Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::3:2 as permitted sender) client-ip=2620:137:e000::3:2;
MIME-Version: 1.0
References: <f1d0b234-e650-0f6e-0f5d-126b3d51d1eb@gmail.com>
 <6E4B8EB9-FA7E-4ABD-81AB-6FB0EF07EA46@oracle.com> <f90866e3-8c54-8a9c-c100-f5550c31b664@gmail.com>
 <9691788F-2B62-4EB5-8879-CEDABD1B9D4B@oracle.com> <78c81a97-4324-c2df-27c0-917436ddee07@gmail.com>
 <6908E735-6912-4AEC-8AD7-995F2F8A144A@oracle.com> <CA1D7203-72FB-4681-AD60-F1D80A57C154@oracle.com>
 <8cb19861-7fd2-4d12-bd76-78bbb8fcee7c@gmail.com> <2660F09A-5823-4581-8977-4F41153724D6@oracle.com>
In-Reply-To: <2660F09A-5823-4581-8977-4F41153724D6@oracle.com>
From:   Olga Kornievskaia <aglo@umich.edu>
Date:   Tue, 26 Sep 2023 17:52:57 -0400
Message-ID: <CAN-5tyHHibZbJh-A=0n3d=Z7GVL1xuYz8H2ygu8HbnsrtUTUxQ@mail.gmail.com>
Subject: Re: Data corruption with 5.10.x client -> 6.5.x server
To:     Chuck Lever III <chuck.lever@oracle.com>
Cc:     =?UTF-8?Q?Mantas_Mikul=C4=97nas?= <grawity@gmail.com>,
        Linux NFS Mailing List <linux-nfs@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Sep 26, 2023 at 10:08=E2=80=AFAM Chuck Lever III <chuck.lever@oracl=
e.com> wrote:
>
>
>
> > On Sep 26, 2023, at 12:41 AM, Mantas Mikul=C4=97nas <grawity@gmail.com>=
 wrote:
> >
> > On 25/09/2023 22.22, Chuck Lever III wrote:
> >>> On Sep 24, 2023, at 2:24 PM, Chuck Lever III <chuck.lever@oracle.com>=
 wrote:
> >>>
> >>>> On Sep 24, 2023, at 12:51 PM, Mantas Mikul=C4=97nas <grawity@gmail.c=
om> wrote:
> >>>>
> >>>> Right, whereas on the server, the first file is filled with "Hello W=
orld (4092 bytes)" as I originally tried to narrow down the issue.
> >>>>
> >>>> Meanwhile, 6.4.x (Arch) clients don't seem to be having any problems=
 with the same server, and with seemingly the same mount options.
> >>>>
> >>>> Thanks for looking into it!<nfs_krb5i_working_6.4client.pcap>
> >>> I found /a/ problem with the nfsd-fixes branch and krb5i, but
> >>> maybe not /your/ problem, and it's with a recent client. Scrounging
> >>> a v5.10-vintage client is a little more work, we'll see if that's
> >>> needed for confirming an eventual fix.
> >> The issue I reproduced appears to be unrelated.
> >>
> >> I'm wondering if I can get you to bisect the server kernel using
> >> your v5.10 client to test? good =3D v6.4, bad =3D v6.5 should do it.
> >
> > Yeah, I will try to bisect but it'll probably take a day or two.
>
> That's great, thank you!
>
> I'm looking into setting up a virtual guest with v5.10 just in case.
> Turns out v5.10 does not build on Fedora latest.
>

I can reproduce this with upstream client do dd if=3D/mnt/4092byteslen
of=3D/dev/null bs=3D4092 count=3D1 iflag=3Ddirect

> --
> Chuck Lever
>
>