From: Yu Kuai <yukuai3@xxxxxxxxxx>
commit 8d28d0ddb986f56920ac97ae704cc3340a699a30 upstream.
After commit ec6bb299c7c3 ("md/md-bitmap: add 'sync_size' into struct
md_bitmap_stats"), following panic is reported:
Oops: general protection fault, probably for non-canonical address
RIP: 0010:bitmap_get_stats+0x2b/0xa0
Call Trace:
<TASK>
md_seq_show+0x2d2/0x5b0
seq_read_iter+0x2b9/0x470
seq_read+0x12f/0x180
proc_reg_read+0x57/0xb0
vfs_read+0xf6/0x380
ksys_read+0x6c/0xf0
do_syscall_64+0x82/0x170
entry_SYSCALL_64_after_hwframe+0x76/0x7e
Root cause is that bitmap_get_stats() can be called at anytime if mddev
is still there, even if bitmap is destroyed, or not fully initialized.
Deferenceing bitmap in this case can crash the kernel. Meanwhile, the
above commit start to deferencing bitmap->storage, make the problem
easier to trigger.
Fix the problem by protecting bitmap_get_stats() with bitmap_info.mutex.
Cc: stable@xxxxxxxxxxxxxxx # v6.12+
Fixes: 32a7627cf3a3 ("[PATCH] md: optimised resync using Bitmap based intent logging")
Reported-and-tested-by: Harshit Mogalapalli <harshit.m.mogalapalli@xxxxxxxxxx>
Closes: https://lore.kernel.org/linux-raid/ca3a91a2-50ae-4f68-b317-abd9889f3907@xxxxxxxxxx/T/#m6e5086c95201135e4941fe38f9efa76daf9666c5
Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
Link: https://lore.kernel.org/r/20250124092055.4050195-1-yukuai1@xxxxxxxxxxxxxxx
Signed-off-by: Song Liu <song@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
drivers/md/md-bitmap.c | 5 ++++-
drivers/md/md.c | 5 +++++
2 files changed, 9 insertions(+), 1 deletion(-)
--- a/drivers/md/md-bitmap.c
+++ b/drivers/md/md-bitmap.c
@@ -2355,7 +2355,10 @@ static int bitmap_get_stats(void *data,
if (!bitmap)
return -ENOENT;
-
+ if (bitmap->mddev->bitmap_info.external)
+ return -ENOENT;
+ if (!bitmap->storage.sb_page) /* no superblock */
+ return -EINVAL;
sb = kmap_local_page(bitmap->storage.sb_page);
stats->sync_size = le64_to_cpu(sb->sync_size);
kunmap_local(sb);
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -8376,6 +8376,10 @@ static int md_seq_show(struct seq_file *
return 0;
spin_unlock(&all_mddevs_lock);
+
+ /* prevent bitmap to be freed after checking */
+ mutex_lock(&mddev->bitmap_info.mutex);
+
spin_lock(&mddev->lock);
if (mddev->pers || mddev->raid_disks || !list_empty(&mddev->disks)) {
seq_printf(seq, "%s : ", mdname(mddev));
@@ -8451,6 +8455,7 @@ static int md_seq_show(struct seq_file *
seq_printf(seq, "\n");
}
spin_unlock(&mddev->lock);
+ mutex_unlock(&mddev->bitmap_info.mutex);
spin_lock(&all_mddevs_lock);
if (mddev == list_last_entry(&all_mddevs, struct mddev, all_mddevs))
Patches currently in stable-queue which might be from yukuai3@xxxxxxxxxx are
queue-6.12/md-md-bitmap-synchronize-bitmap_get_stats-with-bitmap-lifetime.patch
queue-6.12/md-md-bitmap-factor-behind-write-counters-out-from-bitmap_-start-end-write.patch
queue-6.12/md-md-bitmap-move-bitmap_-start-end-write-to-md-upper-layer.patch
queue-6.12/md-add-a-new-callback-pers-bitmap_sector.patch
queue-6.12/md-raid5-implement-pers-bitmap_sector.patch
queue-6.12/md-md-bitmap-remove-the-last-parameter-for-bimtap_ops-endwrite.patch
queue-6.12/nbd-don-t-allow-reconnect-after-disconnect.patch
