_unused)
+{
+   uint32_t learner_id = t->learner_id;
+   struct rte_swx_table_state *ts = &t->table_state[p->n_tables +
+   p->n_selectors + learner_id];
+   struct learner_runtime *l = &t->learners[learner_id];
+   struct learner_statistics *stats = &p->learner_stats[learner_id];
+
+   /* Table. */
+   rte_swx_table_learner_delete(ts->obj, l->mailbox);
+
+   TRACE("[Thread %2u] learner %u forget\n",
+ p->thread_id,
+ learner_id);
+
+   stats->n_pkts_forget += 1;
+}
+
 #endif
-- 
2.17.1

[dpdk-dev] [PATCH V3 09/24] pipeline: create inline functions for extern instruction

[Cristian Dumitrescu]

Create inline functions for the extern instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 22 +++-
 lib/pipeline/rte_swx_pipeline_internal.h | 43 
 2 files changed, 47 insertions(+), 18 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index ca12f34b01..c9e29230c2 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -2317,18 +2317,10 @@ instr_extern_obj_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint32_t obj_id = ip->ext_obj.ext_obj_id;
-   uint32_t func_id = ip->ext_obj.func_id;
-   struct extern_obj_runtime *obj = &t->extern_objs[obj_id];
-   rte_swx_extern_type_member_func_t func = obj->funcs[func_id];
-
-   TRACE("[Thread %2u] extern obj %u member func %u\n",
- p->thread_id,
- obj_id,
- func_id);
+   uint32_t done;
 
/* Extern object member function execute. */
-   uint32_t done = func(obj->obj, obj->mailbox);
+   done = __instr_extern_obj_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc_cond(t, done);
@@ -2340,16 +2332,10 @@ instr_extern_func_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint32_t ext_func_id = ip->ext_func.ext_func_id;
-   struct extern_func_runtime *ext_func = &t->extern_funcs[ext_func_id];
-   rte_swx_extern_func_t func = ext_func->func;
-
-   TRACE("[Thread %2u] extern func %u\n",
- p->thread_id,
- ext_func_id);
+   uint32_t done;
 
/* Extern function execute. */
-   uint32_t done = func(ext_func->mailbox);
+   done = __instr_extern_func_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc_cond(t, done);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 24096a23b6..14d6d88344 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -2015,4 +2015,47 @@ __instr_forget_exec(struct rte_swx_pipeline *p,
stats->n_pkts_forget += 1;
 }
 
+/*
+ * extern.
+ */
+static inline uint32_t
+__instr_extern_obj_exec(struct rte_swx_pipeline *p __rte_unused,
+   struct thread *t,
+   const struct instruction *ip)
+{
+   uint32_t obj_id = ip->ext_obj.ext_obj_id;
+   uint32_t func_id = ip->ext_obj.func_id;
+   struct extern_obj_runtime *obj = &t->extern_objs[obj_id];
+   rte_swx_extern_type_member_func_t func = obj->funcs[func_id];
+   uint32_t done;
+
+   TRACE("[Thread %2u] extern obj %u member func %u\n",
+ p->thread_id,
+ obj_id,
+ func_id);
+
+   done = func(obj->obj, obj->mailbox);
+
+   return done;
+}
+
+static inline uint32_t
+__instr_extern_func_exec(struct rte_swx_pipeline *p __rte_unused,
+struct thread *t,
+const struct instruction *ip)
+{
+   uint32_t ext_func_id = ip->ext_func.ext_func_id;
+   struct extern_func_runtime *ext_func = &t->extern_funcs[ext_func_id];
+   rte_swx_extern_func_t func = ext_func->func;
+   uint32_t done;
+
+   TRACE("[Thread %2u] extern func %u\n",
+ p->thread_id,
+ ext_func_id);
+
+   done = func(ext_func->mailbox);
+
+   return done;
+}
+
 #endif
-- 
2.17.1

[dpdk-dev] [PATCH V3 10/24] pipeline: create inline functions for move instruction

[Cristian Dumitrescu]

Create inline functions for the move instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 26 +++-
 lib/pipeline/rte_swx_pipeline_internal.h | 53 
 2 files changed, 58 insertions(+), 21 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index c9e29230c2..72606f1a06 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -2407,10 +2407,7 @@ instr_mov_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] mov\n",
- p->thread_id);
-
-   MOV(t, ip);
+   __instr_mov_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2422,10 +2419,7 @@ instr_mov_mh_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] mov (mh)\n",
- p->thread_id);
-
-   MOV_MH(t, ip);
+   __instr_mov_mh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2437,10 +2431,7 @@ instr_mov_hm_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] mov (hm)\n",
- p->thread_id);
-
-   MOV_HM(t, ip);
+   __instr_mov_hm_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2452,10 +2443,7 @@ instr_mov_hh_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] mov (hh)\n",
- p->thread_id);
-
-   MOV_HH(t, ip);
+   __instr_mov_hh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2467,11 +2455,7 @@ instr_mov_i_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] mov m.f %" PRIx64 "\n",
- p->thread_id,
- ip->mov.src_val);
-
-   MOV_I(t, ip);
+   __instr_mov_i_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 14d6d88344..1bf94159a9 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -2058,4 +2058,57 @@ __instr_extern_func_exec(struct rte_swx_pipeline *p 
__rte_unused,
return done;
 }
 
+/*
+ * mov.
+ */
+static inline void
+__instr_mov_exec(struct rte_swx_pipeline *p __rte_unused,
+struct thread *t,
+const struct instruction *ip)
+{
+   TRACE("[Thread %2u] mov\n", p->thread_id);
+
+   MOV(t, ip);
+}
+
+static inline void
+__instr_mov_mh_exec(struct rte_swx_pipeline *p __rte_unused,
+   struct thread *t,
+   const struct instruction *ip)
+{
+   TRACE("[Thread %2u] mov (mh)\n", p->thread_id);
+
+   MOV_MH(t, ip);
+}
+
+static inline void
+__instr_mov_hm_exec(struct rte_swx_pipeline *p __rte_unused,
+   struct thread *t,
+   const struct instruction *ip)
+{
+   TRACE("[Thread %2u] mov (hm)\n", p->thread_id);
+
+   MOV_HM(t, ip);
+}
+
+static inline void
+__instr_mov_hh_exec(struct rte_swx_pipeline *p __rte_unused,
+   struct thread *t,
+   const struct instruction *ip)
+{
+   TRACE("[Thread %2u] mov (hh)\n", p->thread_id);
+
+   MOV_HH(t, ip);
+}
+
+static inline void
+__instr_mov_i_exec(struct rte_swx_pipeline *p __rte_unused,
+  struct thread *t,
+  const struct instruction *ip)
+{
+   TRACE("[Thread %2u] mov m.f %" PRIx64 "\n", p->thread_id, 
ip->mov.src_val);
+
+   MOV_I(t, ip);
+}
+
 #endif
-- 
2.17.1

[dpdk-dev] [PATCH V3 11/24] pipeline: create inline functions for DMA instruction

[Cristian Dumitrescu]

Create inline functions for the DMA instruction.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  |  80 ++
 lib/pipeline/rte_swx_pipeline_internal.h | 100 +++
 2 files changed, 123 insertions(+), 57 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 72606f1a06..a06dc8d348 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -2465,46 +2465,12 @@ instr_mov_i_exec(struct rte_swx_pipeline *p)
  * dma.
  */
 static inline void
-__instr_dma_ht_exec(struct rte_swx_pipeline *p, uint32_t n_dma);
-
-static inline void
-__instr_dma_ht_exec(struct rte_swx_pipeline *p, uint32_t n_dma)
+instr_dma_ht_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint8_t *action_data = t->structs[0];
-   uint64_t valid_headers = t->valid_headers;
-   uint32_t i;
-
-   for (i = 0; i < n_dma; i++) {
-   uint32_t header_id = ip->dma.dst.header_id[i];
-   uint32_t struct_id = ip->dma.dst.struct_id[i];
-   uint32_t offset = ip->dma.src.offset[i];
-   uint32_t n_bytes = ip->dma.n_bytes[i];
-
-   struct header_runtime *h = &t->headers[header_id];
-   uint8_t *h_ptr0 = h->ptr0;
-   uint8_t *h_ptr = t->structs[struct_id];
-
-   void *dst = MASK64_BIT_GET(valid_headers, header_id) ?
-   h_ptr : h_ptr0;
-   void *src = &action_data[offset];
-
-   TRACE("[Thread %2u] dma h.s t.f\n", p->thread_id);
 
-   /* Headers. */
-   memcpy(dst, src, n_bytes);
-   t->structs[struct_id] = dst;
-   valid_headers = MASK64_BIT_SET(valid_headers, header_id);
-   }
-
-   t->valid_headers = valid_headers;
-}
-
-static inline void
-instr_dma_ht_exec(struct rte_swx_pipeline *p)
-{
-   __instr_dma_ht_exec(p, 1);
+   __instr_dma_ht_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2513,10 +2479,10 @@ instr_dma_ht_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_dma_ht2_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 2 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_dma_ht_exec(p, 2);
+   __instr_dma_ht2_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2525,10 +2491,10 @@ instr_dma_ht2_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_dma_ht3_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 3 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_dma_ht_exec(p, 3);
+   __instr_dma_ht3_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2537,10 +2503,10 @@ instr_dma_ht3_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_dma_ht4_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 4 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_dma_ht_exec(p, 4);
+   __instr_dma_ht4_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2549,10 +2515,10 @@ instr_dma_ht4_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_dma_ht5_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 5 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_dma_ht_exec(p, 5);
+   __instr_dma_ht5_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2561,10 +2527,10 @@ instr_dma_ht5_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_dma_ht6_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 6 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_dma_ht_exec(p, 6);
+   __instr_dma_ht6_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2573,10 +2539,10 @@ instr_dma_ht6_exec(struct rte_swx_pipeline *p)
 static inline void
 instr_dma_ht7_exec(struct rte_swx_pipeline *p)
 {
-   TRACE("[Thread %2u] *** The next 7 instructions are fused. ***\n",
- p->thread_id);
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
 
-   __instr_dma_ht_exec(p, 7);
+   __instr_dma_ht7_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -2585,10 +2551,10 @@ instr_dma_ht7_exec(struct rte_swx_pipeline *p)
 static in

[dpdk-dev] [PATCH V3 12/24] pipeline: create inline functions for ALU instructions

[Cristian Dumitrescu]

Create inline functions for the ALU instructions.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 348 ++---
 lib/pipeline/rte_swx_pipeline_internal.h | 616 +++
 2 files changed, 660 insertions(+), 304 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index a06dc8d348..8956b6de27 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -3044,10 +3044,8 @@ instr_alu_add_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] add\n", p->thread_id);
-
-   /* Structs. */
-   ALU(t, ip, +);
+   /* Structs */
+   __instr_alu_add_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3059,10 +3057,8 @@ instr_alu_add_mh_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] add (mh)\n", p->thread_id);
-
/* Structs. */
-   ALU_MH(t, ip, +);
+   __instr_alu_add_mh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3074,10 +3070,8 @@ instr_alu_add_hm_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] add (hm)\n", p->thread_id);
-
/* Structs. */
-   ALU_HM(t, ip, +);
+   __instr_alu_add_hm_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3089,10 +3083,8 @@ instr_alu_add_hh_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] add (hh)\n", p->thread_id);
-
/* Structs. */
-   ALU_HH(t, ip, +);
+   __instr_alu_add_hh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3104,10 +3096,8 @@ instr_alu_add_mi_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] add (mi)\n", p->thread_id);
-
/* Structs. */
-   ALU_MI(t, ip, +);
+   __instr_alu_add_mi_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3119,10 +3109,8 @@ instr_alu_add_hi_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] add (hi)\n", p->thread_id);
-
/* Structs. */
-   ALU_HI(t, ip, +);
+   __instr_alu_add_hi_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3134,10 +3122,8 @@ instr_alu_sub_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] sub\n", p->thread_id);
-
/* Structs. */
-   ALU(t, ip, -);
+   __instr_alu_sub_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3149,10 +3135,8 @@ instr_alu_sub_mh_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] sub (mh)\n", p->thread_id);
-
/* Structs. */
-   ALU_MH(t, ip, -);
+   __instr_alu_sub_mh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3164,10 +3148,8 @@ instr_alu_sub_hm_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] sub (hm)\n", p->thread_id);
-
/* Structs. */
-   ALU_HM(t, ip, -);
+   __instr_alu_sub_hm_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3179,10 +3161,8 @@ instr_alu_sub_hh_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] sub (hh)\n", p->thread_id);
-
/* Structs. */
-   ALU_HH(t, ip, -);
+   __instr_alu_sub_hh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3194,10 +3174,8 @@ instr_alu_sub_mi_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] sub (mi)\n", p->thread_id);
-
/* Structs. */
-   ALU_MI(t, ip, -);
+   __instr_alu_sub_mi_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3209,10 +3187,8 @@ instr_alu_sub_hi_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
 
-   TRACE("[Thread %2u] sub (hi)\n", p->thread_id);
-
/* Structs. */
-   ALU_HI(t, ip, -);
+   __instr_alu_sub_hi_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -3224,10 +3200,8 @@ instr_alu_shl_exec(struct rte_swx_pipeline *p)
struct thread *t = &p->threa

[dpdk-dev] [PATCH V3 13/24] pipeline: create inline functions for register instructions

[Cristian Dumitrescu]

Create inline functions for the register instructions.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 320 ++-
 lib/pipeline/rte_swx_pipeline_internal.h | 475 +++
 2 files changed, 502 insertions(+), 293 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 8956b6de27..c7117bb6da 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -3901,134 +3901,14 @@ instr_regadd_translate(struct rte_swx_pipeline *p,
return 0;
 }
 
-static inline uint64_t *
-instr_regarray_regarray(struct rte_swx_pipeline *p, struct instruction *ip)
-{
-   struct regarray_runtime *r = 
&p->regarray_runtime[ip->regarray.regarray_id];
-   return r->regarray;
-}
-
-static inline uint64_t
-instr_regarray_idx_hbo(struct rte_swx_pipeline *p, struct thread *t, struct 
instruction *ip)
-{
-   struct regarray_runtime *r = 
&p->regarray_runtime[ip->regarray.regarray_id];
-
-   uint8_t *idx_struct = t->structs[ip->regarray.idx.struct_id];
-   uint64_t *idx64_ptr = (uint64_t *)&idx_struct[ip->regarray.idx.offset];
-   uint64_t idx64 = *idx64_ptr;
-   uint64_t idx64_mask = UINT64_MAX >> (64 - ip->regarray.idx.n_bits);
-   uint64_t idx = idx64 & idx64_mask & r->size_mask;
-
-   return idx;
-}
-
-#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
-
-static inline uint64_t
-instr_regarray_idx_nbo(struct rte_swx_pipeline *p, struct thread *t, struct 
instruction *ip)
-{
-   struct regarray_runtime *r = 
&p->regarray_runtime[ip->regarray.regarray_id];
-
-   uint8_t *idx_struct = t->structs[ip->regarray.idx.struct_id];
-   uint64_t *idx64_ptr = (uint64_t *)&idx_struct[ip->regarray.idx.offset];
-   uint64_t idx64 = *idx64_ptr;
-   uint64_t idx = (ntoh64(idx64) >> (64 - ip->regarray.idx.n_bits)) & 
r->size_mask;
-
-   return idx;
-}
-
-#else
-
-#define instr_regarray_idx_nbo instr_regarray_idx_hbo
-
-#endif
-
-static inline uint64_t
-instr_regarray_idx_imm(struct rte_swx_pipeline *p, struct instruction *ip)
-{
-   struct regarray_runtime *r = 
&p->regarray_runtime[ip->regarray.regarray_id];
-
-   uint64_t idx = ip->regarray.idx_val & r->size_mask;
-
-   return idx;
-}
-
-static inline uint64_t
-instr_regarray_src_hbo(struct thread *t, struct instruction *ip)
-{
-   uint8_t *src_struct = t->structs[ip->regarray.dstsrc.struct_id];
-   uint64_t *src64_ptr = (uint64_t 
*)&src_struct[ip->regarray.dstsrc.offset];
-   uint64_t src64 = *src64_ptr;
-   uint64_t src64_mask = UINT64_MAX >> (64 - ip->regarray.dstsrc.n_bits);
-   uint64_t src = src64 & src64_mask;
-
-   return src;
-}
-
-#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
-
-static inline uint64_t
-instr_regarray_src_nbo(struct thread *t, struct instruction *ip)
-{
-   uint8_t *src_struct = t->structs[ip->regarray.dstsrc.struct_id];
-   uint64_t *src64_ptr = (uint64_t 
*)&src_struct[ip->regarray.dstsrc.offset];
-   uint64_t src64 = *src64_ptr;
-   uint64_t src = ntoh64(src64) >> (64 - ip->regarray.dstsrc.n_bits);
-
-   return src;
-}
-
-#else
-
-#define instr_regarray_src_nbo instr_regarray_src_hbo
-
-#endif
-
-static inline void
-instr_regarray_dst_hbo_src_hbo_set(struct thread *t, struct instruction *ip, 
uint64_t src)
-{
-   uint8_t *dst_struct = t->structs[ip->regarray.dstsrc.struct_id];
-   uint64_t *dst64_ptr = (uint64_t 
*)&dst_struct[ip->regarray.dstsrc.offset];
-   uint64_t dst64 = *dst64_ptr;
-   uint64_t dst64_mask = UINT64_MAX >> (64 - ip->regarray.dstsrc.n_bits);
-
-   *dst64_ptr = (dst64 & ~dst64_mask) | (src & dst64_mask);
-
-}
-
-#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
-
-static inline void
-instr_regarray_dst_nbo_src_hbo_set(struct thread *t, struct instruction *ip, 
uint64_t src)
-{
-   uint8_t *dst_struct = t->structs[ip->regarray.dstsrc.struct_id];
-   uint64_t *dst64_ptr = (uint64_t 
*)&dst_struct[ip->regarray.dstsrc.offset];
-   uint64_t dst64 = *dst64_ptr;
-   uint64_t dst64_mask = UINT64_MAX >> (64 - ip->regarray.dstsrc.n_bits);
-
-   src = hton64(src) >> (64 - ip->regarray.dstsrc.n_bits);
-   *dst64_ptr = (dst64 & ~dst64_mask) | (src & dst64_mask);
-}
-
-#else
-
-#define instr_regarray_dst_nbo_src_hbo_set instr_regarray_dst_hbo_src_hbo_set
-
-#endif
-
 static inline void
 instr_regprefetch_rh_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   uint64_t *regarray, idx;
-
-   TRACE("[Thread %2u] regprefetch (r[h])\n", p->thread_id);
 
/* Structs. */
-   regarray = instr_regarray_regarray(p, ip);
-   idx = instr_regarray_idx_nbo(p, t, ip);
-   rte_prefetch0(®array[idx]);
+   __instr_regprefetch_rh_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -4039,14 +3919,9 @@ instr_regprefetch_rm_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->t

[dpdk-dev] [PATCH V3 15/24] pipeline: create inline functions for instruction operands

[Cristian Dumitrescu]

Create inline functions to get the instruction operands.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline_internal.h | 29 
 1 file changed, 29 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 791adfb471..efd136196f 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -928,6 +928,35 @@ struct thread {
 #define HEADER_VALID(thread, header_id) \
MASK64_BIT_GET((thread)->valid_headers, header_id)
 
+static inline uint64_t
+instr_operand_hbo(struct thread *t, const struct instr_operand *x)
+{
+   uint8_t *x_struct = t->structs[x->struct_id];
+   uint64_t *x64_ptr = (uint64_t *)&x_struct[x->offset];
+   uint64_t x64 = *x64_ptr;
+   uint64_t x64_mask = UINT64_MAX >> (64 - x->n_bits);
+
+   return x64 & x64_mask;
+}
+
+#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
+
+static inline uint64_t
+instr_operand_nbo(struct thread *t, const struct instr_operand *x)
+{
+   uint8_t *x_struct = t->structs[x->struct_id];
+   uint64_t *x64_ptr = (uint64_t *)&x_struct[x->offset];
+   uint64_t x64 = *x64_ptr;
+
+   return ntoh64(x64) >> (64 - x->n_bits);
+}
+
+#else
+
+#define instr_operand_nbo instr_operand_hbo
+
+#endif
+
 #define ALU(thread, ip, operator)  \
 {  
\
uint8_t *dst_struct = (thread)->structs[(ip)->alu.dst.struct_id];  \
-- 
2.17.1

[dpdk-dev] [PATCH V3 14/24] pipeline: create inline functions for meter instructions

[Cristian Dumitrescu]

Create inline functions for the meter instructions.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 457 +--
 lib/pipeline/rte_swx_pipeline_internal.h | 541 +++
 2 files changed, 558 insertions(+), 440 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index c7117bb6da..8b64c57652 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -4470,119 +4470,14 @@ instr_meter_translate(struct rte_swx_pipeline *p,
CHECK(0, EINVAL);
 }
 
-static inline struct meter *
-instr_meter_idx_hbo(struct rte_swx_pipeline *p, struct thread *t, struct 
instruction *ip)
-{
-   struct metarray_runtime *r = 
&p->metarray_runtime[ip->meter.metarray_id];
-
-   uint8_t *idx_struct = t->structs[ip->meter.idx.struct_id];
-   uint64_t *idx64_ptr = (uint64_t *)&idx_struct[ip->meter.idx.offset];
-   uint64_t idx64 = *idx64_ptr;
-   uint64_t idx64_mask = UINT64_MAX >> (64 - (ip)->meter.idx.n_bits);
-   uint64_t idx = idx64 & idx64_mask & r->size_mask;
-
-   return &r->metarray[idx];
-}
-
-#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
-
-static inline struct meter *
-instr_meter_idx_nbo(struct rte_swx_pipeline *p, struct thread *t, struct 
instruction *ip)
-{
-   struct metarray_runtime *r = 
&p->metarray_runtime[ip->meter.metarray_id];
-
-   uint8_t *idx_struct = t->structs[ip->meter.idx.struct_id];
-   uint64_t *idx64_ptr = (uint64_t *)&idx_struct[ip->meter.idx.offset];
-   uint64_t idx64 = *idx64_ptr;
-   uint64_t idx = (ntoh64(idx64) >> (64 - ip->meter.idx.n_bits)) & 
r->size_mask;
-
-   return &r->metarray[idx];
-}
-
-#else
-
-#define instr_meter_idx_nbo instr_meter_idx_hbo
-
-#endif
-
-static inline struct meter *
-instr_meter_idx_imm(struct rte_swx_pipeline *p, struct instruction *ip)
-{
-   struct metarray_runtime *r = 
&p->metarray_runtime[ip->meter.metarray_id];
-
-   uint64_t idx =  ip->meter.idx_val & r->size_mask;
-
-   return &r->metarray[idx];
-}
-
-static inline uint32_t
-instr_meter_length_hbo(struct thread *t, struct instruction *ip)
-{
-   uint8_t *src_struct = t->structs[ip->meter.length.struct_id];
-   uint64_t *src64_ptr = (uint64_t *)&src_struct[ip->meter.length.offset];
-   uint64_t src64 = *src64_ptr;
-   uint64_t src64_mask = UINT64_MAX >> (64 - (ip)->meter.length.n_bits);
-   uint64_t src = src64 & src64_mask;
-
-   return (uint32_t)src;
-}
-
-#if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
-
-static inline uint32_t
-instr_meter_length_nbo(struct thread *t, struct instruction *ip)
-{
-   uint8_t *src_struct = t->structs[ip->meter.length.struct_id];
-   uint64_t *src64_ptr = (uint64_t *)&src_struct[ip->meter.length.offset];
-   uint64_t src64 = *src64_ptr;
-   uint64_t src = ntoh64(src64) >> (64 - ip->meter.length.n_bits);
-
-   return (uint32_t)src;
-}
-
-#else
-
-#define instr_meter_length_nbo instr_meter_length_hbo
-
-#endif
-
-static inline enum rte_color
-instr_meter_color_in_hbo(struct thread *t, struct instruction *ip)
-{
-   uint8_t *src_struct = t->structs[ip->meter.color_in.struct_id];
-   uint64_t *src64_ptr = (uint64_t 
*)&src_struct[ip->meter.color_in.offset];
-   uint64_t src64 = *src64_ptr;
-   uint64_t src64_mask = UINT64_MAX >> (64 - ip->meter.color_in.n_bits);
-   uint64_t src = src64 & src64_mask;
-
-   return (enum rte_color)src;
-}
-
-static inline void
-instr_meter_color_out_hbo_set(struct thread *t, struct instruction *ip, enum 
rte_color color_out)
-{
-   uint8_t *dst_struct = t->structs[ip->meter.color_out.struct_id];
-   uint64_t *dst64_ptr = (uint64_t 
*)&dst_struct[ip->meter.color_out.offset];
-   uint64_t dst64 = *dst64_ptr;
-   uint64_t dst64_mask = UINT64_MAX >> (64 - ip->meter.color_out.n_bits);
-
-   uint64_t src = (uint64_t)color_out;
-
-   *dst64_ptr = (dst64 & ~dst64_mask) | (src & dst64_mask);
-}
-
 static inline void
 instr_metprefetch_h_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   struct meter *m;
-
-   TRACE("[Thread %2u] metprefetch (h)\n", p->thread_id);
 
/* Structs. */
-   m = instr_meter_idx_nbo(p, t, ip);
-   rte_prefetch0(m);
+   __instr_metprefetch_h_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -4593,13 +4488,9 @@ instr_metprefetch_m_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   struct meter *m;
-
-   TRACE("[Thread %2u] metprefetch (m)\n", p->thread_id);
 
/* Structs. */
-   m = instr_meter_idx_hbo(p, t, ip);
-   rte_prefetch0(m);
+   __instr_metprefetch_m_exec(p, t, ip);
 
/* Thread. */
thread_ip_inc(p);
@@ -4610,13 +4501,9 @@ instr_metprefetch_i_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = 

[dpdk-dev] [PATCH V3 16/24] pipeline: enable persistent instruction meta-data

[Cristian Dumitrescu]

Save the instruction meta-data for later use instead of freeing it up
once the instruction translation is completed.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 9 ++---
 lib/pipeline/rte_swx_pipeline_internal.h | 2 ++
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 8b64c57652..4099e364f5 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -6456,13 +6456,14 @@ instruction_config(struct rte_swx_pipeline *p,
 
if (a) {
a->instructions = instr;
+   a->instruction_data = data;
a->n_instructions = n_instructions;
} else {
p->instructions = instr;
+   p->instruction_data = data;
p->n_instructions = n_instructions;
}
 
-   free(data);
return 0;
 
 error:
@@ -6811,8 +6812,8 @@ action_build(struct rte_swx_pipeline *p)
 {
struct action *action;
 
-   p->action_instructions = calloc(p->n_actions,
-   sizeof(struct instruction *));
+   /* p->action_instructions. */
+   p->action_instructions = calloc(p->n_actions, sizeof(struct instruction 
*));
CHECK(p->action_instructions, ENOMEM);
 
TAILQ_FOREACH(action, &p->actions, node)
@@ -6841,6 +6842,7 @@ action_free(struct rte_swx_pipeline *p)
break;
 
TAILQ_REMOVE(&p->actions, action, node);
+   free(action->instruction_data);
free(action->instructions);
free(action);
}
@@ -8777,6 +8779,7 @@ rte_swx_pipeline_free(struct rte_swx_pipeline *p)
if (!p)
return;
 
+   free(p->instruction_data);
free(p->instructions);
 
metarray_free(p);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index efd136196f..7a02d6cb5f 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -693,6 +693,7 @@ struct action {
struct struct_type *st;
int *args_endianness; /* 0 = Host Byte Order (HBO); 1 = Network Byte 
Order (NBO). */
struct instruction *instructions;
+   struct instruction_data *instruction_data;
uint32_t n_instructions;
uint32_t id;
 };
@@ -1388,6 +1389,7 @@ struct rte_swx_pipeline {
struct regarray_runtime *regarray_runtime;
struct metarray_runtime *metarray_runtime;
struct instruction *instructions;
+   struct instruction_data *instruction_data;
struct thread threads[RTE_SWX_PIPELINE_THREADS_MAX];
 
uint32_t n_structs;
-- 
2.17.1

[dpdk-dev] [PATCH V3 17/24] pipeline: introduce action functions

[Cristian Dumitrescu]

For better performance, the option to run a single function per action
is now provided, which requires a single function call per action that
can be better optimized by the C compiler, as opposed to one function
call per instruction. Special table lookup instructions are added to
to support this feature.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c  | 127 +++
 lib/pipeline/rte_swx_pipeline_internal.h |   6 ++
 2 files changed, 133 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 4099e364f5..0d02548137 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -2097,6 +2097,62 @@ instr_table_exec(struct rte_swx_pipeline *p)
thread_ip_action_call(p, t, action_id);
 }
 
+static inline void
+instr_table_af_exec(struct rte_swx_pipeline *p)
+{
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
+   uint32_t table_id = ip->table.table_id;
+   struct rte_swx_table_state *ts = &t->table_state[table_id];
+   struct table_runtime *table = &t->tables[table_id];
+   struct table_statistics *stats = &p->table_stats[table_id];
+   uint64_t action_id, n_pkts_hit, n_pkts_action;
+   uint8_t *action_data;
+   action_func_t action_func;
+   int done, hit;
+
+   /* Table. */
+   done = table->func(ts->obj,
+  table->mailbox,
+  table->key,
+  &action_id,
+  &action_data,
+  &hit);
+   if (!done) {
+   /* Thread. */
+   TRACE("[Thread %2u] table %u (not finalized)\n",
+ p->thread_id,
+ table_id);
+
+   thread_yield(p);
+   return;
+   }
+
+   action_id = hit ? action_id : ts->default_action_id;
+   action_data = hit ? action_data : ts->default_action_data;
+   action_func = p->action_funcs[action_id];
+   n_pkts_hit = stats->n_pkts_hit[hit];
+   n_pkts_action = stats->n_pkts_action[action_id];
+
+   TRACE("[Thread %2u] table %u (%s, action %u)\n",
+ p->thread_id,
+ table_id,
+ hit ? "hit" : "miss",
+ (uint32_t)action_id);
+
+   t->action_id = action_id;
+   t->structs[0] = action_data;
+   t->hit = hit;
+   stats->n_pkts_hit[hit] = n_pkts_hit + 1;
+   stats->n_pkts_action[action_id] = n_pkts_action + 1;
+
+   /* Thread. */
+   thread_ip_inc(p);
+
+   /* Action. */
+   action_func(p);
+}
+
 static inline void
 instr_selector_exec(struct rte_swx_pipeline *p)
 {
@@ -2193,6 +2249,68 @@ instr_learner_exec(struct rte_swx_pipeline *p)
thread_ip_action_call(p, t, action_id);
 }
 
+static inline void
+instr_learner_af_exec(struct rte_swx_pipeline *p)
+{
+   struct thread *t = &p->threads[p->thread_id];
+   struct instruction *ip = t->ip;
+   uint32_t learner_id = ip->table.table_id;
+   struct rte_swx_table_state *ts = &t->table_state[p->n_tables +
+   p->n_selectors + learner_id];
+   struct learner_runtime *l = &t->learners[learner_id];
+   struct learner_statistics *stats = &p->learner_stats[learner_id];
+   uint64_t action_id, n_pkts_hit, n_pkts_action, time;
+   uint8_t *action_data;
+   action_func_t action_func;
+   int done, hit;
+
+   /* Table. */
+   time = rte_get_tsc_cycles();
+
+   done = rte_swx_table_learner_lookup(ts->obj,
+   l->mailbox,
+   time,
+   l->key,
+   &action_id,
+   &action_data,
+   &hit);
+   if (!done) {
+   /* Thread. */
+   TRACE("[Thread %2u] learner %u (not finalized)\n",
+ p->thread_id,
+ learner_id);
+
+   thread_yield(p);
+   return;
+   }
+
+   action_id = hit ? action_id : ts->default_action_id;
+   action_data = hit ? action_data : ts->default_action_data;
+   action_func = p->action_funcs[action_id];
+   n_pkts_hit = stats->n_pkts_hit[hit];
+   n_pkts_action = stats->n_pkts_action[action_id];
+
+   TRACE("[Thread %2u] learner %u (%s, action %u)\n",
+ p->thread_id,
+ learner_id,
+ hit ? "hit" : "miss",
+ (uint32_t)action_id);
+
+   t->action_id = action_id;
+   t->structs[0] = action_data;
+   t->hit = hit;
+   t->learner_id = learner_id;
+   t->time = time;
+   stats->n_pkts_hit[hit] = n_pkts_hit + 1;
+   stats->n_pkts_action[action_id] = n_pkts_action + 1;
+
+   /* Thread. */
+   thread_ip_action_call(p, t, action_id);
+
+   /* Action */
+

[dpdk-dev] [PATCH V3 18/24] pipeline: introduce custom instructions

[Cristian Dumitrescu]

For better performance, the option to create custom instructions when
the program is translated and add them on-the-fly to the pipeline is
now provided. Multiple regular instructions can now be consolidated
into a single C function optimized by the C compiler directly.

Signed-off-by: Cristian Dumitrescu 
---
V3:
-created per pipeline instance instruction table

 lib/pipeline/rte_swx_pipeline.c  | 39 ++--
 lib/pipeline/rte_swx_pipeline_internal.h | 10 ++
 2 files changed, 46 insertions(+), 3 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 0d02548137..9afe42ed25 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -6590,8 +6590,6 @@ instruction_config(struct rte_swx_pipeline *p,
return err;
 }
 
-typedef void (*instr_exec_t)(struct rte_swx_pipeline *);
-
 static instr_exec_t instruction_table[] = {
[INSTR_RX] = instr_rx_exec,
[INSTR_TX] = instr_tx_exec,
@@ -6782,12 +6780,41 @@ static instr_exec_t instruction_table[] = {
[INSTR_RETURN] = instr_return_exec,
 };
 
+static int
+instruction_table_build(struct rte_swx_pipeline *p)
+{
+   p->instruction_table = 
calloc(RTE_SWX_PIPELINE_INSTRUCTION_TABLE_SIZE_MAX,
+ sizeof(struct instr_exec_t *));
+   if (!p->instruction_table)
+   return -EINVAL;
+
+   memcpy(p->instruction_table, instruction_table, 
sizeof(instruction_table));
+
+   return 0;
+}
+
+static void
+instruction_table_build_free(struct rte_swx_pipeline *p)
+{
+   if (!p->instruction_table)
+   return;
+
+   free(p->instruction_table);
+   p->instruction_table = NULL;
+}
+
+static void
+instruction_table_free(struct rte_swx_pipeline *p)
+{
+   instruction_table_build_free(p);
+}
+
 static inline void
 instr_exec(struct rte_swx_pipeline *p)
 {
struct thread *t = &p->threads[p->thread_id];
struct instruction *ip = t->ip;
-   instr_exec_t instr = instruction_table[ip->type];
+   instr_exec_t instr = p->instruction_table[ip->type];
 
instr(p);
 }
@@ -8916,6 +8943,7 @@ rte_swx_pipeline_free(struct rte_swx_pipeline *p)
selector_free(p);
table_free(p);
action_free(p);
+   instruction_table_free(p);
metadata_free(p);
header_free(p);
extern_func_free(p);
@@ -8985,6 +9013,10 @@ rte_swx_pipeline_build(struct rte_swx_pipeline *p)
if (status)
goto error;
 
+   status = instruction_table_build(p);
+   if (status)
+   goto error;
+
status = action_build(p);
if (status)
goto error;
@@ -9024,6 +9056,7 @@ rte_swx_pipeline_build(struct rte_swx_pipeline *p)
selector_build_free(p);
table_build_free(p);
action_build_free(p);
+   instruction_table_build_free(p);
metadata_build_free(p);
header_build_free(p);
extern_func_build_free(p);
diff --git a/lib/pipeline/rte_swx_pipeline_internal.h 
b/lib/pipeline/rte_swx_pipeline_internal.h
index 3578a10501..64625b40c6 100644
--- a/lib/pipeline/rte_swx_pipeline_internal.h
+++ b/lib/pipeline/rte_swx_pipeline_internal.h
@@ -541,6 +541,9 @@ enum instruction_type {
 * Return from action
 */
INSTR_RETURN,
+
+   /* Start of custom instructions. */
+   INSTR_CUSTOM_0,
 };
 
 struct instr_operand {
@@ -686,6 +689,8 @@ struct instruction_data {
int invalid;
 };
 
+typedef void (*instr_exec_t)(struct rte_swx_pipeline *);
+
 /*
  * Action.
  */
@@ -1363,6 +1368,10 @@ instr_operand_nbo(struct thread *t, const struct 
instr_operand *x)
 #define RTE_SWX_PIPELINE_THREADS_MAX 16
 #endif
 
+#ifndef RTE_SWX_PIPELINE_INSTRUCTION_TABLE_SIZE_MAX
+#define RTE_SWX_PIPELINE_INSTRUCTION_TABLE_SIZE_MAX 256
+#endif
+
 struct rte_swx_pipeline {
struct struct_type_tailq struct_types;
struct port_in_type_tailq port_in_types;
@@ -1396,6 +1405,7 @@ struct rte_swx_pipeline {
struct metarray_runtime *metarray_runtime;
struct instruction *instructions;
struct instruction_data *instruction_data;
+   instr_exec_t *instruction_table;
struct thread threads[RTE_SWX_PIPELINE_THREADS_MAX];
 
uint32_t n_structs;
-- 
2.17.1

[dpdk-dev] [PATCH V3 19/24] pipeline: introduce pipeline compilation

[Cristian Dumitrescu]

Lay the foundation to generate C code for the pipeline: C functions
for actions and custom instructions are generated, built as shared
object library and loaded into the pipeline.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c | 44 +
 1 file changed, 44 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 9afe42ed25..392e2cf1bc 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -8977,6 +8977,9 @@ rte_swx_pipeline_instructions_config(struct 
rte_swx_pipeline *p,
return 0;
 }
 
+static int
+pipeline_compile(struct rte_swx_pipeline *p);
+
 int
 rte_swx_pipeline_build(struct rte_swx_pipeline *p)
 {
@@ -9046,6 +9049,9 @@ rte_swx_pipeline_build(struct rte_swx_pipeline *p)
goto error;
 
p->build_done = 1;
+
+   pipeline_compile(p);
+
return 0;
 
 error:
@@ -9789,3 +9795,41 @@ rte_swx_ctl_meter_stats_read(struct rte_swx_pipeline *p,
 
return 0;
 }
+
+/*
+ * Pipeline compilation.
+ */
+static int
+pipeline_codegen(struct rte_swx_pipeline *p)
+{
+   FILE *f = NULL;
+
+   if (!p)
+   return -EINVAL;
+
+   /* Create the .c file. */
+   f = fopen("/tmp/pipeline.c", "w");
+   if (!f)
+   return -EIO;
+
+   /* Include the .h file. */
+   fprintf(f, "#include \"rte_swx_pipeline_internal.h\"\n");
+
+   /* Close the .c file. */
+   fclose(f);
+
+   return 0;
+}
+
+static int
+pipeline_compile(struct rte_swx_pipeline *p)
+{
+   int status = 0;
+
+   /* Code generation. */
+   status = pipeline_codegen(p);
+   if (status)
+   return status;
+
+   return status;
+}
-- 
2.17.1

[dpdk-dev] [PATCH V3 20/24] pipeline: export pipeline instructions to file

[Cristian Dumitrescu]

Export the array of translated instructions to a C file. There is one
such array per action and one for the pipeline.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c | 1093 +++
 1 file changed, 1093 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 392e2cf1bc..c962283fed 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -9799,9 +9799,1093 @@ rte_swx_ctl_meter_stats_read(struct rte_swx_pipeline 
*p,
 /*
  * Pipeline compilation.
  */
+static const char *
+instr_type_to_name(struct instruction *instr)
+{
+   switch (instr->type) {
+   case INSTR_RX: return "INSTR_RX";
+
+   case INSTR_TX: return "INSTR_TX";
+   case INSTR_TX_I: return "INSTR_TX_I";
+
+   case INSTR_HDR_EXTRACT: return "INSTR_HDR_EXTRACT";
+   case INSTR_HDR_EXTRACT2: return "INSTR_HDR_EXTRACT2";
+   case INSTR_HDR_EXTRACT3: return "INSTR_HDR_EXTRACT3";
+   case INSTR_HDR_EXTRACT4: return "INSTR_HDR_EXTRACT4";
+   case INSTR_HDR_EXTRACT5: return "INSTR_HDR_EXTRACT5";
+   case INSTR_HDR_EXTRACT6: return "INSTR_HDR_EXTRACT6";
+   case INSTR_HDR_EXTRACT7: return "INSTR_HDR_EXTRACT7";
+   case INSTR_HDR_EXTRACT8: return "INSTR_HDR_EXTRACT8";
+
+   case INSTR_HDR_EXTRACT_M: return "INSTR_HDR_EXTRACT_M";
+
+   case INSTR_HDR_LOOKAHEAD: return "INSTR_HDR_LOOKAHEAD";
+
+   case INSTR_HDR_EMIT: return "INSTR_HDR_EMIT";
+   case INSTR_HDR_EMIT_TX: return "INSTR_HDR_EMIT_TX";
+   case INSTR_HDR_EMIT2_TX: return "INSTR_HDR_EMIT2_TX";
+   case INSTR_HDR_EMIT3_TX: return "INSTR_HDR_EMIT3_TX";
+   case INSTR_HDR_EMIT4_TX: return "INSTR_HDR_EMIT4_TX";
+   case INSTR_HDR_EMIT5_TX: return "INSTR_HDR_EMIT5_TX";
+   case INSTR_HDR_EMIT6_TX: return "INSTR_HDR_EMIT6_TX";
+   case INSTR_HDR_EMIT7_TX: return "INSTR_HDR_EMIT7_TX";
+   case INSTR_HDR_EMIT8_TX: return "INSTR_HDR_EMIT8_TX";
+
+   case INSTR_HDR_VALIDATE: return "INSTR_HDR_VALIDATE";
+   case INSTR_HDR_INVALIDATE: return "INSTR_HDR_INVALIDATE";
+
+   case INSTR_MOV: return "INSTR_MOV";
+   case INSTR_MOV_MH: return "INSTR_MOV_MH";
+   case INSTR_MOV_HM: return "INSTR_MOV_HM";
+   case INSTR_MOV_HH: return "INSTR_MOV_HH";
+   case INSTR_MOV_I: return "INSTR_MOV_I";
+
+   case INSTR_DMA_HT: return "INSTR_DMA_HT";
+   case INSTR_DMA_HT2: return "INSTR_DMA_HT2";
+   case INSTR_DMA_HT3: return "INSTR_DMA_HT3";
+   case INSTR_DMA_HT4: return "INSTR_DMA_HT4";
+   case INSTR_DMA_HT5: return "INSTR_DMA_HT5";
+   case INSTR_DMA_HT6: return "INSTR_DMA_HT6";
+   case INSTR_DMA_HT7: return "INSTR_DMA_HT7";
+   case INSTR_DMA_HT8: return "INSTR_DMA_HT8";
+
+   case INSTR_ALU_ADD: return "INSTR_ALU_ADD";
+   case INSTR_ALU_ADD_MH: return "INSTR_ALU_ADD_MH";
+   case INSTR_ALU_ADD_HM: return "INSTR_ALU_ADD_HM";
+   case INSTR_ALU_ADD_HH: return "INSTR_ALU_ADD_HH";
+   case INSTR_ALU_ADD_MI: return "INSTR_ALU_ADD_MI";
+   case INSTR_ALU_ADD_HI: return "INSTR_ALU_ADD_HI";
+
+   case INSTR_ALU_SUB: return "INSTR_ALU_SUB";
+   case INSTR_ALU_SUB_MH: return "INSTR_ALU_SUB_MH";
+   case INSTR_ALU_SUB_HM: return "INSTR_ALU_SUB_HM";
+   case INSTR_ALU_SUB_HH: return "INSTR_ALU_SUB_HH";
+   case INSTR_ALU_SUB_MI: return "INSTR_ALU_SUB_MI";
+   case INSTR_ALU_SUB_HI: return "INSTR_ALU_SUB_HI";
+
+   case INSTR_ALU_CKADD_FIELD: return "INSTR_ALU_CKADD_FIELD";
+   case INSTR_ALU_CKADD_STRUCT20: return "INSTR_ALU_CKADD_STRUCT20";
+   case INSTR_ALU_CKADD_STRUCT: return "INSTR_ALU_CKADD_STRUCT";
+   case INSTR_ALU_CKSUB_FIELD: return "INSTR_ALU_CKSUB_FIELD";
+
+   case INSTR_ALU_AND: return "INSTR_ALU_AND";
+   case INSTR_ALU_AND_MH: return "INSTR_ALU_AND_MH";
+   case INSTR_ALU_AND_HM: return "INSTR_ALU_AND_HM";
+   case INSTR_ALU_AND_HH: return "INSTR_ALU_AND_HH";
+   case INSTR_ALU_AND_I: return "INSTR_ALU_AND_I";
+
+   case INSTR_ALU_OR: return "INSTR_ALU_OR";
+   case INSTR_ALU_OR_MH: return "INSTR_ALU_OR_MH";
+   case INSTR_ALU_OR_HM: return "INSTR_ALU_OR_HM";
+   case INSTR_ALU_OR_HH: return "INSTR_ALU_OR_HH";
+   case INSTR_ALU_OR_I: return "INSTR_ALU_OR_I";
+
+   case INSTR_ALU_XOR: return "INSTR_ALU_XOR";
+   case INSTR_ALU_XOR_MH: return "INSTR_ALU_XOR_MH";
+   case INSTR_ALU_XOR_HM: return "INSTR_ALU_XOR_HM";
+   case INSTR_ALU_XOR_HH: return "INSTR_ALU_XOR_HH";
+   case INSTR_ALU_XOR_I: return "INSTR_ALU_XOR_I";
+
+   case INSTR_ALU_SHL: return "INSTR_ALU_SHL";
+   case INSTR_ALU_SHL_MH: return "INSTR_ALU_SHL_MH";
+   case INSTR_ALU_SHL_HM: return "INSTR_ALU_SHL_HM";
+   case INSTR_ALU_SHL_HH: return "INSTR_ALU_SHL_HH";
+   case INSTR_ALU_SHL_MI: return "INSTR_ALU_SHL_MI";
+   case INSTR_ALU_SHL_HI: return "INSTR_ALU_SHL_HI";
+
+   case INSTR_ALU_SHR: return "INSTR_ALU_SHR";
+   ca

[dpdk-dev] [PATCH V3 21/24] pipeline: generate action functions

[Cristian Dumitrescu]

Generate a C function for each action. For most instructions, the
associated inline function is called directly. Special care is taken
for TX, jump and return instructions.

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c | 662 
 1 file changed, 662 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index c962283fed..ccd26d0f3a 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1376,6 +1376,26 @@ instruction_is_tx(enum instruction_type type)
}
 }
 
+static int
+instruction_does_tx(struct instruction *instr)
+{
+   switch (instr->type) {
+   case INSTR_TX:
+   case INSTR_TX_I:
+   case INSTR_HDR_EMIT_TX:
+   case INSTR_HDR_EMIT2_TX:
+   case INSTR_HDR_EMIT3_TX:
+   case INSTR_HDR_EMIT4_TX:
+   case INSTR_HDR_EMIT5_TX:
+   case INSTR_HDR_EMIT6_TX:
+   case INSTR_HDR_EMIT7_TX:
+   case INSTR_HDR_EMIT8_TX:
+   return 1;
+   default:
+   return 0;
+   }
+}
+
 static int
 instruction_is_jmp(struct instruction *instr)
 {
@@ -10882,6 +10902,644 @@ action_data_codegen(struct action *a, FILE *f)
fprintf(f, "};\n");
 }
 
+static const char *
+instr_type_to_func(struct instruction *instr)
+{
+   switch (instr->type) {
+   case INSTR_RX: return NULL;
+
+   case INSTR_TX: return "__instr_tx_exec";
+   case INSTR_TX_I: return "__instr_tx_i_exec";
+
+   case INSTR_HDR_EXTRACT: return "__instr_hdr_extract_exec";
+   case INSTR_HDR_EXTRACT2: return "__instr_hdr_extract2_exec";
+   case INSTR_HDR_EXTRACT3: return "__instr_hdr_extract3_exec";
+   case INSTR_HDR_EXTRACT4: return "__instr_hdr_extract4_exec";
+   case INSTR_HDR_EXTRACT5: return "__instr_hdr_extract5_exec";
+   case INSTR_HDR_EXTRACT6: return "__instr_hdr_extract6_exec";
+   case INSTR_HDR_EXTRACT7: return "__instr_hdr_extract7_exec";
+   case INSTR_HDR_EXTRACT8: return "__instr_hdr_extract8_exec";
+
+   case INSTR_HDR_EXTRACT_M: return "__instr_hdr_extract_m_exec";
+
+   case INSTR_HDR_LOOKAHEAD: return "__instr_hdr_lookahead_exec";
+
+   case INSTR_HDR_EMIT: return "__instr_hdr_emit_exec";
+   case INSTR_HDR_EMIT_TX: return "__instr_hdr_emit_tx_exec";
+   case INSTR_HDR_EMIT2_TX: return "__instr_hdr_emit2_tx_exec";
+   case INSTR_HDR_EMIT3_TX: return "__instr_hdr_emit3_tx_exec";
+   case INSTR_HDR_EMIT4_TX: return "__instr_hdr_emit4_tx_exec";
+   case INSTR_HDR_EMIT5_TX: return "__instr_hdr_emit5_tx_exec";
+   case INSTR_HDR_EMIT6_TX: return "__instr_hdr_emit6_tx_exec";
+   case INSTR_HDR_EMIT7_TX: return "__instr_hdr_emit7_tx_exec";
+   case INSTR_HDR_EMIT8_TX: return "__instr_hdr_emit8_tx_exec";
+
+   case INSTR_HDR_VALIDATE: return "__instr_hdr_validate_exec";
+   case INSTR_HDR_INVALIDATE: return "__instr_hdr_invalidate_exec";
+
+   case INSTR_MOV: return "__instr_mov_exec";
+   case INSTR_MOV_MH: return "__instr_mov_mh_exec";
+   case INSTR_MOV_HM: return "__instr_mov_hm_exec";
+   case INSTR_MOV_HH: return "__instr_mov_hh_exec";
+   case INSTR_MOV_I: return "__instr_mov_i_exec";
+
+   case INSTR_DMA_HT: return "__instr_dma_ht_exec";
+   case INSTR_DMA_HT2: return "__instr_dma_ht2_exec";
+   case INSTR_DMA_HT3: return "__instr_dma_ht3_exec";
+   case INSTR_DMA_HT4: return "__instr_dma_ht4_exec";
+   case INSTR_DMA_HT5: return "__instr_dma_ht5_exec";
+   case INSTR_DMA_HT6: return "__instr_dma_ht6_exec";
+   case INSTR_DMA_HT7: return "__instr_dma_ht7_exec";
+   case INSTR_DMA_HT8: return "__instr_dma_ht8_exec";
+
+   case INSTR_ALU_ADD: return "__instr_alu_add_exec";
+   case INSTR_ALU_ADD_MH: return "__instr_alu_add_mh_exec";
+   case INSTR_ALU_ADD_HM: return "__instr_alu_add_hm_exec";
+   case INSTR_ALU_ADD_HH: return "__instr_alu_add_hh_exec";
+   case INSTR_ALU_ADD_MI: return "__instr_alu_add_mi_exec";
+   case INSTR_ALU_ADD_HI: return "__instr_alu_add_hi_exec";
+
+   case INSTR_ALU_SUB: return "__instr_alu_sub_exec";
+   case INSTR_ALU_SUB_MH: return "__instr_alu_sub_mh_exec";
+   case INSTR_ALU_SUB_HM: return "__instr_alu_sub_hm_exec";
+   case INSTR_ALU_SUB_HH: return "__instr_alu_sub_hh_exec";
+   case INSTR_ALU_SUB_MI: return "__instr_alu_sub_mi_exec";
+   case INSTR_ALU_SUB_HI: return "__instr_alu_sub_hi_exec";
+
+   case INSTR_ALU_CKADD_FIELD: return "__instr_alu_ckadd_field_exec";
+   case INSTR_ALU_CKADD_STRUCT20: return "__instr_alu_ckadd_struct20_exec";
+   case INSTR_ALU_CKADD_STRUCT: return "__instr_alu_ckadd_struct_exec";
+   case INSTR_ALU_CKSUB_FIELD: return "__instr_alu_cksub_field_exec";
+
+   case INSTR_ALU_AND: return "__instr_alu_and_exec";
+   case INSTR_ALU_AND_MH: return "__instr_alu_and_mh_exec";
+   case INSTR_ALU_AND_HM: return "__instr_alu_and_hm_exec";
+   case INSTR_ALU_A

[dpdk-dev] [PATCH V3 22/24] pipeline: generate custom instruction functions

[Cristian Dumitrescu]

Generate a C function for each custom instruction, which essentially
consolidate multiple regular instructions into a single function call.
The pipeline program is split into groups of instructions, and a
custom instruction is generated for each group that has more than one
instruction. Special care is taken the instructions that can do thread
yield (RX, extern) and for those that can change the instruction
pointer (TX, near/far jump).

Signed-off-by: Cristian Dumitrescu 
---
 lib/pipeline/rte_swx_pipeline.c | 651 +++-
 1 file changed, 645 insertions(+), 6 deletions(-)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index ccd26d0f3a..e669dd09d2 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -1436,6 +1436,24 @@ instruction_is_jmp(struct instruction *instr)
}
 }
 
+static int
+instruction_does_thread_yield(struct instruction *instr)
+{
+   switch (instr->type) {
+   case INSTR_RX:
+   case INSTR_TABLE:
+   case INSTR_TABLE_AF:
+   case INSTR_SELECTOR:
+   case INSTR_LEARNER:
+   case INSTR_LEARNER_AF:
+   case INSTR_EXTERN_OBJ:
+   case INSTR_EXTERN_FUNC:
+   return 1;
+   default:
+   return 0;
+   }
+}
+
 static struct field *
 action_field_parse(struct action *action, const char *name);
 
@@ -11540,15 +11558,623 @@ action_instr_codegen(struct action *a, FILE *f)
fprintf(f, "}\n\n");
 }
 
+struct instruction_group {
+   TAILQ_ENTRY(instruction_group) node;
+
+   uint32_t group_id;
+
+   uint32_t first_instr_id;
+
+   uint32_t last_instr_id;
+
+   instr_exec_t func;
+};
+
+TAILQ_HEAD(instruction_group_list, instruction_group);
+
+static struct instruction_group *
+instruction_group_list_group_find(struct instruction_group_list *igl, uint32_t 
instruction_id)
+{
+   struct instruction_group *g;
+
+   TAILQ_FOREACH(g, igl, node)
+   if ((g->first_instr_id <= instruction_id) && (instruction_id <= 
g->last_instr_id))
+   return g;
+
+   return NULL;
+}
+
+static void
+instruction_group_list_free(struct instruction_group_list *igl)
+{
+   if (!igl)
+   return;
+
+   for ( ; ; ) {
+   struct instruction_group *g;
+
+   g = TAILQ_FIRST(igl);
+   if (!g)
+   break;
+
+   TAILQ_REMOVE(igl, g, node);
+   free(g);
+   }
+
+   free(igl);
+}
+
+static struct instruction_group_list *
+instruction_group_list_create(struct rte_swx_pipeline *p)
+{
+   struct instruction_group_list *igl = NULL;
+   struct instruction_group *g = NULL;
+   uint32_t n_groups = 0, i;
+
+   if (!p || !p->instructions || !p->instruction_data || 
!p->n_instructions)
+   goto error;
+
+   /* List init. */
+   igl = calloc(1, sizeof(struct instruction_group_list));
+   if (!igl)
+   goto error;
+
+   TAILQ_INIT(igl);
+
+   /* Allocate the first group. */
+   g = calloc(1, sizeof(struct instruction_group));
+   if (!g)
+   goto error;
+
+   /* Iteration 1: Separate the instructions into groups based on the 
thread yield
+* instructions. Do not worry about the jump instructions at this point.
+*/
+   for (i = 0; i < p->n_instructions; i++) {
+   struct instruction *instr = &p->instructions[i];
+
+   /* Check for thread yield instructions. */
+   if (!instruction_does_thread_yield(instr))
+   continue;
+
+   /* If the current group contains at least one instruction, then 
finalize it (with
+* the previous instruction), add it to the list and allocate a 
new group (that
+* starts with the current instruction).
+*/
+   if (i - g->first_instr_id) {
+   /* Finalize the group. */
+   g->last_instr_id = i - 1;
+
+   /* Add the group to the list. Advance the number of 
groups. */
+   TAILQ_INSERT_TAIL(igl, g, node);
+   n_groups++;
+
+   /* Allocate a new group. */
+   g = calloc(1, sizeof(struct instruction_group));
+   if (!g)
+   goto error;
+
+   /* Initialize the new group. */
+   g->group_id = n_groups;
+   g->first_instr_id = i;
+   }
+
+   /* Finalize the current group (with the current instruction, 
therefore this group
+* contains just the current thread yield instruction), add it 
to the list and
+* allocate a new group (that starts with the next instruction).
+*/
+
+   /* Finalize the group. */
+   g->last_instr_id = i;
+
+

[dpdk-dev] [PATCH V3 23/24] pipeline: build shared object for pipeline

[Cristian Dumitrescu]

Build the generated C file into a shared object library.

Signed-off-by: Cristian Dumitrescu 
Signed-off-by: Cunming Liang 
---
 lib/pipeline/rte_swx_pipeline.c  | 131 +++
 lib/pipeline/rte_swx_pipeline_internal.h |   1 +
 2 files changed, 132 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index e669dd09d2..02519a05a9 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -5,6 +5,7 @@
 #include 
 #include 
 #include 
+#include 
 
 #include "rte_swx_pipeline_internal.h"
 
@@ -8968,9 +8969,13 @@ rte_swx_pipeline_config(struct rte_swx_pipeline **p, int 
numa_node)
 void
 rte_swx_pipeline_free(struct rte_swx_pipeline *p)
 {
+   void *lib;
+
if (!p)
return;
 
+   lib = p->lib;
+
free(p->instruction_data);
free(p->instructions);
 
@@ -8991,6 +8996,9 @@ rte_swx_pipeline_free(struct rte_swx_pipeline *p)
struct_free(p);
 
free(p);
+
+   if (lib)
+   dlclose(lib);
 }
 
 int
@@ -12205,6 +12213,124 @@ pipeline_codegen(struct rte_swx_pipeline *p, struct 
instruction_group_list *igl)
return 0;
 }
 
+#ifndef RTE_SWX_PIPELINE_CMD_MAX_SIZE
+#define RTE_SWX_PIPELINE_CMD_MAX_SIZE 4096
+#endif
+
+static int
+pipeline_libload(struct rte_swx_pipeline *p, struct instruction_group_list 
*igl)
+{
+   struct action *a;
+   struct instruction_group *g;
+   char *dir_in, *buffer = NULL;
+   const char *dir_out;
+   int status = 0;
+
+   /* Get the environment variables. */
+   dir_in = getenv("RTE_INSTALL_DIR");
+   if (!dir_in) {
+   status = -EINVAL;
+   goto free;
+   }
+
+   dir_out = "/tmp";
+
+   /* Memory allocation for the command buffer. */
+   buffer = malloc(RTE_SWX_PIPELINE_CMD_MAX_SIZE);
+   if (!buffer) {
+   status = -ENOMEM;
+   goto free;
+   }
+
+   snprintf(buffer,
+RTE_SWX_PIPELINE_CMD_MAX_SIZE,
+"gcc -c -O3 -fpic -Wno-deprecated-declarations -o 
%s/pipeline.o %s/pipeline.c "
+"-I %s/lib/pipeline "
+"-I %s/lib/eal/include "
+"-I %s/lib/eal/x86/include "
+"-I %s/lib/eal/include/generic "
+"-I %s/lib/meter "
+"-I %s/lib/port "
+"-I %s/lib/table "
+"-I %s/lib/pipeline "
+"-I %s/config "
+"-I %s/build "
+"-I %s/lib/eal/linux/include "
+">%s/pipeline.log 2>&1 "
+"&& "
+"gcc -shared %s/pipeline.o -o %s/libpipeline.so "
+">>%s/pipeline.log 2>&1",
+dir_out,
+dir_out,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_in,
+dir_out,
+dir_out,
+dir_out,
+dir_out);
+
+   /* Build the shared object library. */
+   status = system(buffer);
+   if (status)
+   goto free;
+
+   /* Open library. */
+   snprintf(buffer,
+RTE_SWX_PIPELINE_CMD_MAX_SIZE,
+"%s/libpipeline.so",
+dir_out);
+
+   p->lib = dlopen(buffer, RTLD_LAZY);
+   if (!p->lib) {
+   status = -EIO;
+   goto free;
+   }
+
+   /* Get the action function symbols. */
+   TAILQ_FOREACH(a, &p->actions, node) {
+   snprintf(buffer, RTE_SWX_PIPELINE_CMD_MAX_SIZE, 
"action_%s_run", a->name);
+
+   p->action_funcs[a->id] = dlsym(p->lib, buffer);
+   if (!p->action_funcs[a->id]) {
+   status = -EINVAL;
+   goto free;
+   }
+   }
+
+   /* Get the pipeline function symbols. */
+   TAILQ_FOREACH(g, igl, node) {
+   if (g->first_instr_id == g->last_instr_id)
+   continue;
+
+   snprintf(buffer, RTE_SWX_PIPELINE_CMD_MAX_SIZE, 
"pipeline_func_%u", g->group_id);
+
+   g->func = dlsym(p->lib, buffer);
+   if (!g->func) {
+   status = -EINVAL;
+   goto free;
+   }
+   }
+
+free:
+   if (status && p->lib) {
+   dlclose(p->lib);
+   p->lib = NULL;
+   }
+
+   free(buffer);
+
+   return status;
+}
+
 static int
 pipeline_compile(struct rte_swx_pipeline *p)
 {
@@ -1,6 +12348,11 @@ pipeline_compile(struct rte_swx_pipeline *p)
if (status)
goto free;
 
+   /* Build and load the shared object library. */
+   status = pipeline_libload(p, igl);
+   if (status)
+   goto free;
+
 free:
i

[dpdk-dev] [PATCH V3 24/24] pipeline: enable pipeline compilation

[Cristian Dumitrescu]

Commit the pipeline changes when the compilation process is
successful: change the table lookup instructions to execute the action
function for each action, replace the regular pipeline instructions
with the custom instructions.

Signed-off-by: Cristian Dumitrescu 
---
V3:
-added more checks

 lib/pipeline/rte_swx_pipeline.c | 94 +
 1 file changed, 94 insertions(+)

diff --git a/lib/pipeline/rte_swx_pipeline.c b/lib/pipeline/rte_swx_pipeline.c
index 02519a05a9..31f0029404 100644
--- a/lib/pipeline/rte_swx_pipeline.c
+++ b/lib/pipeline/rte_swx_pipeline.c
@@ -12177,6 +12177,26 @@ instruction_group_list_codegen(struct 
instruction_group_list *igl,
}
 }
 
+static uint32_t
+instruction_group_list_custom_instructions_count(struct instruction_group_list 
*igl)
+{
+   struct instruction_group *g;
+   uint32_t n_custom_instr = 0;
+
+   /* Groups with a single instruction: no function is generated for this 
group, the group
+* keeps its current instruction. Groups with more than two 
instructions: one function and
+* the associated custom instruction get generated for each such group.
+*/
+   TAILQ_FOREACH(g, igl, node) {
+   if (g->first_instr_id == g->last_instr_id)
+   continue;
+
+   n_custom_instr++;
+   }
+
+   return n_custom_instr;
+}
+
 static int
 pipeline_codegen(struct rte_swx_pipeline *p, struct instruction_group_list 
*igl)
 {
@@ -12331,6 +12351,73 @@ pipeline_libload(struct rte_swx_pipeline *p, struct 
instruction_group_list *igl)
return status;
 }
 
+static int
+pipeline_adjust_check(struct rte_swx_pipeline *p __rte_unused,
+ struct instruction_group_list *igl)
+{
+   uint32_t n_custom_instr = 
instruction_group_list_custom_instructions_count(igl);
+
+   /* Check that enough space is available within the pipeline instruction 
table to store all
+* the custom instructions.
+*/
+   if (INSTR_CUSTOM_0 + n_custom_instr > 
RTE_SWX_PIPELINE_INSTRUCTION_TABLE_SIZE_MAX)
+   return -ENOSPC;
+
+   return 0;
+}
+
+static void
+pipeline_adjust(struct rte_swx_pipeline *p, struct instruction_group_list *igl)
+{
+   struct instruction_group *g;
+   uint32_t i;
+
+   /* Pipeline table instructions. */
+   for (i = 0; i < p->n_instructions; i++) {
+   struct instruction *instr = &p->instructions[i];
+
+   if (instr->type == INSTR_TABLE)
+   instr->type = INSTR_TABLE_AF;
+
+   if (instr->type == INSTR_LEARNER)
+   instr->type = INSTR_LEARNER_AF;
+   }
+
+   /* Pipeline custom instructions. */
+   i = 0;
+   TAILQ_FOREACH(g, igl, node) {
+   struct instruction *instr = &p->instructions[g->first_instr_id];
+   uint32_t j;
+
+   if (g->first_instr_id == g->last_instr_id)
+   continue;
+
+   /* Install a new custom instruction. */
+   p->instruction_table[INSTR_CUSTOM_0 + i] = g->func;
+
+   /* First instruction of the group: change its type to the new 
custom instruction. */
+   instr->type = INSTR_CUSTOM_0 + i;
+
+   /* All the subsequent instructions of the group: invalidate. */
+   for (j = g->first_instr_id + 1; j <= g->last_instr_id; j++) {
+   struct instruction_data *data = &p->instruction_data[j];
+
+   data->invalid = 1;
+   }
+
+   i++;
+   }
+
+   /* Remove the invalidated instructions. */
+   p->n_instructions = instr_compact(p->instructions, p->instruction_data, 
p->n_instructions);
+
+   /* Resolve the jump destination for any "standalone" jump instructions 
(i.e. those jump
+* instructions that are the only instruction within their group, so 
they were left
+* unmodified).
+*/
+   instr_jmp_resolve(p->instructions, p->instruction_data, 
p->n_instructions);
+}
+
 static int
 pipeline_compile(struct rte_swx_pipeline *p)
 {
@@ -12353,6 +12440,13 @@ pipeline_compile(struct rte_swx_pipeline *p)
if (status)
goto free;
 
+   /* Adjust instructions. */
+   status = pipeline_adjust_check(p, igl);
+   if (status)
+   goto free;
+
+   pipeline_adjust(p, igl);
+
 free:
instruction_group_list_free(igl);
 
-- 
2.17.1

Re: [dpdk-dev] [PATCH V3 01/24] pipeline: move data structures to internal header file

[Stephen Hemminger]

On Mon, 13 Sep 2021 17:44:20 +0100
Cristian Dumitrescu  wrote:

> Start to consolidate the data structures and inline functions required
> by the pipeline instructions into an internal header file.
> 
> Signed-off-by: Cristian Dumitrescu 
> ---
> Depends-on: series-18297 ("[V4,1/4] table: add support learner tables")

Won't this change will make future changes to API/ABI harder because more
of the pipeline internals are exposed to application?

Re: [dpdk-dev] [PATCH v3 6/6] bbdev: reduce warning level for one scenario

[Chautru, Nicolas]

> -Original Message-
> From: Tom Rix 
> Sent: Sunday, September 12, 2021 5:55 AM
> To: Chautru, Nicolas ; dev@dpdk.org;
> gak...@marvell.com
> Cc: tho...@monjalon.net; hemant.agra...@nxp.com; Zhang, Mingshan
> ; Joshi, Arun 
> Subject: Re: [PATCH v3 6/6] bbdev: reduce warning level for one scenario
> 
> 
> On 9/7/21 6:15 PM, Nicolas Chautru wrote:
> > Queue setup may genuinely fail when adding incremental queues for a
> > given priority level. In that case application would attempt to
> > configure a queue at a different priority level.
> > Not an actual error.
> >
> > Signed-off-by: Nicolas Chautru 
> > ---
> >   lib/bbdev/rte_bbdev.c | 7 ---
> >   1 file changed, 4 insertions(+), 3 deletions(-)
> >
> > diff --git a/lib/bbdev/rte_bbdev.c b/lib/bbdev/rte_bbdev.c index
> > fc37236..defddcf 100644
> > --- a/lib/bbdev/rte_bbdev.c
> > +++ b/lib/bbdev/rte_bbdev.c
> > @@ -528,9 +528,10 @@ struct rte_bbdev *
> > ret = dev->dev_ops->queue_setup(dev, queue_id, (conf != NULL) ?
> > conf : &dev_info.default_queue_conf);
> > if (ret < 0) {
> > -   rte_bbdev_log(ERR,
> > -   "Device %u queue %u setup failed", dev_id,
> > -   queue_id);
> > +   /* This may happen when trying different priority levels */
> > +   rte_bbdev_log(INFO,
> > +   "Device %u queue %u setup failed",
> > +   dev_id, queue_id);
> 
> This change is just changing the log level, which is fine.
> 
> I am looking at how the error handling is done for the function.
> 
> It seems like the bailing is done in the middle of change the queue state.
> 
> ex/ the block above this one
> 
> /* Release existing queue ... */
> 
> Does this leave the queue in a bad state ?

Hi Tom, 
That would not be related to that change indeed. 

The queue would end up in a not configured when rte_bbdev_queue_configure() 
fails but then can still  be configured again without limitation (worst thing 
than can happen is that queue_release is called, hence leaves the queue in a 
deterministic state, unconfigured but ready to be configured).
Note that queue_release() just removes the configuration of the queue, but the 
queue is still there and can be configured again (actual total number of queues 
unchanged, based on number previously set with rte_bbdev_setup_queues()).

Thanks
Nic

> 
> Tom
> 
> > return ret;
> > }
> >

Re: [dpdk-dev] [PATCH V2 01/24] pipeline: move data structures to internal header file

[Dumitrescu, Cristian]

> -Original Message-
> From: Richardson, Bruce 
> Sent: Friday, September 10, 2021 3:09 PM
> To: Dumitrescu, Cristian 
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH V2 01/24] pipeline: move data structures to
> internal header file
> 
> On Fri, Sep 10, 2021 at 02:36:50PM +0100, Cristian Dumitrescu wrote:
> > Start to consolidate the data structures and inline functions required
> > by the pipeline instructions into an internal header file.
> >
> > Signed-off-by: Cristian Dumitrescu 
> > ---
> > Depends-on: series-18297 ("[V4,1/4] table: add support learner tables")
> >
> >  lib/pipeline/meson.build |4 +
> >  lib/pipeline/rte_swx_pipeline.c  | 1373 +
> >  lib/pipeline/rte_swx_pipeline_internal.h | 1383
> ++
> >  3 files changed, 1388 insertions(+), 1372 deletions(-)
> >  create mode 100644 lib/pipeline/rte_swx_pipeline_internal.h
> >
> > diff --git a/lib/pipeline/meson.build b/lib/pipeline/meson.build
> > index 9132bb517a..ec009631bf 100644
> > --- a/lib/pipeline/meson.build
> > +++ b/lib/pipeline/meson.build
> > @@ -18,3 +18,7 @@ headers = files(
> >  'rte_swx_ctl.h',
> >  )
> >  deps += ['port', 'table', 'meter', 'sched', 'cryptodev']
> > +
> > +indirect_headers += files(
> > +'rte_swx_pipeline_internal.h',
> > +)
> 
> internal headers should not appear in meson.build at all, as they are not
> for installation. "indirect_headers" is for non-internal, public headers
> which are not directly included by apps, but included via another public
> header. For example, rte_ethdev_core.h should not be included directly, but
> by including rte_ethdev.h which will then pull it in. [The reason these are
> separated out in meson.build is for header sanity checking, to check that
> normal, non-indirect public headers include all their dependent headers]

Yes, my bad, fixed in V3 just sent. Thanks, Bruce.

[dpdk-dev] [PATCH v2] net/af_packet: remove timestamp from packet status

[Tudor Cornea]

We should eliminate the timestamp status from the packet
status. This should only matter if timestamping is enabled
on the socket, but we might hit a kernel bug, which is fixed
in newer releases.

For interfaces of type 'veth', the sent skb is forwarded
to the peer and back into the network stack which timestamps
it on the RX path if timestamping is enabled globally
(which happens if any socket enables timestamping).

When the skb is destructed, tpacket_destruct_skb() is called
and it calls __packet_set_timestamp() which doesn't check
the flags on the socket and returns the timestamp if it is
set in the skb (and for veth it is, as mentioned above).

See the following kernel commit for reference [1]:

net: packetmmap: fix only tx timestamp on request

The packetmmap tx ring should only return timestamps if requested
via setsockopt PACKET_TIMESTAMP, as documented. This allows
compatibility with non-timestamp aware user-space code which checks
tp_status == TP_STATUS_AVAILABLE; not expecting additional timestamp
flags to be set in tp_status.

[1] https://www.spinics.net/lists/kernel/msg3959391.html

Signed-off-by: Mihai Pogonaru 
Signed-off-by: Tudor Cornea 

---
v2:
* Remove compile-time check for kernel version
---
 drivers/net/af_packet/rte_eth_af_packet.c | 20 ++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/drivers/net/af_packet/rte_eth_af_packet.c 
b/drivers/net/af_packet/rte_eth_af_packet.c
index b73b211..7ecea4e 100644
--- a/drivers/net/af_packet/rte_eth_af_packet.c
+++ b/drivers/net/af_packet/rte_eth_af_packet.c
@@ -167,6 +167,22 @@ eth_af_packet_rx(void *queue, struct rte_mbuf **bufs, 
uint16_t nb_pkts)
return num_rx;
 }
 
+static inline bool tx_ring_status_unavailable(uint32_t tp_status)
+{
+   /*
+* We eliminate the timestamp status from the packet status.
+* This should only matter if timestamping is enabled on the socket,
+* but there is a bug in the kernel which is fixed in newer releases.
+*
+* See the following kernel commit for reference:
+* commit 171c3b151118a2fe0fc1e2a9d1b5a1570cfe82d2
+* net: packetmmap: fix only tx timestamp on request
+*/
+   tp_status &= ~(TP_STATUS_TS_SOFTWARE | TP_STATUS_TS_RAW_HARDWARE);
+
+   return tp_status != TP_STATUS_AVAILABLE;
+}
+
 /*
  * Callback to handle sending packets through a real NIC.
  */
@@ -212,8 +228,8 @@ eth_af_packet_tx(void *queue, struct rte_mbuf **bufs, 
uint16_t nb_pkts)
}
 
/* point at the next incoming frame */
-   if ((ppd->tp_status != TP_STATUS_AVAILABLE) &&
-   (poll(&pfd, 1, -1) < 0))
+   if (tx_ring_status_unavailable(ppd->tp_status) &&
+   poll(&pfd, 1, -1) < 0)
break;
 
/* copy the tx frame data */
-- 
2.7.4

Re: [dpdk-dev] [PATCH] net/af_packet: remove timestamp from packet status

[Tudor Cornea]

Thanks for the observation.

I have removed the compile-time kernel version check in v2 of the patch


On Mon, 13 Sept 2021 at 18:09, Stephen Hemminger 
wrote:

> On Mon, 13 Sep 2021 17:09:11 +0300
> Tudor Cornea  wrote:
>
> > +static inline bool tx_ring_status_unavailable(uint32_t tp_status)
> > +{
> > +#if KERNEL_VERSION(5, 10, 0) > LINUX_VERSION_CODE
>
> No, having kernel dependent userspace in DPDK is not good practice.
>
> Distribution vendors don't number their kernels the same as upstream.
> RHEL for example, keeps same version over life or release but backports
> many fixes.
>
> Also, the system DPDK runs on is often not the system DPDK is built
> on.
>

[dpdk-dev] [PATCH] crypto/mlx5: fix crypto QP indexing

[Tal Shnaiderman]

The crypto QP consumer (ci) and producer (pi) indexes are increased
with each successful enqueue/dequeue operations.

However the QP pi index is calculated with a wraparound the number
of elements while the QP ci does not.

This is causing incorrect engine calculation for encqueued WQ values
(wq->pi - wq->ci) and eventually the device stops accepting new enqueue
operations.

Fixed by removing the wraparound on QP pi and using a temp calculation
where wraparound values are needed.

Fixes: 8e196c08ab53 ("crypto/mlx5: support enqueue/dequeue operations")
Cc: sta...@dpdk.org

Signed-off-by: Tal Shnaiderman 
Acked-by: Matan Azrad 
---
 drivers/crypto/mlx5/mlx5_crypto.c | 8 +---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/crypto/mlx5/mlx5_crypto.c 
b/drivers/crypto/mlx5/mlx5_crypto.c
index b3d5200ca3..eeec554e6e 100644
--- a/drivers/crypto/mlx5/mlx5_crypto.c
+++ b/drivers/crypto/mlx5/mlx5_crypto.c
@@ -494,6 +494,7 @@ mlx5_crypto_enqueue_burst(void *queue_pair, struct 
rte_crypto_op **ops,
struct rte_crypto_op *op;
uint16_t mask = qp->entries_n - 1;
uint16_t remain = qp->entries_n - (qp->pi - qp->ci);
+   uint32_t idx;
 
if (remain < nb_ops)
nb_ops = remain;
@@ -502,8 +503,9 @@ mlx5_crypto_enqueue_burst(void *queue_pair, struct 
rte_crypto_op **ops,
if (unlikely(remain == 0))
return 0;
do {
+   idx = qp->pi & mask;
op = *ops++;
-   umr = RTE_PTR_ADD(qp->umem_buf, priv->wqe_set_size * qp->pi);
+   umr = RTE_PTR_ADD(qp->umem_buf, priv->wqe_set_size * idx);
if (unlikely(mlx5_crypto_wqe_set(priv, qp, op, umr) == 0)) {
qp->stats.enqueue_err_count++;
if (remain != nb_ops) {
@@ -512,8 +514,8 @@ mlx5_crypto_enqueue_burst(void *queue_pair, struct 
rte_crypto_op **ops,
}
return 0;
}
-   qp->ops[qp->pi] = op;
-   qp->pi = (qp->pi + 1) & mask;
+   qp->ops[idx] = op;
+   qp->pi++;
} while (--remain);
qp->stats.enqueued_count += nb_ops;
rte_io_wmb();
-- 
2.16.1.windows.4