diff mbox

arm64: ubsan: select ARCH_HAS_UBSAN_SANITIZE_ALL

Message ID 1454716218-4918-1-git-send-email-yang.shi@linaro.org (mailing list archive)
State New, archived
Headers show

Commit Message

Yang Shi Feb. 5, 2016, 11:50 p.m. UTC
To enable UBSAN on arm64, ARCH_HAS_UBSAN_SANITIZE_ALL need to be selected.

Basic kernel bootup test is passed on arm64 with CONFIG_UBSAN_SANITIZE_ALL
enabled.

Signed-off-by: Yang Shi <yang.shi@linaro.org>
---
 arch/arm64/Kconfig | 1 +
 1 file changed, 1 insertion(+)

Comments

Andrey Ryabinin Feb. 8, 2016, 7:52 a.m. UTC | #1
On 02/06/2016 02:50 AM, Yang Shi wrote:
> To enable UBSAN on arm64, ARCH_HAS_UBSAN_SANITIZE_ALL need to be selected.
> 
> Basic kernel bootup test is passed on arm64 with CONFIG_UBSAN_SANITIZE_ALL
> enabled.
> 
> Signed-off-by: Yang Shi <yang.shi@linaro.org>
> ---


Acked-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Mark Rutland Feb. 8, 2016, 11:46 a.m. UTC | #2
Hi,

On Fri, Feb 05, 2016 at 03:50:18PM -0800, Yang Shi wrote:
> To enable UBSAN on arm64, ARCH_HAS_UBSAN_SANITIZE_ALL need to be selected.
> 
> Basic kernel bootup test is passed on arm64 with CONFIG_UBSAN_SANITIZE_ALL
> enabled.
> 
> Signed-off-by: Yang Shi <yang.shi@linaro.org>
> ---
>  arch/arm64/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index 8cc6228..1c29e20 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -14,6 +14,7 @@ config ARM64
>  	select ARCH_WANT_OPTIONAL_GPIOLIB
>  	select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
>  	select ARCH_WANT_FRAME_POINTERS
> +	select ARCH_HAS_UBSAN_SANITIZE_ALL
>  	select ARM_AMBA
>  	select ARM_ARCH_TIMER
>  	select ARM_GIC

I gave this a go, and I got a couple of splats (included below) when
booting an Ubuntu 14.04 arm64 rootfs. I'm using Linato 15.08 GCC 5.1 to
compile an arm64 defconfig, and I see the issue with v4.5-rc2 and
v4.5-rc3.

I will dig into that and file a report shortl, unless someone has
already reported the same issue.

Did you see any failures in your testing? For reference, which kernel
version, compiler, and config were you using?

This patch itself looks good, so FWIW:

Tested-by: Mark Rutland <mark.rutland@arm.com>

Thanks,
Mark.

[    3.804750] ================================================================================
[    3.813176] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:2612:15
[    3.819431] shift exponent 4294967295 is too large for 32-bit type 'int'
[    3.826121] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.5.0-rc2+ #48
[    3.832463] Hardware name: AMD Overdrive/Supercharger/Default string, BIOS ROD0085E 11/23/2015
[    3.841060] Call trace:
[    3.843499] [<ffffffc00008d7b8>] dump_backtrace+0x0/0x298
[    3.848887] [<ffffffc00008da64>] show_stack+0x14/0x20
[    3.853929] [<ffffffc00056e0f0>] dump_stack+0xe0/0x178
[    3.859056] [<ffffffc0005b734c>] ubsan_epilogue+0x14/0x50
[    3.864444] [<ffffffc0005b7748>] __ubsan_handle_shift_out_of_bounds+0xe0/0x138
[    3.871655] [<ffffffc0003e1734>] ext4_mb_init+0x84c/0x920
[    3.877043] [<ffffffc0003ba294>] ext4_fill_super+0x2eac/0x4958
[    3.882866] [<ffffffc0002c1008>] mount_bdev+0x180/0x1e8
[    3.888079] [<ffffffc0003adf8c>] ext4_mount+0x14/0x20
[    3.893118] [<ffffffc0002c23f4>] mount_fs+0x44/0x1c8
[    3.898073] [<ffffffc0002ed9c0>] vfs_kern_mount+0x50/0x1a8
[    3.903547] [<ffffffc0002f3d90>] do_mount+0x240/0x1478
[    3.908673] [<ffffffc0002f54d0>] SyS_mount+0x90/0xf8
[    3.913627] [<ffffffc000eb2750>] mount_block_root+0x22c/0x3c4
[    3.919361] [<ffffffc000eb2a08>] mount_root+0x120/0x138
[    3.924574] [<ffffffc000eb2b5c>] prepare_namespace+0x13c/0x184
[    3.930396] [<ffffffc000eb21bc>] kernel_init_freeable+0x390/0x3b4
[    3.936479] [<ffffffc000bb4a78>] kernel_init+0x10/0xe0
[    3.941606] [<ffffffc000086cd0>] ret_from_fork+0x10/0x40
[    3.946905] ================================================================================

[    5.566166] ================================================================================
[    5.574596] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:1274:11
[    5.580851] shift exponent -1 is negative
[    5.584851] CPU: 4 PID: 1028 Comm: mount Not tainted 4.5.0-rc2+ #48
[    5.591105] Hardware name: AMD Overdrive/Supercharger/Default string, BIOS ROD0085E 11/23/2015
[    5.599702] Call trace:
[    5.602142] [<ffffffc00008d7b8>] dump_backtrace+0x0/0x298
[    5.607530] [<ffffffc00008da64>] show_stack+0x14/0x20
[    5.612572] [<ffffffc00056e0f0>] dump_stack+0xe0/0x178
[    5.617700] [<ffffffc0005b734c>] ubsan_epilogue+0x14/0x50
[    5.623088] [<ffffffc0005b7748>] __ubsan_handle_shift_out_of_bounds+0xe0/0x138
[    5.630300] [<ffffffc0003d2a04>] mb_find_order_for_block+0x154/0x1b0
[    5.636641] [<ffffffc0003d2b2c>] mb_find_extent+0xcc/0x548
[    5.642116] [<ffffffc0003de6a8>] ext4_mb_complex_scan_group+0xe8/0x4e8
[    5.648632] [<ffffffc0003ded7c>] ext4_mb_regular_allocator+0x2d4/0x648
[    5.655148] [<ffffffc0003e2b4c>] ext4_mb_new_blocks+0x344/0x7e0
[    5.661056] [<ffffffc0003cbf54>] ext4_ext_map_blocks+0x684/0xf68
[    5.667052] [<ffffffc000393664>] ext4_map_blocks+0x12c/0x500
[    5.672699] [<ffffffc000398df4>] ext4_writepages+0x47c/0xe38
[    5.678348] [<ffffffc00020da20>] do_writepages+0x48/0xc8
[    5.683649] [<ffffffc0001f9100>] __filemap_fdatawrite_range+0x70/0xe8
[    5.690078] [<ffffffc0001f91b0>] filemap_flush+0x18/0x20
[    5.695378] [<ffffffc000394b64>] ext4_alloc_da_blocks+0x3c/0x78
[    5.701285] [<ffffffc0003ac1c8>] ext4_rename+0x690/0xe38
[    5.706585] [<ffffffc0003ac98c>] ext4_rename2+0x1c/0x40
[    5.711800] [<ffffffc0002d0510>] vfs_rename+0x2c0/0xa90
[    5.717013] [<ffffffc0002d661c>] SyS_renameat2+0x464/0x5c0
[    5.722486] [<ffffffc0002d6788>] SyS_renameat+0x10/0x18
[    5.727700] [<ffffffc000086d30>] el0_svc_naked+0x24/0x28
[    5.732998] ================================================================================
Yang Shi Feb. 8, 2016, 5:09 p.m. UTC | #3
On 2/8/2016 3:46 AM, Mark Rutland wrote:
> Hi,
>
> On Fri, Feb 05, 2016 at 03:50:18PM -0800, Yang Shi wrote:
>> To enable UBSAN on arm64, ARCH_HAS_UBSAN_SANITIZE_ALL need to be selected.
>>
>> Basic kernel bootup test is passed on arm64 with CONFIG_UBSAN_SANITIZE_ALL
>> enabled.
>>
>> Signed-off-by: Yang Shi <yang.shi@linaro.org>
>> ---
>>   arch/arm64/Kconfig | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
>> index 8cc6228..1c29e20 100644
>> --- a/arch/arm64/Kconfig
>> +++ b/arch/arm64/Kconfig
>> @@ -14,6 +14,7 @@ config ARM64
>>   	select ARCH_WANT_OPTIONAL_GPIOLIB
>>   	select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
>>   	select ARCH_WANT_FRAME_POINTERS
>> +	select ARCH_HAS_UBSAN_SANITIZE_ALL
>>   	select ARM_AMBA
>>   	select ARM_ARCH_TIMER
>>   	select ARM_GIC
>
> I gave this a go, and I got a couple of splats (included below) when
> booting an Ubuntu 14.04 arm64 rootfs. I'm using Linato 15.08 GCC 5.1 to
> compile an arm64 defconfig, and I see the issue with v4.5-rc2 and
> v4.5-rc3.
>
> I will dig into that and file a report shortl, unless someone has
> already reported the same issue.
>
> Did you see any failures in your testing? For reference, which kernel
> version, compiler, and config were you using?

Yes, I did. I'm using 4.5-rc1 with gcc 5.2. And, I got one more splat 
and was digging into it.

I saw your report to ext4 maintainers. I tried to have a quick fix, but 
it sounds not work well. And, that code does look suspicious. Let's see 
what the ext4 maintainers say.

Thanks,
Yang

>
> This patch itself looks good, so FWIW:
>
> Tested-by: Mark Rutland <mark.rutland@arm.com>
>
> Thanks,
> Mark.
>
> [    3.804750] ================================================================================
> [    3.813176] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:2612:15
> [    3.819431] shift exponent 4294967295 is too large for 32-bit type 'int'
> [    3.826121] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.5.0-rc2+ #48
> [    3.832463] Hardware name: AMD Overdrive/Supercharger/Default string, BIOS ROD0085E 11/23/2015
> [    3.841060] Call trace:
> [    3.843499] [<ffffffc00008d7b8>] dump_backtrace+0x0/0x298
> [    3.848887] [<ffffffc00008da64>] show_stack+0x14/0x20
> [    3.853929] [<ffffffc00056e0f0>] dump_stack+0xe0/0x178
> [    3.859056] [<ffffffc0005b734c>] ubsan_epilogue+0x14/0x50
> [    3.864444] [<ffffffc0005b7748>] __ubsan_handle_shift_out_of_bounds+0xe0/0x138
> [    3.871655] [<ffffffc0003e1734>] ext4_mb_init+0x84c/0x920
> [    3.877043] [<ffffffc0003ba294>] ext4_fill_super+0x2eac/0x4958
> [    3.882866] [<ffffffc0002c1008>] mount_bdev+0x180/0x1e8
> [    3.888079] [<ffffffc0003adf8c>] ext4_mount+0x14/0x20
> [    3.893118] [<ffffffc0002c23f4>] mount_fs+0x44/0x1c8
> [    3.898073] [<ffffffc0002ed9c0>] vfs_kern_mount+0x50/0x1a8
> [    3.903547] [<ffffffc0002f3d90>] do_mount+0x240/0x1478
> [    3.908673] [<ffffffc0002f54d0>] SyS_mount+0x90/0xf8
> [    3.913627] [<ffffffc000eb2750>] mount_block_root+0x22c/0x3c4
> [    3.919361] [<ffffffc000eb2a08>] mount_root+0x120/0x138
> [    3.924574] [<ffffffc000eb2b5c>] prepare_namespace+0x13c/0x184
> [    3.930396] [<ffffffc000eb21bc>] kernel_init_freeable+0x390/0x3b4
> [    3.936479] [<ffffffc000bb4a78>] kernel_init+0x10/0xe0
> [    3.941606] [<ffffffc000086cd0>] ret_from_fork+0x10/0x40
> [    3.946905] ================================================================================
>
> [    5.566166] ================================================================================
> [    5.574596] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:1274:11
> [    5.580851] shift exponent -1 is negative
> [    5.584851] CPU: 4 PID: 1028 Comm: mount Not tainted 4.5.0-rc2+ #48
> [    5.591105] Hardware name: AMD Overdrive/Supercharger/Default string, BIOS ROD0085E 11/23/2015
> [    5.599702] Call trace:
> [    5.602142] [<ffffffc00008d7b8>] dump_backtrace+0x0/0x298
> [    5.607530] [<ffffffc00008da64>] show_stack+0x14/0x20
> [    5.612572] [<ffffffc00056e0f0>] dump_stack+0xe0/0x178
> [    5.617700] [<ffffffc0005b734c>] ubsan_epilogue+0x14/0x50
> [    5.623088] [<ffffffc0005b7748>] __ubsan_handle_shift_out_of_bounds+0xe0/0x138
> [    5.630300] [<ffffffc0003d2a04>] mb_find_order_for_block+0x154/0x1b0
> [    5.636641] [<ffffffc0003d2b2c>] mb_find_extent+0xcc/0x548
> [    5.642116] [<ffffffc0003de6a8>] ext4_mb_complex_scan_group+0xe8/0x4e8
> [    5.648632] [<ffffffc0003ded7c>] ext4_mb_regular_allocator+0x2d4/0x648
> [    5.655148] [<ffffffc0003e2b4c>] ext4_mb_new_blocks+0x344/0x7e0
> [    5.661056] [<ffffffc0003cbf54>] ext4_ext_map_blocks+0x684/0xf68
> [    5.667052] [<ffffffc000393664>] ext4_map_blocks+0x12c/0x500
> [    5.672699] [<ffffffc000398df4>] ext4_writepages+0x47c/0xe38
> [    5.678348] [<ffffffc00020da20>] do_writepages+0x48/0xc8
> [    5.683649] [<ffffffc0001f9100>] __filemap_fdatawrite_range+0x70/0xe8
> [    5.690078] [<ffffffc0001f91b0>] filemap_flush+0x18/0x20
> [    5.695378] [<ffffffc000394b64>] ext4_alloc_da_blocks+0x3c/0x78
> [    5.701285] [<ffffffc0003ac1c8>] ext4_rename+0x690/0xe38
> [    5.706585] [<ffffffc0003ac98c>] ext4_rename2+0x1c/0x40
> [    5.711800] [<ffffffc0002d0510>] vfs_rename+0x2c0/0xa90
> [    5.717013] [<ffffffc0002d661c>] SyS_renameat2+0x464/0x5c0
> [    5.722486] [<ffffffc0002d6788>] SyS_renameat+0x10/0x18
> [    5.727700] [<ffffffc000086d30>] el0_svc_naked+0x24/0x28
> [    5.732998] ================================================================================
>
Mark Rutland Feb. 8, 2016, 5:28 p.m. UTC | #4
On Mon, Feb 08, 2016 at 09:09:46AM -0800, Shi, Yang wrote:
> On 2/8/2016 3:46 AM, Mark Rutland wrote:
> >Hi,
> >
> >On Fri, Feb 05, 2016 at 03:50:18PM -0800, Yang Shi wrote:
> >>To enable UBSAN on arm64, ARCH_HAS_UBSAN_SANITIZE_ALL need to be selected.
> >>
> >>Basic kernel bootup test is passed on arm64 with CONFIG_UBSAN_SANITIZE_ALL
> >>enabled.
> >>
> >>Signed-off-by: Yang Shi <yang.shi@linaro.org>
> >>---
> >>  arch/arm64/Kconfig | 1 +
> >>  1 file changed, 1 insertion(+)
> >>
> >>diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> >>index 8cc6228..1c29e20 100644
> >>--- a/arch/arm64/Kconfig
> >>+++ b/arch/arm64/Kconfig
> >>@@ -14,6 +14,7 @@ config ARM64
> >>  	select ARCH_WANT_OPTIONAL_GPIOLIB
> >>  	select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
> >>  	select ARCH_WANT_FRAME_POINTERS
> >>+	select ARCH_HAS_UBSAN_SANITIZE_ALL
> >>  	select ARM_AMBA
> >>  	select ARM_ARCH_TIMER
> >>  	select ARM_GIC
> >
> >I gave this a go, and I got a couple of splats (included below) when
> >booting an Ubuntu 14.04 arm64 rootfs. I'm using Linato 15.08 GCC 5.1 to
> >compile an arm64 defconfig, and I see the issue with v4.5-rc2 and
> >v4.5-rc3.
> >
> >I will dig into that and file a report shortl, unless someone has
> >already reported the same issue.
> >
> >Did you see any failures in your testing? For reference, which kernel
> >version, compiler, and config were you using?
> 
> Yes, I did. I'm using 4.5-rc1 with gcc 5.2. And, I got one more
> splat and was digging into it.

Ok. Glad to hear it's not just me seeing it.

In future it would be best to point out issues you've spotted so that
reviewers don't confuse them for false positives.

Thanks,
Mark.
Catalin Marinas Feb. 8, 2016, 6:15 p.m. UTC | #5
On Fri, Feb 05, 2016 at 03:50:18PM -0800, Yang Shi wrote:
> To enable UBSAN on arm64, ARCH_HAS_UBSAN_SANITIZE_ALL need to be selected.
> 
> Basic kernel bootup test is passed on arm64 with CONFIG_UBSAN_SANITIZE_ALL
> enabled.
> 
> Signed-off-by: Yang Shi <yang.shi@linaro.org>

Applied. Thanks.
diff mbox

Patch

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 8cc6228..1c29e20 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -14,6 +14,7 @@  config ARM64
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
 	select ARCH_WANT_FRAME_POINTERS
+	select ARCH_HAS_UBSAN_SANITIZE_ALL
 	select ARM_AMBA
 	select ARM_ARCH_TIMER
 	select ARM_GIC