[PATCH rdma-core] verbs: Corrected description of max_sge and max_sge_rd

[PATCH rdma-core] verbs: Corrected description of max_sge and max_sge_rd

[Yishai Hadas]

From: Parav Pandit <parav-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

This patch corrects ambiguity in documentation of max_sge and max_sge_rd
fields. This ambiguity seems to be created due to [1] and due to mention
of "RD and non RD QPs" in man page.

Here RD can be understood as reliable datagram QP of Infiniband
specification as opposed to "read".

max_sge_rd: Maximum number of sges per WR for RDMA Read operations.
max_sge: Maximum number of sges per WR for QP for non RDMA Read
operations.

[1] http://www.rdmamojo.com/2012/07/13/ibv_query_device/

Signed-off-by: Parav Pandit<parav-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Reviewed-by: Yishai Hadas <yishaih-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---

Pull request was sent:
https://github.com/linux-rdma/rdma-core/pull/131

 libibverbs/man/ibv_query_device.3 | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libibverbs/man/ibv_query_device.3 b/libibverbs/man/ibv_query_device.3
index 35daa53..0e9d286 100644
--- a/libibverbs/man/ibv_query_device.3
+++ b/libibverbs/man/ibv_query_device.3
@@ -33,8 +33,8 @@ uint32_t                hw_ver;                 /* Hardware version */
 int                     max_qp;                 /* Maximum number of supported QPs */
 int                     max_qp_wr;              /* Maximum number of outstanding WR on any work queue */
 int                     device_cap_flags;       /* HCA capabilities mask */
-int                     max_sge;                /* Maximum number of s/g per WR for non-RD QPs */
-int                     max_sge_rd;             /* Maximum number of s/g per WR for RD QPs */
+int                     max_sge;                /* Maximum number of s/g per WR for SQ & RQ of QP for non RDMA Read operations */
+int                     max_sge_rd;             /* Maximum number of s/g per WR for RDMA Read operations */
 int                     max_cq;                 /* Maximum number of supported CQs */
 int                     max_cqe;                /* Maximum number of CQE capacity per CQ */
 int                     max_mr;                 /* Maximum number of supported MRs */
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core] verbs: Corrected description of max_sge and max_sge_rd

[Yuval Shaia]

On Wed, May 10, 2017 at 07:24:25PM +0300, Yishai Hadas wrote:
> From: Parav Pandit <parav-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> 
> This patch corrects ambiguity in documentation of max_sge and max_sge_rd
> fields. This ambiguity seems to be created due to [1] and due to mention
> of "RD and non RD QPs" in man page.
> 
> Here RD can be understood as reliable datagram QP of Infiniband
> specification as opposed to "read".
> 
> max_sge_rd: Maximum number of sges per WR for RDMA Read operations.
> max_sge: Maximum number of sges per WR for QP for non RDMA Read
> operations.

So the new comments makes it clear but can you consider change the name to
reflect that? i.e. it is good to have comments describing variables role
but better to have the names self explanatory.

Something like the following will make comments almost unneeded.
max_sge_rdma_rd: Maximum number of sges per WR for RDMA Read operations.
max_sge_rd: Maximum number of sges per WR for QP for non RDMA Read

My 2 cents.

> 
> [1] http://www.rdmamojo.com/2012/07/13/ibv_query_device/
> 
> Signed-off-by: Parav Pandit<parav-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> Reviewed-by: Yishai Hadas <yishaih-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> ---
> 
> Pull request was sent:
> https://github.com/linux-rdma/rdma-core/pull/131
> 
>  libibverbs/man/ibv_query_device.3 | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/libibverbs/man/ibv_query_device.3 b/libibverbs/man/ibv_query_device.3
> index 35daa53..0e9d286 100644
> --- a/libibverbs/man/ibv_query_device.3
> +++ b/libibverbs/man/ibv_query_device.3
> @@ -33,8 +33,8 @@ uint32_t                hw_ver;                 /* Hardware version */
>  int                     max_qp;                 /* Maximum number of supported QPs */
>  int                     max_qp_wr;              /* Maximum number of outstanding WR on any work queue */
>  int                     device_cap_flags;       /* HCA capabilities mask */
> -int                     max_sge;                /* Maximum number of s/g per WR for non-RD QPs */
> -int                     max_sge_rd;             /* Maximum number of s/g per WR for RD QPs */
> +int                     max_sge;                /* Maximum number of s/g per WR for SQ & RQ of QP for non RDMA Read operations */

Should the line be break here as it exceed 80 characters?

> +int                     max_sge_rd;             /* Maximum number of s/g per WR for RDMA Read operations */
>  int                     max_cq;                 /* Maximum number of supported CQs */
>  int                     max_cqe;                /* Maximum number of CQE capacity per CQ */
>  int                     max_mr;                 /* Maximum number of supported MRs */
> -- 
> 1.8.3.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core] verbs: Corrected description of max_sge and max_sge_rd

[Yishai Hadas]

On 5/11/2017 11:33 AM, Yuval Shaia wrote:
> On Wed, May 10, 2017 at 07:24:25PM +0300, Yishai Hadas wrote:
>> From: Parav Pandit <parav-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>>
>> This patch corrects ambiguity in documentation of max_sge and max_sge_rd
>> fields. This ambiguity seems to be created due to [1] and due to mention
>> of "RD and non RD QPs" in man page.
>>
>> Here RD can be understood as reliable datagram QP of Infiniband
>> specification as opposed to "read".
>>
>> max_sge_rd: Maximum number of sges per WR for RDMA Read operations.
>> max_sge: Maximum number of sges per WR for QP for non RDMA Read
>> operations.
>
> So the new comments makes it clear but can you consider change the name to
> reflect that?

Those fields are exposed to user applications and can't be renamed 
without breaking compatibility. Please note that the notation of "rd" is 
used in other fields as well and are documented correctly (e.g. see man 
page for max_qp_rd_atom), so this patch just aligns the man page with 
other fields around and fixes the ambiguity.

>> diff --git a/libibverbs/man/ibv_query_device.3 b/libibverbs/man/ibv_query_device.3
>> index 35daa53..0e9d286 100644
>> --- a/libibverbs/man/ibv_query_device.3
>> +++ b/libibverbs/man/ibv_query_device.3
>> @@ -33,8 +33,8 @@ uint32_t                hw_ver;                 /* Hardware version */
>>  int                     max_qp;                 /* Maximum number of supported QPs */
>>  int                     max_qp_wr;              /* Maximum number of outstanding WR on any work queue */
>>  int                     device_cap_flags;       /* HCA capabilities mask */
>> -int                     max_sge;                /* Maximum number of s/g per WR for non-RD QPs */
>> -int                     max_sge_rd;             /* Maximum number of s/g per WR for RD QPs */
>> +int                     max_sge;                /* Maximum number of s/g per WR for SQ & RQ of QP for non RDMA Read operations */
>
> Should the line be break here as it exceed 80 characters?
Not really required, this is aligned with having one descriptive line 
per field as done for other fields which are even longer, when using 
'man' it looks fine.

Thanks for your review.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma-core] verbs: Corrected description of max_sge and max_sge_rd

[Yuval Shaia]

On Thu, May 11, 2017 at 12:44:19PM +0300, Yishai Hadas wrote:
> On 5/11/2017 11:33 AM, Yuval Shaia wrote:
> >On Wed, May 10, 2017 at 07:24:25PM +0300, Yishai Hadas wrote:
> >>From: Parav Pandit <parav-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> >>
> >>This patch corrects ambiguity in documentation of max_sge and max_sge_rd
> >>fields. This ambiguity seems to be created due to [1] and due to mention
> >>of "RD and non RD QPs" in man page.
> >>
> >>Here RD can be understood as reliable datagram QP of Infiniband
> >>specification as opposed to "read".
> >>
> >>max_sge_rd: Maximum number of sges per WR for RDMA Read operations.
> >>max_sge: Maximum number of sges per WR for QP for non RDMA Read
> >>operations.
> >
> >So the new comments makes it clear but can you consider change the name to
> >reflect that?
> 
> Those fields are exposed to user applications and can't be renamed without
> breaking compatibility. Please note that the notation of "rd" is used in
> other fields as well and are documented correctly (e.g. see man page for
> max_qp_rd_atom), so this patch just aligns the man page with other fields
> around and fixes the ambiguity.

Make sense

> 
> >>diff --git a/libibverbs/man/ibv_query_device.3 b/libibverbs/man/ibv_query_device.3
> >>index 35daa53..0e9d286 100644
> >>--- a/libibverbs/man/ibv_query_device.3
> >>+++ b/libibverbs/man/ibv_query_device.3
> >>@@ -33,8 +33,8 @@ uint32_t                hw_ver;                 /* Hardware version */
> >> int                     max_qp;                 /* Maximum number of supported QPs */
> >> int                     max_qp_wr;              /* Maximum number of outstanding WR on any work queue */
> >> int                     device_cap_flags;       /* HCA capabilities mask */
> >>-int                     max_sge;                /* Maximum number of s/g per WR for non-RD QPs */
> >>-int                     max_sge_rd;             /* Maximum number of s/g per WR for RD QPs */
> >>+int                     max_sge;                /* Maximum number of s/g per WR for SQ & RQ of QP for non RDMA Read operations */
> >
> >Should the line be break here as it exceed 80 characters?
> Not really required, this is aligned with having one descriptive line per
> field as done for other fields which are even longer, when using 'man' it
> looks fine.
> 
> Thanks for your review.

Reviewed-by: Yuval Shaia <yuval.shaia-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html