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(-)
>
next prev 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).