Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Christian Schoenebeck <linux_oss@crudebyte.com>
To:     Stefano Stabellini <sstabellini@kernel.org>,
        Dominique Martinet <asmadeus@codewreck.org>
Cc:     GUO Zihua <guozihua@huawei.com>,
        v9fs-developer@lists.sourceforge.net, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/2] 9p/xen: check logical size for buffer size
Date:   Mon, 21 Nov 2022 15:16:10 +0100
Message-ID: <8926859.ZlztnY0qrB@silver>
In-Reply-To: <Y3hADWgV9JeajmfF@codewreck.org>
References: <20221118135542.63400-1-asmadeus@codewreck.org>
 <alpine.DEB.2.22.394.2211181713420.1049131@ubuntu-linux-20-04-desktop>
 <Y3hADWgV9JeajmfF@codewreck.org>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On Saturday, November 19, 2022 3:31:41 AM CET Dominique Martinet wrote:
[...]
> > I made this change and tried the two patches together. Unfortunately I
> > get the following error as soon as I try to write a file:
> > 
> > /bin/sh: can't create /mnt/file: Input/output error
> > 
> > 
> > Next I reverted the second patch and only kept this patch. With that, it
> > worked as usual. It looks like the second patch is the problem. I have
> > not investigated further.
> 
> Thanks -- it's now obvious I shouldn't send patches without testing
> before bedtime...
> I could reproduce easily with virtio as well, this one was silly as well
> (>= instead of >). . . With another problem when zc requests get
> involved, as we don't actually allocate more than 4k for the rpc itself.
> 
> If I adjust it to also check with the zc 'inlen' as follow it appears to
> work:
> https://github.com/martinetd/linux/commit/162015a0dac40eccc9e8311a5eb031596ad35e82
> But that inlen isn't actually precise, and trans_virtio (the only
> transport implementing zc rpc) actually takes some liberty with the
> actual sg size to better fit hardwre, so that doesn't really make
> sense either and we probably should just trust trans_virtio at this
> point?
> 
> This isn't obvious, so I'll just drop this patch for now.
> Checking witih msize isn't any good but it can wait till we sort it out
> as transports now all already check this one way or another; I'd like to
> get the actual fixes out first.
> 
> (Christian, if you have time to look at it and take over I'd appreciate
> it, but there's no hurry.)

OK, I'll look at this.

Best regards,
Christian Schoenebeck