Skip to content

Commit

Permalink
btrfs: fix u32 overflows when left shifting stripe_nr
Browse files Browse the repository at this point in the history 
[BUG]
David reported an ASSERT() get triggered during fio load on 8 devices
with data/raid6 and metadata/raid1c3:

  fio --rw=randrw --randrepeat=1 --size=3000m \
	  --bsrange=512b-64k --bs_unaligned \
	  --ioengine=libaio --fsync=1024 \
	  --name=job0 --name=job1 \

The ASSERT() is from rbio_add_bio() of raid56.c:

	ASSERT(orig_logical >= full_stripe_start &&
	       orig_logical + orig_len <= full_stripe_start +
	       rbio->nr_data * BTRFS_STRIPE_LEN);

Which is checking if the target rbio is crossing the full stripe
boundary.

  [100.789] assertion failed: orig_logical >= full_stripe_start && orig_logical + orig_len <= full_stripe_start + rbio->nr_data * BTRFS_STRIPE_LEN, in fs/btrfs/raid56.c:1622
  [100.795] ------------[ cut here ]------------
  [100.796] kernel BUG at fs/btrfs/raid56.c:1622!
  [100.797] invalid opcode: 0000 [#1] PREEMPT SMP KASAN
  [100.798] CPU: 1 PID: 100 Comm: kworker/u8:4 Not tainted 6.4.0-rc6-default+ torvalds#124
  [100.799] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552-rebuilt.opensuse.org 04/01/2014
  [100.802] Workqueue: writeback wb_workfn (flush-btrfs-1)
  [100.803] RIP: 0010:rbio_add_bio+0x204/0x210 [btrfs]
  [100.806] RSP: 0018:ffff888104a8f300 EFLAGS: 00010246
  [100.808] RAX: 00000000000000a1 RBX: ffff8881075907e0 RCX: ffffed1020951e01
  [100.809] RDX: 0000000000000000 RSI: 0000000000000008 RDI: 0000000000000001
  [100.811] RBP: 0000000141d20000 R08: 0000000000000001 R09: ffff888104a8f04f
  [100.813] R10: ffffed1020951e09 R11: 0000000000000003 R12: ffff88810e87f400
  [100.815] R13: 0000000041d20000 R14: 0000000144529000 R15: ffff888101524000
  [100.817] FS:  0000000000000000(0000) GS:ffff88811ac00000(0000) knlGS:0000000000000000
  [100.821] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  [100.822] CR2: 000055d54e44c270 CR3: 000000010a9a1006 CR4: 00000000003706a0
  [100.824] Call Trace:
  [100.825]  <TASK>
  [100.825]  ? die+0x32/0x80
  [100.826]  ? do_trap+0x12d/0x160
  [100.827]  ? rbio_add_bio+0x204/0x210 [btrfs]
  [100.827]  ? rbio_add_bio+0x204/0x210 [btrfs]
  [100.829]  ? do_error_trap+0x90/0x130
  [100.830]  ? rbio_add_bio+0x204/0x210 [btrfs]
  [100.831]  ? handle_invalid_op+0x2c/0x30
  [100.833]  ? rbio_add_bio+0x204/0x210 [btrfs]
  [100.835]  ? exc_invalid_op+0x29/0x40
  [100.836]  ? asm_exc_invalid_op+0x16/0x20
  [100.837]  ? rbio_add_bio+0x204/0x210 [btrfs]
  [100.837]  raid56_parity_write+0x64/0x270 [btrfs]
  [100.838]  btrfs_submit_chunk+0x26e/0x800 [btrfs]
  [100.840]  ? btrfs_bio_init+0x80/0x80 [btrfs]
  [100.841]  ? release_pages+0x503/0x6d0
  [100.842]  ? folio_unlock+0x2f/0x60
  [100.844]  ? __folio_put+0x60/0x60
  [100.845]  ? btrfs_do_readpage+0xae0/0xae0 [btrfs]
  [100.847]  btrfs_submit_bio+0x21/0x60 [btrfs]
  [100.847]  submit_one_bio+0x6a/0xb0 [btrfs]
  [100.849]  extent_write_cache_pages+0x395/0x680 [btrfs]
  [100.850]  ? __extent_writepage+0x520/0x520 [btrfs]
  [100.851]  ? mark_usage+0x190/0x190
  [100.852]  extent_writepages+0xdb/0x130 [btrfs]
  [100.853]  ? extent_write_locked_range+0x480/0x480 [btrfs]
  [100.854]  ? mark_usage+0x190/0x190
  [100.854]  ? attach_extent_buffer_page+0x220/0x220 [btrfs]
  [100.855]  ? reacquire_held_locks+0x178/0x280
  [100.856]  ? writeback_sb_inodes+0x245/0x7f0
  [100.857]  do_writepages+0x102/0x2e0
  [100.858]  ? page_writeback_cpu_online+0x10/0x10
  [100.859]  ? __lock_release.isra.0+0x14a/0x4d0
  [100.860]  ? reacquire_held_locks+0x280/0x280
  [100.861]  ? __lock_acquired+0x1e9/0x3d0
  [100.862]  ? do_raw_spin_lock+0x1b0/0x1b0
  [100.863]  __writeback_single_inode+0x94/0x450
  [100.864]  writeback_sb_inodes+0x372/0x7f0
  [100.864]  ? lock_sync+0xd0/0xd0
  [100.865]  ? do_raw_spin_unlock+0x93/0xf0
  [100.866]  ? sync_inode_metadata+0xc0/0xc0
  [100.867]  ? rwsem_optimistic_spin+0x340/0x340
  [100.868]  __writeback_inodes_wb+0x70/0x130
  [100.869]  wb_writeback+0x2d1/0x530
  [100.869]  ? __writeback_inodes_wb+0x130/0x130
  [100.870]  ? lockdep_hardirqs_on_prepare.part.0+0xf1/0x1c0
  [100.870]  wb_do_writeback+0x3eb/0x480
  [100.871]  ? wb_writeback+0x530/0x530
  [100.871]  ? mark_lock_irq+0xcd0/0xcd0
  [100.872]  wb_workfn+0xe0/0x3f0<

[CAUSE]
Commit a97699d ("btrfs: replace map_lookup->stripe_len by
BTRFS_STRIPE_LEN") changes how we calculate the map length, to reduce
u64 division.

Function btrfs_max_io_len() is to get the length to the stripe boundary.

It calculates the full stripe start offset (inside the chunk) by the
following code:

		*full_stripe_start =
			rounddown(*stripe_nr, nr_data_stripes(map)) <<
			BTRFS_STRIPE_LEN_SHIFT;

The calculation itself is fine, but the value returned by rounddown() is
dependent on both @stripe_nr (which is u32) and nr_data_stripes() (which
returned int).

Thus the result is also u32, then we do the left shift, which can
overflow u32.

If such overflow happens, @full_stripe_start will be a value way smaller
than @offset, causing later "full_stripe_len - (offset -
*full_stripe_start)" to underflow, thus make later length calculation to
have no stripe boundary limit, resulting a write bio to exceed stripe
boundary.

There are some other locations like this, with a u32 @stripe_nr got left
shift, which can lead to a similar overflow.

[FIX]
Fix all @stripe_nr with left shift with a type cast to u64 before the
left shift.

Those involved @stripe_nr or similar variables are recording the stripe
number inside the chunk, which is small enough to be contained by u32,
but their offset inside the chunk can not fit into u32.

Thus for those specific left shifts, a type cast to u64 is necessary so
this patch does not touch them and the code will be cleaned up in the
future to keep the fix minimal.

Reported-by: David Sterba <[email protected]>
Fixes: a97699d ("btrfs: replace map_lookup->stripe_len by BTRFS_STRIPE_LEN")
Tested-by: David Sterba <[email protected]>
Signed-off-by: Qu Wenruo <[email protected]>
Signed-off-by: David Sterba <[email protected]>
  • Loading branch information
@adam900710 @kdave
adam900710 authored and kdave committed Jun 21, 2023
1 parent 8a4a0b2 commit 00bc86e
Showing 1 changed file with 7 additions and 5 deletions.
12 changes: 7 additions & 5 deletions fs/btrfs/volumes.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5985,12 +5985,12 @@ struct btrfs_discard_stripe *btrfs_map_discard(struct btrfs_fs_info *fs_info,
stripe_nr = offset >> BTRFS_STRIPE_LEN_SHIFT;

/* stripe_offset is the offset of this block in its stripe */
stripe_offset = offset - (stripe_nr << BTRFS_STRIPE_LEN_SHIFT);
stripe_offset = offset - ((u64)stripe_nr << BTRFS_STRIPE_LEN_SHIFT);

stripe_nr_end = round_up(offset + length, BTRFS_STRIPE_LEN) >>
BTRFS_STRIPE_LEN_SHIFT;
stripe_cnt = stripe_nr_end - stripe_nr;
stripe_end_offset = (stripe_nr_end << BTRFS_STRIPE_LEN_SHIFT) -
stripe_end_offset = ((u64)stripe_nr_end << BTRFS_STRIPE_LEN_SHIFT) -
(offset + length);
/*
* after this, stripe_nr is the number of stripes on this
Expand Down Expand Up @@ -6033,7 +6033,7 @@ struct btrfs_discard_stripe *btrfs_map_discard(struct btrfs_fs_info *fs_info,
for (i = 0; i < *num_stripes; i++) {
stripes[i].physical =
map->stripes[stripe_index].physical +
stripe_offset + (stripe_nr << BTRFS_STRIPE_LEN_SHIFT);
stripe_offset + ((u64)stripe_nr << BTRFS_STRIPE_LEN_SHIFT);
stripes[i].dev = map->stripes[stripe_index].dev;

if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
Expand Down Expand Up @@ -6199,9 +6199,11 @@ static u64 btrfs_max_io_len(struct map_lookup *map, enum btrfs_map_op op,
* not ensured to be power of 2.
*/
*full_stripe_start =
rounddown(*stripe_nr, nr_data_stripes(map)) <<
(u64)rounddown(*stripe_nr, nr_data_stripes(map)) <<
BTRFS_STRIPE_LEN_SHIFT;

ASSERT(*full_stripe_start + full_stripe_len > offset);
ASSERT(*full_stripe_start <= offset);
/*
* For writes to RAID56, allow to write a full stripe set, but
* no straddling of stripe sets.
Expand All @@ -6224,7 +6226,7 @@ static void set_io_stripe(struct btrfs_io_stripe *dst, const struct map_lookup *
{
dst->dev = map->stripes[stripe_index].dev;
dst->physical = map->stripes[stripe_index].physical +
stripe_offset + (stripe_nr << BTRFS_STRIPE_LEN_SHIFT);
stripe_offset + ((u64)stripe_nr << BTRFS_STRIPE_LEN_SHIFT);
}

int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
Expand Down

0 comments on commit 00bc86e

Please sign in to comment.