linux-kernel - Re: [stable] ext4 fscrypt_get_encryption

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] [day] [month] [year] [list]

Date:   Wed, 13 Jan 2021 20:34:12 +0900
From:   Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>
To:     Eric Biggers <ebiggers@...nel.org>
Cc:     Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jaegeuk Kim <jaegeuk@...nel.org>,
        "Theodore Y. Ts'o" <tytso@....edu>,
        Suleiman Souhlal <suleiman@...gle.com>,
        linux-fscrypt@...r.kernel.org, stable@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [stable] ext4 fscrypt_get_encryption_info() circular locking
 dependency

Hello,

Eric, sorry for the delay.

On (20/12/11 10:03), Eric Biggers wrote:
> > [..]
> > 
> > [ 1598.658233]  __rwsem_down_read_failed_common+0x186/0x201
> > [ 1598.658235]  call_rwsem_down_read_failed+0x14/0x30
> > [ 1598.658238]  down_read+0x2e/0x45
> > [ 1598.658240]  rmap_walk_file+0x73/0x1ce
> > [ 1598.658242]  page_referenced+0x10d/0x154
> > [ 1598.658247]  shrink_active_list+0x1d4/0x475
> > [ 1598.658250]  shrink_node+0x27e/0x661
> > [ 1598.658254]  try_to_free_pages+0x425/0x7ec
> > [ 1598.658258]  __alloc_pages_nodemask+0x80b/0x1514
> > [ 1598.658279]  __do_page_cache_readahead+0xd4/0x1a9
> > [ 1598.658282]  filemap_fault+0x346/0x573
> > [ 1598.658287]  ext4_filemap_fault+0x31/0x44
> 
> Could you provide some more information about what is causing these actual
> lockups for you?  Are there more stack traces?

I think I have some leads, and, just like you said, this deos not appear
to be ext4 related.

A likely root cause for the lockups I'm observing, is that kswapd
and virtio_balloon have reverse locking order for THP pages:

	down_write(mapping->i_mmap_rwsem)  -->  page->PG_locked
vs
	page->PG_locked --> down_read(mapping->i_mmap_rwsem)

	-ss