Re: [dpdk-dev] [PATCH v3 1/2] net/i40e: extended list of operations for ddp processing

[Chilikin, Andrey]

Hi Beilei,

> -Original Message-
> From: Xing, Beilei
> Sent: Wednesday, June 28, 2017 7:58 AM
> To: Chilikin, Andrey ; dev@dpdk.org
> Cc: Wu, Jingjing 
> Subject: RE: [PATCH v3 1/2] net/i40e: extended list of operations for ddp
> processing
> 
> Hi Andrey,
> 

> > +   }
> > +   else {
> 
> Should be  } else {  here according to the coding style.

Will fix in v4

> > --- a/drivers/net/i40e/rte_pmd_i40e.h
> > +++ b/drivers/net/i40e/rte_pmd_i40e.h
> > @@ -71,6 +71,8 @@ struct rte_pmd_i40e_mb_event_param {  enum
> > rte_pmd_i40e_package_op {
> > RTE_PMD_I40E_PKG_OP_UNDEFINED = 0,
> > RTE_PMD_I40E_PKG_OP_WR_ADD,   /**< load package and add to
> > info list */
> > +   RTE_PMD_I40E_PKG_OP_WR_DEL, /**< load package and delete
> > from info list */
> > +   RTE_PMD_I40E_PKG_OP_WR_ONLY, /**< load package without
> > modifying info
> > +list */
> 
> The only difference between RTE_PMD_I40E_PKG_OP_WR_ADD and
> RTE_PMD_I40E_PKG_OP_WR_ONLY is if need to change info list, right? Is there
> any difference between the profiles added? Do we need another CLI in testpmd
> to indicate how/when we need to use RTE_PMD_I40E_PKG_OP_WR_ONLY?
> 
I believe that existing testpmd CLI is enough for i40e as this 'if' statement
if (track_id && (op != RTE_PMD_I40E_PKG_OP_WR_ONLY)) {
will not register profiles with track_id == 0

> Beilei

Regards,
Andrey

Re: [dpdk-dev] [PATCH v5 12/19] devargs: generic device representation

[Thomas Monjalon]

21/06/2017 01:35, Gaetan Rivet:
> Remove the dependency of this subsystem upon bus specific device
> representation.
> 
> Devargs only validates that a device declaration is correct and handled
> by a bus. The device interpretation is done afterward within the bus.
> 
> Signed-off-by: Gaetan Rivet 
> ---
>  app/test-pmd/testpmd.c  |  9 ++
>  lib/librte_eal/common/eal_common_devargs.c  | 48 
> +
>  lib/librte_eal/common/eal_common_pci.c  |  9 --
>  lib/librte_eal/common/eal_common_vdev.c | 17 +-
>  lib/librte_eal/common/include/rte_devargs.h | 16 ++
>  lib/librte_eal/common/include/rte_vdev.h|  2 +-
>  test/test/test_devargs.c|  9 ++
>  7 files changed, 30 insertions(+), 80 deletions(-)

Is it correct to add this change when rebasing?

--- a/lib/librte_eal/common/eal_common_pci.c
+++ b/lib/librte_eal/common/eal_common_pci.c
@@ -533,7 +533,7 @@ pci_plug(struct rte_devargs *da)
struct rte_pci_device *pdev;
struct rte_pci_addr *addr;
 
-   addr = &da->pci.addr;
+   da->bus->parse(da->name, &addr);
/*
 * Update eventual pci device in global list.
 * Insert it if none was found.

Re: [dpdk-dev] [PATCH v3 1/2] net/i40e: extended list of operations for ddp processing

[Xing, Beilei]

Hi Andrey,

> -Original Message-
> From: Chilikin, Andrey
> Sent: Wednesday, June 28, 2017 3:40 PM
> To: Xing, Beilei ; dev@dpdk.org
> Cc: Wu, Jingjing 
> Subject: RE: [PATCH v3 1/2] net/i40e: extended list of operations for ddp
> processing
> 
> Hi Beilei,
> 
> > -Original Message-
> > From: Xing, Beilei
> > Sent: Wednesday, June 28, 2017 7:58 AM
> > To: Chilikin, Andrey ; dev@dpdk.org
> > Cc: Wu, Jingjing 
> > Subject: RE: [PATCH v3 1/2] net/i40e: extended list of operations for
> > ddp processing
> >
> > Hi Andrey,
> >
> 
> > > + }
> > > + else {
> >
> > Should be  } else {  here according to the coding style.
> 
> Will fix in v4
> 
> > > --- a/drivers/net/i40e/rte_pmd_i40e.h
> > > +++ b/drivers/net/i40e/rte_pmd_i40e.h
> > > @@ -71,6 +71,8 @@ struct rte_pmd_i40e_mb_event_param {  enum
> > > rte_pmd_i40e_package_op {
> > >   RTE_PMD_I40E_PKG_OP_UNDEFINED = 0,
> > >   RTE_PMD_I40E_PKG_OP_WR_ADD,   /**< load package and add to
> > > info list */
> > > + RTE_PMD_I40E_PKG_OP_WR_DEL, /**< load package and delete
> > > from info list */
> > > + RTE_PMD_I40E_PKG_OP_WR_ONLY, /**< load package without
> > > modifying info
> > > +list */
> >
> > The only difference between RTE_PMD_I40E_PKG_OP_WR_ADD and
> > RTE_PMD_I40E_PKG_OP_WR_ONLY is if need to change info list, right? Is
> > there any difference between the profiles added? Do we need another
> > CLI in testpmd to indicate how/when we need to use
> RTE_PMD_I40E_PKG_OP_WR_ONLY?
> >
> I believe that existing testpmd CLI is enough for i40e as this 'if' statement 
> if
> (track_id && (op != RTE_PMD_I40E_PKG_OP_WR_ONLY)) { will not register
> profiles with track_id == 0
> 

OK, so track_id=0 is for RTE_PMD_I40E_PKG_OP_WR_ONLY. From the patch, it seems 
we needn't new operation RTE_PMD_I40E_PKG_OP_WR_ONLY exposed to application as 
PMD can decide if register WR_ONLY profiles by track_id. 

Re: [dpdk-dev] [PATCH v5 12/19] devargs: generic device representation

[Gaëtan Rivet]

On Wed, Jun 28, 2017 at 09:44:18AM +0200, Thomas Monjalon wrote:
> 21/06/2017 01:35, Gaetan Rivet:
> > Remove the dependency of this subsystem upon bus specific device
> > representation.
> > 
> > Devargs only validates that a device declaration is correct and handled
> > by a bus. The device interpretation is done afterward within the bus.
> > 
> > Signed-off-by: Gaetan Rivet 
> > ---
> >  app/test-pmd/testpmd.c  |  9 ++
> >  lib/librte_eal/common/eal_common_devargs.c  | 48 
> > +
> >  lib/librte_eal/common/eal_common_pci.c  |  9 --
> >  lib/librte_eal/common/eal_common_vdev.c | 17 +-
> >  lib/librte_eal/common/include/rte_devargs.h | 16 ++
> >  lib/librte_eal/common/include/rte_vdev.h|  2 +-
> >  test/test/test_devargs.c|  9 ++
> >  7 files changed, 30 insertions(+), 80 deletions(-)
> 
> Is it correct to add this change when rebasing?
> 
> --- a/lib/librte_eal/common/eal_common_pci.c
> +++ b/lib/librte_eal/common/eal_common_pci.c
> @@ -533,7 +533,7 @@ pci_plug(struct rte_devargs *da)
> struct rte_pci_device *pdev;
> struct rte_pci_addr *addr;
>  
> -   addr = &da->pci.addr;
> +   da->bus->parse(da->name, &addr);
> /*
>  * Update eventual pci device in global list.
>  * Insert it if none was found.
> 

This is the spirit of this API at least :).
bus->parse() will verify that the name is correct, as well as interpret
it to obtain the bus-specific representation of the device.

> > Devargs only validates that a device declaration is correct and handled
> > by a bus. The device interpretation is done afterward within the bus.

So the PCI address can be removed from rte_devargs. As such, it cannot
be intepreted during parsing, and it must be done afterward from within
the bus itself.

I was wondering whether I should update all subsystems relying on this
previous rte_devargs logic at once, EAL + vdev + PCI + tests etc...
But this is the only way to keep compilation atomicity between patches.

-- 
Gaëtan Rivet
6WIND

[dpdk-dev] [PATCH v4 2/2] app/testpmd: enable ddp remove profile feature

[Andrey Chilikin]

New command 'ddp del (port) (profile_path)' removes previously
loaded profile and deletes it from the list of the loaded profiles.

Signed-off-by: Andrey Chilikin 
Acked-by: Beilei Xing 
---
 app/test-pmd/cmdline.c  | 74 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  7 +++
 2 files changed, 81 insertions(+)

diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 81d1f84fe..0fc40a6fd 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -606,6 +606,9 @@ static void cmd_help_long_parsed(void *parsed_result,
"ddp add (port_id) (profile_path[,output_path])\n"
"Load a profile package on a port\n\n"
 
+   "ddp del (port_id) (profile_path)\n"
+   "Delete a profile package from a port\n\n"
+
"ptype mapping get (port_id) (valid_only)\n"
"Get ptype mapping on a port\n\n"
 
@@ -13014,6 +13017,76 @@ cmdline_parse_inst_t cmd_ddp_add = {
},
 };
 
+/* Delete dynamic device personalization*/
+struct cmd_ddp_del_result {
+   cmdline_fixed_string_t ddp;
+   cmdline_fixed_string_t del;
+   uint8_t port_id;
+   char filepath[];
+};
+
+cmdline_parse_token_string_t cmd_ddp_del_ddp =
+   TOKEN_STRING_INITIALIZER(struct cmd_ddp_del_result, ddp, "ddp");
+cmdline_parse_token_string_t cmd_ddp_del_del =
+   TOKEN_STRING_INITIALIZER(struct cmd_ddp_del_result, del, "del");
+cmdline_parse_token_num_t cmd_ddp_del_port_id =
+   TOKEN_NUM_INITIALIZER(struct cmd_ddp_del_result, port_id, UINT8);
+cmdline_parse_token_string_t cmd_ddp_del_filepath =
+   TOKEN_STRING_INITIALIZER(struct cmd_ddp_del_result, filepath, NULL);
+
+static void
+cmd_ddp_del_parsed(
+   void *parsed_result,
+   __attribute__((unused)) struct cmdline *cl,
+   __attribute__((unused)) void *data)
+{
+   struct cmd_ddp_del_result *res = parsed_result;
+   uint8_t *buff;
+   uint32_t size;
+   int ret = -ENOTSUP;
+
+   if (res->port_id > nb_ports) {
+   printf("Invalid port, range is [0, %d]\n", nb_ports - 1);
+   return;
+   }
+
+   if (!all_ports_stopped()) {
+   printf("Please stop all ports first\n");
+   return;
+   }
+
+   buff = open_ddp_package_file(res->filepath, &size);
+   if (!buff)
+   return;
+
+#ifdef RTE_LIBRTE_I40E_PMD
+   if (ret == -ENOTSUP)
+   ret = rte_pmd_i40e_process_ddp_package(res->port_id,
+  buff, size,
+  RTE_PMD_I40E_PKG_OP_WR_DEL);
+#endif
+
+   if (ret == -EACCES)
+   printf("Profile does not exist.\n");
+   else if (ret < 0)
+   printf("Failed to delete profile.\n");
+
+   close_ddp_package_file(buff);
+}
+
+cmdline_parse_inst_t cmd_ddp_del = {
+   .f = cmd_ddp_del_parsed,
+   .data = NULL,
+   .help_str = "ddp del  ",
+   .tokens = {
+   (void *)&cmd_ddp_del_ddp,
+   (void *)&cmd_ddp_del_del,
+   (void *)&cmd_ddp_del_port_id,
+   (void *)&cmd_ddp_del_filepath,
+   NULL,
+   },
+};
+
 /* Get dynamic device personalization profile info */
 struct cmd_ddp_info_result {
cmdline_fixed_string_t ddp;
@@ -13993,6 +14066,7 @@ cmdline_parse_ctx_t main_ctx[] = {
(cmdline_parse_inst_t *)&cmd_strict_link_prio,
(cmdline_parse_inst_t *)&cmd_tc_min_bw,
(cmdline_parse_inst_t *)&cmd_ddp_add,
+   (cmdline_parse_inst_t *)&cmd_ddp_del,
(cmdline_parse_inst_t *)&cmd_ddp_get_list,
(cmdline_parse_inst_t *)&cmd_ddp_get_info,
(cmdline_parse_inst_t *)&cmd_show_vf_stats,
diff --git a/doc/guides/testpmd_app_ug/testpmd_funcs.rst 
b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
index e6b0b514a..b8f47fde8 100644
--- a/doc/guides/testpmd_app_ug/testpmd_funcs.rst
+++ b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
@@ -1232,6 +1232,13 @@ Load a dynamic device personalization (DDP) package::
 
testpmd> ddp add (port_id) (package_path[,output_path])
 
+ddp del
+~~~
+
+Delete a dynamic device personalization package::
+
+   testpmd> ddp del (port_id) (package_path)
+
 ptype mapping
 ~
 
-- 
2.13.0

[dpdk-dev] [PATCH v4 1/2] net/i40e: extended list of operations for ddp processing

[Andrey Chilikin]

This patch adds ability to remove already loaded profile
or write profile without registering it

Signed-off-by: Andrey Chilikin 
---
 drivers/net/i40e/rte_pmd_i40e.c | 86 -
 drivers/net/i40e/rte_pmd_i40e.h |  6 ++-
 2 files changed, 65 insertions(+), 27 deletions(-)

diff --git a/drivers/net/i40e/rte_pmd_i40e.c b/drivers/net/i40e/rte_pmd_i40e.c
index 45cdcfaa3..467238891 100644
--- a/drivers/net/i40e/rte_pmd_i40e.c
+++ b/drivers/net/i40e/rte_pmd_i40e.c
@@ -1554,11 +1554,7 @@ i40e_check_profile_info(uint8_t port, uint8_t 
*profile_info_sec)
 sizeof(struct i40e_profile_section_header));
for (i = 0; i < p_list->p_count; i++) {
p = &p_list->p_info[i];
-   if ((pinfo->track_id == p->track_id) &&
-   !memcmp(&pinfo->version, &p->version,
-   sizeof(struct i40e_ddp_version)) &&
-   !memcmp(&pinfo->name, &p->name,
-   I40E_DDP_NAME_SIZE)) {
+   if (pinfo->track_id == p->track_id) {
PMD_DRV_LOG(INFO, "Profile exists.");
rte_free(buff);
return 1;
@@ -1584,6 +1580,13 @@ rte_pmd_i40e_process_ddp_package(uint8_t port, uint8_t 
*buff,
int is_exist;
enum i40e_status_code status = I40E_SUCCESS;
 
+   if (op != RTE_PMD_I40E_PKG_OP_WR_ADD &&
+   op != RTE_PMD_I40E_PKG_OP_WR_ONLY &&
+   op != RTE_PMD_I40E_PKG_OP_WR_DEL) {
+   PMD_DRV_LOG(ERR, "Operation not supported.");
+   return -ENOTSUP;
+   }
+
RTE_ETH_VALID_PORTID_OR_ERR_RET(port, -ENODEV);
 
dev = &rte_eth_devices[port];
@@ -1620,6 +1623,10 @@ rte_pmd_i40e_process_ddp_package(uint8_t port, uint8_t 
*buff,
return -EINVAL;
}
track_id = ((struct i40e_metadata_segment *)metadata_seg_hdr)->track_id;
+   if (track_id == I40E_DDP_TRACKID_INVALID) {
+   PMD_DRV_LOG(ERR, "Invalid track_id");
+   return -EINVAL;
+   }
 
/* Find profile segment */
profile_seg_hdr = i40e_find_segment_in_package(SEGMENT_TYPE_I40E,
@@ -1639,40 +1646,67 @@ rte_pmd_i40e_process_ddp_package(uint8_t port, uint8_t 
*buff,
return -EINVAL;
}
 
+   /* Check if the profile already loaded */
+   i40e_generate_profile_info_sec(
+   ((struct i40e_profile_segment *)profile_seg_hdr)->name,
+   &((struct i40e_profile_segment *)profile_seg_hdr)->version,
+   track_id, profile_info_sec,
+   op == RTE_PMD_I40E_PKG_OP_WR_ADD);
+   is_exist = i40e_check_profile_info(port, profile_info_sec);
+   if (is_exist < 0) {
+   PMD_DRV_LOG(ERR, "Failed to check profile.");
+   rte_free(profile_info_sec);
+   return -EINVAL;
+   }
+
if (op == RTE_PMD_I40E_PKG_OP_WR_ADD) {
-   /* Check if the profile exists */
-   i40e_generate_profile_info_sec(
-((struct i40e_profile_segment *)profile_seg_hdr)->name,
-&((struct i40e_profile_segment *)profile_seg_hdr)->version,
-track_id, profile_info_sec, 1);
-   is_exist = i40e_check_profile_info(port, profile_info_sec);
-   if (is_exist > 0) {
+   if (is_exist) {
PMD_DRV_LOG(ERR, "Profile already exists.");
rte_free(profile_info_sec);
-   return 1;
-   } else if (is_exist < 0) {
-   PMD_DRV_LOG(ERR, "Failed to check profile.");
+   return -EEXIST;
+   }
+   } else if (op == RTE_PMD_I40E_PKG_OP_WR_DEL) {
+   if (!is_exist) {
+   PMD_DRV_LOG(ERR, "Profile does not exist.");
rte_free(profile_info_sec);
-   return -EINVAL;
+   return -EACCES;
}
+   }
 
-   /* Write profile to HW */
+   if (op == RTE_PMD_I40E_PKG_OP_WR_DEL) {
+   status = i40e_rollback_profile(
+   hw,
+   (struct i40e_profile_segment *)profile_seg_hdr,
+   track_id);
+   if (status) {
+   PMD_DRV_LOG(ERR, "Failed to write profile for delete.");
+   rte_free(profile_info_sec);
+   return status;
+   }
+   } else {
status = i40e_write_profile(
-   hw,
-   (struct i40e_profile_segment *)profile_seg_hdr,
-   track_id);
+   hw,
+   (struct i40e_profile_segment *)profile_seg_hdr,
+   track_id);
if (status) {
-   PMD_DRV_LOG(ERR, "Failed to write

[dpdk-dev] [PATCH v4 0/2] net/i40e: extended list of operations for ddp processing

[Andrey Chilikin]

This patch adds two new operations for dynamic device personalisation:
* remove already loaded profile and delete it from the list
* write profile without registering it

This patchset depends on:
[PATCH v2 00/16] net/i40e: base code update
http://dpdk.org/ml/archives/dev/2017-June/068732.html
http://dpdk.org/dev/patchwork/patch/25705/

v4:
- code style fixed in rte_pmd_i40e.c

v3:
- move testpmd updates to 'ddp add' command to a separate
  patch http://dpdk.org/dev/patchwork/patch/25779/

v2:
- Local static functions replaced by corresponding new
  functions in i40e base code
- Test-pmd command added

Andrey Chilikin (2):
  net/i40e: extended list of operations for ddp processing
  app/testpmd: enable ddp remove profile feature

 app/test-pmd/cmdline.c  | 74 +
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  7 +++
 drivers/net/i40e/rte_pmd_i40e.c | 86 -
 drivers/net/i40e/rte_pmd_i40e.h |  6 +-
 4 files changed, 146 insertions(+), 27 deletions(-)

-- 
2.13.0

Re: [dpdk-dev] [PATCH v4 1/2] net/i40e: extended list of operations for ddp processing

[Xing, Beilei]

> -Original Message-
> From: Chilikin, Andrey
> Sent: Wednesday, June 28, 2017 4:15 PM
> To: dev@dpdk.org
> Cc: Xing, Beilei ; Wu, Jingjing
> ; Chilikin, Andrey 
> Subject: [PATCH v4 1/2] net/i40e: extended list of operations for ddp
> processing
> 
> This patch adds ability to remove already loaded profile or write profile
> without registering it
> 
> Signed-off-by: Andrey Chilikin 

Acked-by: Beilei Xing 

Re: [dpdk-dev] [PATCH v2 5/5] app/testpmd: add command to test NIC restoration

[Wu, Jingjing]

> -Original Message-
> From: Dai, Wei
> Sent: Tuesday, June 27, 2017 10:07 PM
> To: tho...@monjalon.net; Lu, Wenzhuo ; Ananyev,
> Konstantin ; Zhang, Helin
> ; Wu, Jingjing 
> Cc: dev@dpdk.org; Dai, Wei 
> Subject: [PATCH v2 5/5] app/testpmd: add command to test NIC restoration
> 
> When a NIC is reset, a message will show it.
> And then user can run the command "reset_port port_id"
> to reset the port and to keep same port id without any without any
> configuration.
> This patch adds a testpmd command "reconfig_port port_id"
> to test whether the port can be reconfigured to have success Rx and Tx 
> function.
> The new command will configure the port with the simplest setting which
> includes only 1 Rx queue, only 1 Tx queue, Rx mode = None and Tx mode =
> None.
> It check if the port can receive and forward some packets.
> For example 100 packets in this new command.
> Before testing with "reset_port port_id" and then "reconfig_port port_id",
> current forwarding should be stopped first to avoid crash.
> 

Testpmd has mechanism to support reconfigure, you can check the need_reconfig
Field in port struct.
No need to define a new command.


Thanks
Jingjing

Re: [dpdk-dev] [PATCH v2 4/5] app/testpmd: add command to test NIC reset

[Wu, Jingjing]

> -Original Message-
> From: Dai, Wei
> Sent: Tuesday, June 27, 2017 10:07 PM
> To: tho...@monjalon.net; Lu, Wenzhuo ; Ananyev,
> Konstantin ; Zhang, Helin
> ; Wu, Jingjing 
> Cc: dev@dpdk.org; Dai, Wei 
> Subject: [PATCH v2 4/5] app/testpmd: add command to test NIC reset
> 
> When a NIC is reset, a message will show it.
> And then user can run the command "reset_port port_id"
> to process it.
> 
> Signed-off-by: Wei Dai 
> ---
>  app/test-pmd/cmdline.c | 31 +++  app/test-
> pmd/config.c  | 13 +  app/test-pmd/testpmd.h |  1 +
>  3 files changed, 45 insertions(+)
> 
> diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c index
> ff8ffd2..90f6bde 100644
> --- a/app/test-pmd/cmdline.c
> +++ b/app/test-pmd/cmdline.c
> @@ -2588,6 +2588,36 @@ cmdline_parse_inst_t cmd_stop = {
>   },
>  };
> 
> +/* *** reset a port *** */
> +struct cmd_reset_port_result {
> + cmdline_fixed_string_t command;
> + uint8_t port_id;
> +};
> +
> +static void cmd_reset_port_parsed(__attribute__((unused)) void
> *parsed_result,
> + __attribute__((unused)) struct cmdline *cl,
> + __attribute__((unused)) void *data) {
> + struct cmd_reset_port_result *res = parsed_result;
> + reset_port(res->port_id);
> +}
> +
> +cmdline_parse_token_string_t cmd_reset_port_cmd =
> + TOKEN_STRING_INITIALIZER(struct cmd_reset_port_result, command,
> +"reset_port"); cmdline_parse_token_num_t cmd_reset_port_id =
> + TOKEN_NUM_INITIALIZER(struct cmd_reset_port_result, port_id,
> UINT8);
> +
> +cmdline_parse_inst_t cmd_reset_port = {
> + .f = cmd_reset_port_parsed,
> + .data = NULL,
> + .help_str = "reset_port ",

You can just add one more option in the commands
"port start|stop|close " and " port start|stop|close all"
To "port start|stop|close|reset " and " port start|stop|close|reset 
all" instead of defining a new one.

[dpdk-dev] [PATCH] examples/vhost: introduce a new vhost-user-scsi sample application

[Changpeng Liu]

vhost-user protocol is common to many virtio devices, such as
virtio_net/virtio_scsi/virtio_blk. Since DPDK vhost library
removed the NET specific data structures, the vhost library
is common to other virtio devices, such as virtio-scsi.

Here we introduce a simple memory based block device that
can be presented to Guest VM through vhost-user-scsi-pci
controller. Similar with vhost-net, the sample application
will process the I/Os sent via virt rings.

Signed-off-by: Changpeng Liu 
---
 MAINTAINERS |   2 +
 doc/guides/sample_app_ug/vhost_scsi.rst | 134 +
 examples/Makefile   |   2 +-
 examples/vhost_scsi/Makefile|  59 
 examples/vhost_scsi/scsi.c  | 507 
 examples/vhost_scsi/scsi_spec.h | 488 ++
 examples/vhost_scsi/vhost_scsi.c| 472 +
 examples/vhost_scsi/vhost_scsi.h| 250 
 8 files changed, 1913 insertions(+), 1 deletion(-)
 create mode 100644 doc/guides/sample_app_ug/vhost_scsi.rst
 create mode 100644 examples/vhost_scsi/Makefile
 create mode 100644 examples/vhost_scsi/scsi.c
 create mode 100644 examples/vhost_scsi/scsi_spec.h
 create mode 100644 examples/vhost_scsi/vhost_scsi.c
 create mode 100644 examples/vhost_scsi/vhost_scsi.h

diff --git a/MAINTAINERS b/MAINTAINERS
index f6095ef..cd0fb7e 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -431,6 +431,8 @@ F: lib/librte_vhost/
 F: doc/guides/prog_guide/vhost_lib.rst
 F: examples/vhost/
 F: doc/guides/sample_app_ug/vhost.rst
+F: examples/vhost_scsi/
+F: doc/guides/sample_app_ug/vhost_scsi.rst
 
 Vhost PMD
 M: Tetsuya Mukawa 
diff --git a/doc/guides/sample_app_ug/vhost_scsi.rst 
b/doc/guides/sample_app_ug/vhost_scsi.rst
new file mode 100644
index 000..2213bfd
--- /dev/null
+++ b/doc/guides/sample_app_ug/vhost_scsi.rst
@@ -0,0 +1,134 @@
+
+..  BSD LICENSE
+Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+* Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+* Redistributions in binary form must reproduce the above copyright
+notice, this list of conditions and the following disclaimer in
+the documentation and/or other materials provided with the
+distribution.
+* Neither the name of Intel Corporation nor the names of its
+contributors may be used to endorse or promote products derived
+from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
+Vhost_scsi Sample Application
+
+
+The vhost_scsi sample application implemented a simple SCSI block device,
+which used as the  backend of Qemu vhost-user-scsi device. Users can extend
+the exist example to use other type of block device(e.g. AIO) besides
+memory based block device. Similar with vhost-user-net device, the sample
+application used domain socket to communicate with Qemu, and the virtio
+ring was processed by vhost_scsi sample application.
+
+Testing steps
+-
+
+This section shows the steps how to start a VM with the block device as
+fast data path for critical application.
+
+Build
+~
+
+Follow the *Getting Started Guide for Linux* on generic info about
+environment setup and building DPDK from source.
+
+In this example, you need build DPDK both on the host and inside guest.
+Also, you need build this example.
+
+.. code-block:: console
+
+export RTE_SDK=/path/to/dpdk_source
+export RTE_TARGET=x86_64-native-linuxapp-gcc
+
+cd ${RTE_SDK}/examples/vhost_scsi
+make
+
+
+Start the vhost_scsi example
+~
+
+.. code-block:: console
+
+./vhost_scsi -m 1024
+
+Check the `Parameters`_ section for the explanations on what do those
+parameters mean.
+
+.. _vhost_scsi_app_run_vm:
+
+Start the VM
+
+
+.. code-block:: console
+
+qemu-system-x86_64 -machine accel=kvm \
+-m $mem -

Re: [dpdk-dev] [PATCH] net/bond: change link status check to no-wait

[Declan Doherty]

On 26/06/2017 9:53 PM, Chas Williams wrote:

On Mon, 2017-06-26 at 16:13 +0100, Declan Doherty wrote:

In the 802.3ad periodic callback function the link status of all slaves
is checked using rte_eth_link_get function. Depending on the slave
device this function can block for up to 9 seconds and therefore
could cause issues with the LACP Daemon state machine and control
patches handling. This patch changes to the no-wait version of
the link get function.

Fixes: 46fb4368 ("bond:add mode 4")

Signed-off-by: Declan Doherty 
---

...


It seems like this should be done in bandwidth_left() as well.



Yep, your right. I'll make that change in the v2.

Thanks
Declan

Re: [dpdk-dev] [PATCH v5 04/12] net/failsafe: add fail-safe PMD

[Thomas Monjalon]

You need this patch for shared library build:

--- a/drivers/net/failsafe/Makefile
+++ b/drivers/net/failsafe/Makefile
@@ -34,6 +34,10 @@ include $(RTE_SDK)/mk/rte.vars.mk
 # Library name
 LIB = librte_pmd_failsafe.a
 
+EXPORT_MAP := rte_pmd_failsafe_version.map
+
+LIBABIVER := 1
+
 # Sources are stored in SRCS-y
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_FAILSAFE) += failsafe.c
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_FAILSAFE) += failsafe_args.c

--- /dev/null
+++ b/drivers/net/failsafe/rte_pmd_failsafe_version.map
@@ -0,0 +1,4 @@
+DPDK_17.08 {
+
+   local: *;
+};

Re: [dpdk-dev] [PATCH v6 09/11] pci: implement find_device bus operation

[Gaëtan Rivet]

On Tue, Jun 27, 2017 at 04:35:14PM -0700, Stephen Hemminger wrote:
> On Tue, 27 Jun 2017 18:11:16 +0200
> Gaetan Rivet  wrote:
> 
> > +   int start_found = !!(start == NULL);
> > +
> > +   FOREACH_DEVICE_ON_PCIBUS(dev) {
> > +   if (!start_found) {
> > +   if (&dev->device == start)
> > +   start_found = 1;
> > +   continue;
> > +   }
> > +   if (cmp(&dev->device, data) == 0)
> > +   return &dev->device;
> > +   }
> > +   return NULL;
> > +}
> > +
> 
> Why is start_found not a boolean?
> 

Ah, yes, I wrote this a few times over in rte_bus and rte_vdev, and
mostly used the same scheme in the PCI implementation, without checking
for the use of stdbool in the vincinity otherwise.

I would not be opposed to using a bool if I was rewriting it, but I
don't feel this change warrants a new version.

> Do you really need start_found at all? Why not just reuse existing
> pointer?
> 

You are right, it could be reduced. But I find it a little too "clever"
in a sense, and I prefer usually to avoid rewriting input parameters. If this
function had to be refactored later, the writer would need to be careful
about start having changed. The advantage of using one variable less
does not outweight this drawback I think.

>   FOREACH_DEVICE_ON_PCIBUS(dev) {
>   if (start) {
>   if (&dev->device == start)
>   start = NULL
>   continue;
>   }
>   if (cmp(&dev->device, data) == 0)
>   return &dev->device;
>   }
>   return NULL;
> }

-- 
Gaëtan Rivet
6WIND

[dpdk-dev] [PATCH] lib/lib_eal:fix the pointer 'elem' may be NULL bug

[RongQiang Xie]

Signed-off-by: RongQiang Xie 
---
 lib/librte_eal/common/eal_common_memzone.c | 6 ++
 1 file changed, 6 insertions(+)
 mode change 100644 => 100755 lib/librte_eal/common/eal_common_memzone.c

diff --git a/lib/librte_eal/common/eal_common_memzone.c 
b/lib/librte_eal/common/eal_common_memzone.c
old mode 100644
new mode 100755
index 3026e36..8ea781b
--- a/lib/librte_eal/common/eal_common_memzone.c
+++ b/lib/librte_eal/common/eal_common_memzone.c
@@ -238,6 +238,12 @@
}
 
const struct malloc_elem *elem = malloc_elem_from_data(mz_addr);
+   
+   if (elem == NULL) {
+   RTE_LOG(ERR, EAL, "%s(): Cannot malloc elem from data \n", 
__func__);
+   rte_errno = ENOSPC;
+   return NULL;
+   }
 
/* fill the zone in config */
mz = get_next_free_memzone();
-- 
1.8.3.1

Re: [dpdk-dev] [PATCH v4 0/2] net/i40e: extended list of operations for ddp processing

[Ferruh Yigit]

On 6/28/2017 9:15 AM, Andrey Chilikin wrote:
> This patch adds two new operations for dynamic device personalisation:
> * remove already loaded profile and delete it from the list
> * write profile without registering it
> 
> This patchset depends on:
> [PATCH v2 00/16] net/i40e: base code update
> http://dpdk.org/ml/archives/dev/2017-June/068732.html
> http://dpdk.org/dev/patchwork/patch/25705/
> 
> v4:
> - code style fixed in rte_pmd_i40e.c
> 
> v3:
> - move testpmd updates to 'ddp add' command to a separate
>   patch http://dpdk.org/dev/patchwork/patch/25779/
> 
> v2:
> - Local static functions replaced by corresponding new
>   functions in i40e base code
> - Test-pmd command added
> 
> Andrey Chilikin (2):
>   net/i40e: extended list of operations for ddp processing
>   app/testpmd: enable ddp remove profile feature

Series applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH 1/2] bond: avoid allocating mempool on socket LCORE_ID_ANY

[Declan Doherty]

On 26/04/17 22:30, Jan Blunck wrote:

If the numa node is unknown, use the value from rte_socket_id() to avoid
an allocation failure.

Signed-off-by: Eric Kinzie 
CC: Eric Kinzie 
---
  drivers/net/bonding/rte_eth_bond_8023ad.c | 2 ++
  1 file changed, 2 insertions(+)

diff --git a/drivers/net/bonding/rte_eth_bond_8023ad.c 
b/drivers/net/bonding/rte_eth_bond_8023ad.c
index 7b863d6..b8df536 100644
--- a/drivers/net/bonding/rte_eth_bond_8023ad.c
+++ b/drivers/net/bonding/rte_eth_bond_8023ad.c
@@ -887,6 +887,8 @@ bond_mode_8023ad_activate_slave(struct rte_eth_dev 
*bond_dev, uint8_t slave_id)
RTE_ASSERT(port->rx_ring == NULL);
RTE_ASSERT(port->tx_ring == NULL);
socket_id = rte_eth_devices[slave_id].data->numa_node;
+   if (socket_id == (uint8_t)LCORE_ID_ANY)


socket_id is an integer so we shouldn't be casting LCORE_ID_ANY to an 
uint8_t as this condition would never match as LCORE_ID_ANY wouldn't be 
UINT32_MAX after casting



+   socket_id = 0;


Should this not be socket_id = rte_socket_id(); as we don't know we are 
running on


  
  	element_size = sizeof(struct slow_protocol_frame) +

RTE_PKTMBUF_HEADROOM;

Re: [dpdk-dev] [PATCH 2/2] bond: use rte_eth_dev_socket_id()

[Declan Doherty]

On 26/04/17 22:30, Jan Blunck wrote:

The bonded device does not have to be a PCI device. Use the rte_ethdev
functions instead.

Signed-off-by: Wen Chiu 
Signed-off-by: Stephen Hemminger 
Signed-off-by: Jan Blunck 
---
  drivers/net/bonding/rte_eth_bond_8023ad.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/bonding/rte_eth_bond_8023ad.c 
b/drivers/net/bonding/rte_eth_bond_8023ad.c
index b8df536..f84553c 100644
--- a/drivers/net/bonding/rte_eth_bond_8023ad.c
+++ b/drivers/net/bonding/rte_eth_bond_8023ad.c
@@ -886,7 +886,7 @@ bond_mode_8023ad_activate_slave(struct rte_eth_dev 
*bond_dev, uint8_t slave_id)
  
  	RTE_ASSERT(port->rx_ring == NULL);

RTE_ASSERT(port->tx_ring == NULL);
-   socket_id = rte_eth_devices[slave_id].data->numa_node;
+   socket_id = rte_eth_dev_socket_id(slave_id);
if (socket_id == (uint8_t)LCORE_ID_ANY)
socket_id = 0;
  



Acked-by: Declan Doherty 

Re: [dpdk-dev] [PATCH v6 09/11] pci: implement find_device bus operation

[Richardson, Bruce]

> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Gaëtan Rivet
> Sent: Wednesday, June 28, 2017 10:18 AM
> To: Stephen Hemminger 
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v6 09/11] pci: implement find_device bus
> operation
> 
> On Tue, Jun 27, 2017 at 04:35:14PM -0700, Stephen Hemminger wrote:
> > On Tue, 27 Jun 2017 18:11:16 +0200
> > Gaetan Rivet  wrote:
> >
> > > + int start_found = !!(start == NULL);
> > > +
> > > + FOREACH_DEVICE_ON_PCIBUS(dev) {
> > > + if (!start_found) {
> > > + if (&dev->device == start)
> > > + start_found = 1;
> > > + continue;
> > > + }
> > > + if (cmp(&dev->device, data) == 0)
> > > + return &dev->device;
> > > + }
> > > + return NULL;
> > > +}
> > > +
> >
> > Why is start_found not a boolean?
> >
> 
> Ah, yes, I wrote this a few times over in rte_bus and rte_vdev, and mostly
> used the same scheme in the PCI implementation, without checking for the
> use of stdbool in the vincinity otherwise.
> 
> I would not be opposed to using a bool if I was rewriting it, but I don't
> feel this change warrants a new version.
> 
> > Do you really need start_found at all? Why not just reuse existing
> > pointer?
> >
> 
> You are right, it could be reduced. But I find it a little too "clever"
> in a sense, and I prefer usually to avoid rewriting input parameters. If
> this function had to be refactored later, the writer would need to be
> careful about start having changed. The advantage of using one variable
> less does not outweight this drawback I think.
> 
+1 for having an extra variable for clarity.

Re: [dpdk-dev] [PATCH] vfio: allow to map other memory regions

[Wodkowski, PawelX]

> -Original Message-
> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> Sent: Monday, June 19, 2017 11:04 PM
> To: Wodkowski, PawelX 
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH] vfio: allow to map other memory regions
> 
> Hi,
> Some comments below
> 
> 24/05/2017 13:17, Pawel Wodkowski:
> > Currently it is not possible to use memory that is not owned by DPDK to
> > perform DMA. This scenarion might be used in vhost applications (like
> > SPDK) where guest send its own memory table. To fill this gap provide
> > API to allow registering arbitrary address in VFIO container.
> >
> > Signed-off-by: Pawel Wodkowski 
> > ---
> >  lib/librte_eal/linuxapp/eal/Makefile|   3 +
> >  lib/librte_eal/linuxapp/eal/eal_vfio.c  | 142
> +---
> >  lib/librte_eal/linuxapp/eal/eal_vfio.h  |  10 ++
> >  lib/librte_eal/linuxapp/eal/include/rte_iommu.h |  78 +
> >  lib/librte_eal/linuxapp/eal/rte_eal_version.map |   8 ++
> >  5 files changed, 224 insertions(+), 17 deletions(-)
> >  create mode 100644 lib/librte_eal/linuxapp/eal/include/rte_iommu.h
> 
> VFIO is not referenced in the doxygen of these functions.
> Could we use this API for something else than VFIO?

This is for any IOMMU hw/module/driver used in host which require special
care about memory regions used for DMA. It is not restricted to VFIO even though
only VFIO is implemented.

> 
> Any API should be declared in common directory, even if it is not
> implemented for FreeBSD (returning -ENOTSUP).

I think those function should be NOP for FreeBSD (like RTE_VFIO_NOIOMMU do)
or be conditionally compiled/included (like it is now). I decide to take second 
way.
Do you think that I should move rte_iommu.h to common directory and use #ifdef
there?

Pawel

Re: [dpdk-dev] [PATCH v2] net/tap: restore state of remote device when closing

[Ferruh Yigit]

> On 06/27/2017 02:33 PM, Thomas Monjalon wrote:
>> When exiting a DPDK application, the TAP remote was left
>> with the link down even if it was initially up.
>>
>> The device flags of the remote netdevice are saved when probing,
>> and restored when calling the close function.
>> The remote state is not set down when calling the stop function anymore.
>>
>> Signed-off-by: Thomas Monjalon 

> Acked-by: Pascal Mazon 

Applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH] lib/lib_eal:fix the pointer 'elem' may be NULL bug

[Sergio Gonzalez Monroy]

On 28/06/2017 09:12, RongQiang Xie wrote:

Signed-off-by: RongQiang Xie 
---
  lib/librte_eal/common/eal_common_memzone.c | 6 ++
  1 file changed, 6 insertions(+)
  mode change 100644 => 100755 lib/librte_eal/common/eal_common_memzone.c

diff --git a/lib/librte_eal/common/eal_common_memzone.c 
b/lib/librte_eal/common/eal_common_memzone.c
old mode 100644
new mode 100755
index 3026e36..8ea781b
--- a/lib/librte_eal/common/eal_common_memzone.c
+++ b/lib/librte_eal/common/eal_common_memzone.c
@@ -238,6 +238,12 @@
}
  
  	const struct malloc_elem *elem = malloc_elem_from_data(mz_addr);

+   
+   if (elem == NULL) {
+   RTE_LOG(ERR, EAL, "%s(): Cannot malloc elem from data \n", 
__func__);
+   rte_errno = ENOSPC;
+   return NULL;
+   }
  
  	/* fill the zone in config */

mz = get_next_free_memzone();


Hi,

Could you elaborate how do you trigger this error?

The only way you are getting NULL from malloc_elem_from_data is:
A) data is NULL, which cannot be the case as we previously check for it
B) cookies failure id you have DEBUG enabled, otherwise elem would have 
to be NULL


Thanks,
Sergio

[dpdk-dev] [PATCH v2 0/3] net/ark: augment user extension and bug fix

[John Miller]

v2:
* Split the user extension private data per port patch and the
  set_mtu patch into separate patches.
* Add better description to per port private data patch
* stats_reset patch untouched from V1

John Miller (3):
  net/ark: allow unique user data for each port in extension calls
  net/ark: add set_mtu call to user extension API
  net/ark: fix bug in stats_reset operation

 drivers/net/ark/ark_ethdev.c | 72 
 drivers/net/ark/ark_ext.h|  4 +++
 drivers/net/ark/ark_global.h |  5 +--
 3 files changed, 60 insertions(+), 21 deletions(-)

-- 
1.9.1

[dpdk-dev] [PATCH v2 2/3] net/ark: add set_mtu call to user extension API

[John Miller]

Allows a user extension to set a callback for the set_mtu
operation.

Signed-off-by: John Miller 
---
 drivers/net/ark/ark_ethdev.c | 19 +++
 drivers/net/ark/ark_ext.h|  4 
 drivers/net/ark/ark_global.h |  1 +
 3 files changed, 24 insertions(+)

diff --git a/drivers/net/ark/ark_ethdev.c b/drivers/net/ark/ark_ethdev.c
index e5a8ff7..5733ba3 100644
--- a/drivers/net/ark/ark_ethdev.c
+++ b/drivers/net/ark/ark_ethdev.c
@@ -77,6 +77,7 @@ static int eth_ark_macaddr_add(struct rte_eth_dev *dev,
   uint32_t pool);
 static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
   uint32_t index);
+static int  eth_ark_set_mtu(struct rte_eth_dev *dev, uint16_t size);
 
 /*
  * The packet generator is a functional block used to generate packet
@@ -180,6 +181,7 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
.mac_addr_remove = eth_ark_macaddr_remove,
.mac_addr_set = eth_ark_set_default_mac_addr,
 
+   .mtu_set = eth_ark_set_mtu,
 };
 
 static int
@@ -257,6 +259,10 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
(void (*)(struct rte_eth_dev *, struct ether_addr *,
  void *))
dlsym(ark->d_handle, "mac_addr_set");
+   ark->user_ext.set_mtu =
+   (int (*)(struct rte_eth_dev *, uint16_t,
+ void *))
+   dlsym(ark->d_handle, "set_mtu");
 
return found;
 }
@@ -887,6 +893,19 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
   ark->user_data[dev->data->port_id]);
 }
 
+static int
+eth_ark_set_mtu(struct rte_eth_dev *dev, uint16_t  size)
+{
+   struct ark_adapter *ark =
+   (struct ark_adapter *)dev->data->dev_private;
+
+   if (ark->user_ext.set_mtu)
+   return ark->user_ext.set_mtu(dev, size,
+ark->user_data[dev->data->port_id]);
+
+   return -ENOTSUP;
+}
+
 static inline int
 process_pktdir_arg(const char *key, const char *value,
   void *extra_args)
diff --git a/drivers/net/ark/ark_ext.h b/drivers/net/ark/ark_ext.h
index f805f64..63b7a26 100644
--- a/drivers/net/ark/ark_ext.h
+++ b/drivers/net/ark/ark_ext.h
@@ -112,4 +112,8 @@ void mac_addr_set(struct rte_eth_dev *dev,
  struct ether_addr *mac_addr,
  void *user_data);
 
+int set_mtu(struct rte_eth_dev *dev,
+   uint16_t size,
+   void *user_data);
+
 #endif
diff --git a/drivers/net/ark/ark_global.h b/drivers/net/ark/ark_global.h
index 58af8db..2a6375f 100644
--- a/drivers/net/ark/ark_global.h
+++ b/drivers/net/ark/ark_global.h
@@ -106,6 +106,7 @@ struct ark_user_ext {
 void *);
void (*mac_addr_remove)(struct rte_eth_dev *, uint32_t, void *);
void (*mac_addr_set)(struct rte_eth_dev *, struct ether_addr *, void *);
+   int (*set_mtu)(struct rte_eth_dev *, uint16_t, void *);
 };
 
 struct ark_adapter {
-- 
1.9.1

[dpdk-dev] [PATCH v2 1/3] net/ark: allow unique user data for each port in extension calls

[John Miller]

This change allows a user extension to provide unique private
callback data for all ports.

Arkville is a single-function multi-port device.  User_data resides
in the singleton Arkville structure.  This structure is shared across
all ports devices (eth_dev) which are created one per port.  For the
command extension callback we provide an opaque user pointer, which is
currently implemented in the singleton Arkville structure.

With this patch, we are providing a unique user pointer for each port
rather than a common pointer across multiple ports.  The pointers are
stored in an array of size RTE_MAX_ETHPORTS in the arkville structure and
are indexed by port_id in the PMD. The motivation for this change is that
users of the arkville PMD extension have a unique pointer per port rather
then one per function.

Signed-off-by: John Miller 
---
 drivers/net/ark/ark_ethdev.c | 51 
 drivers/net/ark/ark_global.h |  4 ++--
 2 files changed, 35 insertions(+), 20 deletions(-)

diff --git a/drivers/net/ark/ark_ethdev.c b/drivers/net/ark/ark_ethdev.c
index c061f7b..e5a8ff7 100644
--- a/drivers/net/ark/ark_ethdev.c
+++ b/drivers/net/ark/ark_ethdev.c
@@ -179,6 +179,7 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
.mac_addr_add = eth_ark_macaddr_add,
.mac_addr_remove = eth_ark_macaddr_remove,
.mac_addr_set = eth_ark_set_default_mac_addr,
+
 };
 
 static int
@@ -346,8 +347,9 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
}
 
if (ark->user_ext.dev_init) {
-   ark->user_data = ark->user_ext.dev_init(dev, ark->a_bar, 0);
-   if (!ark->user_data) {
+   ark->user_data[dev->data->port_id] =
+   ark->user_ext.dev_init(dev, ark->a_bar, 0);
+   if (!ark->user_data[dev->data->port_id]) {
PMD_DRV_LOG(INFO,
"Failed to initialize PMD extension!"
" continuing without it\n");
@@ -369,7 +371,8 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
 */
if (ark->user_ext.dev_get_port_count)
port_count =
-   ark->user_ext.dev_get_port_count(dev, ark->user_data);
+   ark->user_ext.dev_get_port_count(dev,
+ark->user_data[dev->data->port_id]);
ark->num_ports = port_count;
 
for (p = 0; p < port_count; p++) {
@@ -410,9 +413,10 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
goto error;
}
 
-   if (ark->user_ext.dev_init)
-   ark->user_data =
+   if (ark->user_ext.dev_init) {
+   ark->user_data[eth_dev->data->port_id] =
ark->user_ext.dev_init(dev, ark->a_bar, p);
+   }
}
 
return ret;
@@ -508,7 +512,8 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
return 0;
 
if (ark->user_ext.dev_uninit)
-   ark->user_ext.dev_uninit(dev, ark->user_data);
+   ark->user_ext.dev_uninit(dev,
+ark->user_data[dev->data->port_id]);
 
ark_pktgen_uninit(ark->pg);
ark_pktchkr_uninit(ark->pc);
@@ -529,7 +534,8 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
 
eth_ark_dev_set_link_up(dev);
if (ark->user_ext.dev_configure)
-   return ark->user_ext.dev_configure(dev, ark->user_data);
+   return ark->user_ext.dev_configure(dev,
+  ark->user_data[dev->data->port_id]);
return 0;
 }
 
@@ -592,7 +598,8 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
}
 
if (ark->user_ext.dev_start)
-   ark->user_ext.dev_start(dev, ark->user_data);
+   ark->user_ext.dev_start(dev,
+   ark->user_data[dev->data->port_id]);
 
return 0;
 }
@@ -614,7 +621,8 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
 
/* Stop the extension first */
if (ark->user_ext.dev_stop)
-   ark->user_ext.dev_stop(dev, ark->user_data);
+   ark->user_ext.dev_stop(dev,
+  ark->user_data[dev->data->port_id]);
 
/* Stop the packet generator */
if (ark->start_pg)
@@ -697,7 +705,8 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
uint16_t i;
 
if (ark->user_ext.dev_close)
-   ark->user_ext.dev_close(dev, ark->user_data);
+   ark->user_ext.dev_close(dev,
+ark->user_data[dev->data->port_id]);
 
eth_ark_dev_stop(dev);
eth_ark_udm_force_close(dev);
@@ -765,7 +774,7 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
if (ark->user_ext.link_update) {
return ark->user_ext.link_update

[dpdk-dev] [PATCH v2 3/3] net/ark: fix bug in stats_reset operation

[John Miller]

Repairs a bug in the stats_reset where the wrong queue was
being passed into tx reset.

Signed-off-by: John Miller 
---
 drivers/net/ark/ark_ethdev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ark/ark_ethdev.c b/drivers/net/ark/ark_ethdev.c
index 5733ba3..6db362b 100644
--- a/drivers/net/ark/ark_ethdev.c
+++ b/drivers/net/ark/ark_ethdev.c
@@ -842,7 +842,7 @@ static void eth_ark_macaddr_remove(struct rte_eth_dev *dev,
(struct ark_adapter *)dev->data->dev_private;
 
for (i = 0; i < dev->data->nb_tx_queues; i++)
-   eth_tx_queue_stats_reset(dev->data->rx_queues[i]);
+   eth_tx_queue_stats_reset(dev->data->tx_queues[i]);
for (i = 0; i < dev->data->nb_rx_queues; i++)
eth_rx_queue_stats_reset(dev->data->rx_queues[i]);
if (ark->user_ext.stats_reset)
-- 
1.9.1

[dpdk-dev] [PATCH v2] net/bond: change link status check to no-wait

[Declan Doherty]

In 2 modes (802.3ad and TLB) of the link bonding driver, monitoring of
link status is used to determine the active slaves to use. These
functions are currently using the rte_link_link_get which could block
for up to 9 seconds, depending on the slave device.

In the 802.3ad periodic callback and in the TLB slave callback the
link status of slaves are checked using rte_eth_link_get function.

This patch changes to the no-wait version of the link get function to
avoid the possible issues which could be introduced if this call was
to block for an extended period of time.

Fixes: 46fb4368 ("bond:add mode 4")
Fixes: 7c76a747 ("bond:add mode 5")

Signed-off-by: Declan Doherty 
---
 drivers/net/bonding/rte_eth_bond_8023ad.c | 2 +-
 drivers/net/bonding/rte_eth_bond_pmd.c| 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/bonding/rte_eth_bond_8023ad.c 
b/drivers/net/bonding/rte_eth_bond_8023ad.c
index d2b7592..3a97336 100644
--- a/drivers/net/bonding/rte_eth_bond_8023ad.c
+++ b/drivers/net/bonding/rte_eth_bond_8023ad.c
@@ -758,7 +758,7 @@ bond_mode_8023ad_periodic_cb(void *arg)
uint16_t key;
 
slave_id = internals->active_slaves[i];
-   rte_eth_link_get(slave_id, &link_info);
+   rte_eth_link_get_nowait(slave_id, &link_info);
rte_eth_macaddr_get(slave_id, &slave_addr);
 
if (link_info.link_status != 0) {
diff --git a/drivers/net/bonding/rte_eth_bond_pmd.c 
b/drivers/net/bonding/rte_eth_bond_pmd.c
index a8d9780..eff2a66 100644
--- a/drivers/net/bonding/rte_eth_bond_pmd.c
+++ b/drivers/net/bonding/rte_eth_bond_pmd.c
@@ -654,7 +654,7 @@ bandwidth_left(uint8_t port_id, uint64_t load, uint8_t 
update_idx,
 {
struct rte_eth_link link_status;
 
-   rte_eth_link_get(port_id, &link_status);
+   rte_eth_link_get_nowait(port_id, &link_status);
uint64_t link_bwg = link_status.link_speed * 100ULL / 8;
if (link_bwg == 0)
return;
-- 
2.9.4

Re: [dpdk-dev] [PATCH v9 1/2] mem: balanced allocation of hugepages

[Sergio Gonzalez Monroy]

On 27/06/2017 11:24, Ilya Maximets wrote:

Currently EAL allocates hugepages one by one not paying attention
from which NUMA node allocation was done.

Such behaviour leads to allocation failure if number of available
hugepages for application limited by cgroups or hugetlbfs and
memory requested not only from the first socket.

Example:
# 90 x 1GB hugepages availavle in a system

cgcreate -g hugetlb:/test
# Limit to 32GB of hugepages
cgset -r hugetlb.1GB.limit_in_bytes=34359738368 test
# Request 4GB from each of 2 sockets
cgexec -g hugetlb:test testpmd --socket-mem=4096,4096 ...

EAL: SIGBUS: Cannot mmap more hugepages of size 1024 MB
EAL: 32 not 90 hugepages of size 1024 MB allocated
EAL: Not enough memory available on socket 1!
 Requested: 4096MB, available: 0MB
PANIC in rte_eal_init():
Cannot init memory

This happens beacause all allocated pages are
on socket 0.

Fix this issue by setting mempolicy MPOL_PREFERRED for each hugepage
to one of requested nodes using following schema:

1) Allocate essential hugepages:
1.1) Allocate as many hugepages from numa N to
 only fit requested memory for this numa.
1.2) repeat 1.1 for all numa nodes.
2) Try to map all remaining free hugepages in a round-robin
   fashion.
3) Sort pages and choose the most suitable.

In this case all essential memory will be allocated and all remaining
pages will be fairly distributed between all requested nodes.

New config option RTE_EAL_NUMA_AWARE_HUGEPAGES introduced and
enabled by default for linuxapp except armv7 and dpaa2.
Enabling of this option adds libnuma as a dependency for EAL.

Fixes: 77988fc08dc5 ("mem: fix allocating all free hugepages")

Signed-off-by: Ilya Maximets 
---
  config/common_base|   1 +
  config/common_linuxapp|   1 +
  config/defconfig_arm-armv7a-linuxapp-gcc  |   3 +
  config/defconfig_arm64-dpaa2-linuxapp-gcc |   3 +
  lib/librte_eal/linuxapp/eal/Makefile  |   3 +
  lib/librte_eal/linuxapp/eal/eal_memory.c  | 120 --
  mk/rte.app.mk |   3 +
  7 files changed, 126 insertions(+), 8 deletions(-)


Good stuff Ilya!

Hemant, Jerin, could you also ack the patch if you are happy with it? 
Thanks.


Acked-by: Sergio Gonzalez Monroy 

[dpdk-dev] [PATCH v2 2/2] net/i40e: add hot plug monitor in i40e

[Jeff Guo]

From: "Guo, Jia" 

This patch enable the hot plug feature in i40e, by monitoring the
hot plug uevent of the device. When remove event got, call the app
callback function to handle the detach process.

Signed-off-by: Guo, Jia 
---
v2->v1: remove unused part for current stage.
---
 drivers/net/i40e/i40e_ethdev.c | 18 ++
 1 file changed, 18 insertions(+)

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 4ee1113..122187e 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -1283,6 +1283,7 @@ static inline void i40e_GLQF_reg_init(struct i40e_hw *hw)
 
/* enable uio intr after callback register */
rte_intr_enable(intr_handle);
+
/*
 * Add an ethertype filter to drop all flow control frames transmitted
 * from VSIs. By doing so, we stop VF from sending out PAUSE or PFC
@@ -5832,11 +5833,28 @@ struct i40e_vsi *
 {
struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
struct i40e_hw *hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+   struct rte_uevent event;
uint32_t icr0;
+   struct rte_pci_device *pci_dev;
+   struct rte_intr_handle *intr_handle;
+
+   pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+   intr_handle = &pci_dev->intr_handle;
 
/* Disable interrupt */
i40e_pf_disable_irq0(hw);
 
+   /* check device uevent */
+   if (rte_uevent_get(intr_handle->uevent_fd, &event) > 0) {
+   if (event.subsystem == RTE_UEVENT_SUBSYSTEM_UIO) {
+   if (event.action == RTE_UEVENT_REMOVE) {
+   _rte_eth_dev_callback_process(dev,
+   RTE_ETH_EVENT_INTR_RMV, NULL);
+   }
+   }
+   goto done;
+   }
+
/* read out interrupt causes */
icr0 = I40E_READ_REG(hw, I40E_PFINT_ICR0);
 
-- 
1.8.3.1

[dpdk-dev] [PATCH v2 1/2] eal: add uevent api for hot plug

[Jeff Guo]

From: "Guo, Jia" 

This patch aim to add a variable "uevent_fd" in structure
"rte_intr_handle" for enable kernel object uevent monitoring,
and add some uevent API in rte eal interrupt, that is
“rte_uevent_connect” and “rte_uevent_get”, so that all driver
could use these API to monitor and read out the uevent, then
corresponding to handle these uevent, such as detach or attach
the device.

Signed-off-by: Guo, Jia 
---
v2->v1: remove global variables of hotplug_fd, add uevent_fd
in rte_intr_handle to let each pci device self maintain
to fix dual device fd issue. refine some typo error. 
---
 lib/librte_eal/common/eal_common_pci_uio.c |   6 +-
 lib/librte_eal/linuxapp/eal/eal_interrupts.c   | 143 -
 lib/librte_eal/linuxapp/eal/eal_pci_uio.c  |  12 ++
 .../linuxapp/eal/include/exec-env/rte_interrupts.h |  34 +
 4 files changed, 192 insertions(+), 3 deletions(-)

diff --git a/lib/librte_eal/common/eal_common_pci_uio.c 
b/lib/librte_eal/common/eal_common_pci_uio.c
index 367a681..5b62f70 100644
--- a/lib/librte_eal/common/eal_common_pci_uio.c
+++ b/lib/librte_eal/common/eal_common_pci_uio.c
@@ -117,6 +117,7 @@
 
dev->intr_handle.fd = -1;
dev->intr_handle.uio_cfg_fd = -1;
+   dev->intr_handle.uevent_fd = -1;
dev->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN;
 
/* secondary processes - use already recorded details */
@@ -227,7 +228,10 @@
close(dev->intr_handle.uio_cfg_fd);
dev->intr_handle.uio_cfg_fd = -1;
}
-
+   if (dev->intr_handle.uevent_fd >= 0) {
+   close(dev->intr_handle.uevent_fd);
+   dev->intr_handle.uevent_fd = -1;
+   }
dev->intr_handle.fd = -1;
dev->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN;
 }
diff --git a/lib/librte_eal/linuxapp/eal/eal_interrupts.c 
b/lib/librte_eal/linuxapp/eal/eal_interrupts.c
index 2e3bd12..d596522 100644
--- a/lib/librte_eal/linuxapp/eal/eal_interrupts.c
+++ b/lib/librte_eal/linuxapp/eal/eal_interrupts.c
@@ -65,6 +65,10 @@
 #include 
 #include 
 
+#include 
+#include 
+#include 
+
 #include "eal_private.h"
 #include "eal_vfio.h"
 #include "eal_thread.h"
@@ -74,6 +78,9 @@
 
 static RTE_DEFINE_PER_LCORE(int, _epfd) = -1; /**< epoll fd per thread */
 
+#define RTE_UEVENT_MSG_LEN 4096
+#define RTE_UEVENT_SUBSYSTEM_UIO 1
+
 /**
  * union for pipe fds.
  */
@@ -669,10 +676,13 @@ struct rte_intr_source {
RTE_SET_USED(r);
return -1;
}
+
rte_spinlock_lock(&intr_lock);
TAILQ_FOREACH(src, &intr_sources, next)
-   if (src->intr_handle.fd ==
-   events[n].data.fd)
+   if ((src->intr_handle.fd ==
+   events[n].data.fd) ||
+   (src->intr_handle.uevent_fd ==
+   events[n].data.fd))
break;
if (src == NULL){
rte_spinlock_unlock(&intr_lock);
@@ -858,7 +868,24 @@ static __attribute__((noreturn)) void *
}
else
numfds++;
+
+   /**
+* add device uevent file descriptor
+* into wait list for uevent monitoring.
+*/
+   ev.events = EPOLLIN | EPOLLPRI | EPOLLRDHUP | EPOLLHUP;
+   ev.data.fd = src->intr_handle.uevent_fd;
+   if (epoll_ctl(pfd, EPOLL_CTL_ADD,
+   src->intr_handle.uevent_fd, &ev) < 0){
+   rte_panic("Error adding uevent_fd %d epoll_ctl"
+   ", %s\n",
+   src->intr_handle.uevent_fd,
+   strerror(errno));
+   } else
+   numfds++;
}
+
+
rte_spinlock_unlock(&intr_lock);
/* serve the interrupt */
eal_intr_handle_interrupts(pfd, numfds);
@@ -1255,3 +1282,115 @@ static __attribute__((noreturn)) void *
 
return 0;
 }
+
+int
+rte_uevent_connect(void)
+{
+   struct sockaddr_nl addr;
+   int ret;
+   int netlink_fd = -1;
+   int size = 64 * 1024;
+   int nonblock = 1;
+   memset(&addr, 0, sizeof(addr));
+   addr.nl_family = AF_NETLINK;
+   addr.nl_pid = 0;
+   addr.nl_groups = 0x;
+
+   netlink_fd = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_KOBJECT_UEVENT);
+   if (netlink_fd < 0)
+   return -1;
+
+   RTE_LOG(ERR, EAL,
+   "netlink_fd is %d\n", netlink_fd);
+
+   setsockopt(netlink_fd, SOL_SOCKET, SO_RCVBUFFORCE, &size, sizeof(size));
+
+   ret = ioctl(netlink_fd, FIONBIO, &nonblock);
+  

Re: [dpdk-dev] [PATCH V2 0/6] some local vmxnet3 patches

[Ferruh Yigit]

On 6/15/2017 1:16 PM, Charles (Chas) Williams wrote:
> This series addresses some local issues with the vmxnet3 driver that
> others might find of interest.
> 
> Changes in v2:
> 
> - net/vmxnet3: Implement retrieval of extended stats
> 
> .id field wasn't being filled in xstats.
> 
> - net/vmxnet3: receive queue memory leak
> 
> The leak and buffer reallcation were split.  The buffer reallocation
> patch may be superceded by another commit so isn't included in this
> series while under discussion.
> 

Series applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Thomas Monjalon]

27/06/2017 21:03, Jan Blunck:
> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet  wrote:
> > --- a/lib/librte_eal/common/include/rte_bus.h
> > +++ b/lib/librte_eal/common/include/rte_bus.h
> >  /**
> > + * Implementation specific probe function which is responsible for linking
> > + * devices on that bus with applicable drivers.
> > + * The plugged device might already have been used previously by the bus,
> > + * in which case some buses might prefer to detect and re-use the relevant
> > + * information pertaining to this device.
> > + *
> > + * @param da
> > + * Device declaration.
> > + *
> > + * @return
> > + * The pointer to a valid rte_device usable by the bus on success.
> > + * NULL on error. rte_errno is then set.
> > + */
> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs *da);
> 
> Shouldn't this be orthogonal to unplug() and take a rte_device. You
> should only be able to plug devices that have been found by scan
> before.

Plugging a device that has been scanned before is a special case.
In a true hotplug scenario, we could use this plug function passing
a devargs.
I don't see any issue passing rte_devargs to plug and rte_device to unplug.

Re: [dpdk-dev] [PATCH] vfio: allow to map other memory regions

[Thomas Monjalon]

28/06/2017 11:54, Wodkowski, PawelX:
> > -Original Message-
> > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > Sent: Monday, June 19, 2017 11:04 PM
> > To: Wodkowski, PawelX 
> > Cc: dev@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH] vfio: allow to map other memory regions
> > 
> > Hi,
> > Some comments below
> > 
> > 24/05/2017 13:17, Pawel Wodkowski:
> > > Currently it is not possible to use memory that is not owned by DPDK to
> > > perform DMA. This scenarion might be used in vhost applications (like
> > > SPDK) where guest send its own memory table. To fill this gap provide
> > > API to allow registering arbitrary address in VFIO container.
> > >
> > > Signed-off-by: Pawel Wodkowski 
> > > ---
> > >  lib/librte_eal/linuxapp/eal/Makefile|   3 +
> > >  lib/librte_eal/linuxapp/eal/eal_vfio.c  | 142
> > +---
> > >  lib/librte_eal/linuxapp/eal/eal_vfio.h  |  10 ++
> > >  lib/librte_eal/linuxapp/eal/include/rte_iommu.h |  78 +
> > >  lib/librte_eal/linuxapp/eal/rte_eal_version.map |   8 ++
> > >  5 files changed, 224 insertions(+), 17 deletions(-)
> > >  create mode 100644 lib/librte_eal/linuxapp/eal/include/rte_iommu.h
> > 
> > VFIO is not referenced in the doxygen of these functions.
> > Could we use this API for something else than VFIO?
> 
> This is for any IOMMU hw/module/driver used in host which require special
> care about memory regions used for DMA. It is not restricted to VFIO even 
> though
> only VFIO is implemented.
> 
> > 
> > Any API should be declared in common directory, even if it is not
> > implemented for FreeBSD (returning -ENOTSUP).
> 
> I think those function should be NOP for FreeBSD (like RTE_VFIO_NOIOMMU do)
> or be conditionally compiled/included (like it is now). I decide to take 
> second way.
> Do you think that I should move rte_iommu.h to common directory and use #ifdef
> there?

No #ifdef please.
You must define new functions in a common header and implement it
for Linux and BSD. The BSD function can be return an error.

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Jan Blunck]

On Wed, Jun 28, 2017 at 1:44 PM, Thomas Monjalon  wrote:
> 27/06/2017 21:03, Jan Blunck:
>> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet  wrote:
>> > --- a/lib/librte_eal/common/include/rte_bus.h
>> > +++ b/lib/librte_eal/common/include/rte_bus.h
>> >  /**
>> > + * Implementation specific probe function which is responsible for linking
>> > + * devices on that bus with applicable drivers.
>> > + * The plugged device might already have been used previously by the bus,
>> > + * in which case some buses might prefer to detect and re-use the relevant
>> > + * information pertaining to this device.
>> > + *
>> > + * @param da
>> > + * Device declaration.
>> > + *
>> > + * @return
>> > + * The pointer to a valid rte_device usable by the bus on success.
>> > + * NULL on error. rte_errno is then set.
>> > + */
>> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs *da);
>>
>> Shouldn't this be orthogonal to unplug() and take a rte_device. You
>> should only be able to plug devices that have been found by scan
>> before.
>
> Plugging a device that has been scanned before is a special case.
> In a true hotplug scenario, we could use this plug function passing
> a devargs.
> I don't see any issue passing rte_devargs to plug and rte_device to unplug.

What do you mean by "true hotplug"?

The problem with this is that passing just rte_devargs to plug()
requires the bus to parse and enrich the rte_devargs with bus
specifics. From there it gets folded into the to-be-created bus
specific rte_XXX_device. This makes it unnecessarily complicated and
even worse it adds a second code path how devices come alive.

When we get notified about a hotplug event we already know which bus
this event belongs to:

1. scan the bus for incoming devices
2. plug single device with devargs and probe for drivers

Makes sense?

Re: [dpdk-dev] [PATCH 1/2] bond: avoid allocating mempool on socket LCORE_ID_ANY

[Jan Blunck]

On Wed, Jun 28, 2017 at 11:47 AM, Declan Doherty
 wrote:
> On 26/04/17 22:30, Jan Blunck wrote:
>>
>> If the numa node is unknown, use the value from rte_socket_id() to avoid
>> an allocation failure.
>>
>> Signed-off-by: Eric Kinzie 
>> CC: Eric Kinzie 
>> ---
>>   drivers/net/bonding/rte_eth_bond_8023ad.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/net/bonding/rte_eth_bond_8023ad.c
>> b/drivers/net/bonding/rte_eth_bond_8023ad.c
>> index 7b863d6..b8df536 100644
>> --- a/drivers/net/bonding/rte_eth_bond_8023ad.c
>> +++ b/drivers/net/bonding/rte_eth_bond_8023ad.c
>> @@ -887,6 +887,8 @@ bond_mode_8023ad_activate_slave(struct rte_eth_dev
>> *bond_dev, uint8_t slave_id)
>> RTE_ASSERT(port->rx_ring == NULL);
>> RTE_ASSERT(port->tx_ring == NULL);
>> socket_id = rte_eth_devices[slave_id].data->numa_node;
>> +   if (socket_id == (uint8_t)LCORE_ID_ANY)
>
>
> socket_id is an integer so we shouldn't be casting LCORE_ID_ANY to an
> uint8_t as this condition would never match as LCORE_ID_ANY wouldn't be
> UINT32_MAX after casting
>

Ouch ... let me fix this.

>> +   socket_id = 0;
>
>
> Should this not be socket_id = rte_socket_id(); as we don't know we are
> running on
>
>
>> element_size = sizeof(struct slow_protocol_frame) +
>> RTE_PKTMBUF_HEADROOM;
>>
>

Re: [dpdk-dev] [PATCH 0/2] cxgbe: fix supported speed capabilities

[Ferruh Yigit]

On 6/28/2017 5:37 AM, Rahul Lakkireddy wrote:
> Patch 1 updates supported port module types.
> Patch 2 uses port type to determine the supported speed capabilities.
> 
> Thanks,
> Rahul
> 
> Rahul Lakkireddy (2):
>   cxgbe: update supported port module types
>   cxgbe: fix supported speed capabilities

Series applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Thomas Monjalon]

28/06/2017 13:58, Jan Blunck:
> On Wed, Jun 28, 2017 at 1:44 PM, Thomas Monjalon  wrote:
> > 27/06/2017 21:03, Jan Blunck:
> >> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet  
> >> wrote:
> >> > --- a/lib/librte_eal/common/include/rte_bus.h
> >> > +++ b/lib/librte_eal/common/include/rte_bus.h
> >> >  /**
> >> > + * Implementation specific probe function which is responsible for 
> >> > linking
> >> > + * devices on that bus with applicable drivers.
> >> > + * The plugged device might already have been used previously by the 
> >> > bus,
> >> > + * in which case some buses might prefer to detect and re-use the 
> >> > relevant
> >> > + * information pertaining to this device.
> >> > + *
> >> > + * @param da
> >> > + * Device declaration.
> >> > + *
> >> > + * @return
> >> > + * The pointer to a valid rte_device usable by the bus on success.
> >> > + * NULL on error. rte_errno is then set.
> >> > + */
> >> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs *da);
> >>
> >> Shouldn't this be orthogonal to unplug() and take a rte_device. You
> >> should only be able to plug devices that have been found by scan
> >> before.
> >
> > Plugging a device that has been scanned before is a special case.
> > In a true hotplug scenario, we could use this plug function passing
> > a devargs.
> > I don't see any issue passing rte_devargs to plug and rte_device to unplug.
> 
> What do you mean by "true hotplug"?

I mean a kernel notification of a new device.

> The problem with this is that passing just rte_devargs to plug()
> requires the bus to parse and enrich the rte_devargs with bus
> specifics. From there it gets folded into the to-be-created bus
> specific rte_XXX_device. This makes it unnecessarily complicated and
> even worse it adds a second code path how devices come alive.

Just after the notification, the rte_device does not exist yet.
So the plug function could share the same code as the scan function
to get the metadata and create the device instance.

> When we get notified about a hotplug event we already know which bus
> this event belongs to:
> 
> 1. scan the bus for incoming devices

No need to scan every devices here.

> 2. plug single device with devargs and probe for drivers
> 
> Makes sense?

I want to make sure there is no misunderstanding first :)

[dpdk-dev] [PATCH] mbuf: fix debug checks for headroom and tailroom

[Vasily Philipov]

rte_pktmbuf_headroom() and rte_pktmbuf_tailroom() should be usable
with any segment, not only with headered ones, so is_header should be 0
when we call for sanity check inside them.

Fixes: af75078fece3 ("first public release")
Cc: sta...@dpdk.org

Signed-off-by: Vasily Philipov 
---
 lib/librte_mbuf/rte_mbuf.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index fe605c7..fbe8203 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -1452,7 +1452,7 @@ static inline void rte_pktmbuf_refcnt_update(struct 
rte_mbuf *m, int16_t v)
  */
 static inline uint16_t rte_pktmbuf_headroom(const struct rte_mbuf *m)
 {
-   __rte_mbuf_sanity_check(m, 1);
+   __rte_mbuf_sanity_check(m, 0);
return m->data_off;
 }
 
@@ -1466,7 +1466,7 @@ static inline uint16_t rte_pktmbuf_headroom(const struct 
rte_mbuf *m)
  */
 static inline uint16_t rte_pktmbuf_tailroom(const struct rte_mbuf *m)
 {
-   __rte_mbuf_sanity_check(m, 1);
+   __rte_mbuf_sanity_check(m, 0);
return (uint16_t)(m->buf_len - rte_pktmbuf_headroom(m) -
  m->data_len);
 }
-- 
1.8.3.1

Re: [dpdk-dev] [PATCH v2 0/3] net/ark: augment user extension and bug fix

[Ferruh Yigit]

On 6/28/2017 11:08 AM, John Miller wrote:
> v2:
> * Split the user extension private data per port patch and the
>   set_mtu patch into separate patches.
> * Add better description to per port private data patch
> * stats_reset patch untouched from V1
> 
> John Miller (3):
>   net/ark: allow unique user data for each port in extension calls
>   net/ark: add set_mtu call to user extension API
>   net/ark: fix bug in stats_reset operation

Series applied to dpdk-next-net/master, thanks.

(some patch titles updated)

[dpdk-dev] [PATCH] vhost: fix checking of device features

[Ivan Dyukov]

To compare enabled features in current device we must use bit
mask instead of bit position.

CC: sta...@dpdk.org
Fixes: c843af3aa13e ("vhost: access header only")

Signed-off-by: Ivan Dyukov 
---
 lib/librte_vhost/virtio_net.c | 8 +---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index ebfda1c..4fae4c1 100644
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -601,9 +601,11 @@ static inline bool
 virtio_net_with_host_offload(struct virtio_net *dev)
 {
if (dev->features &
-   (VIRTIO_NET_F_CSUM | VIRTIO_NET_F_HOST_ECN |
-VIRTIO_NET_F_HOST_TSO4 | VIRTIO_NET_F_HOST_TSO6 |
-VIRTIO_NET_F_HOST_UFO))
+   ((1ULL << VIRTIO_NET_F_CSUM) |
+(1ULL << VIRTIO_NET_F_HOST_ECN) |
+(1ULL << VIRTIO_NET_F_HOST_TSO4) |
+(1ULL << VIRTIO_NET_F_HOST_TSO6) |
+(1ULL << VIRTIO_NET_F_HOST_UFO)))
return true;
 
return false;
-- 
2.7.4

Re: [dpdk-dev] 17.05.1 patches review and test

[Kavanagh, Mark B]

>From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Yuanhan Liu
>Sent: Thursday, June 22, 2017 2:08 PM
>To: dpdk stable 
>Cc: dev@dpdk.org
>Subject: [dpdk-dev] 17.05.1 patches review and test
>
>Hi all,
>
>I'm doing an unexpected stable release, which is expected to be released
>in about two months. The reason I'm doing it is there is an import bug
>fix commit that is needed for OVS 2.8 release. The fix is a vhost-user
>commit comes from me. Well, unfortunately, I'm also the author of the
>original bug :/
>
>So I'm doing an urgent release. That also means we will have another
>17.05 release (17.05.2) shortly after v17.08 is out.
>
>And here is a list of patches targeted for this release. Please help
>review and test. The planned date for the final release is 30, June.
>Before that, please shout if anyone has objections.
>
>These patches are located at branch 17.05 of dpdk-stable repo:
>http://dpdk.org/browse/dpdk-stable/
>

Hi Yuanhan,

Thanks for making this release available ahead of schedule - it puts OvS in a 
good position to upgrade the supported version of DPDK in OvS v2.8.0 (due 
~mid-August).

I can confirm that the OvS team has performed selective validation on the 
following DPDK-dependent OvS features:
- Jumbo frames
- vhost NUMA (broken in 17.05 release)
- Rate Limiting
- QoS
- VxLAN tunneling
- Rx checksum offload
- Flow control
- Addition/deletion of vdevs

As such, we're happy to proceed with the 17.05.1 release.

Thanks once again,
Mark

>Thanks.
>
>--yliu
>
>---
>Ajit Khaparde (1):
>  net/bnxt: fix reporting of link status
>
>Alejandro Lucero (1):
>  vfio: fix array bounds check
>
>Andrew Rybchenko (1):
>  net/sfc: add Tx queue flush failed flag for sanity
>
>Andy Moreton (2):
>  net/sfc/base: fix error code usage in common code
>  net/sfc/base: let caller know that queue is already flushed
>
>Beilei Xing (1):
>  app/testpmd: fix creating E-Tag and NVGRE flow rules
>
>Chas Williams (3):
>  net/af_packet: handle possible null pointer
>  net/af_packet: fix packet bytes counting
>  net/ring: fix adding MAC addresses
>
>Dariusz Stojaczyk (2):
>  vhost: fix malloc size too small
>  vhost: fix guest pages memory leak
>
>David Marchand (1):
>  drivers/net: fix vfio kmod dependency
>
>Ferruh Yigit (3):
>  kni: fix build with gcc 7.1
>  net/enic: fix build with gcc 7.1
>  net/mlx5: fix build with gcc 7.1
>
>Harish Patil (1):
>  net/qede: fix VXLAN tunnel Tx offload flag setting
>
>Jerin Jacob (1):
>  examples/vhost: fix uninitialized descriptor indexes
>
>John Miller (4):
>  net/ark: fix buffer not null terminated
>  net/ark: fix return code not checked
>  net/ark: fix null pointer dereference
>  net/ark: fix return value of null not checked
>
>Lee Roberts (1):
>  kni: fix build on RHEL 7.4
>
>Markus Theil (1):
>  net/igb: fix add/delete of flex filters
>
>Nélio Laranjeiro (1):
>  net/mlx5: fix flow application order on stop/start
>
>Pascal Mazon (1):
>  net/tap: fix some flow collision
>
>Qi Zhang (3):
>  net/ixgbe: fix fdir mask not be reset
>  net/i40e: exclude internal packet's byte count
>  net/i40e: fix VF statistics
>
>Rahul Lakkireddy (2):
>  net/cxgbe: fix port statistics
>  net/cxgbe: fix rxq default params for ports under same PF
>
>Shahaf Shuler (1):
>  net/mlx5: fix completion buffer size
>
>Shijith Thotton (1):
>  net/liquidio: fix MTU calculation from port configuration
>
>Tiwei Bie (1):
>  net/virtio: zero the whole memory zone
>
>Wei Dai (1):
>  lpm: fix index of tbl8
>
>Wei Zhao (1):
>  net/igb: fix checksum valid flags
>
>Wenzhuo Lu (1):
>  net/i40e/base: fix Tx error stats on VF
>
>Yongseok Koh (2):
>  net/mlx5: fix exception handling
>  net/mlx5: fix redundant free of Tx buffer
>
>Yuanhan Liu (1):
>  vhost: fix crash on NUMA

Re: [dpdk-dev] [PATCH V2 0/6] some local vmxnet3 patches

[Ferruh Yigit]

On 6/28/2017 12:30 PM, Ferruh Yigit wrote:
> On 6/15/2017 1:16 PM, Charles (Chas) Williams wrote:
>> This series addresses some local issues with the vmxnet3 driver that
>> others might find of interest.
>>
>> Changes in v2:
>>
>> - net/vmxnet3: Implement retrieval of extended stats
>>
>> .id field wasn't being filled in xstats.
>>
>> - net/vmxnet3: receive queue memory leak
>>
>> The leak and buffer reallcation were split.  The buffer reallocation
>> patch may be superceded by another commit so isn't included in this
>> series while under discussion.
>>
> 
> Series applied to dpdk-next-net/master, thanks.

Hi Charles,

Patches require vmxnet3.ini update, "Extended stats" and "Link status
event" ones I guess. I missed them to before getting patchset.

Would you mind sending another patch to update features document?

Thanks,
ferruh

Re: [dpdk-dev] [PATCH] vhost: fix checking of device features

[Maxime Coquelin]

On 06/28/2017 02:40 PM, Ivan Dyukov wrote:

To compare enabled features in current device we must use bit
mask instead of bit position.

CC: sta...@dpdk.org
Fixes: c843af3aa13e ("vhost: access header only")

Signed-off-by: Ivan Dyukov 
---
  lib/librte_vhost/virtio_net.c | 8 +---
  1 file changed, 5 insertions(+), 3 deletions(-)


Thanks for the fix Ivan, and sorry for introducing this bug.
Out of curiosity, did you noticed it because it broke offloading,
or just by code review?

Reviewed-by: Maxime Coquelin 


diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index ebfda1c..4fae4c1 100644
--- a/lib/librte_vhost/virtio_net.c
+++ b/lib/librte_vhost/virtio_net.c
@@ -601,9 +601,11 @@ static inline bool
  virtio_net_with_host_offload(struct virtio_net *dev)
  {
if (dev->features &
-   (VIRTIO_NET_F_CSUM | VIRTIO_NET_F_HOST_ECN |
-VIRTIO_NET_F_HOST_TSO4 | VIRTIO_NET_F_HOST_TSO6 |
-VIRTIO_NET_F_HOST_UFO))
+   ((1ULL << VIRTIO_NET_F_CSUM) |
+(1ULL << VIRTIO_NET_F_HOST_ECN) |
+(1ULL << VIRTIO_NET_F_HOST_TSO4) |
+(1ULL << VIRTIO_NET_F_HOST_TSO6) |
+(1ULL << VIRTIO_NET_F_HOST_UFO)))
return true;
  
  	return false;

Re: [dpdk-dev] [PATCH] net/ixgbe: link check enhancement on VF

[Ferruh Yigit]

On 6/28/2017 3:22 AM, Wenzhuo Lu wrote:
> From: Liang-min Wang 
> 
> In current implementation, when checking VF link state, PF state
> is checked too, although the function has a parameter to tell
> if PF state checking is needed.
> But in some scenario, user may not care about the PF state.
> This patch enables the unused parameter to only check the VF
> link state.
> 
> Signed-off-by: Liang-min Wang 
> Acked-by: Wenzhuo Lu 

Applied to dpdk-next-net/master, thanks.

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Jan Blunck]

On Wed, Jun 28, 2017 at 2:11 PM, Thomas Monjalon  wrote:
> 28/06/2017 13:58, Jan Blunck:
>> On Wed, Jun 28, 2017 at 1:44 PM, Thomas Monjalon  wrote:
>> > 27/06/2017 21:03, Jan Blunck:
>> >> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet  
>> >> wrote:
>> >> > --- a/lib/librte_eal/common/include/rte_bus.h
>> >> > +++ b/lib/librte_eal/common/include/rte_bus.h
>> >> >  /**
>> >> > + * Implementation specific probe function which is responsible for 
>> >> > linking
>> >> > + * devices on that bus with applicable drivers.
>> >> > + * The plugged device might already have been used previously by the 
>> >> > bus,
>> >> > + * in which case some buses might prefer to detect and re-use the 
>> >> > relevant
>> >> > + * information pertaining to this device.
>> >> > + *
>> >> > + * @param da
>> >> > + * Device declaration.
>> >> > + *
>> >> > + * @return
>> >> > + * The pointer to a valid rte_device usable by the bus on success.
>> >> > + * NULL on error. rte_errno is then set.
>> >> > + */
>> >> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs *da);
>> >>
>> >> Shouldn't this be orthogonal to unplug() and take a rte_device. You
>> >> should only be able to plug devices that have been found by scan
>> >> before.
>> >
>> > Plugging a device that has been scanned before is a special case.
>> > In a true hotplug scenario, we could use this plug function passing
>> > a devargs.
>> > I don't see any issue passing rte_devargs to plug and rte_device to unplug.
>>
>> What do you mean by "true hotplug"?
>
> I mean a kernel notification of a new device.
>

Does a "false hotplug" exist, too? :)

>> The problem with this is that passing just rte_devargs to plug()
>> requires the bus to parse and enrich the rte_devargs with bus
>> specifics. From there it gets folded into the to-be-created bus
>> specific rte_XXX_device. This makes it unnecessarily complicated and
>> even worse it adds a second code path how devices come alive.
>
> Just after the notification, the rte_device does not exist yet.
> So the plug function could share the same code as the scan function
> to get the metadata and create the device instance.
>

Exactly this is what I want to avoid. The plug() function would become
a "scan-one and probe". From my point of view plug() and unplug()
should be orthogonal. The plug() and unplug() should only be
responsible for adding drivers with optional arguments. The EAL should
allow the drivers to get unplugged/re-plugged at run-time. I want to
be able to change arguments ... or even drivers :)

>> When we get notified about a hotplug event we already know which bus
>> this event belongs to:
>>
>> 1. scan the bus for incoming devices
>
> No need to scan every devices here.
>

This is a readdir followed by open+read+close for any new device. This
code belongs here anyway. Its lightweight if nothing changed. The scan
itself should be idempotent anyway.

>> 2. plug single device with devargs and probe for drivers
>>
>> Makes sense?
>
> I want to make sure there is no misunderstanding first :)

Which makes sense. That is probably my fault due to being too
distracted with other things and not communicating well enough while
Gaetan consumed my code.

Re: [dpdk-dev] [PATCH V2 0/6] some local vmxnet3 patches

[Charles (Chas) Williams]

On 06/28/2017 08:52 AM, Ferruh Yigit wrote:

On 6/28/2017 12:30 PM, Ferruh Yigit wrote:

On 6/15/2017 1:16 PM, Charles (Chas) Williams wrote:

This series addresses some local issues with the vmxnet3 driver that
others might find of interest.

Changes in v2:

- net/vmxnet3: Implement retrieval of extended stats

.id field wasn't being filled in xstats.

- net/vmxnet3: receive queue memory leak

The leak and buffer reallcation were split.  The buffer reallocation
patch may be superceded by another commit so isn't included in this
series while under discussion.



Series applied to dpdk-next-net/master, thanks.


Hi Charles,

Patches require vmxnet3.ini update, "Extended stats" and "Link status
event" ones I guess. I missed them to before getting patchset.

Would you mind sending another patch to update features document?

Thanks,
ferruh



I will get to it today.

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Thomas Monjalon]

28/06/2017 15:09, Jan Blunck:
> On Wed, Jun 28, 2017 at 2:11 PM, Thomas Monjalon  wrote:
> > 28/06/2017 13:58, Jan Blunck:
> >> On Wed, Jun 28, 2017 at 1:44 PM, Thomas Monjalon  
> >> wrote:
> >> > 27/06/2017 21:03, Jan Blunck:
> >> >> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet  
> >> >> wrote:
> >> >> > --- a/lib/librte_eal/common/include/rte_bus.h
> >> >> > +++ b/lib/librte_eal/common/include/rte_bus.h
> >> >> >  /**
> >> >> > + * Implementation specific probe function which is responsible for 
> >> >> > linking
> >> >> > + * devices on that bus with applicable drivers.
> >> >> > + * The plugged device might already have been used previously by the 
> >> >> > bus,
> >> >> > + * in which case some buses might prefer to detect and re-use the 
> >> >> > relevant
> >> >> > + * information pertaining to this device.
> >> >> > + *
> >> >> > + * @param da
> >> >> > + * Device declaration.
> >> >> > + *
> >> >> > + * @return
> >> >> > + * The pointer to a valid rte_device usable by the bus on 
> >> >> > success.
> >> >> > + * NULL on error. rte_errno is then set.
> >> >> > + */
> >> >> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs 
> >> >> > *da);
> >> >>
> >> >> Shouldn't this be orthogonal to unplug() and take a rte_device. You
> >> >> should only be able to plug devices that have been found by scan
> >> >> before.
> >> >
> >> > Plugging a device that has been scanned before is a special case.
> >> > In a true hotplug scenario, we could use this plug function passing
> >> > a devargs.
> >> > I don't see any issue passing rte_devargs to plug and rte_device to 
> >> > unplug.
> >>
> >> What do you mean by "true hotplug"?
> >
> > I mean a kernel notification of a new device.
> 
> Does a "false hotplug" exist, too? :)

The false hotplug was the original attach function which was just
adding a new ethdev interface.

> >> The problem with this is that passing just rte_devargs to plug()
> >> requires the bus to parse and enrich the rte_devargs with bus
> >> specifics. From there it gets folded into the to-be-created bus
> >> specific rte_XXX_device. This makes it unnecessarily complicated and
> >> even worse it adds a second code path how devices come alive.
> >
> > Just after the notification, the rte_device does not exist yet.
> > So the plug function could share the same code as the scan function
> > to get the metadata and create the device instance.
> 
> Exactly this is what I want to avoid. 

Why do you want to avoid that?
I think you mean it is not what you had in mind.

> The plug() function would become a "scan-one and probe".

Yes

> From my point of view plug() and unplug() should be orthogonal.
> The plug() and unplug() should only be responsible for adding drivers
> with optional arguments. The EAL should allow the drivers to get
> unplugged/re-plugged at run-time. I want to be able to change arguments
> ... or even drivers :)

It is a totally different thing.
We are talking about hotplug of a device,
and you are talking about changing drivers dynamically.

So I still don't understand what is the issue with the plug/unplug
functions proposed here.

> >> When we get notified about a hotplug event we already know which bus
> >> this event belongs to:
> >>
> >> 1. scan the bus for incoming devices
> >
> > No need to scan every devices here.
> 
> This is a readdir followed by open+read+close for any new device. This
> code belongs here anyway. Its lightweight if nothing changed. The scan
> itself should be idempotent anyway.
> 
> >> 2. plug single device with devargs and probe for drivers
> >>
> >> Makes sense?
> >
> > I want to make sure there is no misunderstanding first :)
> 
> Which makes sense. That is probably my fault due to being too
> distracted with other things and not communicating well enough while
> Gaetan consumed my code.

Your ideas are probably interesting, and I want to understand them.
In the meantime, we need to progress on 17.08-rc1 which must be done
in following days. Please let's separate the ideas which are not
yet implemented from what we are already able to deliver.

Re: [dpdk-dev] [PATCH 00/20 v2] next-eventdev: NXP DPAA2 eventdev PMD

[Nipun Gupta]

Hi Jerin,

Is there anything pending for this patch series to be merged?
Do you want me to rebase it on the dpdk main repository as dpdk-next-eventdev 
branch is not currently in sync with main repo?

Regards,
Nipun

> -Original Message-
> From: Nipun Gupta [mailto:nipun.gu...@nxp.com]
> Sent: Thursday, June 15, 2017 14:29
> To: dev@dpdk.org
> Cc: Hemant Agrawal ;
> jerin.ja...@caviumnetworks.com; harry.van.haa...@intel.com;
> bruce.richard...@intel.com; gage.e...@intel.com; Shreyansh Jain
> ; Nipun Gupta 
> Subject: [PATCH 00/20 v2] next-eventdev: NXP DPAA2 eventdev PMD
> 
> The following patch set adds NXP DPAA2 HW based eventdev
> implementation to the next-eventdev tree.
> 
> The patchset consists of the following:
> - DPCON support required by DPAA2 eventdev
> - DPCI support in fslmc bus
> - Interrupt support in QBMAN
> - Implemetation of eventdev API's for DPAA2
> - Documentation of DPAA2 eventdev PMD
> 
> Changes in v2:
>   Rebase to latest dpdk-next-eventdev
>   Update to support only atomic and ordered queues
>   Fix checkpatch errors
> 
> Hemant Agrawal (2):
>   bus/fslmc: generic framework for mc object creation
>   bus/fslmc: export qbman dqrr funcs for eventdev usages
> 
> Nipun Gupta (18):
>   event/dpaa2: add basic build infrastructure
>   bus/fslmc: integrating dpio and dpbp to object framework
>   bus/fslmc: adding basic dpcon support
>   event/dpaa2: register dpcon as dpaa2 device for bus scan
>   bus/fslmc: adding basic dpci support
>   bus/fslmc: register dpci as dpaa2 device for bus scan
>   bus/fslmc: adding cpu support in stashing config
>   event/dpaa2: add initialization of event device
>   bus/fslmc: add support for static dequeue from portal
>   event/dpaa2: add configuration functions
>   bus/fslmc: support enqueue with multiple enqueue descriptors
>   bus/fslmc: add callback per queue to enable
>   bus/fslmc: change func argument to const to avoid warning
>   event/dpaa2: add enqueue and dequeue functionality
>   fslmc/bus: add interrupt enabling routine
>   bus/fslmc: enable portal interrupt handling
>   event/dpaa2: handle timeout using interrupts in dequeue
>   doc: add NXP DPAA2 EVENTDEV details
> 
>  MAINTAINERS|   5 +
>  config/defconfig_arm64-dpaa2-linuxapp-gcc  |   6 +
>  doc/guides/eventdevs/dpaa2.rst | 157 +
>  drivers/bus/fslmc/Makefile |   5 +-
>  drivers/bus/fslmc/fslmc_vfio.c |  69 ++-
>  drivers/bus/fslmc/fslmc_vfio.h |  47 +-
>  drivers/bus/fslmc/mc/dpci.c| 307 +
>  drivers/bus/fslmc/mc/dpcon.c   | 230 +++
>  drivers/bus/fslmc/mc/dpio.c|  44 ++
>  drivers/bus/fslmc/mc/fsl_dpci.h| 404 
>  drivers/bus/fslmc/mc/fsl_dpci_cmd.h| 147 +
>  drivers/bus/fslmc/mc/fsl_dpcon.h   | 238 +++
>  drivers/bus/fslmc/mc/fsl_dpcon_cmd.h   | 175 ++
>  drivers/bus/fslmc/mc/fsl_dpio.h|  30 +
>  drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c   |  14 +-
>  drivers/bus/fslmc/portal/dpaa2_hw_dpci.c   | 179 ++
>  drivers/bus/fslmc/portal/dpaa2_hw_dpio.c   | 133 +++-
>  drivers/bus/fslmc/portal/dpaa2_hw_dpio.h   |   5 +
>  drivers/bus/fslmc/portal/dpaa2_hw_pvt.h|  28 +-
>  drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h |  46 +-
>  drivers/bus/fslmc/qbman/qbman_portal.c |  94 ++-
>  drivers/bus/fslmc/rte_bus_fslmc_version.map|  23 +
>  drivers/event/Makefile |   2 +
>  drivers/event/dpaa2/Makefile   |  64 ++
>  drivers/event/dpaa2/dpaa2_eventdev.c   | 686 
> +
>  drivers/event/dpaa2/dpaa2_eventdev.h   | 113 
>  drivers/event/dpaa2/dpaa2_hw_dpcon.c   | 139 +
>  .../event/dpaa2/rte_pmd_dpaa2_event_version.map|   3 +
>  mk/rte.app.mk  |   1 +
>  29 files changed, 3353 insertions(+), 41 deletions(-)
>  create mode 100644 doc/guides/eventdevs/dpaa2.rst
>  create mode 100644 drivers/bus/fslmc/mc/dpci.c
>  create mode 100644 drivers/bus/fslmc/mc/dpcon.c
>  create mode 100644 drivers/bus/fslmc/mc/fsl_dpci.h
>  create mode 100644 drivers/bus/fslmc/mc/fsl_dpci_cmd.h
>  create mode 100644 drivers/bus/fslmc/mc/fsl_dpcon.h
>  create mode 100644 drivers/bus/fslmc/mc/fsl_dpcon_cmd.h
>  create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_dpci.c
>  create mode 100644 drivers/event/dpaa2/Makefile
>  create mode 100644 drivers/event/dpaa2/dpaa2_eventdev.c
>  create mode 100644 drivers/event/dpaa2/dpaa2_eventdev.h
>  create mode 100644 drivers/event/dpaa2/dpaa2_hw_dpcon.c
>  create mode 100644 drivers/event/dpaa2/rte_pmd_dpaa2_event_version.map
> 
> --
> 1.9.1

[dpdk-dev] [PATCH v5 4/4] app/testpmd: add isolated mode parameter

[Vasily Philipov]

In case if --isolated-mode parameter was added to the command line
the rte flow isolate action will be applied on each port before
the device configuring.

Signed-off-by: Vasily Philipov 
---
 app/test-pmd/parameters.c |  3 +++
 app/test-pmd/testpmd.c| 14 ++
 app/test-pmd/testpmd.h|  1 +
 3 files changed, 18 insertions(+)

diff --git a/app/test-pmd/parameters.c b/app/test-pmd/parameters.c
index fbe6284..e313871 100644
--- a/app/test-pmd/parameters.c
+++ b/app/test-pmd/parameters.c
@@ -623,6 +623,7 @@
{ "tx-queue-stats-mapping", 1, 0, 0 },
{ "rx-queue-stats-mapping", 1, 0, 0 },
{ "no-flush-rx",0, 0, 0 },
+   { "isolated-mode",  0, 0, 0 },
{ "txpkts", 1, 0, 0 },
{ "disable-link-check", 0, 0, 0 },
{ "no-lsc-interrupt",   0, 0, 0 },
@@ -1081,6 +1082,8 @@
lsc_interrupt = 0;
if (!strcmp(lgopts[opt_idx].name, "no-rmv-interrupt"))
rmv_interrupt = 0;
+   if (!strcmp(lgopts[opt_idx].name, "isolated-mode"))
+   isolated_mode = 1;
if (!strcmp(lgopts[opt_idx].name, "print-event"))
if (parse_event_printing_config(optarg, 1)) {
rte_exit(EXIT_FAILURE,
diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
index d1041af..610e675 100644
--- a/app/test-pmd/testpmd.c
+++ b/app/test-pmd/testpmd.c
@@ -267,6 +267,11 @@ struct fwd_engine * fwd_engines[] = {
 uint8_t no_flush_rx = 0; /* flush by default */
 
 /*
+ * Flow API isolated mode.
+ */
+uint8_t isolated_mode;
+
+/*
  * Avoids to check link status when starting/stopping a port.
  */
 uint8_t no_link_check = 0; /* check by default */
@@ -1422,6 +1427,15 @@ static void eth_event_callback(uint8_t port_id,
if (port->need_reconfig > 0) {
port->need_reconfig = 0;
 
+   if (isolated_mode) {
+   int ret = port_flow_isolate(pi, 1);
+   if (ret) {
+   printf("Failed to apply isolated"
+  " mode on port %d\n", pi);
+   return -1;
+   }
+   }
+
printf("Configuring Port %d (socket %u)\n", pi,
port->socket_id);
/* configure port */
diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
index e6c43ba..dbe9898 100644
--- a/app/test-pmd/testpmd.h
+++ b/app/test-pmd/testpmd.h
@@ -303,6 +303,7 @@ struct queue_stats_mappings {
 extern uint8_t  numa_support; /**< set by "--numa" parameter */
 extern uint16_t port_topology; /**< set by "--port-topology" parameter */
 extern uint8_t no_flush_rx; /**

[dpdk-dev] [PATCH v5 3/4] net/mlx4: support for the RSS flow action

[Vasily Philipov]

The isolated mode should be enabled.
The number of queues in RSS ring must be power of 2.
The sharing a queue between several RSS rings is impossible.

Signed-off-by: Vasily Philipov 
---
 drivers/net/mlx4/mlx4.c  |  21 +++--
 drivers/net/mlx4/mlx4.h  |   5 ++
 drivers/net/mlx4/mlx4_flow.c | 197 ++-
 drivers/net/mlx4/mlx4_flow.h |   3 +-
 4 files changed, 211 insertions(+), 15 deletions(-)

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 22fa7c6..6ab7241 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -554,9 +554,9 @@ void priv_unlock(struct priv *priv)
  *   The number of entries in queues[].
  *
  * @return
- *   0 on success, negative errno value on failure.
+ *   Pointer to a parent rxq structure, NULL on failure.
  */
-static int
+struct rxq *
 priv_create_parent(struct priv *priv,
   uint16_t queues[],
   uint16_t children_n)
@@ -568,13 +568,15 @@ void priv_unlock(struct priv *priv)
parent = rte_zmalloc("parent queue",
 sizeof(*parent),
 RTE_CACHE_LINE_SIZE);
-   if (!parent)
-   return -ENOMEM;
+   if (!parent) {
+   ERROR("cannot allocate memory for RSS parent queue");
+   return NULL;
+   }
ret = rxq_setup(priv->dev, parent, 0, 0, 0,
NULL, NULL, children_n, NULL);
if (ret) {
rte_free(parent);
-   return -ret;
+   return NULL;
}
parent->rss.queues_n = children_n;
if (queues) {
@@ -587,7 +589,7 @@ void priv_unlock(struct priv *priv)
parent->rss.queues[i] = i;
}
LIST_INSERT_HEAD(&priv->parents, parent, next);
-   return 0;
+   return parent;
 }
 
 /**
@@ -639,7 +641,6 @@ void priv_unlock(struct priv *priv)
unsigned int rxqs_n = dev->data->nb_rx_queues;
unsigned int txqs_n = dev->data->nb_tx_queues;
unsigned int tmp;
-   int ret;
 
priv->rxqs = (void *)dev->data->rx_queues;
priv->txqs = (void *)dev->data->tx_queues;
@@ -696,14 +697,12 @@ void priv_unlock(struct priv *priv)
priv->rxqs_n = rxqs_n;
if (priv->isolated)
return 0;
-   ret = priv_create_parent(priv, NULL, priv->rxqs_n);
-   if (!ret)
+   if (priv_create_parent(priv, NULL, priv->rxqs_n))
return 0;
/* Failure, rollback. */
priv->rss = 0;
priv->rxqs_n = tmp;
-   assert(ret > 0);
-   return ret;
+   return ENOMEM;
 }
 
 /**
diff --git a/drivers/net/mlx4/mlx4.h b/drivers/net/mlx4/mlx4.h
index b5fe1b4..f45e017 100644
--- a/drivers/net/mlx4/mlx4.h
+++ b/drivers/net/mlx4/mlx4.h
@@ -370,4 +370,9 @@ struct priv {
 void
 rxq_parent_cleanup(struct rxq *parent);
 
+struct rxq *
+priv_create_parent(struct priv *priv,
+  uint16_t queues[],
+  uint16_t children_n);
+
 #endif /* RTE_PMD_MLX4_H_ */
diff --git a/drivers/net/mlx4/mlx4_flow.c b/drivers/net/mlx4/mlx4_flow.c
index 3fd2716..9c0fba1 100644
--- a/drivers/net/mlx4/mlx4_flow.c
+++ b/drivers/net/mlx4/mlx4_flow.c
@@ -112,6 +112,7 @@ struct rte_flow_drop {
 static const enum rte_flow_action_type valid_actions[] = {
RTE_FLOW_ACTION_TYPE_DROP,
RTE_FLOW_ACTION_TYPE_QUEUE,
+   RTE_FLOW_ACTION_TYPE_RSS,
RTE_FLOW_ACTION_TYPE_END,
 };
 
@@ -672,6 +673,76 @@ struct rte_flow_drop {
if (!queue || (queue->index > (priv->rxqs_n - 1)))
goto exit_action_not_supported;
action.queue = 1;
+   action.queues_n = 1;
+   action.queues[0] = queue->index;
+   } else if (actions->type == RTE_FLOW_ACTION_TYPE_RSS) {
+   int i;
+   int ierr;
+   const struct rte_flow_action_rss *rss =
+   (const struct rte_flow_action_rss *)
+   actions->conf;
+
+   if (!priv->hw_rss) {
+   rte_flow_error_set(error, ENOTSUP,
+  RTE_FLOW_ERROR_TYPE_ACTION,
+  actions,
+  "RSS cannot be used with "
+  "the current configuration");
+   return -rte_errno;
+   }
+   if (!priv->isolated) {
+   rte_flow_error_set(error, ENOTSUP,
+  RTE_FLOW_ERROR_TYPE_ACTION,
+  actions,
+  "RSS cannot be used without "
+  "isolated mode");
+   return -rte_errno;
+ 

[dpdk-dev] [PATCH v5 1/4] net/mlx4: RSS parent queues new method maintenance

[Vasily Philipov]

Insert just created parent queue in a list, keep the list in private
structure.

Signed-off-by: Vasily Philipov 
---
The series depends on:

http://dpdk.org/ml/archives/dev/2017-April/064327.html
http://dpdk.org/dev/patchwork/patch/23741/
---
 drivers/net/mlx4/mlx4.c | 377 +---
 drivers/net/mlx4/mlx4.h |  17 ++-
 2 files changed, 274 insertions(+), 120 deletions(-)

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index ec4419a..9f3c746 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -533,13 +533,94 @@ void priv_unlock(struct priv *priv)
 
 static int
 rxq_setup(struct rte_eth_dev *dev, struct rxq *rxq, uint16_t desc,
- unsigned int socket, int inactive, const struct rte_eth_rxconf *conf,
- struct rte_mempool *mp);
+ unsigned int socket, int inactive,
+ const struct rte_eth_rxconf *conf,
+ struct rte_mempool *mp, int children_n,
+ struct rxq *rxq_parent);
 
 static void
 rxq_cleanup(struct rxq *rxq);
 
 /**
+ * Create RSS parent queue.
+ *
+ * The new created strucutre will be on the head of priv parents list.
+ *
+ * @param priv
+ *   Pointer to private structure.
+ * @param queues
+ *   queues indices array, if NULL use all Rx queues.
+ * @param children_n
+ *   The number of entries in queues[].
+ *
+ * @return
+ *   0 on success, negative errno value on failure.
+ */
+static int
+priv_create_parent(struct priv *priv,
+  uint16_t queues[],
+  uint16_t children_n)
+{
+   int ret;
+   uint16_t i;
+   struct rxq *parent;
+
+   parent = rte_zmalloc("parent queue",
+sizeof(*parent),
+RTE_CACHE_LINE_SIZE);
+   if (!parent)
+   return -ENOMEM;
+   ret = rxq_setup(priv->dev, parent, 0, 0, 0,
+   NULL, NULL, children_n, NULL);
+   if (ret) {
+   rte_free(parent);
+   return -ret;
+   }
+   parent->rss.queues_n = children_n;
+   if (queues) {
+   for (i = 0; i < children_n; ++i)
+   parent->rss.queues[i] = queues[i];
+   } else {
+   /* the default RSS ring case */
+   assert(priv->rxqs_n == children_n);
+   for (i = 0; i < priv->rxqs_n; ++i)
+   parent->rss.queues[i] = i;
+   }
+   LIST_INSERT_HEAD(&priv->parents, parent, next);
+   return 0;
+}
+
+/**
+ * Cleanup RX queue parent structure.
+ *
+ * @param parent
+ *   RX queue parent structure.
+ */
+void
+rxq_parent_cleanup(struct rxq *parent)
+{
+   LIST_REMOVE(parent, next);
+   rxq_cleanup(parent);
+   rte_free(parent);
+}
+
+/**
+ * Clean up parent structures from the parents list.
+ *
+ * @param priv
+ *   Pointer to private structure.
+ */
+static void
+priv_parents_list_cleanup(struct priv *priv)
+{
+   while (!LIST_EMPTY(&priv->parents)) {
+   struct rxq *parent = LIST_FIRST(&priv->parents);
+
+   rxq_parent_cleanup(parent);
+   }
+}
+
+/**
  * Ethernet device configuration.
  *
  * Prepare the driver for a given number of TX and RX queues.
@@ -588,7 +669,7 @@ void priv_unlock(struct priv *priv)
for (i = 0; (i != priv->rxqs_n); ++i)
if ((*priv->rxqs)[i] != NULL)
return EINVAL;
-   rxq_cleanup(&priv->rxq_parent);
+   priv_parents_list_cleanup(priv);
priv->rss = 0;
priv->rxqs_n = 0;
}
@@ -613,7 +694,7 @@ void priv_unlock(struct priv *priv)
priv->rss = 1;
tmp = priv->rxqs_n;
priv->rxqs_n = rxqs_n;
-   ret = rxq_setup(dev, &priv->rxq_parent, 0, 0, 0, NULL, NULL);
+   ret = priv_create_parent(priv, NULL, priv->rxqs_n);
if (!ret)
return 0;
/* Failure, rollback. */
@@ -2503,7 +2584,7 @@ struct txq_mp2mr_mbuf_check_data {
if (!BITFIELD_ISSET(priv->mac_configured, mac_index))
return;
if (priv->rss) {
-   rxq_mac_addr_del(&priv->rxq_parent, mac_index);
+   rxq_mac_addr_del(LIST_FIRST(&priv->parents), mac_index);
goto end;
}
for (i = 0; (i != priv->dev->data->nb_rx_queues); ++i)
@@ -2570,7 +2651,7 @@ struct txq_mp2mr_mbuf_check_data {
goto end;
}
if (priv->rss) {
-   ret = rxq_mac_addr_add(&priv->rxq_parent, mac_index);
+   ret = rxq_mac_addr_add(LIST_FIRST(&priv->parents), mac_index);
if (ret)
return ret;
goto end;
@@ -2752,8 +2833,9 @@ struct txq_mp2mr_mbuf_check_data {
rxq_promiscuous_disable(rxq);
rxq_allmulticast_disable(rxq);
rxq_mac_addrs_del(rxq);
-   claim_zero(ibv_destroy_qp(rxq->qp));
}
+   if (rxq->qp != NULL)
+   claim_zero(i

[dpdk-dev] [PATCH v5 2/4] net/mlx4: implement isolated mode from flow API

[Vasily Philipov]

The user must request isolated mode before device configuration,
the default RSS ring isn't created in this case.

Signed-off-by: Vasily Philipov 
---
 drivers/net/mlx4/mlx4.c  | 58 +++-
 drivers/net/mlx4/mlx4.h  |  1 +
 drivers/net/mlx4/mlx4_flow.c | 39 +
 drivers/net/mlx4/mlx4_flow.h |  5 
 4 files changed, 92 insertions(+), 11 deletions(-)

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 9f3c746..22fa7c6 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -650,7 +650,7 @@ void priv_unlock(struct priv *priv)
}
if (rxqs_n == priv->rxqs_n)
return 0;
-   if (!rte_is_power_of_2(rxqs_n)) {
+   if (!rte_is_power_of_2(rxqs_n) && !priv->isolated) {
unsigned n_active;
 
n_active = rte_align32pow2(rxqs_n + 1) >> 1;
@@ -694,6 +694,8 @@ void priv_unlock(struct priv *priv)
priv->rss = 1;
tmp = priv->rxqs_n;
priv->rxqs_n = rxqs_n;
+   if (priv->isolated)
+   return 0;
ret = priv_create_parent(priv, NULL, priv->rxqs_n);
if (!ret)
return 0;
@@ -2580,6 +2582,7 @@ struct txq_mp2mr_mbuf_check_data {
 {
unsigned int i;
 
+   assert(!priv->isolated);
assert(mac_index < elemof(priv->mac));
if (!BITFIELD_ISSET(priv->mac_configured, mac_index))
return;
@@ -2829,7 +2832,7 @@ struct txq_mp2mr_mbuf_check_data {
rxq->if_cq,
¶ms));
}
-   if (rxq->qp != NULL) {
+   if (rxq->qp != NULL && !rxq->priv->isolated) {
rxq_promiscuous_disable(rxq);
rxq_allmulticast_disable(rxq);
rxq_mac_addrs_del(rxq);
@@ -3530,7 +3533,7 @@ struct txq_mp2mr_mbuf_check_data {
return 0;
}
/* Remove attached flows if RSS is disabled (no parent queue). */
-   if (!priv->rss) {
+   if (!priv->rss && !priv->isolated) {
rxq_allmulticast_disable(&tmpl);
rxq_promiscuous_disable(&tmpl);
rxq_mac_addrs_del(&tmpl);
@@ -3575,7 +3578,7 @@ struct txq_mp2mr_mbuf_check_data {
return err;
}
/* Reconfigure flows. Do not care for errors. */
-   if (!priv->rss) {
+   if (!priv->rss && !priv->isolated) {
rxq_mac_addrs_add(&tmpl);
if (priv->promisc)
rxq_promiscuous_enable(&tmpl);
@@ -3734,7 +3737,7 @@ struct txq_mp2mr_mbuf_check_data {
  strerror(ret));
return ret;
}
-   if (parent || !priv->rss) {
+   if (!priv->isolated && (parent || !priv->rss)) {
/* Configure MAC and broadcast addresses. */
ret = rxq_mac_addrs_add(rxq);
if (ret) {
@@ -4013,7 +4016,7 @@ struct txq_mp2mr_mbuf_check_data {
return -ENOMEM;
}
}
-   if (priv->rss) {
+   if (priv->rss && !priv->isolated) {
/* The list consists of the single default one. */
parent = LIST_FIRST(&priv->parents);
if (idx >= rte_align32pow2(priv->rxqs_n + 1) >> 1)
@@ -4109,7 +4112,10 @@ struct txq_mp2mr_mbuf_check_data {
}
DEBUG("%p: attaching configured flows to all RX queues", (void *)dev);
priv->started = 1;
-   if (priv->rss) {
+   if (priv->isolated) {
+   rxq = NULL;
+   r = 1;
+   } else if (priv->rss) {
rxq = LIST_FIRST(&priv->parents);
r = 1;
} else {
@@ -4192,7 +4198,10 @@ struct txq_mp2mr_mbuf_check_data {
}
DEBUG("%p: detaching flows from all RX queues", (void *)dev);
priv->started = 0;
-   if (priv->rss) {
+   if (priv->isolated) {
+   rxq = NULL;
+   r = 1;
+   } else if (priv->rss) {
rxq = LIST_FIRST(&priv->parents);
r = 1;
} else {
@@ -4620,6 +4629,8 @@ struct txq_mp2mr_mbuf_check_data {
if (mlx4_is_secondary())
return;
priv_lock(priv);
+   if (priv->isolated)
+   goto end;
DEBUG("%p: removing MAC address from index %" PRIu32,
  (void *)dev, index);
/* Last array entry is reserved for broadcast. */
@@ -4653,6 +4664,12 @@ struct txq_mp2mr_mbuf_check_data {
return -ENOTSUP;
(void)vmdq;
priv_lock(priv);
+   if (priv->isolated) {
+   DEBUG("%p: cannot add MAC address, "
+ "device is in isolated mode", (void *)dev);
+   re = EPERM;
+   goto end;
+   }
DEBUG("%p: adding MAC address at index %" PRIu32,
  (void *)dev, index);
/* Last array entry is reserved for broadcast. */
@@ -4700,6 +4717,12 @@ str

[dpdk-dev] [PATCH v4 0/6] Reproducible build

[lboccass]

From: Luca Boccassi 

In the past couple of years a concerted effort among almost all Linux
distros has been striving toward achieving reproducible builds. [1]
This involves changes to the toolchain, new tools and CI systems. [2]

v1 fixed the documentation, examples and linker script generation.
v2 fixes all problems, which were caused by unstable order of headers
inclusion, source files listing and object file listing when passing
them to the compiler.
DPDK's build, at least with the default configuration, is fully
reproducible with this patch series as tested by the Reproducible
Builds developers experimental toolchain. [3]

v3 restores the first patch, which was eaten by git send-email.

v4 drops the patch that reorders rebuilds, and adds a patch to make
the inclusion of headers deterministic with regards to GCC embedding
the full file path when expading __FILE__ and when writing the
directory listing in the DWARF objects.
It also drops the first 2 patches which have already been merged.

[1] https://reproducible-builds.org/
[2] https://reproducible-builds.org/tools/
[3] https://wiki.debian.org/ReproducibleBuilds/ExperimentalToolchain#Us

Luca Boccassi (6):
  mk: sort list of shared objects in linker script
  mk: sort list of files in examples.dox
  mk: sort headers before wildcard inclusion
  mk: sort source files before passing them to the compiler
  mk: sort object files when building deps lists
  mk: set -ISCDIR before -IRTE_OUT/include in CFLAGS

 drivers/net/cxgbe/Makefile |  2 +-
 drivers/net/e1000/Makefile |  2 +-
 drivers/net/fm10k/Makefile |  2 +-
 drivers/net/i40e/Makefile  |  2 +-
 drivers/net/ixgbe/Makefile |  2 +-
 drivers/net/qede/Makefile  |  2 +-
 drivers/net/sfc/Makefile   |  2 +-
 drivers/net/thunderx/Makefile  |  2 +-
 examples/ip_pipeline/Makefile  |  2 +-
 examples/multi_process/client_server_mp/mp_server/Makefile |  2 +-
 examples/server_node_efd/server/Makefile   |  2 +-
 lib/librte_acl/Makefile| 10 --
 lib/librte_bitratestats/Makefile   |  9 -
 lib/librte_cmdline/Makefile|  9 -
 lib/librte_distributor/Makefile| 10 --
 lib/librte_eal/common/Makefile |  2 +-
 lib/librte_efd/Makefile| 10 --
 lib/librte_hash/Makefile   | 10 --
 lib/librte_ip_frag/Makefile| 10 --
 lib/librte_jobstats/Makefile   | 10 --
 lib/librte_kni/Makefile|  9 -
 lib/librte_kvargs/Makefile |  9 -
 lib/librte_latencystats/Makefile   |  9 -
 lib/librte_lpm/Makefile| 10 --
 lib/librte_mbuf/Makefile   |  9 -
 lib/librte_mempool/Makefile|  9 -
 lib/librte_metrics/Makefile|  9 -
 lib/librte_net/Makefile|  9 -
 lib/librte_pdump/Makefile  | 10 --
 lib/librte_power/Makefile  |  9 -
 lib/librte_reorder/Makefile| 10 --
 lib/librte_ring/Makefile   |  9 -
 lib/librte_timer/Makefile  |  9 -
 lib/librte_vhost/Makefile  |  9 -
 mk/rte.app.mk  |  4 ++--
 mk/rte.combinedlib.mk  |  2 +-
 mk/rte.hostapp.mk  |  4 ++--
 mk/rte.sdkdoc.mk   |  2 +-
 mk/rte.shared.mk   |  4 ++--
 39 files changed, 196 insertions(+), 51 deletions(-)

-- 
2.11.0

[dpdk-dev] [PATCH v4 1/6] mk: sort list of shared objects in linker script

[lboccass]

From: Luca Boccassi 

The output of wildcard might not be stable and depend on the
filesystem and other factors.
This means the content libdpdk.so linker script might change between
builds from the same sources.
Run the list through sort to ensure reproducibility.

Signed-off-by: Luca Boccassi 
---
 mk/rte.combinedlib.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mk/rte.combinedlib.mk b/mk/rte.combinedlib.mk
index 449358b33..2ab7ee8a1 100644
--- a/mk/rte.combinedlib.mk
+++ b/mk/rte.combinedlib.mk
@@ -42,7 +42,7 @@ endif
 RTE_LIBNAME := dpdk
 COMBINEDLIB := lib$(RTE_LIBNAME)$(EXT)
 
-LIBS := $(filter-out $(COMBINEDLIB), $(notdir $(wildcard 
$(RTE_OUTPUT)/lib/*$(EXT
+LIBS := $(filter-out $(COMBINEDLIB), $(sort $(notdir $(wildcard 
$(RTE_OUTPUT)/lib/*$(EXT)
 
 all: FORCE
$(Q)echo "GROUP ( $(LIBS) )" > $(RTE_OUTPUT)/lib/$(COMBINEDLIB)
-- 
2.11.0

[dpdk-dev] [PATCH v4 2/6] mk: sort list of files in examples.dox

[lboccass]

From: Luca Boccassi 

The result of find might not be stable depending on external
conditions.
Pipe it through LC_ALL=C sort to ensure reproducible results when
generating examples.dox.

Signed-off-by: Luca Boccassi 
---
 mk/rte.sdkdoc.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mk/rte.sdkdoc.mk b/mk/rte.sdkdoc.mk
index c0eaa3502..de31b78cf 100644
--- a/mk/rte.sdkdoc.mk
+++ b/mk/rte.sdkdoc.mk
@@ -93,7 +93,7 @@ $(API_EXAMPLES): api-html-clean
$(Q)mkdir -p $(@D)
@printf '/**\n' > $(API_EXAMPLES)
@printf '@page examples DPDK Example Programs\n\n' >> $(API_EXAMPLES)
-   @find examples -type f -name '*.c' -printf '@example %p\n' >> 
$(API_EXAMPLES)
+   @find examples -type f -name '*.c' -printf '@example %p\n' | LC_ALL=C 
sort >> $(API_EXAMPLES)
@printf '*/\n' >> $(API_EXAMPLES)
 
 guides-pdf-clean: guides-pdf-img-clean
-- 
2.11.0

[dpdk-dev] [PATCH v4 3/6] mk: sort headers before wildcard inclusion

[lboccass]

From: Luca Boccassi 

In order to achieve fully reproducible builds, always use the same
inclusion order for headers in the Makefiles.

Signed-off-by: Luca Boccassi 
---
 examples/ip_pipeline/Makefile  | 2 +-
 examples/multi_process/client_server_mp/mp_server/Makefile | 2 +-
 examples/server_node_efd/server/Makefile   | 2 +-
 lib/librte_eal/common/Makefile | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/examples/ip_pipeline/Makefile b/examples/ip_pipeline/Makefile
index dc7e0ddd7..12ce0a1d5 100644
--- a/examples/ip_pipeline/Makefile
+++ b/examples/ip_pipeline/Makefile
@@ -43,7 +43,7 @@ APP = ip_pipeline
 
 VPATH += $(SRCDIR)/pipeline
 
-INC += $(wildcard *.h) $(wildcard pipeline/*.h)
+INC += $(sort $(wildcard *.h)) $(sort $(wildcard pipeline/*.h))
 
 # all source are stored in SRCS-y
 SRCS-$(CONFIG_RTE_LIBRTE_PIPELINE) := main.c
diff --git a/examples/multi_process/client_server_mp/mp_server/Makefile 
b/examples/multi_process/client_server_mp/mp_server/Makefile
index 5552999b5..160c17b68 100644
--- a/examples/multi_process/client_server_mp/mp_server/Makefile
+++ b/examples/multi_process/client_server_mp/mp_server/Makefile
@@ -49,7 +49,7 @@ APP = mp_server
 # all source are stored in SRCS-y
 SRCS-y := main.c init.c args.c
 
-INC := $(wildcard *.h)
+INC := $(sort $(wildcard *.h))
 
 CFLAGS += $(WERROR_FLAGS) -O3
 CFLAGS += -I$(SRCDIR)/../shared
diff --git a/examples/server_node_efd/server/Makefile 
b/examples/server_node_efd/server/Makefile
index a2f2f361b..9f1fe2894 100644
--- a/examples/server_node_efd/server/Makefile
+++ b/examples/server_node_efd/server/Makefile
@@ -49,7 +49,7 @@ APP = server
 # all source are stored in SRCS-y
 SRCS-y := main.c init.c args.c
 
-INC := $(wildcard *.h)
+INC := $(sort $(wildcard *.h))
 
 CFLAGS += $(WERROR_FLAGS) -O3
 CFLAGS += -I$(SRCDIR)/../shared
diff --git a/lib/librte_eal/common/Makefile b/lib/librte_eal/common/Makefile
index a5bd1089a..4b712600a 100644
--- a/lib/librte_eal/common/Makefile
+++ b/lib/librte_eal/common/Makefile
@@ -48,7 +48,7 @@ GENERIC_INC += rte_vect.h rte_io.h
 
 # defined in mk/arch/$(RTE_ARCH)/rte.vars.mk
 ARCH_DIR ?= $(RTE_ARCH)
-ARCH_INC := $(notdir $(wildcard 
$(RTE_SDK)/lib/librte_eal/common/include/arch/$(ARCH_DIR)/*.h))
+ARCH_INC := $(sort $(notdir $(wildcard 
$(RTE_SDK)/lib/librte_eal/common/include/arch/$(ARCH_DIR)/*.h)))
 
 SYMLINK-$(CONFIG_RTE_LIBRTE_EAL)-include := $(addprefix include/,$(INC))
 SYMLINK-$(CONFIG_RTE_LIBRTE_EAL)-include += \
-- 
2.11.0

[dpdk-dev] [PATCH v4 5/6] mk: sort object files when building deps lists

[lboccass]

From: Luca Boccassi 

In order to achieve reproducible builds, always use the same
order when listing object files to build dependencies lists.

Signed-off-by: Luca Boccassi 
---
 mk/rte.app.mk | 4 ++--
 mk/rte.hostapp.mk | 4 ++--
 mk/rte.shared.mk  | 4 ++--
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/mk/rte.app.mk b/mk/rte.app.mk
index bcaf1b382..54134dea4 100644
--- a/mk/rte.app.mk
+++ b/mk/rte.app.mk
@@ -263,8 +263,8 @@ LDLIBS_NAMES += $(patsubst -Wl$(comma)-l%,lib%.a,$(filter 
-Wl$(comma)-l%,$(LDLIB
 
 # list of found libraries files (useful for deps). If not found, the
 # library is silently ignored and dep won't be checked
-LDLIBS_FILES := $(wildcard $(foreach dir,$(LDLIBS_PATH),\
-   $(addprefix $(dir)/,$(LDLIBS_NAMES
+LDLIBS_FILES := $(sort $(wildcard $(foreach dir,$(LDLIBS_PATH),\
+   $(addprefix $(dir)/,$(LDLIBS_NAMES)
 
 #
 # Compile executable file if needed
diff --git a/mk/rte.hostapp.mk b/mk/rte.hostapp.mk
index 5cb4909cb..f58173c31 100644
--- a/mk/rte.hostapp.mk
+++ b/mk/rte.hostapp.mk
@@ -69,9 +69,9 @@ O_TO_EXE_DO = @set -e; \
 -include .$(HOSTAPP).cmd
 
 # list of .a files that are linked to this application
-LDLIBS_FILES := $(wildcard \
+LDLIBS_FILES := $(sort $(wildcard \
$(addprefix $(RTE_OUTPUT)/lib/, \
-   $(patsubst -l%,lib%.a,$(filter -l%,$(LDLIBS)
+   $(patsubst -l%,lib%.a,$(filter -l%,$(LDLIBS))
 
 #
 # Compile executable file if needed
diff --git a/mk/rte.shared.mk b/mk/rte.shared.mk
index 87ccf0ba4..4e680bc03 100644
--- a/mk/rte.shared.mk
+++ b/mk/rte.shared.mk
@@ -85,8 +85,8 @@ LDLIBS_NAMES += $(patsubst -Wl$(comma)-l%,lib%.a,$(filter 
-Wl$(comma)-l%,$(LDLIB
 
 # list of found libraries files (useful for deps). If not found, the
 # library is silently ignored and dep won't be checked
-LDLIBS_FILES := $(wildcard $(foreach dir,$(LDLIBS_PATH),\
-   $(addprefix $(dir)/,$(LDLIBS_NAMES
+LDLIBS_FILES := $(sort $(wildcard $(foreach dir,$(LDLIBS_PATH),\
+   $(addprefix $(dir)/,$(LDLIBS_NAMES)
 
 #
 # Archive objects in .so file if needed
-- 
2.11.0

[dpdk-dev] [PATCH v4 4/6] mk: sort source files before passing them to the compiler

[lboccass]

From: Luca Boccassi 

In order to achieve reproducible builds, always use the same
order when listing files for compilation.

Signed-off-by: Luca Boccassi 
---
 drivers/net/cxgbe/Makefile| 2 +-
 drivers/net/e1000/Makefile| 2 +-
 drivers/net/fm10k/Makefile| 2 +-
 drivers/net/i40e/Makefile | 2 +-
 drivers/net/ixgbe/Makefile| 2 +-
 drivers/net/qede/Makefile | 2 +-
 drivers/net/sfc/Makefile  | 2 +-
 drivers/net/thunderx/Makefile | 2 +-
 8 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/net/cxgbe/Makefile b/drivers/net/cxgbe/Makefile
index 7cef6279c..b4666b5af 100644
--- a/drivers/net/cxgbe/Makefile
+++ b/drivers/net/cxgbe/Makefile
@@ -67,7 +67,7 @@ endif
 # Add extra flags for base driver files (also known as shared code)
 # to disable warnings in them
 #
-BASE_DRIVER_OBJS=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+BASE_DRIVER_OBJS=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(BASE_DRIVER_OBJS), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
diff --git a/drivers/net/e1000/Makefile b/drivers/net/e1000/Makefile
index ffdf36d37..59d96bca1 100644
--- a/drivers/net/e1000/Makefile
+++ b/drivers/net/e1000/Makefile
@@ -68,7 +68,7 @@ endif
 # Add extra flags for base driver files (also known as shared code)
 # to disable warnings in them
 #
-BASE_DRIVER_OBJS=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+BASE_DRIVER_OBJS=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(BASE_DRIVER_OBJS), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
diff --git a/drivers/net/fm10k/Makefile b/drivers/net/fm10k/Makefile
index e0024f052..0bc124eb1 100644
--- a/drivers/net/fm10k/Makefile
+++ b/drivers/net/fm10k/Makefile
@@ -80,7 +80,7 @@ endif
 #
 # Add extra flags for base driver source files to disable warnings in them
 #
-BASE_DRIVER_OBJS=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+BASE_DRIVER_OBJS=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(BASE_DRIVER_OBJS), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
diff --git a/drivers/net/i40e/Makefile b/drivers/net/i40e/Makefile
index 56f210d6d..06eedc592 100644
--- a/drivers/net/i40e/Makefile
+++ b/drivers/net/i40e/Makefile
@@ -78,7 +78,7 @@ endif
 
 CFLAGS_i40e_lan_hmc.o += -Wno-error
 endif
-OBJS_BASE_DRIVER=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+OBJS_BASE_DRIVER=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(OBJS_BASE_DRIVER), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
diff --git a/drivers/net/ixgbe/Makefile b/drivers/net/ixgbe/Makefile
index 6a651b923..f5c370ce5 100644
--- a/drivers/net/ixgbe/Makefile
+++ b/drivers/net/ixgbe/Makefile
@@ -87,7 +87,7 @@ endif
 # Add extra flags for base driver files (also known as shared code)
 # to disable warnings in them
 #
-BASE_DRIVER_OBJS=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+BASE_DRIVER_OBJS=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(BASE_DRIVER_OBJS), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
diff --git a/drivers/net/qede/Makefile b/drivers/net/qede/Makefile
index f03441d9a..83ff95474 100644
--- a/drivers/net/qede/Makefile
+++ b/drivers/net/qede/Makefile
@@ -78,7 +78,7 @@ endif
 # to disable warnings in them
 #
 #
-BASE_DRIVER_OBJS=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+BASE_DRIVER_OBJS=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(BASE_DRIVER_OBJS), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
diff --git a/drivers/net/sfc/Makefile b/drivers/net/sfc/Makefile
index 57aa963ba..8cfd14d45 100644
--- a/drivers/net/sfc/Makefile
+++ b/drivers/net/sfc/Makefile
@@ -71,7 +71,7 @@ endif
 # List of base driver object files for which
 # special CFLAGS above should be applied
 #
-BASE_DRIVER_OBJS=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+BASE_DRIVER_OBJS=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(BASE_DRIVER_OBJS), \
   $(eval CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
diff --git a/drivers/net/thunderx/Makefile b/drivers/net/thunderx/Makefile
index 706250b8b..ac384a624 100644
--- a/drivers/net/thunderx/Makefile
+++ b/drivers/net/thunderx/Makefile
@@ -45,7 +45,7 @@ EXPORT_MAP := rte_pmd_thunderx_nicvf_version.map
 
 LIBABIVER := 1
 
-OBJS_BASE_DRIVER=$(patsubst %.c,%.o,$(notdir $(wildcard $(SRCDIR)/base/*.c)))
+OBJS_BASE_DRIVER=$(sort $(patsubst %.c,%.o,$(notdir $(wildcard 
$(SRCDIR)/base/*.c
 $(foreach obj, $(OBJS_BASE_DRIVER), $(eval 
CFLAGS_$(obj)+=$(CFLAGS_BASE_DRIVER)))
 
 VPATH += $(SRCDIR)/base
-- 
2.11.0

[dpdk-dev] [PATCH v4 6/6] mk: set -ISCDIR before -IRTE_OUT/include in CFLAGS

[lboccass]

From: Luca Boccassi 

A race condition can happen during parallel builds, where a header
might be installed in RTE_OUT/include before CFLAGS is recursively
expanded. This causes GCC to sometimes pick the header path as
SRCDIR/... and sometimes as RTE_OUT/include/... making the build
unreproducible, as the full path is used for the expansion of
__FILE__ and in the DWARF directory listing.
Always pass -ISRCDIR first to CFLAGS so that it's deterministic.

Signed-off-by: Luca Boccassi 
---
 lib/librte_acl/Makefile  | 10 --
 lib/librte_bitratestats/Makefile |  9 -
 lib/librte_cmdline/Makefile  |  9 -
 lib/librte_distributor/Makefile  | 10 --
 lib/librte_efd/Makefile  | 10 --
 lib/librte_hash/Makefile | 10 --
 lib/librte_ip_frag/Makefile  | 10 --
 lib/librte_jobstats/Makefile | 10 --
 lib/librte_kni/Makefile  |  9 -
 lib/librte_kvargs/Makefile   |  9 -
 lib/librte_latencystats/Makefile |  9 -
 lib/librte_lpm/Makefile  | 10 --
 lib/librte_mbuf/Makefile |  9 -
 lib/librte_mempool/Makefile  |  9 -
 lib/librte_metrics/Makefile  |  9 -
 lib/librte_net/Makefile  |  9 -
 lib/librte_pdump/Makefile| 10 --
 lib/librte_power/Makefile|  9 -
 lib/librte_reorder/Makefile  | 10 --
 lib/librte_ring/Makefile |  9 -
 lib/librte_timer/Makefile|  9 -
 lib/librte_vhost/Makefile|  9 -
 22 files changed, 176 insertions(+), 31 deletions(-)

diff --git a/lib/librte_acl/Makefile b/lib/librte_acl/Makefile
index e2dacd606..470ef6d21 100644
--- a/lib/librte_acl/Makefile
+++ b/lib/librte_acl/Makefile
@@ -34,8 +34,14 @@ include $(RTE_SDK)/mk/rte.vars.mk
 # library name
 LIB = librte_acl.a
 
-CFLAGS += -O3
-CFLAGS += $(WERROR_FLAGS) -I$(SRCDIR)
+# Include the source dir first, so that headers paths are always picked
+# from there. By including it last race conditions might happen during paralle
+# builds, and headers might be already installed in RTE_OUT/include when the
+# variable is recursively expanded, thus causing GCC to sometimes use the
+# SRCDIR path and sometimes the RTE_OUT/include, making the builds not
+# reproducible.
+CFLAGS := -I$(SRCDIR) $(CFLAGS)
+CFLAGS += $(WERROR_FLAGS) -O3
 
 EXPORT_MAP := rte_acl_version.map
 
diff --git a/lib/librte_bitratestats/Makefile b/lib/librte_bitratestats/Makefile
index 58a20ea09..26a3f4908 100644
--- a/lib/librte_bitratestats/Makefile
+++ b/lib/librte_bitratestats/Makefile
@@ -34,7 +34,14 @@ include $(RTE_SDK)/mk/rte.vars.mk
 # library name
 LIB = librte_bitratestats.a
 
-CFLAGS += $(WERROR_FLAGS) -I$(SRCDIR) -O3
+# Include the source dir first, so that headers paths are always picked
+# from there. By including it last race conditions might happen during paralle
+# builds, and headers might be already installed in RTE_OUT/include when the
+# variable is recursively expanded, thus causing GCC to sometimes use the
+# SRCDIR path and sometimes the RTE_OUT/include, making the builds not
+# reproducible.
+CFLAGS := -I$(SRCDIR) $(CFLAGS)
+CFLAGS += $(WERROR_FLAGS) -O3
 
 EXPORT_MAP := rte_bitratestats_version.map
 
diff --git a/lib/librte_cmdline/Makefile b/lib/librte_cmdline/Makefile
index 644f68e47..9dd75f2d2 100644
--- a/lib/librte_cmdline/Makefile
+++ b/lib/librte_cmdline/Makefile
@@ -34,7 +34,14 @@ include $(RTE_SDK)/mk/rte.vars.mk
 # library name
 LIB = librte_cmdline.a
 
-CFLAGS += $(WERROR_FLAGS) -I$(SRCDIR) -O3
+# Include the source dir first, so that headers paths are always picked
+# from there. By including it last race conditions might happen during paralle
+# builds, and headers might be already installed in RTE_OUT/include when the
+# variable is recursively expanded, thus causing GCC to sometimes use the
+# SRCDIR path and sometimes the RTE_OUT/include, making the builds not
+# reproducible.
+CFLAGS := -I$(SRCDIR) $(CFLAGS)
+CFLAGS += $(WERROR_FLAGS) -O3
 
 EXPORT_MAP := rte_cmdline_version.map
 
diff --git a/lib/librte_distributor/Makefile b/lib/librte_distributor/Makefile
index 3ffb911ce..87418f254 100644
--- a/lib/librte_distributor/Makefile
+++ b/lib/librte_distributor/Makefile
@@ -34,8 +34,14 @@ include $(RTE_SDK)/mk/rte.vars.mk
 # library name
 LIB = librte_distributor.a
 
-CFLAGS += -O3
-CFLAGS += $(WERROR_FLAGS) -I$(SRCDIR)
+# Include the source dir first, so that headers paths are always picked
+# from there. By including it last race conditions might happen during paralle
+# builds, and headers might be already installed in RTE_OUT/include when the
+# variable is recursively expanded, thus causing GCC to sometimes use the
+# SRCDIR path and sometimes the RTE_OUT/include, making the builds not
+# reproducible.
+CFLAGS := -I$(SRCDIR) $(CFLAGS)
+CFLAGS += $(WERROR_FLAGS) -O3
 
 EXPORT_MAP := rte_distributor_version.map
 
diff --git a/lib/librte_efd/Makefile b/lib/librte_efd/Makefile

Re: [dpdk-dev] [PATCH v3 7/8] mk: sort object files when building deps lists

[Luca Boccassi]

On Tue, 2017-06-27 at 18:14 +0200, Thomas Monjalon wrote:
> 27/06/2017 16:51, Luca Boccassi:
> > On Tue, 2017-06-27 at 15:52 +0200, Thomas Monjalon wrote:
> > > 27/06/2017 12:43, Luca Boccassi:
> > > > On Tue, 2017-06-27 at 01:20 +0200, Thomas Monjalon wrote:
> > > > > 23/06/2017 20:41, lbocc...@brocade.com:
> > > > > > From: Luca Boccassi 
> > > > > > 
> > > > > > In order to achieve reproducible builds, always use the
> > > > > > same
> > > > > > order when listing object files to build dependencies
> > > > > > lists.
> > > > > > 
> > > > > > Signed-off-by: Luca Boccassi 
> > > > > > ---
> > > > > >  mk/rte.app.mk | 4 ++--
> > > > > >  mk/rte.hostapp.mk | 4 ++--
> > > > > >  mk/rte.shared.mk  | 4 ++--
> > > > > >  3 files changed, 6 insertions(+), 6 deletions(-)
> > > > > > 
> > > > > > --- a/mk/rte.app.mk
> > > > > > +++ b/mk/rte.app.mk
> > > > > > @@ -263,8 +263,8 @@ LDLIBS_NAMES += $(patsubst -Wl$(comma)-
> > > > > > l%,lib%.a,$(filter -Wl$(comma)-l%,$(LDLIB
> > > > > >  
> > > > > >  # list of found libraries files (useful for deps). If not
> > > > > > found,
> > > > > > the
> > > > > >  # library is silently ignored and dep won't be checked
> > > > > > -LDLIBS_FILES := $(wildcard $(foreach dir,$(LDLIBS_PATH),\
> > > > > > -   $(addprefix $(dir)/,$(LDLIBS_NAMES
> > > > > > +LDLIBS_FILES := $(sort $(wildcard $(foreach
> > > > > > dir,$(LDLIBS_PATH),\
> > > > > > +   $(addprefix $(dir)/,$(LDLIBS_NAMES)
> > > > > 
> > > > > You cannot sort libraries.
> > > > > Check - for instance - this comment above in this file:
> > > > >   # Eliminate duplicates without sorting, only keep the
> > > > > last
> > > > > occurrence
> > > > >   filter-libs = \
> > > > 
> > > > Not sure I follow - what's the reason for avoiding to sort the
> > > > list
> > > > of
> > > > libs to link against?
> > > 
> > > Sorry, the ordering issue is with LDLIBS not LDLIBS_FILES.
> > > 
> > > > > Why sorting them?
> > > > > What is random in libraries list?
> > > > 
> > > > The issue is that the output of wildcard is not fully
> > > > deterministic. It
> > > > can depend on the filesystem, and even on the locale settings
> > > > [1].
> > > > Before GNU Make 3.82 (2009) it used to automatically sort the
> > > > output,
> > > > but that was removed (to make it faster... sigh). [2]
> > > 
> > > It is not a true wildcard here. It is just filtering files which
> > > do not exist.
> > > I think you do not need this patch for deterministic build.
> > 
> > But then those lists are passed down in the .SECONDEXPANSION rule
> > right?
> 
> I do not follow you.
> Please explain what is the benefit of the patch in the next version.

I thought that these lists are used to determine which files to
recompile - and incidentally, in which order as they are passed in this
snippet in rte.compile-pre.mk:

.SECONDEXPANSION:
%.o: %.c $$(wildcard $$(dep_$$@)) $$(DEP_$$(@)) FORCE
@[ -d $(dir $@) ] || mkdir -p $(dir $@)
$(if $(D),\
@echo -n "$< -> $@ " ; \
echo -n "file_missing=$(call boolean,$(file_missing)) " ; \
echo -n "cmdline_changed=$(call boolean,$(call 
cmdline_changed,$(C_TO_O))) " ; \
echo -n "depfile_missing=$(call boolean,$(depfile_missing)) " ; 
\
echo "depfile_newer=$(call boolean,$(depfile_newer))")
$(if $(or \
$(file_missing),\
$(call cmdline_changed,$(C_TO_O)),\
$(depfile_missing),\
$(depfile_newer)),\
$(C_TO_O_DO))

Did I get that wrong? (It is a bit convoluted :-P )


But nevertheless, I've finally found the root cause for the "wandering
header" - when building the libraries, CFLAGS lists
-I$(RTE_OUTDIR)/include first and then -I$(SRCDIR) last. There is a
race, and sometimes when GCC is called the public header has already
been installed in RTE_OUTDIR and sometimes it has not. This causes the
instability, and causes the expansion of __FILE__ and the directory
listing in the DWARF objects to randomly list the full path to either
RTE_OUTDIR/include or SRCDIR for each library.

By always passing -I$(SRCDIR) first this instability is solved.

I've now dropped this patch and added this new fix in v4.
This might mean that partial rebuilds are not stable, if I understand
correctly how rte.compile-pre.mk works (eg: change 2 files, rebuild,
change them again, rebuild -> output final binary _might_ be
different), but a full rebuild from scratch now seems to be always
reproducible.

> > I am trying to find out an alternative solution. The problem to
> > solve
> > is that the build system picks the public headers path (which is
> > embedded in the object files as notation and in the debug symbol)
> > from
> > a seemingly random location between the make install path and the
> > source path (build/include/rte_foo.h vs lib/librte_foo/rte_foo.h)
> > and
> > this makes the build not reproducible.
> > 
> > Nonetheless, while I work more on the last 4 patches, c

[dpdk-dev] [PATCH] examples/ethtool: explicit PCI header dependency

[Thomas Monjalon]

In devargs rework, rte_pci.h won't be included by rte_ethdev.h
(via rte_devargs.h) anymore.

rte_ethtool_get_drvinfo() could use rte_devargs.name instead of
creating equivalent bus specific name.
For now, it is workarounded by just including rte_pci.h.

Signed-off-by: Thomas Monjalon 
---
 examples/ethtool/lib/rte_ethtool.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/examples/ethtool/lib/rte_ethtool.c 
b/examples/ethtool/lib/rte_ethtool.c
index fabfcb2ba..252382cb5 100644
--- a/examples/ethtool/lib/rte_ethtool.c
+++ b/examples/ethtool/lib/rte_ethtool.c
@@ -36,6 +36,7 @@
 #include 
 #include 
 #include 
+#include 
 #ifdef RTE_LIBRTE_IXGBE_PMD
 #include 
 #endif
@@ -73,6 +74,7 @@ rte_ethtool_get_drvinfo(uint8_t port_id, struct 
ethtool_drvinfo *drvinfo)
dev_info.driver_name);
snprintf(drvinfo->version, sizeof(drvinfo->version), "%s",
rte_version());
+   /* TODO: replace bus_info by rte_devargs.name */
if (dev_info.pci_dev)
snprintf(drvinfo->bus_info, sizeof(drvinfo->bus_info),
"%04x:%02x:%02x.%x",
-- 
2.13.1

Re: [dpdk-dev] [PATCH v3 7/8] mk: sort object files when building deps lists

[Thomas Monjalon]

28/06/2017 16:07, Luca Boccassi:
> On Tue, 2017-06-27 at 18:14 +0200, Thomas Monjalon wrote:
> > 27/06/2017 16:51, Luca Boccassi:
> > > On Tue, 2017-06-27 at 15:52 +0200, Thomas Monjalon wrote:
> > > > 27/06/2017 12:43, Luca Boccassi:
> > > > > On Tue, 2017-06-27 at 01:20 +0200, Thomas Monjalon wrote:
> > > > > > 23/06/2017 20:41, lbocc...@brocade.com:
> > > > > > > From: Luca Boccassi 
> > > > > > > 
> > > > > > > In order to achieve reproducible builds, always use the
> > > > > > > same
> > > > > > > order when listing object files to build dependencies
> > > > > > > lists.
> > > > > > > 
> > > > > > > Signed-off-by: Luca Boccassi 
> > > > > > > ---
> > > > > > >  mk/rte.app.mk | 4 ++--
> > > > > > >  mk/rte.hostapp.mk | 4 ++--
> > > > > > >  mk/rte.shared.mk  | 4 ++--
> > > > > > >  3 files changed, 6 insertions(+), 6 deletions(-)
> > > > > > > 
> > > > > > > --- a/mk/rte.app.mk
> > > > > > > +++ b/mk/rte.app.mk
> > > > > > > @@ -263,8 +263,8 @@ LDLIBS_NAMES += $(patsubst -Wl$(comma)-
> > > > > > > l%,lib%.a,$(filter -Wl$(comma)-l%,$(LDLIB
> > > > > > >  
> > > > > > >  # list of found libraries files (useful for deps). If not
> > > > > > > found,
> > > > > > > the
> > > > > > >  # library is silently ignored and dep won't be checked
> > > > > > > -LDLIBS_FILES := $(wildcard $(foreach dir,$(LDLIBS_PATH),\
> > > > > > > - $(addprefix $(dir)/,$(LDLIBS_NAMES
> > > > > > > +LDLIBS_FILES := $(sort $(wildcard $(foreach
> > > > > > > dir,$(LDLIBS_PATH),\
> > > > > > > + $(addprefix $(dir)/,$(LDLIBS_NAMES)
> > > > > > 
> > > > > > You cannot sort libraries.
> > > > > > Check - for instance - this comment above in this file:
> > > > > > # Eliminate duplicates without sorting, only keep the
> > > > > > last
> > > > > > occurrence
> > > > > > filter-libs = \
> > > > > 
> > > > > Not sure I follow - what's the reason for avoiding to sort the
> > > > > list
> > > > > of
> > > > > libs to link against?
> > > > 
> > > > Sorry, the ordering issue is with LDLIBS not LDLIBS_FILES.
> > > > 
> > > > > > Why sorting them?
> > > > > > What is random in libraries list?
> > > > > 
> > > > > The issue is that the output of wildcard is not fully
> > > > > deterministic. It
> > > > > can depend on the filesystem, and even on the locale settings
> > > > > [1].
> > > > > Before GNU Make 3.82 (2009) it used to automatically sort the
> > > > > output,
> > > > > but that was removed (to make it faster... sigh). [2]
> > > > 
> > > > It is not a true wildcard here. It is just filtering files which
> > > > do not exist.
> > > > I think you do not need this patch for deterministic build.
> > > 
> > > But then those lists are passed down in the .SECONDEXPANSION rule
> > > right?
> > 
> > I do not follow you.
> > Please explain what is the benefit of the patch in the next version.
> 
> I thought that these lists are used to determine which files to
> recompile - and incidentally, in which order as they are passed in this
> snippet in rte.compile-pre.mk:
> 
> .SECONDEXPANSION:
> %.o: %.c $$(wildcard $$(dep_$$@)) $$(DEP_$$(@)) FORCE
>   @[ -d $(dir $@) ] || mkdir -p $(dir $@)
>   $(if $(D),\
>   @echo -n "$< -> $@ " ; \
>   echo -n "file_missing=$(call boolean,$(file_missing)) " ; \
>   echo -n "cmdline_changed=$(call boolean,$(call 
> cmdline_changed,$(C_TO_O))) " ; \
>   echo -n "depfile_missing=$(call boolean,$(depfile_missing)) " ; 
> \
>   echo "depfile_newer=$(call boolean,$(depfile_newer))")
>   $(if $(or \
>   $(file_missing),\
>   $(call cmdline_changed,$(C_TO_O)),\
>   $(depfile_missing),\
>   $(depfile_newer)),\
>   $(C_TO_O_DO))
> 
> Did I get that wrong? (It is a bit convoluted :-P )

LDLIBS_FILES are just used to link the executable AFAIK.
The libraries are built in separate recursive make calls.

> But nevertheless, I've finally found the root cause for the "wandering
> header" - when building the libraries, CFLAGS lists
> -I$(RTE_OUTDIR)/include first and then -I$(SRCDIR) last. There is a
> race, and sometimes when GCC is called the public header has already
> been installed in RTE_OUTDIR and sometimes it has not. This causes the
> instability, and causes the expansion of __FILE__ and the directory
> listing in the DWARF objects to randomly list the full path to either
> RTE_OUTDIR/include or SRCDIR for each library.
> 
> By always passing -I$(SRCDIR) first this instability is solved.

Good news.

Re: [dpdk-dev] [PATCH v3 7/8] mk: sort object files when building deps lists

[Luca Boccassi]

On Wed, 2017-06-28 at 16:37 +0200, Thomas Monjalon wrote:
> 28/06/2017 16:07, Luca Boccassi:
> > On Tue, 2017-06-27 at 18:14 +0200, Thomas Monjalon wrote:
> > > 27/06/2017 16:51, Luca Boccassi:
> > > > On Tue, 2017-06-27 at 15:52 +0200, Thomas Monjalon wrote:
> > > > > 27/06/2017 12:43, Luca Boccassi:
> > > > > > On Tue, 2017-06-27 at 01:20 +0200, Thomas Monjalon wrote:
> > > > > > > 23/06/2017 20:41, lbocc...@brocade.com:
> > > > > > > > From: Luca Boccassi 
> > > > > > > > 
> > > > > > > > In order to achieve reproducible builds, always use the
> > > > > > > > same
> > > > > > > > order when listing object files to build dependencies
> > > > > > > > lists.
> > > > > > > > 
> > > > > > > > Signed-off-by: Luca Boccassi 
> > > > > > > > ---
> > > > > > > >  mk/rte.app.mk | 4 ++--
> > > > > > > >  mk/rte.hostapp.mk | 4 ++--
> > > > > > > >  mk/rte.shared.mk  | 4 ++--
> > > > > > > >  3 files changed, 6 insertions(+), 6 deletions(-)
> > > > > > > > 
> > > > > > > > --- a/mk/rte.app.mk
> > > > > > > > +++ b/mk/rte.app.mk
> > > > > > > > @@ -263,8 +263,8 @@ LDLIBS_NAMES += $(patsubst
> > > > > > > > -Wl$(comma)-
> > > > > > > > l%,lib%.a,$(filter -Wl$(comma)-l%,$(LDLIB
> > > > > > > >  
> > > > > > > >  # list of found libraries files (useful for deps). If
> > > > > > > > not
> > > > > > > > found,
> > > > > > > > the
> > > > > > > >  # library is silently ignored and dep won't be checked
> > > > > > > > -LDLIBS_FILES := $(wildcard $(foreach
> > > > > > > > dir,$(LDLIBS_PATH),\
> > > > > > > > -   $(addprefix $(dir)/,$(LDLIBS_NAMES
> > > > > > > > +LDLIBS_FILES := $(sort $(wildcard $(foreach
> > > > > > > > dir,$(LDLIBS_PATH),\
> > > > > > > > +   $(addprefix $(dir)/,$(LDLIBS_NAMES)
> > > > > > > 
> > > > > > > You cannot sort libraries.
> > > > > > > Check - for instance - this comment above in this file:
> > > > > > >   # Eliminate duplicates without sorting, only keep the
> > > > > > > last
> > > > > > > occurrence
> > > > > > >   filter-libs = \
> > > > > > 
> > > > > > Not sure I follow - what's the reason for avoiding to sort
> > > > > > the
> > > > > > list
> > > > > > of
> > > > > > libs to link against?
> > > > > 
> > > > > Sorry, the ordering issue is with LDLIBS not LDLIBS_FILES.
> > > > > 
> > > > > > > Why sorting them?
> > > > > > > What is random in libraries list?
> > > > > > 
> > > > > > The issue is that the output of wildcard is not fully
> > > > > > deterministic. It
> > > > > > can depend on the filesystem, and even on the locale
> > > > > > settings
> > > > > > [1].
> > > > > > Before GNU Make 3.82 (2009) it used to automatically sort
> > > > > > the
> > > > > > output,
> > > > > > but that was removed (to make it faster... sigh). [2]
> > > > > 
> > > > > It is not a true wildcard here. It is just filtering files
> > > > > which
> > > > > do not exist.
> > > > > I think you do not need this patch for deterministic build.
> > > > 
> > > > But then those lists are passed down in the .SECONDEXPANSION
> > > > rule
> > > > right?
> > > 
> > > I do not follow you.
> > > Please explain what is the benefit of the patch in the next
> > > version.
> > 
> > I thought that these lists are used to determine which files to
> > recompile - and incidentally, in which order as they are passed in
> > this
> > snippet in rte.compile-pre.mk:
> > 
> > .SECONDEXPANSION:
> > %.o: %.c $$(wildcard $$(dep_$$@)) $$(DEP_$$(@)) FORCE
> > @[ -d $(dir $@) ] || mkdir -p $(dir $@)
> > $(if $(D),\
> > @echo -n "$< -> $@ " ; \
> > echo -n "file_missing=$(call boolean,$(file_missing)) "
> > ; \
> > echo -n "cmdline_changed=$(call boolean,$(call
> > cmdline_changed,$(C_TO_O))) " ; \
> > echo -n "depfile_missing=$(call
> > boolean,$(depfile_missing)) " ; \
> > echo "depfile_newer=$(call boolean,$(depfile_newer))")
> > $(if $(or \
> > $(file_missing),\
> > $(call cmdline_changed,$(C_TO_O)),\
> > $(depfile_missing),\
> > $(depfile_newer)),\
> > $(C_TO_O_DO))
> > 
> > Did I get that wrong? (It is a bit convoluted :-P )
> 
> LDLIBS_FILES are just used to link the executable AFAIK.
> The libraries are built in separate recursive make calls.
> 
> > But nevertheless, I've finally found the root cause for the
> > "wandering
> > header" - when building the libraries, CFLAGS lists
> > -I$(RTE_OUTDIR)/include first and then -I$(SRCDIR) last. There is a
> > race, and sometimes when GCC is called the public header has
> > already
> > been installed in RTE_OUTDIR and sometimes it has not. This causes
> > the
> > instability, and causes the expansion of __FILE__ and the directory
> > listing in the DWARF objects to randomly list the full path to
> > either
> > RTE_OUTDIR/include or SRCDIR for each library.
> > 
> > By always passing -I$(SRCDIR) first this instability is solved.
> 
> Good news.

BTW as an example here's the diff between 2 builds, if anyone is
interested:

https://paste.debian.

[dpdk-dev] [PATCH] bug: Use correct arguments in run.py

[McCullough, Harrison]

When using run.py it would occasionally ignore the given command line
arguments because the wrong variable was used.  Fixed this, along with
minor changes to reflect more idiomatic Python usage.

Signed-off-by: Harrison McCullough 
---
 tools/run.py | 11 +--
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/tools/run.py b/tools/run.py
index ec6f45d..f5ec22c 100755
--- a/tools/run.py
+++ b/tools/run.py
@@ -96,7 +96,7 @@ def find_file(arg, t):
 for f in file_list('.', t):
 if os.path.basename(f) == fn:
 return f
-return "" 
+return None
 
 def parse_args():
 ''' Parse the command arguments '''
@@ -118,7 +118,7 @@ def parse_args():
 print("Run '%s --usage' for further information" % sys.argv[0])
 sys.exit(1)
 
-for opt, arg in opts:
+for opt, _ in opts:
 if opt == "--help" or opt == "-h" or opt == "--usage" or opt == "-u":
 usage()
 sys.exit(0)
@@ -128,13 +128,12 @@ def parse_args():
 if opt == "--setup" or opt == "-s":
 run_flag = False
 
-arg = sys.argv[1:]
-if arg == "":
+if not args or len(args) > 1:
 usage()
 sys.exit(1)
 
-fn = find_file(arg[0], cfg_ext)
-if fn != "":
+fn = find_file(args[0], cfg_ext)
+if fn:
 cfg_file = fn
 
 def load_cfg():
-- 
1.9.1

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Jan Blunck]

On Wed, Jun 28, 2017 at 3:30 PM, Thomas Monjalon  wrote:
> 28/06/2017 15:09, Jan Blunck:
>> On Wed, Jun 28, 2017 at 2:11 PM, Thomas Monjalon  wrote:
>> > 28/06/2017 13:58, Jan Blunck:
>> >> On Wed, Jun 28, 2017 at 1:44 PM, Thomas Monjalon  
>> >> wrote:
>> >> > 27/06/2017 21:03, Jan Blunck:
>> >> >> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet  
>> >> >> wrote:
>> >> >> > --- a/lib/librte_eal/common/include/rte_bus.h
>> >> >> > +++ b/lib/librte_eal/common/include/rte_bus.h
>> >> >> >  /**
>> >> >> > + * Implementation specific probe function which is responsible for 
>> >> >> > linking
>> >> >> > + * devices on that bus with applicable drivers.
>> >> >> > + * The plugged device might already have been used previously by 
>> >> >> > the bus,
>> >> >> > + * in which case some buses might prefer to detect and re-use the 
>> >> >> > relevant
>> >> >> > + * information pertaining to this device.
>> >> >> > + *
>> >> >> > + * @param da
>> >> >> > + * Device declaration.
>> >> >> > + *
>> >> >> > + * @return
>> >> >> > + * The pointer to a valid rte_device usable by the bus on 
>> >> >> > success.
>> >> >> > + * NULL on error. rte_errno is then set.
>> >> >> > + */
>> >> >> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs 
>> >> >> > *da);
>> >> >>
>> >> >> Shouldn't this be orthogonal to unplug() and take a rte_device. You
>> >> >> should only be able to plug devices that have been found by scan
>> >> >> before.
>> >> >
>> >> > Plugging a device that has been scanned before is a special case.
>> >> > In a true hotplug scenario, we could use this plug function passing
>> >> > a devargs.
>> >> > I don't see any issue passing rte_devargs to plug and rte_device to 
>> >> > unplug.
>> >>
>> >> What do you mean by "true hotplug"?
>> >
>> > I mean a kernel notification of a new device.
>>
>> Does a "false hotplug" exist, too? :)
>
> The false hotplug was the original attach function which was just
> adding a new ethdev interface.
>
>> >> The problem with this is that passing just rte_devargs to plug()
>> >> requires the bus to parse and enrich the rte_devargs with bus
>> >> specifics. From there it gets folded into the to-be-created bus
>> >> specific rte_XXX_device. This makes it unnecessarily complicated and
>> >> even worse it adds a second code path how devices come alive.
>> >
>> > Just after the notification, the rte_device does not exist yet.
>> > So the plug function could share the same code as the scan function
>> > to get the metadata and create the device instance.
>>
>> Exactly this is what I want to avoid.
>
> Why do you want to avoid that?
> I think you mean it is not what you had in mind.
>
>> The plug() function would become a "scan-one and probe".
>
> Yes
>
>> From my point of view plug() and unplug() should be orthogonal.
>> The plug() and unplug() should only be responsible for adding drivers
>> with optional arguments. The EAL should allow the drivers to get
>> unplugged/re-plugged at run-time. I want to be able to change arguments
>> ... or even drivers :)
>
> It is a totally different thing.
> We are talking about hotplug of a device,
> and you are talking about changing drivers dynamically.
>
> So I still don't understand what is the issue with the plug/unplug
> functions proposed here.
>

I don't agree with the notion that plug() means "scan-one and probe".

>> >> When we get notified about a hotplug event we already know which bus
>> >> this event belongs to:
>> >>
>> >> 1. scan the bus for incoming devices
>> >
>> > No need to scan every devices here.
>>
>> This is a readdir followed by open+read+close for any new device. This
>> code belongs here anyway. Its lightweight if nothing changed. The scan
>> itself should be idempotent anyway.
>>
>> >> 2. plug single device with devargs and probe for drivers
>> >>
>> >> Makes sense?
>> >
>> > I want to make sure there is no misunderstanding first :)
>>
>> Which makes sense. That is probably my fault due to being too
>> distracted with other things and not communicating well enough while
>> Gaetan consumed my code.
>
> Your ideas are probably interesting, and I want to understand them.
> In the meantime, we need to progress on 17.08-rc1 which must be done
> in following days. Please let's separate the ideas which are not
> yet implemented from what we are already able to deliver.

Sorry, if you got the impression that this is taken from thin air but
this is how my original patch series looked like and I'm just
providing the rational behind it.

[dpdk-dev] [PATCH 01/21 v3] drivers: add bus dependency for event

[Nipun Gupta]

From: Hemant Agrawal 

Signed-off-by: Hemant Agrawal 
---
 drivers/Makefile | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/Makefile b/drivers/Makefile
index a04a01f..7fef66d 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -39,5 +39,6 @@ DEPDIRS-net := bus mempool
 DIRS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += crypto
 DEPDIRS-crypto := mempool
 DIRS-$(CONFIG_RTE_LIBRTE_EVENTDEV) += event
+DEPDIRS-event := bus
 
 include $(RTE_SDK)/mk/rte.subdir.mk
-- 
1.9.1

[dpdk-dev] [PATCH 00/21 v3] next-eventdev: NXP DPAA2 eventdev PMD

[Nipun Gupta]

The following patch set adds NXP DPAA2 HW based eventdev
implementation to the next-eventdev tree.

The patchset consists of the following:
- DPCON support required by DPAA2 eventdev
- DPCI support in fslmc bus
- Interrupt support in QBMAN
- Implemetation of eventdev API's for DPAA2
- Documentation of DPAA2 eventdev PMD

Changes in v3:
  Rebased to dpdk master branch
  Fix shared library parallel build (patch 01/21)
  Fixed a checkpatch error

Changes in v2:
  Rebase to latest dpdk-next-eventdev
  Update to support only atomic and ordered queues
  Fix checkpatch errors

Hemant Agrawal (3):
  drivers: add bus dependency for event
  bus/fslmc: generic framework for mc object creation
  bus/fslmc: export qbman dqrr funcs for eventdev usages

Nipun Gupta (18):
  event/dpaa2: add basic build infrastructure
  bus/fslmc: integrating dpio and dpbp to object framework
  bus/fslmc: adding basic dpcon support
  event/dpaa2: register dpcon as dpaa2 device for bus scan
  bus/fslmc: adding basic dpci support
  bus/fslmc: register dpci as dpaa2 device for bus scan
  bus/fslmc: adding cpu support in stashing config
  event/dpaa2: add initialization of event device
  bus/fslmc: add support for static dequeue from portal
  event/dpaa2: add configuration functions
  bus/fslmc: support enqueue with multiple enqueue descriptors
  bus/fslmc: add callback per queue to enable
  bus/fslmc: change func argument to const to avoid warning
  event/dpaa2: add enqueue and dequeue functionality
  fslmc/bus: add interrupt enabling routine
  bus/fslmc: enable portal interrupt handling
  event/dpaa2: handle timeout using interrupts in dequeue
  doc: add NXP DPAA2 EVENTDEV details

 MAINTAINERS|   5 +
 config/defconfig_arm64-dpaa2-linuxapp-gcc  |   6 +
 doc/guides/eventdevs/dpaa2.rst | 175 ++
 drivers/Makefile   |   1 +
 drivers/bus/fslmc/Makefile |   5 +-
 drivers/bus/fslmc/fslmc_vfio.c |  69 +-
 drivers/bus/fslmc/fslmc_vfio.h |  47 +-
 drivers/bus/fslmc/mc/dpci.c| 307 +
 drivers/bus/fslmc/mc/dpcon.c   | 230 +++
 drivers/bus/fslmc/mc/dpio.c|  44 ++
 drivers/bus/fslmc/mc/fsl_dpci.h| 404 
 drivers/bus/fslmc/mc/fsl_dpci_cmd.h| 147 +
 drivers/bus/fslmc/mc/fsl_dpcon.h   | 238 +++
 drivers/bus/fslmc/mc/fsl_dpcon_cmd.h   | 175 ++
 drivers/bus/fslmc/mc/fsl_dpio.h|  30 +
 drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c   |  14 +-
 drivers/bus/fslmc/portal/dpaa2_hw_dpci.c   | 179 ++
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c   | 133 +++-
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h   |   5 +
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h|  28 +-
 drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h |  46 +-
 drivers/bus/fslmc/qbman/qbman_portal.c |  94 ++-
 drivers/bus/fslmc/rte_bus_fslmc_version.map|  18 +
 drivers/event/Makefile |   2 +
 drivers/event/dpaa2/Makefile   |  60 ++
 drivers/event/dpaa2/dpaa2_eventdev.c   | 691 +
 drivers/event/dpaa2/dpaa2_eventdev.h   | 114 
 drivers/event/dpaa2/dpaa2_hw_dpcon.c   | 139 +
 .../event/dpaa2/rte_pmd_dpaa2_event_version.map|   3 +
 mk/rte.app.mk  |   1 +
 30 files changed, 3369 insertions(+), 41 deletions(-)
 create mode 100644 doc/guides/eventdevs/dpaa2.rst
 create mode 100644 drivers/bus/fslmc/mc/dpci.c
 create mode 100644 drivers/bus/fslmc/mc/dpcon.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpci.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpci_cmd.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpcon.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpcon_cmd.h
 create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_dpci.c
 create mode 100644 drivers/event/dpaa2/Makefile
 create mode 100644 drivers/event/dpaa2/dpaa2_eventdev.c
 create mode 100644 drivers/event/dpaa2/dpaa2_eventdev.h
 create mode 100644 drivers/event/dpaa2/dpaa2_hw_dpcon.c
 create mode 100644 drivers/event/dpaa2/rte_pmd_dpaa2_event_version.map

-- 
1.9.1

[dpdk-dev] [PATCH 03/21 v3] bus/fslmc: generic framework for mc object creation

[Nipun Gupta]

From: Hemant Agrawal 

There are muliple help mc object, which are not an independent
device, but they are required for dpaa2 based devices.
This framework allows registration and handling of all such
mc devices.

Signed-off-by: Hemant Agrawal 
Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/fslmc_vfio.c  | 11 +
 drivers/bus/fslmc/fslmc_vfio.h  | 37 +
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  1 +
 3 files changed, 49 insertions(+)

diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 8471a9a..b9dd2a9 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -80,6 +80,17 @@
 void *(*rte_mcp_ptr_list);
 static uint32_t mcp_id;
 static int is_dma_done;
+static struct rte_fslmc_object_list fslmc_obj_list =
+   TAILQ_HEAD_INITIALIZER(fslmc_obj_list);
+
+/*register a fslmc bus based dpaa2 driver */
+void
+rte_fslmc_object_register(struct rte_dpaa2_object *object)
+{
+   RTE_VERIFY(object);
+
+   TAILQ_INSERT_TAIL(&fslmc_obj_list, object, next);
+}
 
 static int vfio_connect_container(struct fslmc_vfio_group *vfio_group)
 {
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index 53dd0b7..58b3acd 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -63,6 +63,24 @@
struct fslmc_vfio_group *group_list[VFIO_MAX_GRP];
 } fslmc_vfio_container;
 
+struct rte_dpaa2_object;
+
+TAILQ_HEAD(rte_fslmc_object_list, rte_dpaa2_object);
+
+typedef int (*rte_fslmc_obj_create_t)(struct fslmc_vfio_device *vdev,
+struct vfio_device_info *obj_info,
+int object_id);
+
+/**
+ * A structure describing a DPAA2 driver.
+ */
+struct rte_dpaa2_object {
+   TAILQ_ENTRY(rte_dpaa2_object) next; /**< Next in list. */
+   const char *name;/**< Name of Object. */
+   uint16_t object_id; /**< DPAA2 Object ID */
+   rte_fslmc_obj_create_t create;
+};
+
 int vfio_dmamap_mem_region(
uint64_t vaddr,
uint64_t iova,
@@ -79,4 +97,23 @@ int dpaa2_create_dpio_device(struct fslmc_vfio_device *vdev,
 
 int dpaa2_create_dpbp_device(int dpbp_id);
 
+/**
+ * Register a DPAA2 MC Object driver.
+ *
+ * @param mc_object
+ *   A pointer to a rte_dpaa_object structure describing the mc object
+ *   to be registered.
+ */
+void rte_fslmc_object_register(struct rte_dpaa2_object *object);
+
+/** Helper for DPAA2 object registration */
+#define RTE_PMD_REGISTER_DPAA2_OBJECT(nm, dpaa2_obj) \
+RTE_INIT(dpaa2objinitfn_ ##nm); \
+static void dpaa2objinitfn_ ##nm(void) \
+{\
+   (dpaa2_obj).name = RTE_STR(nm);\
+   rte_fslmc_object_register(&dpaa2_obj); \
+} \
+RTE_PMD_EXPORT_NAME(nm, __COUNTER__)
+
 #endif /* _FSLMC_VFIO_H_ */
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index 43f3354..90db78c 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -56,6 +56,7 @@ DPDK_17.08 {
mc_get_soc_version;
mc_get_version;
qbman_result_SCN_state_in_mem;
+   rte_fslmc_object_register;
rte_global_active_dqs_list;
 
 } DPDK_17.05;
-- 
1.9.1

[dpdk-dev] [PATCH 02/21 v3] event/dpaa2: add basic build infrastructure

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
Acked-by: Jerin Jacob 
---
 MAINTAINERS|  4 ++
 config/defconfig_arm64-dpaa2-linuxapp-gcc  |  6 ++
 drivers/event/Makefile |  2 +
 drivers/event/dpaa2/Makefile   | 52 +++
 drivers/event/dpaa2/dpaa2_eventdev.c   | 74 ++
 drivers/event/dpaa2/dpaa2_eventdev.h   | 51 +++
 .../event/dpaa2/rte_pmd_dpaa2_event_version.map|  3 +
 mk/rte.app.mk  |  1 +
 8 files changed, 193 insertions(+)
 create mode 100644 drivers/event/dpaa2/Makefile
 create mode 100644 drivers/event/dpaa2/dpaa2_eventdev.c
 create mode 100644 drivers/event/dpaa2/dpaa2_eventdev.h
 create mode 100644 drivers/event/dpaa2/rte_pmd_dpaa2_event_version.map

diff --git a/MAINTAINERS b/MAINTAINERS
index f6095ef..fe1a25b 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -580,6 +580,10 @@ F: drivers/event/sw/
 F: test/test/test_eventdev_sw.c
 F: doc/guides/eventdevs/sw.rst
 
+NXP DPAA2 Eventdev PMD
+M: Hemant Agrawal 
+M: Nipun Gupta 
+F: drivers/event/dpaa2/
 
 Packet processing
 -
diff --git a/config/defconfig_arm64-dpaa2-linuxapp-gcc 
b/config/defconfig_arm64-dpaa2-linuxapp-gcc
index 2304ab6..45e1eb5 100644
--- a/config/defconfig_arm64-dpaa2-linuxapp-gcc
+++ b/config/defconfig_arm64-dpaa2-linuxapp-gcc
@@ -80,3 +80,9 @@ CONFIG_RTE_LIBRTE_DPAA2_SEC_DEBUG_RX=n
 # on a single DPAA2 SEC device.
 #
 CONFIG_RTE_DPAA2_SEC_PMD_MAX_NB_SESSIONS=2048
+
+#
+# Compile schedule-oriented NXP DPAA2 EVENTDEV driver
+#
+CONFIG_RTE_LIBRTE_PMD_DPAA2_EVENTDEV=y
+CONFIG_RTE_LIBRTE_PMD_DPAA2_EVENTDEV_DEBUG=n
diff --git a/drivers/event/Makefile b/drivers/event/Makefile
index 1cf389e..0f12cc9 100644
--- a/drivers/event/Makefile
+++ b/drivers/event/Makefile
@@ -39,5 +39,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV) += sw
 DEPDIRS-sw = $(core-libs) librte_kvargs librte_ring
 DIRS-$(CONFIG_RTE_LIBRTE_PMD_OCTEONTX_SSOVF) += octeontx
 DEPDIRS-octeontx = $(core-libs)
+DIRS-$(CONFIG_RTE_LIBRTE_PMD_DPAA2_EVENTDEV) += dpaa2
+DEPDIRS-dpaa2 = $(core-libs) librte_bus_fslmc
 
 include $(RTE_SDK)/mk/rte.subdir.mk
diff --git a/drivers/event/dpaa2/Makefile b/drivers/event/dpaa2/Makefile
new file mode 100644
index 000..8a021ca
--- /dev/null
+++ b/drivers/event/dpaa2/Makefile
@@ -0,0 +1,52 @@
+#   BSD LICENSE
+#
+#   Copyright(c) 2017 NXP.
+#
+#   Redistribution and use in source and binary forms, with or without
+#   modification, are permitted provided that the following conditions
+#   are met:
+#
+# * Redistributions of source code must retain the above copyright
+#   notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above copyright
+#   notice, this list of conditions and the following disclaimer in
+#   the documentation and/or other materials provided with the
+#   distribution.
+# * Neither the name of NXP nor the names of its
+#   contributors may be used to endorse or promote products derived
+#   from this software without specific prior written permission.
+#
+#   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+#   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+#   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+#   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+#   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+#   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+#   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+#   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+#   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+#   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_pmd_dpaa2_event.a
+
+CFLAGS += $(WERROR_FLAGS)
+
+CFLAGS += -I$(RTE_SDK)/drivers/event/dpaa2
+# versioning export map
+EXPORT_MAP := rte_pmd_dpaa2_event_version.map
+
+LIBABIVER := 1
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-$(CONFIG_RTE_LIBRTE_PMD_DPAA2_EVENTDEV) += dpaa2_eventdev.c
+
+include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/event/dpaa2/dpaa2_eventdev.c 
b/drivers/event/dpaa2/dpaa2_eventdev.c
new file mode 100644
index 000..191901e
--- /dev/null
+++ b/drivers/event/dpaa2/dpaa2_eventdev.c
@@ -0,0 +1,74 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright 2017 NXP.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions

[dpdk-dev] [PATCH 04/21 v3] bus/fslmc: integrating dpio and dpbp to object framework

[Nipun Gupta]

This patch removes the existing static call for dpio and dpbp
create and add them to object registration framework.

This patch also changes the vfio mc object processing to use
the framework.

Signed-off-by: Hemant Agrawal 
Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/fslmc_vfio.c   | 24 ++--
 drivers/bus/fslmc/fslmc_vfio.h   |  9 ++---
 drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c | 14 +++---
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c | 11 +--
 4 files changed, 32 insertions(+), 26 deletions(-)

diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index b9dd2a9..6ebf779 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -374,7 +374,6 @@ int fslmc_vfio_process_group(void)
char path[PATH_MAX];
int64_t v_addr;
int ndev_count;
-   int dpio_count = 0, dpbp_count = 0;
struct fslmc_vfio_group *group = &vfio_groups[0];
static int process_once;
 
@@ -513,18 +512,17 @@ int fslmc_vfio_process_group(void)
  object_type, object_id);
 
fslmc_bus_add_device(dev);
-   }
-   if (!strcmp(object_type, "dpio")) {
-   ret = dpaa2_create_dpio_device(vdev,
-  &device_info,
+   } else {
+   /* Parse all other objects */
+   struct rte_dpaa2_object *object;
+
+   TAILQ_FOREACH(object, &fslmc_obj_list, next) {
+   if (!strcmp(object_type, object->name))
+   object->create(vdev, &device_info,
   object_id);
-   if (!ret)
-   dpio_count++;
-   }
-   if (!strcmp(object_type, "dpbp")) {
-   ret = dpaa2_create_dpbp_device(object_id);
-   if (!ret)
-   dpbp_count++;
+   else
+   continue;
+   }
}
}
closedir(d);
@@ -533,8 +531,6 @@ int fslmc_vfio_process_group(void)
if (ret)
FSLMC_VFIO_LOG(DEBUG, "Error in affining qbman swp %d", ret);
 
-   FSLMC_VFIO_LOG(DEBUG, "DPAA2: Added dpbp_count = %d dpio_count=%d",
- dpbp_count, dpio_count);
return 0;
 
 FAILURE:
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index 58b3acd..ffed62e 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -39,6 +39,8 @@
 #define DPAA2_VENDOR_ID0x1957
 #define DPAA2_MC_DPNI_DEVID7
 #define DPAA2_MC_DPSECI_DEVID  3
+#define DPAA2_MC_DPIO_DEVID9
+#define DPAA2_MC_DPBP_DEVID10
 
 #define VFIO_MAX_GRP 1
 
@@ -90,13 +92,6 @@ int vfio_dmamap_mem_region(
 int fslmc_vfio_process_group(void);
 int rte_fslmc_vfio_dmamap(void);
 
-/* create dpio device */
-int dpaa2_create_dpio_device(struct fslmc_vfio_device *vdev,
-struct vfio_device_info *obj_info,
-int object_id);
-
-int dpaa2_create_dpbp_device(int dpbp_id);
-
 /**
  * Register a DPAA2 MC Object driver.
  *
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c
index a665ec5..b55335a 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpbp.c
@@ -57,9 +57,10 @@
 static struct dpbp_dev_list dpbp_dev_list
= TAILQ_HEAD_INITIALIZER(dpbp_dev_list); /*!< DPBP device list */
 
-int
-dpaa2_create_dpbp_device(
-   int dpbp_id)
+static int
+dpaa2_create_dpbp_device(struct fslmc_vfio_device *vdev __rte_unused,
+struct vfio_device_info *obj_info __rte_unused,
+int dpbp_id)
 {
struct dpaa2_dpbp_dev *dpbp_node;
int ret;
@@ -127,3 +128,10 @@ void dpaa2_free_dpbp_dev(struct dpaa2_dpbp_dev *dpbp)
}
}
 }
+
+static struct rte_dpaa2_object rte_dpaa2_dpbp_obj = {
+   .object_id = DPAA2_MC_DPBP_DEVID,
+   .create = dpaa2_create_dpbp_device,
+};
+
+RTE_PMD_REGISTER_DPAA2_OBJECT(dpbp, rte_dpaa2_dpbp_obj);
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
index 3213237..730555f 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -334,10 +334,10 @@ static inline struct dpaa2_dpio_dev 
*dpaa2_get_qbman_swp(void)
}
 }
 
-int
+static int
 dpaa2_create_dpio_device(struct fslmc_vfio_device *vdev,
 struct vfio_device_info *obj_info,
-   int object_id)
+int object_id)
 {
struct dpaa2_dpio_dev *dpio_dev;
struct vfio_region_info reg_info =

[dpdk-dev] [PATCH 05/21 v3] bus/fslmc: adding basic dpcon support

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/Makefile  |   3 +-
 drivers/bus/fslmc/fslmc_vfio.h  |   1 +
 drivers/bus/fslmc/mc/dpcon.c| 230 +++
 drivers/bus/fslmc/mc/fsl_dpcon.h| 238 
 drivers/bus/fslmc/mc/fsl_dpcon_cmd.h| 175 
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   2 +
 6 files changed, 648 insertions(+), 1 deletion(-)
 create mode 100644 drivers/bus/fslmc/mc/dpcon.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpcon.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpcon_cmd.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 7d9f4b6..38d884e 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -66,7 +66,8 @@ SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
mc/dpmng.c \
 mc/dpbp.c \
 mc/dpio.c \
-mc/mc_sys.c
+mc/mc_sys.c \
+   mc/dpcon.c \
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpbp.c
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index ffed62e..eddce31 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -39,6 +39,7 @@
 #define DPAA2_VENDOR_ID0x1957
 #define DPAA2_MC_DPNI_DEVID7
 #define DPAA2_MC_DPSECI_DEVID  3
+#define DPAA2_MC_DPCON_DEVID   5
 #define DPAA2_MC_DPIO_DEVID9
 #define DPAA2_MC_DPBP_DEVID10
 
diff --git a/drivers/bus/fslmc/mc/dpcon.c b/drivers/bus/fslmc/mc/dpcon.c
new file mode 100644
index 000..b078dff
--- /dev/null
+++ b/drivers/bus/fslmc/mc/dpcon.c
@@ -0,0 +1,230 @@
+/* Copyright 2013-2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#include 
+#include 
+#include 
+#include 
+
+int dpcon_open(struct fsl_mc_io *mc_io,
+  uint32_t cmd_flags,
+  int dpcon_id,
+  uint16_t *token)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPCON_CMDID_OPEN,
+ cmd_flags,
+ 0);
+   DPCON_CMD_OPEN(cmd, dpcon_id);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   *token = MC_CMD_HDR_READ_TOKEN(cmd.header);
+
+   return 0;
+}
+
+int dpcon_close(struct fsl_mc_io *mc_io,
+   uint32_t cmd_flags,
+   uint16_t token)
+{
+   struct mc_command cmd = { 0 };
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPCON_CMDID_CLOSE,
+ cmd_flags,
+ token);
+
+   /* send command to mc*/
+   return mc_send_command(mc_io, &cmd);
+}
+
+int dpcon_create(struct fsl_mc_io *mc_io,
+uint16_t dprc_token,
+   uint32_t cmd_flags,
+   const struct dpcon_cfg *cfg,
+   uint32_t *obj_id)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.

[dpdk-dev] [PATCH 06/21 v3] bus/fslmc: export qbman dqrr funcs for eventdev usages

[Nipun Gupta]

From: Hemant Agrawal 

Signed-off-by: Hemant Agrawal 
---
 drivers/bus/fslmc/rte_bus_fslmc_version.map | 8 
 1 file changed, 8 insertions(+)

diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index 783c3e5..01a7939 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -57,7 +57,15 @@ DPDK_17.08 {
dpcon_get_attributes;
mc_get_soc_version;
mc_get_version;
+   qbman_eq_desc_set_dca;
+   qbman_get_dqrr_from_idx;
+   qbman_get_dqrr_idx;
+   qbman_result_DQ_fqd_ctx;
qbman_result_SCN_state_in_mem;
+   qbman_swp_dqrr_consume;
+   qbman_swp_dqrr_next;
+   qbman_swp_push_set;
+   rte_dpaa2_alloc_dpci_dev;
rte_fslmc_object_register;
rte_global_active_dqs_list;
 
-- 
1.9.1

[dpdk-dev] [PATCH 07/21 v3] event/dpaa2: register dpcon as dpaa2 device for bus scan

[Nipun Gupta]

Registering dpcon as dpaa2 type device handling initialization,
allocation and freeing of the device

Signed-off-by: Nipun Gupta 
---
 drivers/event/dpaa2/Makefile |   8 ++
 drivers/event/dpaa2/dpaa2_eventdev.h |  18 +
 drivers/event/dpaa2/dpaa2_hw_dpcon.c | 139 +++
 3 files changed, 165 insertions(+)
 create mode 100644 drivers/event/dpaa2/dpaa2_hw_dpcon.c

diff --git a/drivers/event/dpaa2/Makefile b/drivers/event/dpaa2/Makefile
index 8a021ca..0566643 100644
--- a/drivers/event/dpaa2/Makefile
+++ b/drivers/event/dpaa2/Makefile
@@ -38,7 +38,14 @@ LIB = librte_pmd_dpaa2_event.a
 
 CFLAGS += $(WERROR_FLAGS)
 
+CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc
+CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/qbman/include
+CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/mc
+CFLAGS += -I$(RTE_SDK)/drivers/bus/fslmc/portal
+CFLAGS += -I$(RTE_SDK)/drivers/mempool/dpaa2
 CFLAGS += -I$(RTE_SDK)/drivers/event/dpaa2
+CFLAGS += -I$(RTE_SDK)/lib/librte_eal/linuxapp/eal
+
 # versioning export map
 EXPORT_MAP := rte_pmd_dpaa2_event_version.map
 
@@ -47,6 +54,7 @@ LIBABIVER := 1
 #
 # all source are stored in SRCS-y
 #
+SRCS-$(CONFIG_RTE_LIBRTE_PMD_DPAA2_EVENTDEV) += dpaa2_hw_dpcon.c
 SRCS-$(CONFIG_RTE_LIBRTE_PMD_DPAA2_EVENTDEV) += dpaa2_eventdev.c
 
 include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/event/dpaa2/dpaa2_eventdev.h 
b/drivers/event/dpaa2/dpaa2_eventdev.h
index 959f443..b151502 100644
--- a/drivers/event/dpaa2/dpaa2_eventdev.h
+++ b/drivers/event/dpaa2/dpaa2_eventdev.h
@@ -34,6 +34,10 @@
 #define __DPAA2_EVENTDEV_H__
 
 #include 
+#include 
+#include 
+#include 
+
 #define EVENTDEV_NAME_DPAA2_PMDevent_dpaa2
 
 #ifdef RTE_LIBRTE_PMD_DPAA2_EVENTDEV_DEBUG
@@ -48,4 +52,18 @@
 #define PMD_DRV_ERR(fmt, args...) \
RTE_LOG(ERR, PMD, "%s(): " fmt "\n", __func__, ## args)
 
+struct dpaa2_dpcon_dev {
+   TAILQ_ENTRY(dpaa2_dpcon_dev) next;
+   struct fsl_mc_io dpcon;
+   uint16_t token;
+   rte_atomic16_t in_use;
+   uint32_t dpcon_id;
+   uint16_t qbman_ch_id;
+   uint8_t num_priorities;
+   uint8_t channel_index;
+};
+
+struct dpaa2_dpcon_dev *rte_dpaa2_alloc_dpcon_dev(void);
+void rte_dpaa2_free_dpcon_dev(struct dpaa2_dpcon_dev *dpcon);
+
 #endif /* __DPAA2_EVENTDEV_H__ */
diff --git a/drivers/event/dpaa2/dpaa2_hw_dpcon.c 
b/drivers/event/dpaa2/dpaa2_hw_dpcon.c
new file mode 100644
index 000..27f5bcb
--- /dev/null
+++ b/drivers/event/dpaa2/dpaa2_hw_dpcon.c
@@ -0,0 +1,139 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright 2017 NXP.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of NXP nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include "dpaa2_eventdev.h"
+
+TAILQ_HEAD(dpcon_dev_list, dpaa2_dpcon_dev);
+static struct dpcon_dev_list dpcon_dev_list
+   = TAILQ_HEAD_INITIALIZER(dpcon_dev_list); /*!< DPCON device list */
+
+static int
+rte_dpaa2_create_dpcon_device(struct fslmc_vfio_device *vdev __rte_unused,
+ struct vfio_device_info *obj_info __rte_unused,
+int dpcon_id)
+{
+   struct dpaa2_dpcon_dev *dpcon_node;
+   struct dpcon_attr attr;
+   int ret;
+
+   /* Allocate DPAA2 dpcon handle */
+   dpcon_node = rte_malloc(NULL, sizeof(struct dpaa2_dpcon_dev), 0);
+   if (!dpcon_node) {
+   PMD_DRV_LOG(

[dpdk-dev] [PATCH 08/21 v3] bus/fslmc: adding basic dpci support

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/Makefile  |   1 +
 drivers/bus/fslmc/mc/dpci.c | 307 +
 drivers/bus/fslmc/mc/fsl_dpci.h | 404 
 drivers/bus/fslmc/mc/fsl_dpci_cmd.h | 147 ++
 drivers/bus/fslmc/rte_bus_fslmc_version.map |   1 +
 5 files changed, 860 insertions(+)
 create mode 100644 drivers/bus/fslmc/mc/dpci.c
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpci.h
 create mode 100644 drivers/bus/fslmc/mc/fsl_dpci_cmd.h

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 38d884e..4884d87 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -68,6 +68,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 mc/dpio.c \
 mc/mc_sys.c \
mc/dpcon.c \
+   mc/dpci.c
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpbp.c
diff --git a/drivers/bus/fslmc/mc/dpci.c b/drivers/bus/fslmc/mc/dpci.c
new file mode 100644
index 000..0ea7837
--- /dev/null
+++ b/drivers/bus/fslmc/mc/dpci.c
@@ -0,0 +1,307 @@
+/*-
+ * This file is provided under a dual BSD/GPLv2 license. When using or
+ * redistributing this file, you may do so under either license.
+ *
+ *   BSD LICENSE
+ *
+ * Copyright 2013-2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of the above-listed copyright holders nor the
+ * names of any contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *   GPL LICENSE SUMMARY
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#include 
+#include 
+#include 
+#include 
+
+int dpci_open(struct fsl_mc_io *mc_io,
+ uint32_t cmd_flags,
+ int dpci_id,
+ uint16_t *token)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPCI_CMDID_OPEN,
+ cmd_flags,
+ 0);
+   DPCI_CMD_OPEN(cmd, dpci_id);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   *token = MC_CMD_HDR_READ_TOKEN(cmd.header);
+
+   return 0;
+}
+
+int dpci_close(struct fsl_mc_io *mc_io,
+  uint32_t cmd_flags,
+  uint16_t token)
+{
+   struct mc_command cmd = { 0 };
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPCI_CMDID_CLOSE,
+ cmd_flags,
+ token);
+
+   /* send command to mc*/
+   return mc_send_command(mc_io, &cmd);
+}
+
+int dpci_create(struct fsl_mc_io *mc_io,
+   uint16_t dprc_token,
+   uint32_t cmd_flags,
+   const struct dpci_cfg *cfg,
+   uint32_t *obj_id)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPCI_CMDID_CREATE,
+ cmd_flags,
+ dprc_token);
+   DPCI_CMD_CREATE(cmd, cfg);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   CMD_CRE

[dpdk-dev] [PATCH 09/21 v3] bus/fslmc: register dpci as dpaa2 device for bus scan

[Nipun Gupta]

Registering dpci as dpaa2 type device handling initialization,
allocation and freeing of the device

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/Makefile   |   1 +
 drivers/bus/fslmc/fslmc_vfio.h   |   1 +
 drivers/bus/fslmc/portal/dpaa2_hw_dpci.c | 179 +++
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h  |  15 +++
 4 files changed, 196 insertions(+)
 create mode 100644 drivers/bus/fslmc/portal/dpaa2_hw_dpci.c

diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index 4884d87..a156847 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -72,6 +72,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
 
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpbp.c
+SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += portal/dpaa2_hw_dpci.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_vfio.c
 SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc_bus.c
 
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index eddce31..7c725f4 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -42,6 +42,7 @@
 #define DPAA2_MC_DPCON_DEVID   5
 #define DPAA2_MC_DPIO_DEVID9
 #define DPAA2_MC_DPBP_DEVID10
+#define DPAA2_MC_DPCI_DEVID11
 
 #define VFIO_MAX_GRP 1
 
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpci.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpci.c
new file mode 100644
index 000..d222f26
--- /dev/null
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpci.c
@@ -0,0 +1,179 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright 2017 NXP.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ *   notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ *   notice, this list of conditions and the following disclaimer in
+ *   the documentation and/or other materials provided with the
+ *   distribution.
+ * * Neither the name of Freescale Semiconductor, Inc nor the names of its
+ *   contributors may be used to endorse or promote products derived
+ *   from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include "portal/dpaa2_hw_pvt.h"
+#include "portal/dpaa2_hw_dpio.h"
+
+TAILQ_HEAD(dpci_dev_list, dpaa2_dpci_dev);
+static struct dpci_dev_list dpci_dev_list
+   = TAILQ_HEAD_INITIALIZER(dpci_dev_list); /*!< DPCI device list */
+
+static int
+rte_dpaa2_create_dpci_device(struct fslmc_vfio_device *vdev __rte_unused,
+struct vfio_device_info *obj_info __rte_unused,
+   int dpci_id)
+{
+   struct dpaa2_dpci_dev *dpci_node;
+   struct dpci_attr attr;
+   struct dpci_rx_queue_cfg rx_queue_cfg;
+   struct dpci_rx_queue_attr rx_attr;
+   int ret, i;
+
+   /* Allocate DPAA2 dpci handle */
+   dpci_node = rte_malloc(NULL, sizeof(struct dpaa2_dpci_dev), 0);
+   if (!dpci_node) {
+   PMD_INIT_LOG(ERR, "Memory allocation failed for DPCI Device");
+   return -1;
+   }
+
+   /* Open the dpci object */
+   dpci_node->dpci.regs = rte_mcp_ptr_list[MC_PORTAL_INDEX];
+   ret = dpci_open(&dpci_node->dpci,
+   CMD_PRI_LOW, dpci_id, &dpci_node->token);
+   if (ret) {
+   PMD_INIT_LOG(ERR, "Resource alloc failure with err code: %d",
+ret);
+   rte_free(dpci_node);
+   return -1;
+   }
+
+   /* Get the device attributes */
+   ret = dpci_get_attributes(&dpci_node->dpci,
+ CMD_PRI_LOW, dpci_node->token, &attr);
+   if (ret != 0) {
+   PMD_INIT_LOG(ERR, "Reading device failed with err code: %d"

[dpdk-dev] [PATCH 10/21 v3] bus/fslmc: adding cpu support in stashing config

[Nipun Gupta]

Stashing can also be configured by other drivers (for instance
event driver) passing cpu_id as an argument. This change
facilitates the same.

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c| 14 ++
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h|  4 
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  2 ++
 3 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
index 730555f..63378f0 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -172,10 +172,9 @@
 }
 
 static int
-dpaa2_configure_stashing(struct dpaa2_dpio_dev *dpio_dev)
+dpaa2_configure_stashing(struct dpaa2_dpio_dev *dpio_dev, int cpu_id)
 {
-   int sdest;
-   int cpu_id, ret;
+   int sdest, ret;
static int first_time;
 
/* find the SoC type for the first time */
@@ -194,7 +193,6 @@
}
 
/* Set the Stashing Destination */
-   cpu_id = rte_lcore_id();
if (cpu_id < 0) {
cpu_id = rte_get_master_lcore();
if (cpu_id < 0) {
@@ -220,7 +218,7 @@
return 0;
 }
 
-static inline struct dpaa2_dpio_dev *dpaa2_get_qbman_swp(void)
+struct dpaa2_dpio_dev *dpaa2_get_qbman_swp(int cpu_id)
 {
struct dpaa2_dpio_dev *dpio_dev = NULL;
int ret;
@@ -236,7 +234,7 @@ static inline struct dpaa2_dpio_dev 
*dpaa2_get_qbman_swp(void)
PMD_DRV_LOG(DEBUG, "New Portal=0x%x (%d) affined thread - %lu",
dpio_dev, dpio_dev->index, syscall(SYS_gettid));
 
-   ret = dpaa2_configure_stashing(dpio_dev);
+   ret = dpaa2_configure_stashing(dpio_dev, cpu_id);
if (ret)
PMD_DRV_LOG(ERR, "dpaa2_configure_stashing failed");
 
@@ -276,7 +274,7 @@ static inline struct dpaa2_dpio_dev 
*dpaa2_get_qbman_swp(void)
}
 
/* Populate the dpaa2_io_portal structure */
-   dpaa2_io_portal[lcore_id].dpio_dev = dpaa2_get_qbman_swp();
+   dpaa2_io_portal[lcore_id].dpio_dev = dpaa2_get_qbman_swp(lcore_id);
 
if (dpaa2_io_portal[lcore_id].dpio_dev) {
RTE_PER_LCORE(_dpaa2_io).dpio_dev
@@ -322,7 +320,7 @@ static inline struct dpaa2_dpio_dev 
*dpaa2_get_qbman_swp(void)
}
 
/* Populate the dpaa2_io_portal structure */
-   dpaa2_io_portal[lcore_id].sec_dpio_dev = dpaa2_get_qbman_swp();
+   dpaa2_io_portal[lcore_id].sec_dpio_dev = dpaa2_get_qbman_swp(lcore_id);
 
if (dpaa2_io_portal[lcore_id].sec_dpio_dev) {
RTE_PER_LCORE(_dpaa2_io).sec_dpio_dev
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
index f2e1168..4269800 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
@@ -53,6 +53,10 @@ struct dpaa2_io_portal_t {
 #define DPAA2_PER_LCORE_SEC_DPIO RTE_PER_LCORE(_dpaa2_io).sec_dpio_dev
 #define DPAA2_PER_LCORE_SEC_PORTAL DPAA2_PER_LCORE_SEC_DPIO->sw_portal
 
+extern struct dpaa2_io_portal_t dpaa2_io_portal[RTE_MAX_LCORE];
+
+struct dpaa2_dpio_dev *dpaa2_get_qbman_swp(int cpu_id);
+
 /* Affine a DPIO portal to current processing thread */
 int dpaa2_affine_qbman_swp(void);
 
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index cf7c0ee..4509051 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -53,6 +53,8 @@ DPDK_17.05 {
 DPDK_17.08 {
global:
 
+   dpaa2_io_portal;
+   dpaa2_get_qbman_swp;
dpci_set_rx_queue;
dpcon_open;
dpcon_get_attributes;
-- 
1.9.1

[dpdk-dev] [PATCH 11/21 v3] event/dpaa2: add initialization of event device

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 drivers/event/dpaa2/dpaa2_eventdev.c | 157 ++-
 drivers/event/dpaa2/dpaa2_eventdev.h |  23 +
 2 files changed, 176 insertions(+), 4 deletions(-)

diff --git a/drivers/event/dpaa2/dpaa2_eventdev.c 
b/drivers/event/dpaa2/dpaa2_eventdev.c
index 191901e..b8cc3f8 100644
--- a/drivers/event/dpaa2/dpaa2_eventdev.c
+++ b/drivers/event/dpaa2/dpaa2_eventdev.c
@@ -30,17 +30,168 @@
  *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+
+#include 
+#include 
+#include 
+#include 
+#include 
 #include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
+#include 
 #include 
 
+#include 
+#include 
+#include 
+#include 
 #include "dpaa2_eventdev.h"
+#include 
+#include 
+
+/* Clarifications
+ * Evendev = SoC Instance
+ * Eventport = DPIO Instance
+ * Eventqueue = DPCON Instance
+ * 1 Eventdev can have N Eventqueue
+ * Soft Event Flow is DPCI Instance
+ */
+
+static uint16_t
+dpaa2_eventdev_enqueue_burst(void *port, const struct rte_event ev[],
+uint16_t nb_events)
+{
+   RTE_SET_USED(port);
+   RTE_SET_USED(ev);
+   RTE_SET_USED(nb_events);
+
+   return 0;
+}
+
+static uint16_t
+dpaa2_eventdev_enqueue(void *port, const struct rte_event *ev)
+{
+   return dpaa2_eventdev_enqueue_burst(port, ev, 1);
+}
+
+static uint16_t
+dpaa2_eventdev_dequeue_burst(void *port, struct rte_event ev[],
+uint16_t nb_events, uint64_t timeout_ticks)
+{
+   RTE_SET_USED(port);
+   RTE_SET_USED(ev);
+   RTE_SET_USED(nb_events);
+   RTE_SET_USED(timeout_ticks);
+
+   return 0;
+}
+
+static uint16_t
+dpaa2_eventdev_dequeue(void *port, struct rte_event *ev,
+  uint64_t timeout_ticks)
+{
+   return dpaa2_eventdev_dequeue_burst(port, ev, 1, timeout_ticks);
+}
+
+static const struct rte_eventdev_ops dpaa2_eventdev_ops;
+
+static int
+dpaa2_eventdev_setup_dpci(struct dpaa2_dpci_dev *dpci_dev,
+ struct dpaa2_dpcon_dev *dpcon_dev)
+{
+   struct dpci_rx_queue_cfg rx_queue_cfg;
+   int ret, i;
+
+   /*Do settings to get the frame on a DPCON object*/
+   rx_queue_cfg.options = DPCI_QUEUE_OPT_DEST;
+   rx_queue_cfg.dest_cfg.dest_type = DPCI_DEST_DPCON;
+   rx_queue_cfg.dest_cfg.dest_id = dpcon_dev->dpcon_id;
+   rx_queue_cfg.dest_cfg.priority = DPAA2_EVENT_DEFAULT_DPCI_PRIO;
+
+   for (i = 0 ; i < DPAA2_EVENT_DPCI_MAX_QUEUES; i++) {
+   rx_queue_cfg.user_ctx = (uint64_t)(&dpci_dev->queue[i]);
+   ret = dpci_set_rx_queue(&dpci_dev->dpci,
+   CMD_PRI_LOW,
+   dpci_dev->token, i,
+   &rx_queue_cfg);
+   if (ret) {
+   PMD_DRV_LOG(ERR, PMD,
+   "set_rx_q failed with err code: %d", ret);
+   return ret;
+   }
+   }
+   return 0;
+}
 
 static int
 dpaa2_eventdev_create(const char *name)
 {
-   RTE_SET_USED(name);
+   struct rte_eventdev *eventdev;
+   struct dpaa2_eventdev *priv;
+   struct dpaa2_dpcon_dev *dpcon_dev = NULL;
+   struct dpaa2_dpci_dev *dpci_dev = NULL;
+   int ret;
+
+   eventdev = rte_event_pmd_vdev_init(name,
+  sizeof(struct dpaa2_eventdev),
+  rte_socket_id());
+   if (eventdev == NULL) {
+   PMD_DRV_ERR("Failed to create eventdev vdev %s", name);
+   goto fail;
+   }
+
+   eventdev->dev_ops   = &dpaa2_eventdev_ops;
+   eventdev->schedule  = NULL;
+   eventdev->enqueue   = dpaa2_eventdev_enqueue;
+   eventdev->enqueue_burst = dpaa2_eventdev_enqueue_burst;
+   eventdev->dequeue   = dpaa2_eventdev_dequeue;
+   eventdev->dequeue_burst = dpaa2_eventdev_dequeue_burst;
+
+   /* For secondary processes, the primary has done all the work */
+   if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+   return 0;
+
+   priv = eventdev->data->dev_private;
+   priv->max_event_queues = 0;
+
+   do {
+   dpcon_dev = rte_dpaa2_alloc_dpcon_dev();
+   if (!dpcon_dev)
+   break;
+   priv->evq_info[priv->max_event_queues].dpcon = dpcon_dev;
+
+   dpci_dev = rte_dpaa2_alloc_dpci_dev();
+   if (!dpci_dev) {
+   rte_dpaa2_free_dpcon_dev(dpcon_dev);
+   break;
+   }
+   priv->evq_info[priv->max_event_queues].dpci = dpci_dev;
+
+   ret = dpaa2_eventdev_setup_dpci(dpci_dev, dpcon_dev);
+   if (ret) {
+   PMD_DRV_LOG(ERR, PMD,
+   "dpci setup failed wi

[dpdk-dev] [PATCH 12/21 v3] bus/fslmc: add support for static dequeue from portal

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/mc/dpio.c | 44 +
 drivers/bus/fslmc/mc/fsl_dpio.h | 30 
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.h|  1 +
 drivers/bus/fslmc/rte_bus_fslmc_version.map |  2 ++
 4 files changed, 77 insertions(+)

diff --git a/drivers/bus/fslmc/mc/dpio.c b/drivers/bus/fslmc/mc/dpio.c
index d84232a..3553952 100644
--- a/drivers/bus/fslmc/mc/dpio.c
+++ b/drivers/bus/fslmc/mc/dpio.c
@@ -257,6 +257,50 @@ int dpio_get_stashing_destination(struct fsl_mc_io *mc_io,
return 0;
 }
 
+int dpio_add_static_dequeue_channel(struct fsl_mc_io *mc_io,
+   uint32_t cmd_flags,
+   uint16_t token,
+   int dpcon_id,
+   uint8_t *channel_index)
+{
+   struct mc_command cmd = { 0 };
+   int err;
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(DPIO_CMDID_ADD_STATIC_DEQUEUE_CHANNEL,
+ cmd_flags,
+ token);
+   DPIO_CMD_ADD_STATIC_DEQUEUE_CHANNEL(cmd, dpcon_id);
+
+   /* send command to mc*/
+   err = mc_send_command(mc_io, &cmd);
+   if (err)
+   return err;
+
+   /* retrieve response parameters */
+   DPIO_RSP_ADD_STATIC_DEQUEUE_CHANNEL(cmd, *channel_index);
+
+   return 0;
+}
+
+int dpio_remove_static_dequeue_channel(struct fsl_mc_io *mc_io,
+  uint32_t cmd_flags,
+  uint16_t token,
+  int dpcon_id)
+{
+   struct mc_command cmd = { 0 };
+
+   /* prepare command */
+   cmd.header = mc_encode_cmd_header(
+   DPIO_CMDID_REMOVE_STATIC_DEQUEUE_CHANNEL,
+   cmd_flags,
+   token);
+   DPIO_CMD_REMOVE_STATIC_DEQUEUE_CHANNEL(cmd, dpcon_id);
+
+   /* send command to mc*/
+   return mc_send_command(mc_io, &cmd);
+}
+
 int dpio_get_api_version(struct fsl_mc_io *mc_io,
 uint32_t cmd_flags,
   uint16_t *major_ver,
diff --git a/drivers/bus/fslmc/mc/fsl_dpio.h b/drivers/bus/fslmc/mc/fsl_dpio.h
index 6d86f07..39b572d 100644
--- a/drivers/bus/fslmc/mc/fsl_dpio.h
+++ b/drivers/bus/fslmc/mc/fsl_dpio.h
@@ -230,6 +230,36 @@ int dpio_get_stashing_destination(struct fsl_mc_io *mc_io,
  uint8_t   *sdest);
 
 /**
+ * dpio_add_static_dequeue_channel() - Add a static dequeue channel.
+ * @mc_io: Pointer to MC portal's I/O object
+ * @cmd_flags: Command flags; one or more of 'MC_CMD_FLAG_'
+ * @token: Token of DPIO object
+ * @dpcon_id:  DPCON object ID
+ * @channel_index: Returned channel index to be used in qbman API
+ *
+ * Return: '0' on Success; Error code otherwise.
+ */
+int dpio_add_static_dequeue_channel(struct fsl_mc_io   *mc_io,
+   uint32_tcmd_flags,
+   uint16_ttoken,
+   int dpcon_id,
+   uint8_t *channel_index);
+
+/**
+ * dpio_remove_static_dequeue_channel() - Remove a static dequeue channel.
+ * @mc_io: Pointer to MC portal's I/O object
+ * @cmd_flags: Command flags; one or more of 'MC_CMD_FLAG_'
+ * @token: Token of DPIO object
+ * @dpcon_id:  DPCON object ID
+ *
+ * Return: '0' on Success; Error code otherwise.
+ */
+int dpio_remove_static_dequeue_channel(struct fsl_mc_io*mc_io,
+  uint32_t cmd_flags,
+  uint16_t token,
+  int  dpcon_id);
+
+/**
  * struct dpio_attr - Structure representing DPIO attributes
  * @id: DPIO object ID
  * @qbman_portal_ce_offset: offset of the software portal cache-enabled area
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
index 4269800..77efe37 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
@@ -42,6 +42,7 @@ struct dpaa2_io_portal_t {
struct dpaa2_dpio_dev *sec_dpio_dev;
uint64_t net_tid;
uint64_t sec_tid;
+   void *eventdev;
 };
 
 /*! Global per thread DPIO portal */
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index 4509051..c879e2f 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -58,6 +58,8 @@ DPDK_17.08 {
dpci_set_rx_queue;
dpcon_open;
dpcon_get_attributes;
+   dpio_add_static_dequeue_channel;
+   dpio_remove_static_dequeue_channel;
mc_get_soc_version;
mc_get_version;
 

[dpdk-dev] [PATCH 13/21 v3] event/dpaa2: add configuration functions

[Nipun Gupta]

This patch adds all the configuration API's for DPAA2 eventdev
including device config, start, stop & port and queue
related API's

Signed-off-by: Nipun Gupta 
---
 drivers/event/dpaa2/dpaa2_eventdev.c | 283 ++-
 drivers/event/dpaa2/dpaa2_eventdev.h |  22 +++
 2 files changed, 304 insertions(+), 1 deletion(-)

diff --git a/drivers/event/dpaa2/dpaa2_eventdev.c 
b/drivers/event/dpaa2/dpaa2_eventdev.c
index b8cc3f8..c00db7a 100644
--- a/drivers/event/dpaa2/dpaa2_eventdev.c
+++ b/drivers/event/dpaa2/dpaa2_eventdev.c
@@ -106,7 +106,288 @@
return dpaa2_eventdev_dequeue_burst(port, ev, 1, timeout_ticks);
 }
 
-static const struct rte_eventdev_ops dpaa2_eventdev_ops;
+static void
+dpaa2_eventdev_info_get(struct rte_eventdev *dev,
+   struct rte_event_dev_info *dev_info)
+{
+   struct dpaa2_eventdev *priv = dev->data->dev_private;
+
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+
+   memset(dev_info, 0, sizeof(struct rte_event_dev_info));
+   dev_info->min_dequeue_timeout_ns =
+   DPAA2_EVENT_MIN_DEQUEUE_TIMEOUT;
+   dev_info->max_dequeue_timeout_ns =
+   DPAA2_EVENT_MAX_DEQUEUE_TIMEOUT;
+   dev_info->dequeue_timeout_ns =
+   DPAA2_EVENT_MIN_DEQUEUE_TIMEOUT;
+   dev_info->max_event_queues = priv->max_event_queues;
+   dev_info->max_event_queue_flows =
+   DPAA2_EVENT_MAX_QUEUE_FLOWS;
+   dev_info->max_event_queue_priority_levels =
+   DPAA2_EVENT_MAX_QUEUE_PRIORITY_LEVELS;
+   dev_info->max_event_priority_levels =
+   DPAA2_EVENT_MAX_EVENT_PRIORITY_LEVELS;
+   dev_info->max_event_ports = RTE_MAX_LCORE;
+   dev_info->max_event_port_dequeue_depth =
+   DPAA2_EVENT_MAX_PORT_DEQUEUE_DEPTH;
+   dev_info->max_event_port_enqueue_depth =
+   DPAA2_EVENT_MAX_PORT_ENQUEUE_DEPTH;
+   dev_info->max_num_events = DPAA2_EVENT_MAX_NUM_EVENTS;
+   dev_info->event_dev_cap = RTE_EVENT_DEV_CAP_DISTRIBUTED_SCHED;
+}
+
+static int
+dpaa2_eventdev_configure(const struct rte_eventdev *dev)
+{
+   struct dpaa2_eventdev *priv = dev->data->dev_private;
+   struct rte_event_dev_config *conf = &dev->data->dev_conf;
+
+   PMD_DRV_FUNC_TRACE();
+
+   priv->dequeue_timeout_ns = conf->dequeue_timeout_ns;
+   priv->nb_event_queues = conf->nb_event_queues;
+   priv->nb_event_ports = conf->nb_event_ports;
+   priv->nb_event_queue_flows = conf->nb_event_queue_flows;
+   priv->nb_event_port_dequeue_depth = conf->nb_event_port_dequeue_depth;
+   priv->nb_event_port_enqueue_depth = conf->nb_event_port_enqueue_depth;
+   priv->event_dev_cfg = conf->event_dev_cfg;
+
+   PMD_DRV_LOG(DEBUG, "Configured eventdev devid=%d", dev->data->dev_id);
+   return 0;
+}
+
+static int
+dpaa2_eventdev_start(struct rte_eventdev *dev)
+{
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+
+   return 0;
+}
+
+static void
+dpaa2_eventdev_stop(struct rte_eventdev *dev)
+{
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+}
+
+static int
+dpaa2_eventdev_close(struct rte_eventdev *dev)
+{
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+
+   return 0;
+}
+
+static void
+dpaa2_eventdev_queue_def_conf(struct rte_eventdev *dev, uint8_t queue_id,
+ struct rte_event_queue_conf *queue_conf)
+{
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+   RTE_SET_USED(queue_id);
+   RTE_SET_USED(queue_conf);
+
+   queue_conf->nb_atomic_flows = DPAA2_EVENT_QUEUE_ATOMIC_FLOWS;
+   queue_conf->event_queue_cfg = RTE_EVENT_QUEUE_CFG_ATOMIC_ONLY |
+ RTE_EVENT_QUEUE_CFG_PARALLEL_ONLY;
+   queue_conf->priority = RTE_EVENT_DEV_PRIORITY_NORMAL;
+}
+
+static void
+dpaa2_eventdev_queue_release(struct rte_eventdev *dev, uint8_t queue_id)
+{
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+   RTE_SET_USED(queue_id);
+}
+
+static int
+dpaa2_eventdev_queue_setup(struct rte_eventdev *dev, uint8_t queue_id,
+  const struct rte_event_queue_conf *queue_conf)
+{
+   struct dpaa2_eventdev *priv = dev->data->dev_private;
+   struct evq_info_t *evq_info =
+   &priv->evq_info[queue_id];
+
+   PMD_DRV_FUNC_TRACE();
+
+   evq_info->event_queue_cfg = queue_conf->event_queue_cfg;
+
+   return 0;
+}
+
+static void
+dpaa2_eventdev_port_def_conf(struct rte_eventdev *dev, uint8_t port_id,
+struct rte_event_port_conf *port_conf)
+{
+   PMD_DRV_FUNC_TRACE();
+
+   RTE_SET_USED(dev);
+   RTE_SET_USED(port_id);
+   RTE_SET_USED(port_conf);
+
+   port_conf->new_event_threshold =
+   DPAA2_EVENT_MAX_NUM_EVENTS;
+   port_conf->dequeue_depth =
+   DPAA2_EVENT_MAX_PORT_DEQUEUE_DEPTH;
+   port_conf->enqueue_depth =
+   DPAA2_EVENT_MAX_PORT_ENQUEUE_DEPTH;
+}
+
+static void
+dpaa2_

[dpdk-dev] [PATCH 14/21 v3] bus/fslmc: support enqueue with multiple enqueue descriptors

[Nipun Gupta]

This patch adds the QBMAN API which support multiple enqueue
descriptors.

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h | 14 +
 drivers/bus/fslmc/qbman/qbman_portal.c | 70 ++
 drivers/bus/fslmc/rte_bus_fslmc_version.map|  1 +
 3 files changed, 85 insertions(+)

diff --git a/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h 
b/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h
index 7731772..39407c8 100644
--- a/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h
+++ b/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h
@@ -883,6 +883,20 @@ void qbman_eq_desc_set_dca(struct qbman_eq_desc *d, int 
enable,
  */
 int qbman_swp_enqueue(struct qbman_swp *s, const struct qbman_eq_desc *d,
  const struct qbman_fd *fd);
+/**
+ * qbman_swp_enqueue_multiple_eqdesc() - Enqueue multiple frames with separte
+ * enqueue descriptors.
+ * @s: the software portal used for enqueue.
+ * @d: the enqueue descriptors
+ * @fd: the frame descriptor to be enqueued.
+ * @num_frames: the number of the frames to be enqueued.
+ *
+ * Return the number of enqueued frames, -EBUSY if the EQCR is not ready.
+ */
+int qbman_swp_enqueue_multiple_eqdesc(struct qbman_swp *s,
+  const struct qbman_eq_desc *d,
+  const struct qbman_fd *fd,
+  int num_frames);
 
 /* TODO:
  * qbman_swp_enqueue_thresh() - Set threshold for EQRI interrupt.
diff --git a/drivers/bus/fslmc/qbman/qbman_portal.c 
b/drivers/bus/fslmc/qbman/qbman_portal.c
index be4e2e5..137b55d 100644
--- a/drivers/bus/fslmc/qbman/qbman_portal.c
+++ b/drivers/bus/fslmc/qbman/qbman_portal.c
@@ -574,6 +574,76 @@ int qbman_swp_enqueue(struct qbman_swp *s, const struct 
qbman_eq_desc *d,
return qbman_swp_enqueue_ring_mode(s, d, fd);
 }
 
+int qbman_swp_enqueue_multiple_eqdesc(struct qbman_swp *s,
+  const struct qbman_eq_desc *d,
+  const struct qbman_fd *fd,
+  int num_frames)
+{
+   uint32_t *p;
+   const uint32_t *cl = qb_cl(d);
+   uint32_t eqcr_ci, eqcr_pi;
+   uint8_t diff;
+   int i, num_enqueued = 0;
+   uint64_t addr_cena;
+
+   if (!s->eqcr.available) {
+   eqcr_ci = s->eqcr.ci;
+   s->eqcr.ci = qbman_cena_read_reg(&s->sys,
+   QBMAN_CENA_SWP_EQCR_CI) & 0xF;
+   diff = qm_cyc_diff(QBMAN_EQCR_SIZE,
+  eqcr_ci, s->eqcr.ci);
+   s->eqcr.available += diff;
+   if (!diff)
+   return 0;
+   }
+
+   eqcr_pi = s->eqcr.pi;
+   num_enqueued = (s->eqcr.available < num_frames) ?
+   s->eqcr.available : num_frames;
+   s->eqcr.available -= num_enqueued;
+   /* Fill in the EQCR ring */
+   for (i = 0; i < num_enqueued; i++) {
+   p = qbman_cena_write_start_wo_shadow(&s->sys,
+   QBMAN_CENA_SWP_EQCR(eqcr_pi & 7));
+   memcpy(&p[1], &cl[1], 28);
+   memcpy(&p[8], &fd[i], sizeof(*fd));
+   eqcr_pi++;
+   eqcr_pi &= 0xF;
+   /*Pointing to the next enqueue descriptor*/
+   cl += (sizeof(struct qbman_eq_desc) / sizeof(uint32_t));
+   }
+
+   lwsync();
+
+   /* Set the verb byte, have to substitute in the valid-bit */
+   eqcr_pi = s->eqcr.pi;
+   cl = qb_cl(d);
+   for (i = 0; i < num_enqueued; i++) {
+   p = qbman_cena_write_start_wo_shadow(&s->sys,
+   QBMAN_CENA_SWP_EQCR(eqcr_pi & 7));
+   p[0] = cl[0] | s->eqcr.pi_vb;
+   eqcr_pi++;
+   eqcr_pi &= 0xF;
+   if (!(eqcr_pi & 7))
+   s->eqcr.pi_vb ^= QB_VALID_BIT;
+   /*Pointing to the next enqueue descriptor*/
+   cl += (sizeof(struct qbman_eq_desc) / sizeof(uint32_t));
+   }
+
+   /* Flush all the cacheline without load/store in between */
+   eqcr_pi = s->eqcr.pi;
+   addr_cena = (uint64_t)s->sys.addr_cena;
+   for (i = 0; i < num_enqueued; i++) {
+   dcbf((uint64_t *)(addr_cena +
+   QBMAN_CENA_SWP_EQCR(eqcr_pi & 7)));
+   eqcr_pi++;
+   eqcr_pi &= 0xF;
+   }
+   s->eqcr.pi = eqcr_pi;
+
+   return num_enqueued;
+}
+
 /*/
 /* Static (push) dequeue */
 /*/
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map 
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index c879e2f..9950557 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -69,6 +69,7 @@ DPDK_17.08 {
qbman_result_SCN_state_in_mem;
qbman_swp_dqrr_consume;
qbman_swp_dqrr_next;
+   qbman_swp_enqueue_mu

[dpdk-dev] [PATCH 15/21 v3] bus/fslmc: add callback per queue to enable

[Nipun Gupta]

Dequeue from event device needs to process the event on
the basis of the hardware queue from which it is dequeued.
A callback is added into dpaa2_queue structure, to enable
event dequeue functionality to call that processing routine.

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 10 ++
 1 file changed, 10 insertions(+)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h 
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 0026ba9..975e431 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -34,6 +34,8 @@
 #ifndef _DPAA2_HW_PVT_H_
 #define _DPAA2_HW_PVT_H_
 
+#include 
+
 #include 
 #include 
 
@@ -101,6 +103,8 @@ struct dpaa2_dpio_dev {
uintptr_t ci_size; /**< Size of the CI region */
int32_t vfio_fd; /**< File descriptor received via VFIO */
int32_t hw_id; /**< An unique ID of this DPIO device instance */
+   uint64_t dqrr_held;
+   uint8_t dqrr_size;
 };
 
 struct dpaa2_dpbp_dev {
@@ -119,6 +123,11 @@ struct queue_storage_info_t {
int toggle;
 };
 
+typedef void (dpaa2_queue_cb_dqrr_t)(struct qbman_swp *swp,
+   const struct qbman_fd *fd,
+   const struct qbman_result *dq,
+   struct rte_event *ev);
+
 struct dpaa2_queue {
struct rte_mempool *mb_pool; /**< mbuf pool to populate RX ring. */
void *dev;
@@ -133,6 +142,7 @@ struct dpaa2_queue {
struct queue_storage_info_t *q_storage;
struct qbman_result *cscn;
};
+   dpaa2_queue_cb_dqrr_t *cb;
 };
 
 struct swp_active_dqs {
-- 
1.9.1

[dpdk-dev] [PATCH 17/21 v3] event/dpaa2: add enqueue and dequeue functionality

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 drivers/event/dpaa2/dpaa2_eventdev.c | 170 +--
 1 file changed, 163 insertions(+), 7 deletions(-)

diff --git a/drivers/event/dpaa2/dpaa2_eventdev.c 
b/drivers/event/dpaa2/dpaa2_eventdev.c
index c00db7a..a7f8516 100644
--- a/drivers/event/dpaa2/dpaa2_eventdev.c
+++ b/drivers/event/dpaa2/dpaa2_eventdev.c
@@ -49,6 +49,7 @@
 #include 
 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
@@ -74,11 +75,85 @@
 dpaa2_eventdev_enqueue_burst(void *port, const struct rte_event ev[],
 uint16_t nb_events)
 {
+   struct rte_eventdev *ev_dev =
+   ((struct dpaa2_io_portal_t *)port)->eventdev;
+   struct dpaa2_eventdev *priv = ev_dev->data->dev_private;
+   uint32_t queue_id = ev[0].queue_id;
+   struct evq_info_t *evq_info = &priv->evq_info[queue_id];
+   uint32_t fqid;
+   struct qbman_swp *swp;
+   struct qbman_fd fd_arr[MAX_TX_RING_SLOTS];
+   uint32_t loop, frames_to_send;
+   struct qbman_eq_desc eqdesc[MAX_TX_RING_SLOTS];
+   uint16_t num_tx = 0;
+   int ret;
+
RTE_SET_USED(port);
-   RTE_SET_USED(ev);
-   RTE_SET_USED(nb_events);
 
-   return 0;
+   if (unlikely(!DPAA2_PER_LCORE_DPIO)) {
+   ret = dpaa2_affine_qbman_swp();
+   if (ret) {
+   PMD_DRV_LOG(ERR, PMD, "Failure in affining portal\n");
+   return 0;
+   }
+   }
+
+   swp = DPAA2_PER_LCORE_PORTAL;
+
+   while (nb_events) {
+   frames_to_send = (nb_events >> 3) ?
+   MAX_TX_RING_SLOTS : nb_events;
+
+   for (loop = 0; loop < frames_to_send; loop++) {
+   const struct rte_event *event = &ev[num_tx + loop];
+
+   if (event->sched_type != RTE_SCHED_TYPE_ATOMIC)
+   fqid = evq_info->dpci->queue[
+   DPAA2_EVENT_DPCI_PARALLEL_QUEUE].fqid;
+   else
+   fqid = evq_info->dpci->queue[
+   DPAA2_EVENT_DPCI_ATOMIC_QUEUE].fqid;
+
+   /* Prepare enqueue descriptor */
+   qbman_eq_desc_clear(&eqdesc[loop]);
+   qbman_eq_desc_set_fq(&eqdesc[loop], fqid);
+   qbman_eq_desc_set_no_orp(&eqdesc[loop], 0);
+   qbman_eq_desc_set_response(&eqdesc[loop], 0, 0);
+
+   if (event->impl_opaque) {
+   uint8_t dqrr_index = event->impl_opaque - 1;
+
+   qbman_eq_desc_set_dca(&eqdesc[loop], 1,
+ dqrr_index, 0);
+   DPAA2_PER_LCORE_DPIO->dqrr_size--;
+   DPAA2_PER_LCORE_DPIO->dqrr_held &=
+   ~(1 << dqrr_index);
+   }
+
+   memset(&fd_arr[loop], 0, sizeof(struct qbman_fd));
+
+   /*
+* todo - need to align with hw context data
+* to avoid copy
+*/
+   struct rte_event *ev_temp = rte_malloc(NULL,
+   sizeof(struct rte_event), 0);
+   rte_memcpy(ev_temp, event, sizeof(struct rte_event));
+   DPAA2_SET_FD_ADDR((&fd_arr[loop]), ev_temp);
+   DPAA2_SET_FD_LEN((&fd_arr[loop]),
+sizeof(struct rte_event));
+   }
+   loop = 0;
+   while (loop < frames_to_send) {
+   loop += qbman_swp_enqueue_multiple_eqdesc(swp,
+   &eqdesc[loop], &fd_arr[loop],
+   frames_to_send - loop);
+   }
+   num_tx += frames_to_send;
+   nb_events -= frames_to_send;
+   }
+
+   return num_tx;
 }
 
 static uint16_t
@@ -87,16 +162,91 @@
return dpaa2_eventdev_enqueue_burst(port, ev, 1);
 }
 
+static void dpaa2_eventdev_process_parallel(struct qbman_swp *swp,
+   const struct qbman_fd *fd,
+   const struct qbman_result *dq,
+   struct rte_event *ev)
+{
+   struct rte_event *ev_temp =
+   (struct rte_event *)DPAA2_GET_FD_ADDR(fd);
+   rte_memcpy(ev, ev_temp, sizeof(struct rte_event));
+   rte_free(ev_temp);
+
+   qbman_swp_dqrr_consume(swp, dq);
+}
+
+static void dpaa2_eventdev_process_atomic(struct qbman_swp *swp,
+ const struct qbman_fd *fd,
+ const struct qbman_result *dq,
+ struct rte_event

[dpdk-dev] [PATCH 16/21 v3] bus/fslmc: change func argument to const to avoid warning

[Nipun Gupta]

qbman_get_dqrr_idx() API is required with constant dqrr entry
in the eventdev driver. Also, this routine is not updating the
dqrr. So, this patch updates its input argument to a const type.

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h | 2 +-
 drivers/bus/fslmc/qbman/qbman_portal.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h 
b/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h
index 39407c8..06bd063 100644
--- a/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h
+++ b/drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h
@@ -349,7 +349,7 @@ void qbman_pull_desc_set_channel(struct qbman_pull_desc *d, 
uint32_t chid,
  *
  * Return dqrr index.
  */
-uint8_t qbman_get_dqrr_idx(struct qbman_result *dqrr);
+uint8_t qbman_get_dqrr_idx(const struct qbman_result *dqrr);
 
 /**
  * qbman_get_dqrr_from_idx() - Use index to get the dqrr entry from the
diff --git a/drivers/bus/fslmc/qbman/qbman_portal.c 
b/drivers/bus/fslmc/qbman/qbman_portal.c
index 137b55d..8002690 100644
--- a/drivers/bus/fslmc/qbman/qbman_portal.c
+++ b/drivers/bus/fslmc/qbman/qbman_portal.c
@@ -1463,7 +1463,7 @@ int qbman_swp_CDAN_set_context_enable(struct qbman_swp 
*s, uint16_t channelid,
  1, ctx);
 }
 
-uint8_t qbman_get_dqrr_idx(struct qbman_result *dqrr)
+uint8_t qbman_get_dqrr_idx(const struct qbman_result *dqrr)
 {
return QBMAN_IDX_FROM_DQRR(dqrr);
 }
-- 
1.9.1

[dpdk-dev] [PATCH 18/21 v3] fslmc/bus: add interrupt enabling routine

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/fslmc_vfio.c | 34 ++
 drivers/bus/fslmc/fslmc_vfio.h |  3 +++
 2 files changed, 37 insertions(+)

diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 6ebf779..d8e3add 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -359,6 +359,40 @@ static int64_t vfio_map_mcp_obj(struct fslmc_vfio_group 
*group, char *mcp_obj)
}
 }
 
+#define IRQ_SET_BUF_LEN  (sizeof(struct vfio_irq_set) + sizeof(int))
+
+int rte_dpaa2_intr_enable(struct rte_intr_handle *intr_handle,
+ uint32_t index)
+{
+   struct vfio_irq_set *irq_set;
+   char irq_set_buf[IRQ_SET_BUF_LEN];
+   int *fd_ptr, fd, ret;
+
+   /* Prepare vfio_irq_set structure and SET the IRQ in VFIO */
+   /* Give the eventfd to VFIO */
+   fd = eventfd(0, 0);
+   irq_set = (struct vfio_irq_set *)irq_set_buf;
+   irq_set->argsz = sizeof(irq_set_buf);
+   irq_set->count = 1;
+   irq_set->flags = VFIO_IRQ_SET_DATA_EVENTFD |
+VFIO_IRQ_SET_ACTION_TRIGGER;
+   irq_set->index = index;
+   irq_set->start = 0;
+   fd_ptr = (int *)&irq_set->data;
+   *fd_ptr = fd;
+
+   ret = ioctl(intr_handle->vfio_dev_fd, VFIO_DEVICE_SET_IRQS, irq_set);
+   if (ret < 0) {
+   FSLMC_VFIO_LOG(ERR, "Unable to set IRQ in VFIO, ret: %d\n",
+  ret);
+   return -1;
+   }
+
+   /* Set the FD and update the flags */
+   intr_handle->fd = fd;
+   return 0;
+}
+
 /* Following function shall fetch total available list of MC devices
  * from VFIO container & populate private list of devices and other
  * data structures
diff --git a/drivers/bus/fslmc/fslmc_vfio.h b/drivers/bus/fslmc/fslmc_vfio.h
index 7c725f4..ebca2b0 100644
--- a/drivers/bus/fslmc/fslmc_vfio.h
+++ b/drivers/bus/fslmc/fslmc_vfio.h
@@ -90,6 +90,9 @@ int vfio_dmamap_mem_region(
uint64_t iova,
uint64_t size);
 
+int rte_dpaa2_intr_enable(struct rte_intr_handle *intr_handle,
+ uint32_t index);
+
 int fslmc_vfio_setup_group(void);
 int fslmc_vfio_process_group(void);
 int rte_fslmc_vfio_dmamap(void);
-- 
1.9.1

[dpdk-dev] [PATCH 19/21 v3] bus/fslmc: enable portal interrupt handling

[Nipun Gupta]

Eventdev requires portal interrupts to handle timeout in the
event dequeue. This patch provides mechanism to enable the
portal interrupts.

Signed-off-by: Nipun Gupta 
---
 drivers/bus/fslmc/portal/dpaa2_hw_dpio.c   | 108 -
 drivers/bus/fslmc/portal/dpaa2_hw_pvt.h|   3 +-
 drivers/bus/fslmc/qbman/include/fsl_qbman_portal.h |  30 ++
 drivers/bus/fslmc/qbman/qbman_portal.c |  22 +
 drivers/bus/fslmc/rte_bus_fslmc_version.map|   1 +
 5 files changed, 158 insertions(+), 6 deletions(-)

diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c 
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
index 63378f0..5d53342 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -46,6 +46,8 @@
 #include 
 #include 
 #include 
+#include 
+#include
 
 #include 
 #include 
@@ -106,6 +108,95 @@
return dpaa2_core_cluster_base + x;
 }
 
+static void dpaa2_affine_dpio_intr_to_respective_core(int32_t dpio_id)
+{
+#define STRING_LEN 28
+#define COMMAND_LEN50
+   uint32_t cpu_mask = 1;
+   int ret;
+   size_t len = 0;
+   char *temp = NULL, *token = NULL;
+   char string[STRING_LEN], command[COMMAND_LEN];
+   FILE *file;
+
+   snprintf(string, STRING_LEN, "dpio.%d", dpio_id);
+   file = fopen("/proc/interrupts", "r");
+   if (!file) {
+   PMD_DRV_LOG(WARN, "Failed to open /proc/interrupts file\n");
+   return;
+   }
+   while (getline(&temp, &len, file) != -1) {
+   if ((strstr(temp, string)) != NULL) {
+   token = strtok(temp, ":");
+   break;
+   }
+   }
+
+   if (!token) {
+   PMD_DRV_LOG(WARN, "Failed to get interrupt id for dpio.%d\n",
+   dpio_id);
+   if (temp)
+   free(temp);
+   fclose(file);
+   return;
+   }
+
+   cpu_mask = cpu_mask << rte_lcore_id();
+   snprintf(command, COMMAND_LEN, "echo %X > /proc/irq/%s/smp_affinity",
+cpu_mask, token);
+   ret = system(command);
+   if (ret < 0)
+   PMD_DRV_LOG(WARN,
+   "Failed to affine interrupts on respective core\n");
+   else
+   PMD_DRV_LOG(WARN, " %s command is executed\n", command);
+
+   free(temp);
+   fclose(file);
+}
+
+static int dpaa2_dpio_intr_init(struct dpaa2_dpio_dev *dpio_dev)
+{
+   struct epoll_event epoll_ev;
+   int eventfd, dpio_epoll_fd, ret;
+   int threshold = 0x3, timeout = 0xFF;
+
+   dpio_epoll_fd = epoll_create(1);
+   ret = rte_dpaa2_intr_enable(&dpio_dev->intr_handle, 0);
+   if (ret) {
+   PMD_DRV_LOG(ERR, "Interrupt registeration failed\n");
+   return -1;
+   }
+
+   if (getenv("DPAA2_PORTAL_INTR_THRESHOLD"))
+   threshold = atoi(getenv("DPAA2_PORTAL_INTR_THRESHOLD"));
+
+   if (getenv("DPAA2_PORTAL_INTR_TIMEOUT"))
+   sscanf(getenv("DPAA2_PORTAL_INTR_TIMEOUT"), "%x", &timeout);
+
+   qbman_swp_interrupt_set_trigger(dpio_dev->sw_portal,
+   QBMAN_SWP_INTERRUPT_DQRI);
+   qbman_swp_interrupt_clear_status(dpio_dev->sw_portal, 0x);
+   qbman_swp_interrupt_set_inhibit(dpio_dev->sw_portal, 0);
+   qbman_swp_dqrr_thrshld_write(dpio_dev->sw_portal, threshold);
+   qbman_swp_intr_timeout_write(dpio_dev->sw_portal, timeout);
+
+   eventfd = dpio_dev->intr_handle.fd;
+   epoll_ev.events = EPOLLIN | EPOLLPRI | EPOLLET;
+   epoll_ev.data.fd = eventfd;
+
+   ret = epoll_ctl(dpio_epoll_fd, EPOLL_CTL_ADD, eventfd, &epoll_ev);
+   if (ret < 0) {
+   PMD_DRV_LOG(ERR, "epoll_ctl failed\n");
+   return -1;
+   }
+   dpio_dev->epoll_fd = dpio_epoll_fd;
+
+   dpaa2_affine_dpio_intr_to_respective_core(dpio_dev->hw_id);
+
+   return 0;
+}
+
 static int
 configure_dpio_qbman_swp(struct dpaa2_dpio_dev *dpio_dev)
 {
@@ -215,6 +306,11 @@
return -1;
}
 
+   if (dpaa2_dpio_intr_init(dpio_dev)) {
+   PMD_DRV_LOG(ERR, "Interrupt registration failed for dpio\n");
+   return -1;
+   }
+
return 0;
 }
 
@@ -339,6 +435,7 @@ struct dpaa2_dpio_dev *dpaa2_get_qbman_swp(int cpu_id)
 {
struct dpaa2_dpio_dev *dpio_dev;
struct vfio_region_info reg_info = { .argsz = sizeof(reg_info)};
+   int vfio_dev_fd;
 
if (obj_info->num_regions < NUM_DPIO_REGIONS) {
PMD_INIT_LOG(ERR, "ERROR, Not sufficient number "
@@ -355,13 +452,14 @@ struct dpaa2_dpio_dev *dpaa2_get_qbman_swp(int cpu_id)
 
dpio_dev->dpio = NULL;
dpio_dev->hw_id = object_id;
-   dpio_dev->vfio_fd = vdev->fd;
+   dpio_dev->intr_handle.vfio_dev_fd = vdev->fd;
rte_atomic16_init(&dpio_dev->ref_count);
/* Using single portal  for all devices */
 

[dpdk-dev] [PATCH 20/21 v3] event/dpaa2: handle timeout using interrupts in dequeue

[Nipun Gupta]

This patch adds support for interrupt handling on the event port.
These interrupts facilitates managing of timeout ticks in the
event dequeue functions.

Signed-off-by: Nipun Gupta 
---
 drivers/event/dpaa2/dpaa2_eventdev.c | 37 +---
 1 file changed, 34 insertions(+), 3 deletions(-)

diff --git a/drivers/event/dpaa2/dpaa2_eventdev.c 
b/drivers/event/dpaa2/dpaa2_eventdev.c
index a7f8516..80ee65c 100644
--- a/drivers/event/dpaa2/dpaa2_eventdev.c
+++ b/drivers/event/dpaa2/dpaa2_eventdev.c
@@ -162,6 +162,32 @@
return dpaa2_eventdev_enqueue_burst(port, ev, 1);
 }
 
+static void dpaa2_eventdev_dequeue_wait(uint64_t timeout_ticks)
+{
+   struct epoll_event epoll_ev;
+   int ret, i = 0;
+
+   qbman_swp_interrupt_clear_status(DPAA2_PER_LCORE_PORTAL,
+QBMAN_SWP_INTERRUPT_DQRI);
+
+RETRY:
+   ret = epoll_wait(DPAA2_PER_LCORE_DPIO->epoll_fd,
+&epoll_ev, 1, timeout_ticks);
+   if (ret < 1) {
+   /* sometimes due to some spurious interrupts epoll_wait fails
+* with errno EINTR. so here we are retrying epoll_wait in such
+* case to avoid the problem.
+*/
+   if (errno == EINTR) {
+   PMD_DRV_LOG(DEBUG, PMD, "epoll_wait fails\n");
+   if (i++ > 10)
+   PMD_DRV_LOG(DEBUG, PMD,
+   "Dequeue burst Failed\n");
+   goto RETRY;
+   }
+   }
+}
+
 static void dpaa2_eventdev_process_parallel(struct qbman_swp *swp,
const struct qbman_fd *fd,
const struct qbman_result *dq,
@@ -204,7 +230,6 @@ static void dpaa2_eventdev_process_atomic(struct qbman_swp 
*swp,
int num_pkts = 0, ret, i = 0;
 
RTE_SET_USED(port);
-   RTE_SET_USED(timeout_ticks);
 
if (unlikely(!DPAA2_PER_LCORE_DPIO)) {
ret = dpaa2_affine_qbman_swp();
@@ -229,8 +254,14 @@ static void dpaa2_eventdev_process_atomic(struct qbman_swp 
*swp,
 
do {
dq = qbman_swp_dqrr_next(swp);
-   if (!dq)
-   return 0;
+   if (!dq) {
+   if (!num_pkts && timeout_ticks) {
+   dpaa2_eventdev_dequeue_wait(timeout_ticks);
+   timeout_ticks = 0;
+   continue;
+   }
+   return num_pkts;
+   }
 
fd = qbman_result_DQ_fd(dq);
 
-- 
1.9.1

[dpdk-dev] [PATCH 21/21 v3] doc: add NXP DPAA2 EVENTDEV details

[Nipun Gupta]

Signed-off-by: Nipun Gupta 
---
 MAINTAINERS|   1 +
 doc/guides/eventdevs/dpaa2.rst | 175 +
 2 files changed, 176 insertions(+)
 create mode 100644 doc/guides/eventdevs/dpaa2.rst

diff --git a/MAINTAINERS b/MAINTAINERS
index fe1a25b..d9dbf8f 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -584,6 +584,7 @@ NXP DPAA2 Eventdev PMD
 M: Hemant Agrawal 
 M: Nipun Gupta 
 F: drivers/event/dpaa2/
+F: doc/guides/eventdevs/dpaa2.rst
 
 Packet processing
 -
diff --git a/doc/guides/eventdevs/dpaa2.rst b/doc/guides/eventdevs/dpaa2.rst
new file mode 100644
index 000..15dccdd
--- /dev/null
+++ b/doc/guides/eventdevs/dpaa2.rst
@@ -0,0 +1,175 @@
+..  BSD LICENSE
+Copyright 2017 NXP.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+* Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+* Redistributions in binary form must reproduce the above copyright
+notice, this list of conditions and the following disclaimer in
+the documentation and/or other materials provided with the
+distribution.
+* Neither the name of NXP nor the names of its
+contributors may be used to endorse or promote products derived
+from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+NXP DPAA2 Eventdev Driver
+==
+
+The dpaa2 eventdev is an implementation of the eventdev API, that provides a
+wide range of the eventdev features. The eventdev relies on a dpaa2 hw to
+perform event scheduling.
+
+More information can be found at `NXP Official Website
+`_.
+
+Features
+
+
+The DPAA2 EVENTDEV implements many features in the eventdev API;
+
+- Hardware based event scheduler
+- 8 event ports
+- 8 event queues
+- Parallel flows
+- Atomic flows
+
+Supported DPAA2 SoCs
+
+
+- LS2080A/LS2040A
+- LS2084A/LS2044A
+- LS2088A/LS2048A
+- LS1088A/LS1048A
+
+Prerequisites
+-
+
+There are three main pre-requisities for executing DPAA2 EVENTDEV on a DPAA2
+compatible board:
+
+1. **ARM 64 Tool Chain**
+
+   For example, the `*aarch64* Linaro Toolchain 
`_.
+
+2. **Linux Kernel**
+
+   It can be obtained from `NXP's Github hosting 
`_.
+
+3. **Rootfile system**
+
+   Any *aarch64* supporting filesystem can be used. For example,
+   Ubuntu 15.10 (Wily) or 16.04 LTS (Xenial) userland which can be obtained
+   from `here 
`_.
+
+As an alternative method, DPAA2 EVENTDEV can also be executed using images 
provided
+as part of SDK from NXP. The SDK includes all the above prerequisites necessary
+to bring up a DPAA2 board.
+
+The following dependencies are not part of DPDK and must be installed
+separately:
+
+- **NXP Linux SDK**
+
+  NXP Linux software development kit (SDK) includes support for family
+  of QorIQ® ARM-Architecture-based system on chip (SoC) processors
+  and corresponding boards.
+
+  It includes the Linux board support packages (BSPs) for NXP SoCs,
+  a fully operational tool chain, kernel and board specific modules.
+
+  SDK and related information can be obtained from:  `NXP QorIQ SDK  
`_.
+
+- **DPDK Helper Scripts**
+
+  DPAA2 based resources can be configured easily with the help of ready scripts
+  as provided in the DPDK helper repository.
+
+  `DPDK Helper Scripts `_.
+
+Currently supported by DPDK:
+
+- NXP SDK **2.0+**.
+- MC Firmware version **10.0.0** and higher.
+- Supported architectures:  **arm64 LE**.
+
+- Follow the DPDK :ref:`Getting Started Guide for Linux ` to setup 

Re: [dpdk-dev] [PATCH v6 05/11] bus: introduce hotplug functionality

[Bruce Richardson]

On Wed, Jun 28, 2017 at 05:11:46PM +0200, Jan Blunck wrote:
> On Wed, Jun 28, 2017 at 3:30 PM, Thomas Monjalon  wrote:
> > 28/06/2017 15:09, Jan Blunck:
> >> On Wed, Jun 28, 2017 at 2:11 PM, Thomas Monjalon  
> >> wrote:
> >> > 28/06/2017 13:58, Jan Blunck:
> >> >> On Wed, Jun 28, 2017 at 1:44 PM, Thomas Monjalon  
> >> >> wrote:
> >> >> > 27/06/2017 21:03, Jan Blunck:
> >> >> >> On Tue, Jun 27, 2017 at 6:11 PM, Gaetan Rivet 
> >> >> >>  wrote:
> >> >> >> > --- a/lib/librte_eal/common/include/rte_bus.h
> >> >> >> > +++ b/lib/librte_eal/common/include/rte_bus.h
> >> >> >> >  /**
> >> >> >> > + * Implementation specific probe function which is responsible 
> >> >> >> > for linking
> >> >> >> > + * devices on that bus with applicable drivers.
> >> >> >> > + * The plugged device might already have been used previously by 
> >> >> >> > the bus,
> >> >> >> > + * in which case some buses might prefer to detect and re-use the 
> >> >> >> > relevant
> >> >> >> > + * information pertaining to this device.
> >> >> >> > + *
> >> >> >> > + * @param da
> >> >> >> > + * Device declaration.
> >> >> >> > + *
> >> >> >> > + * @return
> >> >> >> > + * The pointer to a valid rte_device usable by the bus on 
> >> >> >> > success.
> >> >> >> > + * NULL on error. rte_errno is then set.
> >> >> >> > + */
> >> >> >> > +typedef struct rte_device * (*rte_bus_plug_t)(struct rte_devargs 
> >> >> >> > *da);
> >> >> >>
> >> >> >> Shouldn't this be orthogonal to unplug() and take a rte_device. You
> >> >> >> should only be able to plug devices that have been found by scan
> >> >> >> before.
> >> >> >
> >> >> > Plugging a device that has been scanned before is a special case.
> >> >> > In a true hotplug scenario, we could use this plug function passing
> >> >> > a devargs.
> >> >> > I don't see any issue passing rte_devargs to plug and rte_device to 
> >> >> > unplug.
> >> >>
> >> >> What do you mean by "true hotplug"?
> >> >
> >> > I mean a kernel notification of a new device.
> >>
> >> Does a "false hotplug" exist, too? :)
> >
> > The false hotplug was the original attach function which was just
> > adding a new ethdev interface.
> >
> >> >> The problem with this is that passing just rte_devargs to plug()
> >> >> requires the bus to parse and enrich the rte_devargs with bus
> >> >> specifics. From there it gets folded into the to-be-created bus
> >> >> specific rte_XXX_device. This makes it unnecessarily complicated and
> >> >> even worse it adds a second code path how devices come alive.
> >> >
> >> > Just after the notification, the rte_device does not exist yet.
> >> > So the plug function could share the same code as the scan function
> >> > to get the metadata and create the device instance.
> >>
> >> Exactly this is what I want to avoid.
> >
> > Why do you want to avoid that?
> > I think you mean it is not what you had in mind.
> >
> >> The plug() function would become a "scan-one and probe".
> >
> > Yes
> >
> >> From my point of view plug() and unplug() should be orthogonal.
> >> The plug() and unplug() should only be responsible for adding drivers
> >> with optional arguments. The EAL should allow the drivers to get
> >> unplugged/re-plugged at run-time. I want to be able to change arguments
> >> ... or even drivers :)
> >
> > It is a totally different thing.
> > We are talking about hotplug of a device,
> > and you are talking about changing drivers dynamically.
> >
> > So I still don't understand what is the issue with the plug/unplug
> > functions proposed here.
> >
> 
> I don't agree with the notion that plug() means "scan-one and probe".

What do you see as plug doing then? What do you see as the use-case and
then logic flow for hotplug?

Re: [dpdk-dev] [PATCH 22/38] net/dpaa: add NXP DPAA PMD driver skeleton

[Ferruh Yigit]

On 6/16/2017 6:40 AM, Shreyansh Jain wrote:
> A skeleton which would be called after bus device scan. It currently
> fails to identify the device>
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

> +
> +/* Initialise a network interface */
> +static int dpaa_eth_dev_init(struct rte_eth_dev *eth_dev __rte_unused)

__rte_unused can be removed

<...>

> +
> +static int
> +rte_dpaa_probe(struct rte_dpaa_driver *dpaa_drv __rte_unused,
> +struct rte_dpaa_device *dpaa_dev)
> +{
> + int diag;
> + int ret;
> + struct rte_eth_dev *eth_dev;
> + char ethdev_name[RTE_ETH_NAME_MAX_LEN];
> +
> + PMD_INIT_FUNC_TRACE();
> +
> + if (!is_global_init) {
> + /* One time load of Qman/Bman drivers */
> + ret = qman_global_init();
> + if (ret) {
> + PMD_DRV_LOG(ERR, "QMAN initialization failed: %d",
> + ret);
> + return ret;
> + }
> + ret = bman_global_init();
> + if (ret) {
> + PMD_DRV_LOG(ERR, "BMAN initialization failed: %d",
> + ret);
> + return ret;
> + }
> +
> + is_global_init = 1;
> + }
> +
> + sprintf(ethdev_name, "%s", dpaa_dev->name);

snprintf can be preferred

> +
> + ret = rte_dpaa_portal_init((void *)1);
> + if (ret) {
> + PMD_DRV_LOG(ERR, "Unable to initialize portal");
> + return ret;
> + }
> +
> + eth_dev = rte_eth_dev_allocate(ethdev_name);

If this is done without RTE_PROC_PRIMARY check, this will cause
secondary to memset all device data.

I am adding this because of below check, it multi process support is
intended, this also be protected.

> + if (eth_dev == NULL)
> + return -ENOMEM;
> +
> + if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
> + eth_dev->data->dev_private = rte_zmalloc(
> + "ethdev private structure",
> + sizeof(struct dpaa_if),
> + RTE_CACHE_LINE_SIZE);
> + if (!eth_dev->data->dev_private) {
> + PMD_INIT_LOG(CRIT, "Cannot allocate memzone for"
> +  " private port data\n");
> + rte_eth_dev_release_port(eth_dev);
> + return -ENOMEM;
> + }
> + }
> +
> + eth_dev->device = &dpaa_dev->device;
> + dpaa_dev->eth_dev = eth_dev;

I thought "struct rte_dpaa_device" is bus device, like "struct
rte_pci_device", if so why it has link to the eth_dev?

> + eth_dev->data->rx_mbuf_alloc_failed = 0;

not required, data already memset via rte_eth_dev_allocate()

> +
> + /* Invoke PMD device initialization function */
> + diag = dpaa_eth_dev_init(eth_dev);
> + if (diag) {
> + PMD_DRV_LOG(ERR, "Eth dev initialization failed: %d", ret);
> + return diag;
> + }
> +
> + PMD_DRV_LOG(DEBUG, "Eth dev initialized: %d\n", diag);
> +
> + return 0;
> +}
> +
> +static int
> +rte_dpaa_remove(struct rte_dpaa_device *dpaa_dev)
> +{
> + struct rte_eth_dev *eth_dev;
> +
> + PMD_INIT_FUNC_TRACE();
> +
> + eth_dev = dpaa_dev->eth_dev;

can be:
eth_dev = rte_eth_dev_allocated(dpaa_dev->device.name);

> +
> + if (rte_eal_process_type() == RTE_PROC_PRIMARY)
> + rte_free(eth_dev->data->dev_private);
> +

no pmd uninit() ?

> + rte_eth_dev_release_port(eth_dev);
> +
> + return 0;
> +}

<...>

Re: [dpdk-dev] [PATCH 24/38] net/dpaa: add support for Tx and Rx queue setup

[Ferruh Yigit]

On 6/16/2017 6:40 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 
> ---
>  doc/guides/nics/features/dpaa.ini |   1 +
>  drivers/net/dpaa/Makefile |   4 +
>  drivers/net/dpaa/dpaa_ethdev.c| 279 -
>  drivers/net/dpaa/dpaa_ethdev.h|   6 +
>  drivers/net/dpaa/dpaa_rxtx.c  | 313 
> ++
>  drivers/net/dpaa/dpaa_rxtx.h  |  61 

This patch adds initial rx/tx support, as well as rx/tx queue support as
mentioned in patch subject.

I would be for splitting patch, but even if patch not splitted, I would
suggest updating patch suject and commit log to cover patch content.

<...>
> --- a/doc/guides/nics/features/dpaa.ini
> +++ b/doc/guides/nics/features/dpaa.ini
> @@ -4,5 +4,6 @@
>  ; Refer to default.ini for the full list of available PMD features.
>  ;
>  [Features]
> +Queue start/stop = Y

This requires following dev_ops implemented:
rx_queue_start, rx_queue_stop, tx_queue_start, tx_queue_stop

>  ARMv8= Y
>  Usage doc= Y

<...>

> +
> + /* Initialize Rx FQ's */
> + if (getenv("DPAA_NUM_RX_QUEUES"))

I think this was disscussed before, should a PMD get config options from
enviroment variable? Altough this works, I am for a more explicit
method, like dev_args.

<...>
> +
> + dpaa_intf->rx_queues = rte_zmalloc(NULL,
> + sizeof(struct qman_fq) * num_rx_fqs, MAX_CACHELINE);

A NULL check perhaps?

And if multi-process support desired, this should be done only for
primary process.

<...>
> + /* Allocate memory for storing MAC addresses */
> + eth_dev->data->mac_addrs = rte_zmalloc("mac_addr",
> + ETHER_ADDR_LEN * DPAA_MAX_MAC_FILTER, 0);
> + if (eth_dev->data->mac_addrs == NULL) {
> + PMD_INIT_LOG(ERR, "Failed to allocate %d bytes needed to "
> + "store MAC addresses",
> + ETHER_ADDR_LEN * DPAA_MAX_MAC_FILTER);

Anything to cleanup before exit?

> + return -ENOMEM;
> + }

<...>
> +uint16_t dpaa_eth_queue_rx(void *q,
> +struct rte_mbuf **bufs,
> +uint16_t nb_bufs)
> +{
> + struct qman_fq *fq = q;
> + struct qm_dqrr_entry *dq;
> + uint32_t num_rx = 0, ifid = ((struct dpaa_if *)fq->dpaa_intf)->ifid;
> + int ret;
> +
> + ret = rte_dpaa_portal_init((void *)0);
> + if (ret) {
> + PMD_DRV_LOG(ERR, "Failure in affining portal");
> + return 0;
> + }

This is rx_pkt_burst function, right? Is it Ok to call
rte_dpaa_portal_init() in Rx data path?

<...>
> + buf = (uint64_t)rte_dpaa_mem_ptov(bufs.addr) - bp_info->meta_data_size;
> + if (!buf)
> + goto out;

goto is not required here.

> +
> +out:
> + return (void *)buf;
> +}
> +

<...>
> +uint16_t dpaa_eth_tx_drop_all(void *q  __rte_unused,
> +   struct rte_mbuf **bufs __rte_unused,
> + uint16_t nb_bufs __rte_unused)
> +{
> + PMD_TX_LOG(DEBUG, "Drop all packets");

Should mbufs freed here?

> +
> + /* Drop all incoming packets. No need to free packets here
> +  * because the rte_eth f/w frees up the packets through tx_buffer
> +  * callback in case this functions returns count less than nb_bufs
> +  */
> + return 0;
> +}

<...>

Re: [dpdk-dev] [PATCH 25/38] net/dpaa: add support for MTU update

[Ferruh Yigit]

On 6/16/2017 6:40 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

>  static int
> +dpaa_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
> +{
> + struct dpaa_if *dpaa_intf = dev->data->dev_private;
> +
> + PMD_INIT_FUNC_TRACE();
> +
> + if (mtu < ETHER_MIN_MTU)
> + return -EINVAL;
> +
> + fman_if_set_maxfrm(dpaa_intf->fif, mtu);
> +
> + if (mtu > ETHER_MAX_LEN)
> + return -1

Is it OK to have this check after fman_if_set_maxfrm() ?

> + dev->data->dev_conf.rxmode.jumbo_frame = 0;
> +
> + dev->data->dev_conf.rxmode.max_rx_pkt_len = mtu;

I think this only makes sense when jumbo_frame is 1, although not hurts
to set...

> + return 0;
> +}
<...>

Re: [dpdk-dev] [PATCH 27/38] net/dpaa: add support for link status update

[Ferruh Yigit]

On 6/16/2017 6:40 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

> --- a/doc/guides/nics/features/dpaa.ini
> +++ b/doc/guides/nics/features/dpaa.ini
> @@ -4,6 +4,8 @@
>  ; Refer to default.ini for the full list of available PMD features.
>  ;
>  [Features]
> +Speed capabilities   = P

Speed capabilities feature is not "link->link_speed", this feature means
providing "dev_info->speed_capa" (in dpaa_eth_dev_info())

> +Link status  = Y
>  Queue start/stop = Y
>  Jumbo frame  = Y
>  MTU update   = Y
<...>

Re: [dpdk-dev] [PATCH 30/38] net/dpaa: add support for multicast toggle

[Ferruh Yigit]

On 6/16/2017 6:41 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

> diff --git a/doc/guides/nics/features/dpaa.ini 
> b/doc/guides/nics/features/dpaa.ini
> index a6984a4..80dd3ca 100644
> --- a/doc/guides/nics/features/dpaa.ini
> +++ b/doc/guides/nics/features/dpaa.ini
> @@ -10,5 +10,7 @@ Queue start/stop = Y
>  Jumbo frame  = Y
>  MTU update   = Y
>  Promiscuous mode = Y
> +Allmulticast mode= Y
> +Unicast MAC filter   = Y

"Unicast MAC filter" means implementing "mac_addr_set, mac_addr_add,
mac_addr_remove" dev_ops

>  ARMv8= Y
>  Usage doc= Y

<...>

Re: [dpdk-dev] [PATCH 33/38] net/dpaa: add support for flow control

[Ferruh Yigit]

On 6/16/2017 6:41 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

>  static int
> +dpaa_flow_ctrl_set(struct rte_eth_dev *dev,
> +struct rte_eth_fc_conf *fc_conf)
> +{
> + struct dpaa_if *dpaa_intf = dev->data->dev_private;
> + struct rte_eth_fc_conf *net_fc;
> +
> + PMD_INIT_FUNC_TRACE();
> +
> + if (!(dpaa_intf->fc_conf)) {
> + dpaa_intf->fc_conf = rte_zmalloc(NULL,
> + sizeof(struct rte_eth_fc_conf), MAX_CACHELINE);

Should this be freed in rte_dpaa_remove()

> + if (!dpaa_intf->fc_conf) {
> + PMD_DRV_LOG(ERR, "unable to save flow control info");
> + return -ENOMEM;
> + }
> + }
> + net_fc = dpaa_intf->fc_conf;
> +
<...>

Re: [dpdk-dev] [PATCH 34/38] net/dpaa: add support for hashed RSS

[Ferruh Yigit]

On 6/16/2017 6:41 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

Just to confirm:

Is no HW configuration required to enable RSS?
Is HW updates mbuf->rss automatically, without driver involvement?

<...>

>  Promiscuous mode = Y
>  Allmulticast mode= Y
>  Unicast MAC filter   = Y
> +RSS hash = Y
>  Flow control = Y
>  Basic stats  = Y
>  ARMv8= Y

<...>

Re: [dpdk-dev] [PATCH 35/38] net/dpaa: add support for packet type parsing

[Ferruh Yigit]

On 6/16/2017 6:41 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

> +static const uint32_t *
> +dpaa_supported_ptypes_get(struct rte_eth_dev *dev)
> +{
> + static const uint32_t ptypes[] = {
> + /*todo -= add more types */
> + RTE_PTYPE_L2_ETHER,
> + RTE_PTYPE_L3_IPV4,
> + RTE_PTYPE_L3_IPV4_EXT,
> + RTE_PTYPE_L3_IPV6,
> + RTE_PTYPE_L3_IPV6_EXT,
> + RTE_PTYPE_L4_TCP,
> + RTE_PTYPE_L4_UDP,
> + RTE_PTYPE_L4_SCTP
> + };
> +
> + PMD_INIT_FUNC_TRACE();
> +
> + if (dev->rx_pkt_burst == dpaa_eth_queue_rx)

Isn't this only rx function exists? Is this check required?

> + return ptypes;
> + return NULL;
> +}
>  
>  static int dpaa_eth_dev_start(struct rte_eth_dev *dev)
>  {
> @@ -159,6 +180,10 @@ static void dpaa_eth_dev_info(struct rte_eth_dev *dev,
>   dev_info->max_vfs = 0;
>   dev_info->max_vmdq_pools = ETH_16_POOLS;
>   dev_info->flow_type_rss_offloads = DPAA_RSS_OFFLOAD_ALL;
> + dev_info->rx_offload_capa =
> + (DEV_RX_OFFLOAD_IPV4_CKSUM |
> + DEV_RX_OFFLOAD_UDP_CKSUM  |
> + DEV_RX_OFFLOAD_TCP_CKSUM);

I guess this patch also enable L3/L4 Rx checksum offload, can you please
update commit log.

And should ol_flags set with one of the PKT_RX_IP_CKSUM_BAD,
PKT_RX_IP_CKSUM_GOOD, PKT_RX_IP_CKSUM_NONE? Also with L4 versions of these?

<...>

> +
> + m->tx_offload = annot->parse.ip_off[0];
> + m->tx_offload |= (annot->parse.l4_off - annot->parse.ip_off[0])
> + << DPAA_PKT_L3_LEN_SHIFT;

This is a received mbuf right? Is it required to set tx_offload flag?

> +
> + /* Set the hash values */
> + m->hash.rss = (uint32_t)(rte_be_to_cpu_64(annot->hash));> + 
> m->ol_flags = PKT_RX_RSS_HASH;
> +
> + /* Check if Vlan is present */
> + if (prs & DPAA_PARSE_VLAN_MASK)
> + m->ol_flags |= PKT_RX_VLAN_PKT;

I guess PKT_RX_VLAN_STRIPPED is the preferred flag now.

<...>

Re: [dpdk-dev] [PATCH 36/38] net/dpaa: add support for checksum offload

[Ferruh Yigit]

On 6/16/2017 6:41 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

<...>

> @@ -363,6 +439,18 @@ dpaa_eth_queue_tx(void *q, struct rte_mbuf **bufs, 
> uint16_t nb_bufs)
>   }
>   rte_pktmbuf_free(mbuf);
>   }
> + if (mbuf->ol_flags & 
> DPAA_TX_CKSUM_OFFLOAD_MASK) {
> + if (mbuf->data_off < 
> DEFAULT_TX_ICEOF +
> + sizeof(struct 
> dpaa_eth_parse_results_t)) {
> + PMD_DRV_LOG(DEBUG, 
> "Checksum offload Err: "
> + "Not enough 
> Headroom "
> + "space for 
> correct Checksum offload."
> + "So Calculating 
> checksum in Software.");
> + dpaa_checksum(mbuf);
> + } else
> + 
> dpaa_checksum_offload(mbuf, &fd_arr[loop],
> + mbuf->buf_addr);
> + }

There is a tx_pkt_prepare() dev_ops.
Does it make sense to move this calculations to that function?

>   } else {
>   PMD_DRV_LOG(DEBUG, "Number of Segments 
> not supported");
>   /* Set frames_to_send & nb_bufs so that

<...>

Re: [dpdk-dev] [PATCH 38/38] net/dpaa: add packet dump for debugging

[Ferruh Yigit]

On 6/16/2017 6:41 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 

Is there a driver documentation, I haven't see any in net/dpaa patches?

<...>
> +CONFIG_RTE_LIBRTE_DPAA_DEBUG_DRIVER_DISPLAY=n
> +CONFIG_RTE_LIBRTE_DPAA_CHECKING=n

This config option is not used at all, can be removed.

<...>
> +#ifdef RTE_LIBRTE_DPAA_DEBUG_DRIVER
> + dpaa_debug_queue_init(&dpaa_intf->debug_queues[
> + DPAA_DEBUG_FQ_RX_ERROR], fman_intf->fqid_rx_err);

Out of curiosity, what exactly done here. Is this a special queue, what
is does? It can be useful if documented more in commit log.

> + dpaa_intf->debug_queues[DPAA_DEBUG_FQ_RX_ERROR].dpaa_intf = dpaa_intf;
> + dpaa_debug_queue_init(&dpaa_intf->debug_queues[
> + DPAA_DEBUG_FQ_TX_ERROR], fman_intf->fqid_tx_err);
> + dpaa_intf->debug_queues[DPAA_DEBUG_FQ_TX_ERROR].dpaa_intf = dpaa_intf;
> +#endif
> +
<...>

Re: [dpdk-dev] [PATCH 18/38] doc: add NXP DPAA PMD documentation

[Ferruh Yigit]

On 6/16/2017 6:40 AM, Shreyansh Jain wrote:
> Signed-off-by: Hemant Agrawal 
> Signed-off-by: Shreyansh Jain 
> ---
>  MAINTAINERS   |   2 +
>  doc/guides/nics/dpaa.rst  | 360 
> ++

As a reminder, you may need to send a web page patch to add dpaa as
supported nic:
http://dpdk.org/doc/nics

>  doc/guides/nics/features/dpaa.ini |   8 +
>  doc/guides/nics/index.rst |   1 +

Re: [dpdk-dev] [PATCH] bug: Use correct arguments in run.py

[Wiles, Keith]

> On Jun 28, 2017, at 7:54 AM, McCullough, Harrison 
>  wrote:
> 
> When using run.py it would occasionally ignore the given command line
> arguments because the wrong variable was used.  Fixed this, along with
> minor changes to reflect more idiomatic Python usage.

Thanks I will integrate these changes into the next version of Pktgen.

> 
> Signed-off-by: Harrison McCullough 
> ---
> tools/run.py | 11 +--
> 1 file changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/tools/run.py b/tools/run.py
> index ec6f45d..f5ec22c 100755
> --- a/tools/run.py
> +++ b/tools/run.py
> @@ -96,7 +96,7 @@ def find_file(arg, t):
> for f in file_list('.', t):
> if os.path.basename(f) == fn:
> return f
> -return "" 
> +return None
> 
> def parse_args():
> ''' Parse the command arguments '''
> @@ -118,7 +118,7 @@ def parse_args():
> print("Run '%s --usage' for further information" % sys.argv[0])
> sys.exit(1)
> 
> -for opt, arg in opts:
> +for opt, _ in opts:
> if opt == "--help" or opt == "-h" or opt == "--usage" or opt == "-u":
> usage()
> sys.exit(0)
> @@ -128,13 +128,12 @@ def parse_args():
> if opt == "--setup" or opt == "-s":
> run_flag = False
> 
> -arg = sys.argv[1:]
> -if arg == "":
> +if not args or len(args) > 1:
> usage()
> sys.exit(1)
> 
> -fn = find_file(arg[0], cfg_ext)
> -if fn != "":
> +fn = find_file(args[0], cfg_ext)
> +if fn:
> cfg_file = fn
> 
> def load_cfg():
> -- 
> 1.9.1
> 

Regards,
Keith