linux-kernel - Re: [PATCH] mm/filemap: allocate folios according to the blocksize

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: <ZDzGrbXBUuITlxwI@casper.infradead.org>
Date:   Mon, 17 Apr 2023 05:10:21 +0100
From:   Matthew Wilcox <willy@...radead.org>
To:     kernel test robot <oliver.sang@...el.com>
Cc:     Hannes Reinecke <hare@...e.de>, oe-lkp@...ts.linux.dev,
        lkp@...el.com, linux-fsdevel@...r.kernel.org,
        Pankaj Raghav <p.raghav@...sung.com>,
        linux-kernel@...r.kernel.org, mcgrof@...nel.org
Subject: Re: [PATCH] mm/filemap: allocate folios according to the blocksize

On Mon, Apr 17, 2023 at 10:18:37AM +0800, kernel test robot wrote:
> commit: 3f1c33c25c31221a7a27d302ce6aac8e9b71edbb ("[PATCH] mm/filemap: allocate folios according to the blocksize")

> [ 96.797305][ T3511] WARNING: CPU: 3 PID: 3511 at include/linux/pagemap.h:344 split_huge_page_to_list (include/linux/pagemap.h:344 mm/huge_memory.c:2767) 

Oh, funny.  That's:

        WARN_ON_ONCE(mapping_large_folio_support(mapping) == 0);

so it's found a large folio in a mapping which doesn't claim to support
large folios.  This is a good warning; thank you, bot!

> [   96.994758][ T3511] Call Trace:
> [   96.997888][ T3511]  <TASK>
> [ 97.016651][ T3511] truncate_inode_partial_folio (mm/truncate.c:243) 
> [ 97.022376][ T3511] truncate_inode_pages_range (mm/truncate.c:380) 
> [ 97.090958][ T3511] truncate_pagecache (mm/truncate.c:744) 
> [ 97.095638][ T3511] smb3_simple_falloc+0xcbf/0x1840 cifs

Ah, it's coming from smb/cifs.  I bet it's set i_blkbits to something
large.  Looks like we need to do something like I suggested to set the
minimum folio size in the mapping, rather than basing it on
mapping->host->i_blkbits.