linux-kernel - Re: [REGRESSION] 9pfs issues on 6.12-rc1

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: <20250810051011.1387079-1-arnout@bzzt.net>
Date: Sun, 10 Aug 2025 07:10:11 +0200
From: Arnout Engelen <arnout@...t.net>
To: dhowells@...hat.com
Cc: antony@...nome.org,
	brauner@...nel.org,
	linux-fsdevel@...r.kernel.org,
	linux-kernel@...r.kernel.org,
	maximilian@...sch.me,
	regressions@...ts.linux.dev,
	sedat.dilek@...il.com
Subject: Re: [REGRESSION] 9pfs issues on 6.12-rc1

> Can you tell me what parameters you're mounting 9p with?  Looking at the
> backtrace:
> 
> [   32.390878]  bad_page+0x70/0x110
> [   32.391056]  free_unref_page+0x363/0x4f0
> [   32.391257]  p9_release_pages+0x41/0x90 [9pnet]
> [   32.391627]  p9_virtio_zc_request+0x3d4/0x720 [9pnet_virtio]
> [   32.391896]  ? p9pdu_finalize+0x32/0xa0 [9pnet]
> [   32.392153]  p9_client_zc_rpc.constprop.0+0x102/0x310 [9pnet]
> [   32.392447]  ? kmem_cache_free+0x36/0x370
> [   32.392703]  p9_client_read_once+0x1a6/0x310 [9pnet]
> [   32.392992]  p9_client_read+0x56/0x80 [9pnet]
> [   32.393238]  v9fs_issue_read+0x50/0xd0 [9p]
> [   32.393467]  netfs_read_to_pagecache+0x20c/0x480 [netfs]
> [   32.393832]  netfs_readahead+0x225/0x330 [netfs]
> [   32.394154]  read_pages+0x6a/0x250
> 
> it's using buffered I/O, but when I try and use 9p from qemu, it wants to use
> unbuffered/direct I/O.

The NixOS integration tests mount 9p with 'cache=loose' which triggers the
buffered I/O.

(I'm still seeing an issue in this area on 6.16-rc6, which remains also with
'only' 'cache=lookahead' - I'll do some more analysis before sharing more
about that, though)


Kind regards,

Arnout