v1 can be found here:
http://lkml.kernel.org/g/[email protected]
Unfortunately, I failed to recognize that the very same issue fixed by v1
appears again at another place in fs/ext4/mballoc.c.
Hence a v2 with the [2/2] addressing that second occurence.
Applicable to linux-next-20160318.
Changes to v1:
[1/2] ("fs/ext4: mb_find_order_for_block(): silence UBSAN")
- corrected GCC version from 4.6.0 to 6.0.0 in commit message
[2/2] ("fs/ext4: ext4_mb_init(): silence UBSAN")
- new.
Nicolai Stange (2):
fs/ext4: mb_find_order_for_block(): silence UBSAN
fs/ext4: ext4_mb_init(): silence UBSAN
fs/ext4/mballoc.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
--
2.7.3
Currently, in ext4_mb_init(), there's a loop like the following:
do {
...
offset += 1 << (sb->s_blocksize_bits - i);
i++;
} while (i <= sb->s_blocksize_bits + 1);
Note that the updated offset is used in the loop's next iteration only.
However, at the last iteration, that is at i == sb->s_blocksize_bits + 1,
the shift count becomes equal to (unsigned)-1 > 31 (c.f. C99 6.5.7(3))
and UBSAN reports
UBSAN: Undefined behaviour in fs/ext4/mballoc.c:2621:15
shift exponent 4294967295 is too large for 32-bit type 'int'
[...]
Call Trace:
[<ffffffff818c4d25>] dump_stack+0xbc/0x117
[<ffffffff818c4c69>] ? _atomic_dec_and_lock+0x169/0x169
[<ffffffff819411ab>] ubsan_epilogue+0xd/0x4e
[<ffffffff81941cac>] __ubsan_handle_shift_out_of_bounds+0x1fb/0x254
[<ffffffff81941ab1>] ? __ubsan_handle_load_invalid_value+0x158/0x158
[<ffffffff814b6dc1>] ? kmem_cache_alloc+0x101/0x390
[<ffffffff816fc13b>] ? ext4_mb_init+0x13b/0xfd0
[<ffffffff814293c7>] ? create_cache+0x57/0x1f0
[<ffffffff8142948a>] ? create_cache+0x11a/0x1f0
[<ffffffff821c2168>] ? mutex_lock+0x38/0x60
[<ffffffff821c23ab>] ? mutex_unlock+0x1b/0x50
[<ffffffff814c26ab>] ? put_online_mems+0x5b/0xc0
[<ffffffff81429677>] ? kmem_cache_create+0x117/0x2c0
[<ffffffff816fcc49>] ext4_mb_init+0xc49/0xfd0
[...]
Observe that the mentioned shift exponent, 4294967295, equals (unsigned)-1.
Unless compilers start to do some fancy transformations (which at least
GCC 6.0.0 doesn't currently do), the issue is of cosmetic nature only: the
such calculated value of offset is never used again.
Silence UBSAN by introducing another variable, offset_incr, holding the
next increment to apply to offset and adjust that one by right shifting it
by one position per loop iteration.
Signed-off-by: Nicolai Stange <[email protected]>
---
fs/ext4/mballoc.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 4bc89fe..8dc0d9b 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2585,7 +2585,7 @@ int ext4_mb_init(struct super_block *sb)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
unsigned i, j;
- unsigned offset;
+ unsigned offset, offset_incr;
unsigned max;
int ret;
@@ -2614,11 +2614,13 @@ int ext4_mb_init(struct super_block *sb)
i = 1;
offset = 0;
+ offset_incr = 1 << (sb->s_blocksize_bits - 1);
max = sb->s_blocksize << 2;
do {
sbi->s_mb_offsets[i] = offset;
sbi->s_mb_maxs[i] = max;
- offset += 1 << (sb->s_blocksize_bits - i);
+ offset += offset_incr;
+ offset_incr = offset_incr >> 1;
max = max >> 1;
i++;
} while (i <= sb->s_blocksize_bits + 1);
--
2.7.3
Currently, in mb_find_order_for_block(), there's a loop like the following:
while (order <= e4b->bd_blkbits + 1) {
...
bb += 1 << (e4b->bd_blkbits - order);
}
Note that the updated bb is used in the loop's next iteration only.
However, at the last iteration, that is at order == e4b->bd_blkbits + 1,
the shift count becomes negative (c.f. C99 6.5.7(3)) and UBSAN reports
UBSAN: Undefined behaviour in fs/ext4/mballoc.c:1281:11
shift exponent -1 is negative
[...]
Call Trace:
[<ffffffff818c4d35>] dump_stack+0xbc/0x117
[<ffffffff818c4c79>] ? _atomic_dec_and_lock+0x169/0x169
[<ffffffff819411bb>] ubsan_epilogue+0xd/0x4e
[<ffffffff81941cbc>] __ubsan_handle_shift_out_of_bounds+0x1fb/0x254
[<ffffffff81941ac1>] ? __ubsan_handle_load_invalid_value+0x158/0x158
[<ffffffff816e93a0>] ? ext4_mb_generate_from_pa+0x590/0x590
[<ffffffff816502c8>] ? ext4_read_block_bitmap_nowait+0x598/0xe80
[<ffffffff816e7b7e>] mb_find_order_for_block+0x1ce/0x240
[...]
Unless compilers start to do some fancy transformations (which at least
GCC 6.0.0 doesn't currently do), the issue is of cosmetic nature only: the
such calculated value of bb is never used again.
Silence UBSAN by introducing another variable, bb_incr, holding the next
increment to apply to bb and adjust that one by right shifting it by one
position per loop iteration.
Signed-off-by: Nicolai Stange <[email protected]>
---
fs/ext4/mballoc.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 50e05df..4bc89fe 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -1266,6 +1266,7 @@ static void ext4_mb_unload_buddy(struct ext4_buddy *e4b)
static int mb_find_order_for_block(struct ext4_buddy *e4b, int block)
{
int order = 1;
+ int bb_incr = 1 << (e4b->bd_blkbits - 1);
void *bb;
BUG_ON(e4b->bd_bitmap == e4b->bd_buddy);
@@ -1278,7 +1279,8 @@ static int mb_find_order_for_block(struct ext4_buddy *e4b, int block)
/* this block is part of buddy of order 'order' */
return order;
}
- bb += 1 << (e4b->bd_blkbits - order);
+ bb += bb_incr;
+ bb_incr >>= 1;
order++;
}
return 0;
--
2.7.3