Reply-To: dang@redhat.com
Subject: Re: Questions about pynfs:testLargeData-WRT5
To: "J. Bruce Fields" <bfields@redhat.com>,
        Seiichi Ikarashi <s.ikarashi@jp.fujitsu.com>
Cc: trondmy@primarydata.com, luxy.fnst@cn.fujitsu.com,
        linux-nfs@vger.kernel.org
References: <20180209183112.GC30030@parsley.fieldses.org>
 <6c3159af-17c5-5556-d416-574968bb5e14@cn.fujitsu.com>
 <738139ab-6a55-4c93-008d-56a2332774c9@jp.fujitsu.com>
 <20180213161540.GA26270@parsley.fieldses.org>
 <20180214154909.GA32456@parsley.fieldses.org>
From: Daniel Gryniewicz <dang@redhat.com>
Message-ID: <e2eeabab-7bf2-13cc-ba9d-c8db16f3e49f@redhat.com>
Date: Wed, 14 Feb 2018 11:20:33 -0500
MIME-Version: 1.0
In-Reply-To: <20180214154909.GA32456@parsley.fieldses.org>
Content-Type: text/plain; charset=utf-8; format=flowed
Sender: linux-nfs-owner@vger.kernel.org

This looks sane, and like a good improvement.

Daniel

On 02/14/2018 10:49 AM, J. Bruce Fields wrote:
> On Tue, Feb 13, 2018 at 11:15:40AM -0500, J. Bruce Fields wrote:
>> WRT5 is wrong in any case, I'll fix it.
> 
> How about this?
> 
> --b.
> 
> commit 30f1fff66da3
> Author: J. Bruce Fields <bfields@redhat.com>
> Date:   Tue Feb 13 16:30:27 2018 -0500
> 
>      4.0 server tests: replace WRT5 by better tests
>      
>      This test is sending a write that (at least in the knfsd case) is larger
>      than the server's maximum write size, and expecting it to succeed.
>      That's weird.
>      
>      Remove WRT5 and replace it by two tests.  Both first query the maximum
>      write size.  One does a large (but not too large) write and checks that
>      the data was written correctly.  The other does a too-large write and
>      expects nothing.  (A wide variety of server behavior would be in spec;
>      we do this just in case a buggy server might crash.)
>      
>      Signed-off-by: J. Bruce Fields <bfields@redhat.com>
> 
> diff --git a/nfs4.0/servertests/st_write.py b/nfs4.0/servertests/st_write.py
> index 710452efa11d..004562d2f52b 100644
> --- a/nfs4.0/servertests/st_write.py
> +++ b/nfs4.0/servertests/st_write.py
> @@ -120,17 +120,22 @@ def testNoData(t, env):
>       if compareTimes(time_prior,time_after) != 0:
>           t.fail("WRITE with no data affected time_modify")
>   
> +#WRT5 permanently retired
> +
>   def testLargeData(t, env):
> -    """WRITE with a large amount of data
> +    """WRITE with the maximum size, READ it back and compare
>   
>       FLAGS: write read all
>       DEPEND: MKFILE
> -    CODE: WRT5
> +    CODE: WRT5a
>       """
>       c = env.c1
>       c.init_connection()
> +    maxread, maxwrite = _get_iosize(t, c, c.homedir)
> +    maxio = min(maxread, maxwrite)
>       fh, stateid = c.create_confirm(t.code)
> -    data = "abcdefghijklmnopq" * 0x10000
> +    pattern="abcdefghijklmnop"
> +    data = pattern * (maxio / len(pattern)) + "q" * (maxio % len(pattern))
>       # Write the data
>       pos = 0
>       while pos < len(data):
> @@ -150,6 +155,27 @@ def testLargeData(t, env):
>       if data != newdata:
>           t.fail("READ did not correspond to WRITE with large dataset")
>   
> +def testTooLargeData(t, env):
> +    """WRITE with more than the maximum size
> +
> +    FLAGS: write read all
> +    DEPEND: MKFILE
> +    CODE: WRT5b
> +    """
> +    c = env.c1
> +    c.init_connection()
> +    maxread, maxwrite = _get_iosize(t, c, c.homedir)
> +    fh, stateid = c.create_confirm(t.code)
> +    data = "a" * (maxwrite + 1000000)
> +    try:
> +        # We don't care much what the server does, this is just a check
> +        # to make sure it doesn't crash.
> +        res = c.write_file(fh, data, 0, stateid)
> +    except IOError:
> +        # Linux knfsd closes the socket when the write is too large.
> +        # That's OK.
> +        pass
> +
>   def testDir(t, env):
>       """WRITE to a dir should return NFS4ERR_ISDIR
>   
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>