linux-kernel - Re: frequent slab corruption (since a long time)

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: <20060801.220538.89280517.davem@davemloft.net>
Date:	Tue, 01 Aug 2006 22:05:38 -0700 (PDT)
From:	David Miller <davem@...emloft.net>
To:	davej@...hat.com
Cc:	linux-kernel@...r.kernel.org
Subject: Re: frequent slab corruption (since a long time)

From: Dave Jones <davej@...hat.com>
Date: Tue, 1 Aug 2006 22:16:17 -0400

> Anyone have any clues where that value could be coming from?

Some of the dumps in there looks like ethernet headers.  For example,
in comment #7:

000: 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a 5a
010: 5a 5a 00 11 85 6a 0f ef 00 e0 52 cf 6c 00 08 00
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

That looks like an ethernet header for an IPv4 packet to ethernet
destination MAC 00:11:85:6a:0f:ef from ethernet source MAC
00:e0:52:cf:6c:00

But this chunk is OK since we are looking at a neighbouring
INUSE object.

The reocurring corruption:

0b0: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b ff ff ff ff
0c0: 00 00 00 00 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b

is less identifyable.  Although tg3_alloc_rx_skb() shows up
consistently in the neighbouring objects, the actually
corrupted piece is marked by release_mem() which is the
TTY layer.

The corruption is always a 32-bit 0xffffffff followed by
a 32-bit 0x00000000, 12 bytes into the object.

If it's sitting next to RX ethernet packets, it's probably
something in the vacinity of 1500+ bytes in size as that's
how large the RX skb data areas will be that the tg3 driver
allocates unless it is in Jumbo MTU mode.

By default, do_tty_write() will use a chunk size of 2048 for the write
buffer, tty->write_buf, and this is freed up as part of release_mem()
processing.

Another possibility, is the struct tty_struct itself since that is a
sizable structure too.  And this theory is supported by
alloc_tty_struct() being in some of the triggering backtraces.

Perhaps a TTY refcounting problem or race condition of some sort.

What is 12-bytes into the tty_struct on x86?  The struct tty_ldisc,
"ldisc" is.  Oddly enough, this doesn't match up, since we'd expect
TTY_LDISC_MAGIC (0x5403) and instead we see 0xffffffff there.
Also, after the magic, we'd expect the address of the "n_tty"
string in tty_ldisc_N_TTY to show up in the next word, instead
we find NULL (0x00000000) there.

Something is clobbering the ldisc member of this free'd tty_struct is
seems.  Maybe there is some problem with ldisc refcounting.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/