From: "Li, Zhijian" <lizhijian@cn.fujitsu.com>
To: Leon Romanovsky <leon@kernel.org>,
Jason Gunthorpe <jgg@nvidia.com>,
"Li Zhijian" <lizhijian@cn.fujitsu.com>
Cc: "dledford@redhat.com" <dledford@redhat.com>,
"linux-rdma@vger.kernel.org" <linux-rdma@vger.kernel.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] RDMA/mlx5: return the EFAULT per ibv_advise_mr(3)
Date: Sat, 21 Aug 2021 17:44:43 +0800 [thread overview]
Message-ID: <7b930773-0071-5b96-2a85-718d0ca07bfa@cn.fujitsu.com> (raw)
In-Reply-To: <6b372500-ebc5-bc42-11c5-99de381b2e50@fujitsu.com>
convert to text and send again
Hi Jason & Leon
It reminds me that ibv_advise_mr doesn't mention ENOENT any more which value the API actually returns now.
the ENOENT cases/situations returned by kernel mlx5 implementation is most likely same with EINVALL as its manpage[1].
So shall we return EINVAL instead of ENOENT in kernel side when get_prefetchable_mr returns NULL?
1781 static int mlx5_ib_prefetch_sg_list(struct ib_pd *pd,
1782 enum ib_uverbs_advise_mr_advice advice,
1783 u32 pf_flags, struct ib_sge *sg_list,
1784 u32 num_sge)
1785 {
1786 u32 bytes_mapped = 0;
1787 int ret = 0;
1788 u32 i;
1789
1790 for (i = 0; i < num_sge; ++i) {
1791 struct mlx5_ib_mr *mr;
1792
1793 mr = get_prefetchable_mr(pd, advice, sg_list[i].lkey);
1794 if (!mr)
1795 return -ENOENT;
1796 ret = pagefault_mr(mr, sg_list[i].addr, sg_list[i].length,
1797 &bytes_mapped, pf_flags);
=============
RETURN VALUE
ibv_advise_mr() returns 0 when the call was successful, or the value of errno on failure (which indicates the failure reason).
EOPNOTSUPP
libibverbs or provider driver doesn’t support the ibv_advise_mr() verb (ENOSYS may sometimes be returned by old versions of libibverbs).
ENOTSUP
The advise operation isn’t supported.
EFAULT In one of the following: o When the range requested is out of the MR bounds, or when parts of it are not part of the process address space. o One of the lkeys provided in the scatter gather
list is invalid or with wrong write access.
EINVAL In one of the following: o The PD is invalid. o The flags are invalid.
[1]:https://github.com/linux-rdma/rdma-core/blob/master/libibverbs/man/ibv_advise_mr.3.md
Thanks
on 2021/8/16 10:59, Li, Zhijian wrote:
> On 05/08/2021 14:43, Leon Romanovsky wrote:
>> On Wed, Aug 04, 2021 at 03:50:22PM -0300, Jason Gunthorpe wrote:
>>> On Wed, Aug 04, 2021 at 08:35:30AM +0300, Leon Romanovsky wrote:
>>>> On Tue, Aug 03, 2021 at 03:13:41PM -0300, Jason Gunthorpe wrote:
>>>>> On Tue, Aug 03, 2021 at 08:56:54PM +0300, Leon Romanovsky wrote:
>>>>>> On Tue, Aug 03, 2021 at 01:25:07PM -0300, Jason Gunthorpe wrote:
>>>>>>> On Sun, Aug 01, 2021 at 05:20:50PM +0800, Li Zhijian wrote:
>>>>>>>> ibv_advise_mr(3) says:
>>>>>>>> EFAULT In one of the following: o When the range requested is out of the MR bounds,
>>>>>>>> or when parts of it are not part of the process address space. o One of the
>>>>>>>> lkeys provided in the scatter gather list is invalid or with wrong write access
>>>>>>>>
>>>>>>>> Actually get_prefetchable_mr() will return NULL if it see above conditions
>>>>>>> No, get_prefetchable_mr() returns NULL if the mkey is invalid
>>>>>> And what is this?
>>>>>> 1701 static struct mlx5_ib_mr *
>>>>>> 1702 get_prefetchable_mr(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice,
>>>>>> 1703 u32 lkey)
>>>>>>
>>>>>> ...
>>>>>>
>>>>>> 1721 /* prefetch with write-access must be supported by the MR */
>>>>>> 1722 if (advice == IB_UVERBS_ADVISE_MR_ADVICE_PREFETCH_WRITE &&
>>>>>> 1723 !mr->umem->writable) {
>>>>>> 1724 mr = NULL;
>>>>>> 1725 goto end;
>>>>>> 1726 }
>>>>> I would say that is an invalid mkey
>>>> I see it is as wrong write access.
>>> It just means the man page is wrong
>> ok, it can be a solution too.
> It sounds good. I will try to update the manpage ibv_advise_mr(3) instead.
>
>
> Thanks
> Zhijian
next prev parent reply other threads:[~2021-08-21 9:44 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-01 9:20 [PATCH] RDMA/mlx5: return the EFAULT per ibv_advise_mr(3) Li Zhijian
2021-08-01 11:27 ` Leon Romanovsky
2021-08-03 16:25 ` Jason Gunthorpe
2021-08-03 17:56 ` Leon Romanovsky
2021-08-03 18:13 ` Jason Gunthorpe
2021-08-04 5:35 ` Leon Romanovsky
2021-08-04 18:50 ` Jason Gunthorpe
2021-08-05 6:43 ` Leon Romanovsky
2021-08-16 2:59 ` lizhijian
2021-08-21 9:44 ` Li, Zhijian [this message]
2021-08-25 17:28 ` Jason Gunthorpe
2021-08-26 1:18 ` lizhijian
2021-08-28 8:42 ` Li, Zhijian
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=7b930773-0071-5b96-2a85-718d0ca07bfa@cn.fujitsu.com \
--to=lizhijian@cn.fujitsu.com \
--cc=dledford@redhat.com \
--cc=jgg@nvidia.com \
--cc=leon@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rdma@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).