linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Anna Schumaker <Anna.Schumaker@netapp.com>
To: Daniel Wagner <wagi@monom.org>, <linux-nfs@vger.kernel.org>
Cc: Trond Myklebust <trond.myklebust@primarydata.com>,
	<linux-kernel@vger.kernel.org>,
	Daniel Wagner <daniel.wagner@bmw-carit.de>
Subject: Re: [PATCH 0/2] NFS: Use complete() instead complete_all()
Date: Fri, 23 Sep 2016 09:48:17 -0400	[thread overview]
Message-ID: <ca2d158b-ebdd-a709-a63a-421a11d8ca0a@Netapp.com> (raw)
In-Reply-To: <1474545269-8694-1-git-send-email-wagi@monom.org>

Hi Daniel,

On 09/22/2016 07:54 AM, Daniel Wagner wrote:
> From: Daniel Wagner <daniel.wagner@bmw-carit.de>
> 
> Hi,
> 
> Using complete_all() is not wrong per se but it suggest that there
> might be more than one waiter. For -rt I am reviewing all
> complete_all() users and would like to leave only the real ones in the
> tree. The main problem for -rt about complete_all() is that it can be
> uses inside IRQ context and that can lead to unbounded amount work
> inside the interrupt handler. That is a no no for -rt.
> 
> Besides trying to analys all the code paths to the wait_for_completion()
> call and convince myself that there is only one waiter, I also run
> a few tests:
> 
>  - some fio benchmarks
>  - pynfs
>   -cthon04

Thanks for the patches, and for the extensive testing!  I haven't tried them with xfstests yet, but They look okay to me otherwise.  Assuming I don't see any new failures there I'll plan on adding them for v4.9.

Thanks,
Anna

> 
> Non of them exposed any blocking threads etc. pynfs did showed some
> failures but I am guessing that is just missing implementation:
> 
> pynfs/nfs4.1/testserver.py random:/home/nfs-test --force -v all
>  
> EID50    st_exchange_id.testSSV                                   : FAILURE
>            NFS4Error: OP_EXCHANGE_ID should return NFS4_OK,
>            instead got NFS4ERR_ENCR_ALG_UNSUPP
> LKPP1a   st_lookupp.testLink                                      : RUNNING
> LKPP1a   st_lookupp.testLink                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_SYMLINK, instead got NFS4ERR_NOENT
> LKPP1b   st_lookupp.testBlock                                     : RUNNING
> LKPP1b   st_lookupp.testBlock                                     : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1c   st_lookupp.testChar                                      : RUNNING
> LKPP1c   st_lookupp.testChar                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1d   st_lookupp.testLookupp                                   : RUNNING
> LKPP1d   st_lookupp.testLookupp                                   : PASS
> LKPP1f   st_lookupp.testFifo                                      : RUNNING
> LKPP1f   st_lookupp.testFifo                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1r   st_lookupp.testFile                                      : RUNNING
> LKPP1r   st_lookupp.testFile                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1s   st_lookupp.testSock                                      : RUNNING
> LKPP1s   st_lookupp.testSock                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> 
> 
> PUTFH1a  st_putfh.testLink                                        : RUNNING
> PUTFH1a  st_putfh.testLink                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1b  st_putfh.testBlock                                       : RUNNING
> PUTFH1b  st_putfh.testBlock                                       : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1c  st_putfh.testChar                                        : RUNNING
> PUTFH1c  st_putfh.testChar                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1d  st_putfh.testDir                                         : RUNNING
> PUTFH1d  st_putfh.testDir                                         : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1f  st_putfh.testFifo                                        : RUNNING
> PUTFH1f  st_putfh.testFifo                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1r  st_putfh.testFile                                        : RUNNING
> PUTFH1r  st_putfh.testFile                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1s  st_putfh.testSocket                                      : RUNNING
> PUTFH1s  st_putfh.testSocket                                      : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> 
> 
> RNM1b    st_rename.testValidBlock                                 : RUNNING
> RNM1b    st_rename.testValidBlock                                 : FAILURE
>            OP_CREATE should return NFS4_OK, instead got
>            NFS4ERR_PERM
> RNM1c    st_rename.testValidChar                                  : RUNNING
> RNM1c    st_rename.testValidChar                                  : FAILURE
>            OP_CREATE should return NFS4_OK, instead got
>            NFS4ERR_PERM
> 
> RNM2a    st_rename.testSfhLink                                    : RUNNING
> RNM2a    st_rename.testSfhLink                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2b    st_rename.testSfhBlock                                   : RUNNING
> RNM2b    st_rename.testSfhBlock                                   : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2c    st_rename.testSfhChar                                    : RUNNING
> RNM2c    st_rename.testSfhChar                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2f    st_rename.testSfhFifo                                    : RUNNING
> RNM2f    st_rename.testSfhFifo                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2r    st_rename.testSfhFile                                    : RUNNING
> RNM2r    st_rename.testSfhFile                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2s    st_rename.testSfhSocket                                  : RUNNING
> RNM2s    st_rename.testSfhSocket                                  : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3a    st_rename.testCfhLink                                    : RUNNING
> RNM3a    st_rename.testCfhLink                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3b    st_rename.testCfhBlock                                   : RUNNING
> RNM3b    st_rename.testCfhBlock                                   : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3c    st_rename.testCfhChar                                    : RUNNING
> RNM3c    st_rename.testCfhChar                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3f    st_rename.testCfhFifo                                    : RUNNING
> RNM3f    st_rename.testCfhFifo                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3r    st_rename.testCfhFile                                    : RUNNING
> RNM3r    st_rename.testCfhFile                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3s    st_rename.testCfhSocket                                  : RUNNING
> RNM3s    st_rename.testCfhSocket                                  : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> 
> EID50    st_exchange_id.testSSV                                   : FAILURE
>            NFS4Error: OP_EXCHANGE_ID should return NFS4_OK,
>            instead got NFS4ERR_ENCR_ALG_UNSUPP
> LKPP1a   st_lookupp.testLink                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_SYMLINK, instead got NFS4ERR_NOENT
> LKPP1b   st_lookupp.testBlock                                     : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1c   st_lookupp.testChar                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> 
> LKPP1f   st_lookupp.testFifo                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1r   st_lookupp.testFile                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> LKPP1s   st_lookupp.testSock                                      : FAILURE
>            LOOKUPP with non-dir <cfh> should return
>            NFS4ERR_NOTDIR, instead got NFS4ERR_NOENT
> 
> PUTFH1a  st_putfh.testLink                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1b  st_putfh.testBlock                                       : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1c  st_putfh.testChar                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1d  st_putfh.testDir                                         : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1f  st_putfh.testFifo                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1r  st_putfh.testFile                                        : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> PUTFH1s  st_putfh.testSocket                                      : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> 
> RNM1b    st_rename.testValidBlock                                 : FAILURE
>            OP_CREATE should return NFS4_OK, instead got
>            NFS4ERR_PERM
> RNM1c    st_rename.testValidChar                                  : FAILURE
>            OP_CREATE should return NFS4_OK, instead got
>            NFS4ERR_PERM
> 
> RNM2a    st_rename.testSfhLink                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2b    st_rename.testSfhBlock                                   : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2c    st_rename.testSfhChar                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2f    st_rename.testSfhFifo                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2r    st_rename.testSfhFile                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM2s    st_rename.testSfhSocket                                  : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3a    st_rename.testCfhLink                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3b    st_rename.testCfhBlock                                   : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3c    st_rename.testCfhChar                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3f    st_rename.testCfhFifo                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3r    st_rename.testCfhFile                                    : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> RNM3s    st_rename.testCfhSocket                                  : FAILURE
>            OP_LOOKUP should return NFS4_OK, instead got
>            NFS4ERR_NOENT
> 
> VF1r     st_verify.testMandFile                                   : FAILURE
> 
> cheers,
> daniel
> 
> Daniel Wagner (2):
>   NFS: direct: use complete() instead of complete_all()
>   NFS: cache_lib: use complete() instead of complete_all()
> 
>  fs/nfs/cache_lib.c | 2 +-
>  fs/nfs/direct.c    | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
> 


  parent reply	other threads:[~2016-09-23 13:48 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-09-22 11:54 [PATCH 0/2] NFS: Use complete() instead complete_all() Daniel Wagner
2016-09-22 11:54 ` [PATCH 1/2] NFS: direct: use complete() instead of complete_all() Daniel Wagner
2016-09-22 11:54 ` [PATCH 2/2] NFS: cache_lib: " Daniel Wagner
2016-09-23 13:48 ` Anna Schumaker [this message]
2016-09-26  5:33   ` [PATCH 0/2] NFS: Use complete() instead complete_all() Daniel Wagner
2016-09-26 13:31     ` Anna Schumaker
2016-09-27  7:42       ` Daniel Wagner
2016-09-27  8:33         ` Daniel Wagner
2016-09-27 17:15           ` Anna Schumaker
2016-09-27 17:18           ` Anna Schumaker

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=ca2d158b-ebdd-a709-a63a-421a11d8ca0a@Netapp.com \
    --to=anna.schumaker@netapp.com \
    --cc=daniel.wagner@bmw-carit.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-nfs@vger.kernel.org \
    --cc=trond.myklebust@primarydata.com \
    --cc=wagi@monom.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).