Re: [PATCH 1/1] iscsi_ibft: KASAN false positive failure occurs in ibft_init()
From: George Kennedy
Date: Fri Feb 12 2021 - 13:07:23 EST
On 2/12/2021 10:36 AM, David Hildenbrand wrote:
On 12.02.21 14:51, Dmitry Vyukov wrote:
On Fri, Feb 12, 2021 at 2:31 PM George Kennedy
<george.kennedy@xxxxxxxxxx> wrote:
On 2/10/2021 4:51 PM, George Kennedy wrote:
On 2/3/2021 2:35 PM, Dmitry Vyukov wrote:
On Wed, Feb 3, 2021 at 8:29 PM Konrad Rzeszutek Wilk
<konrad@xxxxxxxxxx> wrote:
Hey Dmitry, Rafael, George, please see below..
On Wed, Jan 27, 2021 at 10:10:07PM +0100, Dmitry Vyukov wrote:
On Wed, Jan 27, 2021 at 9:01 PM George Kennedy
<george.kennedy@xxxxxxxxxx> wrote:
Hi Dmitry,
On 1/27/2021 1:48 PM, Dmitry Vyukov wrote:
On Wed, Jan 27, 2021 at 7:44 PM Konrad Rzeszutek Wilk
<konrad.wilk@xxxxxxxxxx> wrote:
On Tue, Jan 26, 2021 at 01:03:21PM -0500, George Kennedy wrote:
During boot of kernel with CONFIG_KASAN the following KASAN false
positive failure will occur when ibft_init() reads the
ACPI iBFT table: BUG: KASAN: use-after-free in ibft_init
The ACPI iBFT table is not allocated, and the iscsi driver uses
a pointer to it to calculate checksum, etc. KASAN complains
about this pointer with use-after-free, which this is not.
Andrey, Alexander, Dmitry,
I think this is the right way for this, but was wondering if
you have
other suggestions?
Thanks!
Hi George, Konrad,
Please provide a sample KASAN report and kernel version to match
line numbers.
5.4.17-2102.200.0.0.20210106_0000
[ 24.413536] iBFT detected.
[ 24.414074]
==================================================================
[ 24.407342] BUG: KASAN: use-after-free in ibft_init+0x134/0xb8b
[ 24.407342] Read of size 4 at addr ffff8880be452004 by task
swapper/0/1
[ 24.407342]
[ 24.407342] CPU: 1 PID: 1 Comm: swapper/0 Not tainted
5.4.17-2102.200.0.0.20210106_0000.syzk #1
[ 24.407342] Hardware name: QEMU Standard PC (i440FX + PIIX,
1996), BIOS 0.0.0 02/06/2015
[ 24.407342] Call Trace:
[ 24.407342] dump_stack+0xd4/0x119
[ 24.407342] ? ibft_init+0x134/0xb8b
[ 24.407342] print_address_description.constprop.6+0x20/0x220
[ 24.407342] ? ibft_init+0x134/0xb8b
[ 24.407342] ? ibft_init+0x134/0xb8b
[ 24.407342] __kasan_report.cold.9+0x37/0x77
[ 24.407342] ? ibft_init+0x134/0xb8b
[ 24.407342] kasan_report+0x14/0x1b
[ 24.407342] __asan_report_load_n_noabort+0xf/0x11
[ 24.407342] ibft_init+0x134/0xb8b
[ 24.407342] ? dmi_sysfs_init+0x1a5/0x1a5
[ 24.407342] ? dmi_walk+0x72/0x89
[ 24.407342] ? ibft_check_initiator_for+0x159/0x159
[ 24.407342] ? rvt_init_port+0x110/0x101
[ 24.407342] ? ibft_check_initiator_for+0x159/0x159
[ 24.407342] do_one_initcall+0xc3/0x44d
[ 24.407342] ? perf_trace_initcall_level+0x410/0x405
[ 24.407342] kernel_init_freeable+0x551/0x673
[ 24.407342] ? start_kernel+0x94b/0x94b
[ 24.407342] ? __sanitizer_cov_trace_const_cmp1+0x1a/0x1c
[ 24.407342] ? __kasan_check_write+0x14/0x16
[ 24.407342] ? rest_init+0xe6/0xe6
[ 24.407342] kernel_init+0x16/0x1bd
[ 24.407342] ? rest_init+0xe6/0xe6
[ 24.407342] ret_from_fork+0x2b/0x36
[ 24.407342]
[ 24.407342] The buggy address belongs to the page:
[ 24.407342] page:ffffea0002f91480 refcount:0 mapcount:0
mapping:0000000000000000 index:0x1
[ 24.407342] flags: 0xfffffc0000000()
[ 24.407342] raw: 000fffffc0000000 ffffea0002fca588
ffffea0002fb1a88 0000000000000000
[ 24.407342] raw: 0000000000000001 0000000000000000
00000000ffffffff 0000000000000000
[ 24.407342] page dumped because: kasan: bad access detected
[ 24.407342]
[ 24.407342] Memory state around the buggy address:
[ 24.407342] ffff8880be451f00: ff ff ff ff ff ff ff ff ff ff ff
ff ff ff ff ff
[ 24.407342] ffff8880be451f80: ff ff ff ff ff ff ff ff ff ff ff
ff ff ff ff ff
[ 24.407342] >ffff8880be452000: ff ff ff ff ff ff ff ff ff ff ff
ff ff ff ff ff
[ 24.407342] ^
[ 24.407342] ffff8880be452080: ff ff ff ff ff ff ff ff ff ff ff
ff ff ff ff ff
[ 24.407342] ffff8880be452100: ff ff ff ff ff ff ff ff ff ff ff
ff ff ff ff ff
[ 24.407342]
==================================================================
[ 24.407342] Disabling lock debugging due to kernel taint
[ 24.451021] Kernel panic - not syncing: panic_on_warn set ...
[ 24.452002] CPU: 1 PID: 1 Comm: swapper/0 Tainted: G B
5.4.17-2102.200.0.0.20210106_0000.syzk #1
[ 24.452002] Hardware name: QEMU Standard PC (i440FX + PIIX,
1996), BIOS 0.0.0 02/06/2015
[ 24.452002] Call Trace:
[ 24.452002] dump_stack+0xd4/0x119
[ 24.452002] ? ibft_init+0x102/0xb8b
[ 24.452002] panic+0x28f/0x6e0
[ 24.452002] ? __warn_printk+0xe0/0xe0
[ 24.452002] ? ibft_init+0x134/0xb8b
[ 24.452002] ? add_taint+0x68/0xb3
[ 24.452002] ? add_taint+0x68/0xb3
[ 24.452002] ? ibft_init+0x134/0xb8b
[ 24.452002] ? ibft_init+0x134/0xb8b
[ 24.452002] end_report+0x4c/0x54
[ 24.452002] __kasan_report.cold.9+0x55/0x77
[ 24.452002] ? ibft_init+0x134/0xb8b
[ 24.452002] kasan_report+0x14/0x1b
[ 24.452002] __asan_report_load_n_noabort+0xf/0x11
[ 24.452002] ibft_init+0x134/0xb8b
[ 24.452002] ? dmi_sysfs_init+0x1a5/0x1a5
[ 24.452002] ? dmi_walk+0x72/0x89
[ 24.452002] ? ibft_check_initiator_for+0x159/0x159
[ 24.452002] ? rvt_init_port+0x110/0x101
[ 24.452002] ? ibft_check_initiator_for+0x159/0x159
[ 24.452002] do_one_initcall+0xc3/0x44d
[ 24.452002] ? perf_trace_initcall_level+0x410/0x405
[ 24.452002] kernel_init_freeable+0x551/0x673
[ 24.452002] ? start_kernel+0x94b/0x94b
[ 24.452002] ? __sanitizer_cov_trace_const_cmp1+0x1a/0x1c
[ 24.452002] ? __kasan_check_write+0x14/0x16
[ 24.452002] ? rest_init+0xe6/0xe6
[ 24.452002] kernel_init+0x16/0x1bd
[ 24.452002] ? rest_init+0xe6/0xe6
[ 24.452002] ret_from_fork+0x2b/0x36
[ 24.452002] Dumping ftrace buffer:
[ 24.452002] ---------------------------------
[ 24.452002] swapper/-1 1.... 24564337us : rdmaip_init:
2924: rdmaip_init: Active Bonding is DISABLED
[ 24.452002] ---------------------------------
[ 24.452002] Kernel Offset: disabled
[ 24.452002] Rebooting in 1 seconds..
Why does KASAN think the address is freed? For that to happen that
memory should have been freed. I don't remember any similar false
positives from KASAN, so this looks a bit suspicious.
I'm not sure why KASAN thinks the address is freed. There are
other modules where KASAN/KCOV is disabled on boot.
Could this be for a similar reason?
Most of these files are disabled because they cause recursion in
instrumentation, or execute too early in bootstrap process (before
kasan_init).
Somehow the table pointer in ibft_init points to a freed page. I
tracked it down to here:
https://elixir.bootlin.com/linux/v5.4.17/source/drivers/acpi/acpica/tbutils.c#L399
but I can't find where this table_desc->pointer comes from.
Perhaps it
It is what the BIOS generated. It usually points to some memory
location in right under 4GB and the BIOS stashes the DSDT, iBFT, and
other tables in there.
uses some allocation method that's not supported by KASAN? However,
it's the only such case that I've seen, so it's a bit weird.
Could it
use something like memblock_alloc? Or maybe that page was in fact
freed?... Too bad KASAN does not print free stack for pages, maybe
it's not too hard to do if CONFIG_PAGE_OWNER is enabled...
Hm, there is a comment in the acpi_get_table speaking about the
requirement of having a acpi_put_table and:
* DESCRIPTION: Finds and verifies an ACPI table. Table must be
in the
* RSDT/XSDT.
* Note that an early stage acpi_get_table() call
must
be paired
* with an early stage acpi_put_table() call.
otherwise the table
* pointer mapped by the early stage mapping
implementation may be
* erroneously unmapped by the late stage unmapping
implementation
* in an acpi_put_table() invoked during the late
stage.
*
Which would imply that I should use acpi_put_table in the error path
(see below a patch), but also copy the structure instead of
depending
on ACPI keeping it mapped for me. I think.
Hi Konrad,
Thanks for looking into this.
If ACPI unmaps this page, that would perfectly explain the KASAN
report.
George, does this patch eliminate the KASAN report for you?
Hi Dmitry,
No luck with the patch. Tried high level bisect instead. Here are the
results:
BUG: KASAN: use-after-free in ibft_init+0x134/0xc49
Bisect status:
v5.11-rc6 Sun Jan 31 13:50:09 2021 -0800 Failed
v5.11-rc1 Sun Dec 27 15:30:22 2020 -0800 Failed
v5.10 Sun Dec 13 14:41:30 2020 -0800 Failed
v5.10-rc6 Sun Nov 29 15:50:50 2020 -0800 Failed
v5.10-rc5 Sun Nov 22 15:36:08 2020 -0800 Failed
v5.10-rc4 Sun Nov 15 16:44:31 2020 -0800 Failed
v5.10-rc3 Sun Nov 8 16:10:16 2020 -0800 Failed
v5.10-rc2 Sun Nov 1 14:43:52 2020 -0800 Failed
v5.10-rc1 Sun Oct 25 15:14:11 2020 -0700 Failed
v5.9 Sun Oct 11 14:15:50 2020 -0700 OK - 10 reboots so
far w/o kasan failure
So, will look at what changed between v5.9 and v5.10-rc1
git bisect has identified the following as the offending commit:
2020-10-16 torvalds@xxxxxxxxxxxxxxxxxxxx - 7fef431 2020-10-15 David
Hildenbrand mm/page_alloc: place pages to tail in __free_pages_core()
I think this suggests that this is a real use-after-free and the
page_alloc change just exposed a latent bug.
That commit is ridiculously good at finding latent BUGs. Memory gets
allocated and used in the reverse order. So could be that this is
exposing yet another latent bug.
For example:
https://lkml.kernel.org/r/20201213225517.3838501-1-linus.walleij@xxxxxxxxxx
Or a longer discussion at:
https://lkml.kernel.org/r/87361onphy.fsf_-_@xxxxxxxxxxxxxx
Anyhow, let me know in case it looks like this time it's actually my
commit that broke something (although somewhat unlikely I have to say).
Thank you Dmitry & David,
Good read and agree - great smoke test for latent bugs.
Continuing to search for the real bug.
George