Re: [PATCH -next] PM: hibernate: Get block device exclusively when do swsusp_check

On Thu, Oct 21, 2021 at 4:38 PM yebin [off-list ref] wrote:

On 2021/10/21 21:27, Rafael J. Wysocki wrote:

quoted

On Thu, Oct 21, 2021 at 1:38 PM yebin [off-list ref] wrote:

quoted

On 2021/10/21 19:00, Rafael J. Wysocki wrote:

quoted

On Wed, Oct 13, 2021 at 2:06 PM Ye Bin [off-list ref] wrote:

quoted

We got follow issue:
[   89.266592] ------------[ cut here ]------------
[   89.267427] kernel BUG at fs/buffer.c:3020!
[   89.268264] invalid opcode: 0000 [#1] SMP KASAN PTI
[   89.269116] CPU: 7 PID: 1750 Comm: kmmpd-loop0 Not tainted 5.10.0-862.14.0.6.x86_64-08610-gc932cda3cef4-dirty #20
[   89.273169] RIP: 0010:submit_bh_wbc.isra.0+0x538/0x6d0
[   89.277157] RSP: 0018:ffff888105ddfd08 EFLAGS: 00010246
[   89.278093] RAX: 0000000000000005 RBX: ffff888124231498 RCX: ffffffffb2772612
[   89.279332] RDX: 1ffff11024846293 RSI: 0000000000000008 RDI: ffff888124231498
[   89.280591] RBP: ffff8881248cc000 R08: 0000000000000001 R09: ffffed1024846294
[   89.281851] R10: ffff88812423149f R11: ffffed1024846293 R12: 0000000000003800
[   89.283095] R13: 0000000000000001 R14: 0000000000000000 R15: ffff8881161f7000
[   89.284342] FS:  0000000000000000(0000) GS:ffff88839b5c0000(0000) knlGS:0000000000000000
[   89.285711] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   89.286701] CR2: 00007f166ebc01a0 CR3: 0000000435c0e000 CR4: 00000000000006e0
[   89.287919] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   89.289138] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[   89.290368] Call Trace:
[   89.290842]  write_mmp_block+0x2ca/0x510
[   89.292218]  kmmpd+0x433/0x9a0
[   89.294902]  kthread+0x2dd/0x3e0
[   89.296268]  ret_from_fork+0x22/0x30
[   89.296906] Modules linked in:

We can reproduce this issue as follow:
1. mkfs.ext4 -O mmp  /dev/sda -b 1024
2. mount /dev/sda /home/test
3. echo "/dev/sda" > /sys/power/resume
4. wait a moment we will get exception

The sequence of issue is as follows:
         Thread1                       Thread2
mount /dev/sda /home/test
get s_mmp_bh  --> has mapped flag
start kmmpd thread
                                  echo "/dev/sda" > /sys/power/resume
                                    resume_store
                                      software_resume
                                        swsusp_check
                                          set_blocksize
                                            truncate_inode_pages_range
                                              truncate_cleanup_page
                                                block_invalidatepage
                                                  discard_buffer --> clean mapped flag
write_mmp_block
    submit_bh
      submit_bh_wbc
        BUG_ON(!buffer_mapped(bh)) --> trigger bug_on

To solve this issue, get block device exclusively when do swsusp_check.

And why exactly is this going to help?

If a  block device is already  mounted, then do  resume, it will discard
buffer
which file system being used.  That can cause the file system to crash and
may even cause the system to reset abnormally.

I see.

Is there a way to indicate in swsusp_check() that the access will be
read-only and so it doesn't care whether or not the block device is in
use by someone else?

The root reason is that call  'set_blocksize' in 'swsusp_check' which
maybe invalidate
block device's page.

I see.

Moreover, the file system and swap partition share storage space, which
will lead to
data confusion.

quoted

quoted

If the "/sys/power/resume" sysfs interface is unrestricted, it will provide
criminals with a way to attack the system.

quoted

quoted

Signed-off-by: Ye Bin <redacted>
---
   kernel/power/swap.c | 5 +++--
   1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/kernel/power/swap.c b/kernel/power/swap.c
index 9ec418955556..26c0bd2a50da 100644
--- a/kernel/power/swap.c
+++ b/kernel/power/swap.c

@@ -1521,9 +1521,10 @@ int swsusp_read(unsigned int *flags_p)
   int swsusp_check(void)
   {
          int error;
+       void *holder;

          hib_resume_bdev = blkdev_get_by_dev(swsusp_resume_device,
-                                           FMODE_READ, NULL);
+                                           FMODE_READ | FMODE_EXCL, &holder);

So you need to explain to me how this works.

As we call 'blkdev_get_by_path' in 'mount_bdev'  pass mode includes
FMODE_EXCL.
So if 'swsusp_check' try to get block device with mode includes
FMODE_EXCL will failed.

I think if a block device is used as a swap partition, it cannot be
mounted at the
same time. Conversely, if a block device is already mounted, it cannot
be used as
a swap partition.

Well, what if a hibernation image is located in a swap file?  This is
one of the cases that need to be supported.

I don't know much about this, but I think the process can call
swsusp_check, the
hibernation image must be loaded into the block device.

Yes, it is invalid to call swsusp_check() without a hibernation image
in the target swap partition or swap file if that's what you mean.

However, in the case of a swap file, the image may be there in theory
even if the filesystem holding the swap file is mounted.

I refer to the process of swapon. If the swap partition is a block
device, it is also
opened exclusively.

swapon
     claim_swapfile

There are two cases, though, the S_ISBLK() one and the S_ISREG() one,
and the latter is the swap file case I believe.

static int claim_swapfile(struct swap_info_struct *p, struct inode *inode)
{
         int error;

         if (S_ISBLK(inode->i_mode)) {    -->If a block device is
specified, it is also opened exclusively here.
                 p->bdev = blkdev_get_by_dev(inode->i_rdev,
                                    FMODE_READ | FMODE_WRITE |
FMODE_EXCL, p);
                 if (IS_ERR(p->bdev)) {
                         error = PTR_ERR(p->bdev);
                         p->bdev = NULL;
                         return error;
                 }
                 p->old_block_size = block_size(p->bdev);
                 error = set_blocksize(p->bdev, PAGE_SIZE);
                 if (error < 0)
                         return error;
                 /*
                  * Zoned block devices contain zones that have a sequential
                  * write only restriction.  Hence zoned block devices
are not
                  * suitable for swapping.  Disallow them here.
                  */
                 if (blk_queue_is_zoned(p->bdev->bd_disk->queue))
                         return -EINVAL;
                 p->flags |= SWP_BLKDEV;
         } else if (S_ISREG(inode->i_mode)) {
                 p->bdev = inode->i_sb->s_bdev;

Here, it is not exclusive.

Anyway, I think that the change made by the patch should be fine,
because it mostly affects restore and I'm not expecting anyone to
restore the system from an image in a swap file while the filesystem
holding it is mounted.

So I'm going to apply it with some changelog edits.

Thanks!

         }

         return 0;
}

quoted

quoted

quoted

quoted

          if (!IS_ERR(hib_resume_bdev)) {
                  set_blocksize(hib_resume_bdev, PAGE_SIZE);
                  clear_page(swsusp_header);

@@ -1545,7 +1546,7 @@ int swsusp_check(void)

   put:
                  if (error)
-                       blkdev_put(hib_resume_bdev, FMODE_READ);
+                       blkdev_put(hib_resume_bdev, FMODE_READ | FMODE_EXCL);
                  else
                          pr_debug("Image signature found, resuming\n");
          } else {
--

.

.