Re: OVERFLOW_KUNIT_TEST fails with BUG: KASAN: stack-out-of-bounds in string_nocheck+0x168/0x1c8 (kernel 6.11-rc2, PowerMac G4 DP)

[Kees Cook <kees@kernel.org>]

On Fri, Aug 09, 2024 at 11:15:37PM +0200, Erhard Furtner wrote:
> Greetings!
> 
> When KASAN is enabled the Overflow KUnit test fails:
> 
> [...]
>     ok 16 shift_nonsense_test
>     # overflow_allocation_test: 11 allocation overflow tests finished
> ==================================================================
> BUG: KASAN: stack-out-of-bounds in string_nocheck+0x168/0x1c8
> Read of size 1 at addr c976be40 by task kunit_try_catch/1843
> 
> CPU: 0 UID: 0 PID: 1843 Comm: kunit_try_catch Tainted: G                 N 6.11.0-rc2-PMacG4 #1
> Tainted: [N]=TEST
> Hardware name: PowerMac3,6 7455 0x80010303 PowerMac
> Call Trace:
> [c992fb80] [c16651c0] dump_stack_lvl+0x80/0xac (unreliable)
> [c992fba0] [c04e0420] print_report+0xdc/0x504
> [c992fc00] [c04e01d8] kasan_report+0xf8/0x108
> [c992fc80] [c16ae4c8] string_nocheck+0x168/0x1c8
> [c992fcf0] [c16b37a4] string+0xa8/0xbc
> [c992fd60] [c16b8134] vsnprintf+0x868/0x1750
> [c992fdf0] [c0b8490c] kvasprintf+0xa4/0x13c
> [c992fe60] [c0b84c3c] kasprintf+0xb4/0xc8
> [c992fed0] [c0f4c954] module_remove_driver+0x1f0/0x2fc
> [c992ff00] [c0f21628] bus_remove_driver+0x1d0/0x240
> [c992ff30] [bfd0cd40] kunit_put_resource+0x128/0x134 [kunit]
> [c992ff50] [bfd0a120] kunit_cleanup+0x140/0x144 [kunit]
> [c992ff90] [bfd10d64] kunit_generic_run_threadfn_adapter+0xf8/0x148 [kunit]
> [c992ffc0] [c00f57e0] kthread+0x36c/0x37c
> [c992fff0] [c0028304] start_kernel_thread+0x10/0x14
> 
> The buggy address belongs to the physical page:
> page: refcount:0 mapcount:0 mapping:00000000 index:0x0 pfn:0x976b
> flags: 0x0(zone=0)
> raw: 00000000 00000000 eef2bb10 00000000 00000000 00000000 ffffffff 00000000
> raw: 00000000
> page dumped because: kasan: bad access detected
> 
> Memory state around the buggy address:
>  c976bd00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>  c976bd80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> >c976be00: 00 00 00 00 00 00 00 00 f1 f1 f1 f1 04 f2 04 f2
>                                    ^
>  c976be80: 00 04 f3 f3 00 00 00 00 00 00 00 00 00 00 00 00
>  c976bf00: 00 00 f1 f1 f1 f1 00 f3 f3 f3 00 00 00 00 00 00
> ==================================================================
> Disabling lock debugging due to kernel taint
>     not ok 17 overflow_allocation_test
>     # overflow_size_helpers_test: 43 overflow size helper tests finished
>     ok 18 overflow_size_helpers_test
>     # overflows_type_test: 378 overflows_type() tests finished
>     ok 19 overflows_type_test
>     # same_type_test: 0 __same_type() tests finished
>     ok 20 same_type_test
>     # castable_to_type_test: 75 castable_to_type() tests finished
>     ok 21 castable_to_type_test
>     ok 22 DEFINE_FLEX_test
> # overflow: pass:21 fail:1 skip:0 total:22
> # Totals: pass:21 fail:1 skip:0 total:22
> not ok 1 overflow
> 
> 
> This is reproducible on my machine and always happens when running the test via 'modprobe -v overflow_kunit'. Without KASAN enabled (but KFENCE) overflow_allocation_test passes.

Hmm, this implies some kind of corruption is sneaking in and the kunit
resource freeing code is exploding. I don't immediately see the problem,
though.

-- 
Kees Cook