linux-kernel - Re: [PATCH V3 4/4] 9p: convert to the new mount API

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aTLHoPiC93HTc-VM@pilgrim>
Date: Fri, 5 Dec 2025 12:53:04 +0100
From: Remi Pommarel <repk@...plefau.lt>
To: Dominique Martinet <asmadeus@...ewreck.org>
Cc: Eric Sandeen <sandeen@...hat.com>, v9fs@...ts.linux.dev,
 linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
 ericvh@...nel.org, lucho@...kov.net, linux_oss@...debyte.com,
 eadavis@...com
Subject: Re: [PATCH V3 4/4] 9p: convert to the new mount API

On Thu, Dec 04, 2025 at 12:13:33AM +0900, Dominique Martinet wrote:
> Eric Sandeen wrote on Tue, Dec 02, 2025 at 04:12:36PM -0600:
> > Working on this, but something that confuses me about the current
> > (not for-next) code:
> > 
> > If I mount with "cache=loose" I see this in /proc/mounts:
> > 
> > 127.0.0.1 /mnt 9p rw,relatime,uname=fsgqa,aname=/tmp/9,cache=f,access=user,trans=tcp 0 0
> > 
> > note the "cache=f" thanks to show_options printing "cache=%x"
> > 
> > "mount -o cache=f" is rejected, though, because "f" is not a parseable
> > number.
> > 
> > Shouldn't it be printing "cache=0xf" instead of "cache=f?"
> 
> Definitely should be!
> 
> > (for some reason, though, in my test "remount -o,ro" does still work even with
> > "cache=f" in /proc/mounts but that seems to be a side effect of mount.9p trying
> > to use the new mount API when it shouldn't, or ...???)
> 
> ... and Remi explicitly had cache=loose in his command line, so I'm also
> surprised it worked...
> 
> > I'll send my fix-up patch with a (maybe?) extra bugfix of printing
> > "cache=0x%x" in show_options, and you can see what you think... it could
> > be moved into a pure bugfix patch first if you agree.
> 
> Thank you! I would have been happy to see both together but it does make
> more sense separately, I've just tested and pushed both your patches to
> -next
> 
> 
> I also agree the other show_options look safe enough as they either
> print a string or int. . . .
> Ah, actually I spotted another one:
>         if (v9ses->debug)
>                 seq_printf(m, ",debug=%x", v9ses->debug);
> This needs to be prefixed by 0x as well -- Eric, do you mind if I amend
> your patch 5 with that as well?
> 
> 
> Remi - I did check rootfstype=9p as well and all seems fine but I'd
> appreciate if you could test as well

I just tried your 9p-next branch and the issue is gone for rootfstype=9p
using cache=loose (I've also made sure that I reproduce the issue without
the last two commits of your branch).

So yes for me that fixes it, thanks for the patches.

-- 
Remi