Re: [PATCH] loop: Fix freeze if configured block size is not supported

On Mon, Aug 21, 2017 at 09:11:32PM +0200, Milan Broz wrote:

On 08/21/2017 08:47 PM, Omar Sandoval wrote:

quoted

On Fri, Aug 18, 2017 at 03:07:33PM +0200, Milan Broz wrote:

quoted

The commit f2c6df7dbf9a60e1cd9941f9fb376d4d9ad1e8dd

    loop: support 4k physical blocksize

adds support for loop block size with only specific block sizes.

If the size is not supported, the code returns -EINVAL keeping
the loop queue frozen. This causes that device could be locked
for a long time by processes trying to scan device (udev).
(And also causing subsequent LOOP_CLR_FD operations noop.)

Fix it by using goto to proper exit location with queue unfreeze.

(The same bug is for setting crypt attribute but this code is
probably no more used. Patch fixes it as well though.)

Signed-off-by: Milan Broz <redacted>

Heh, I sent the same patch [1] only hours before :) v2 is here [2] if
you want to give it a reviewed-by.

Yes, and I noticed it 10 seconds after I sent my patch :)
You can add reviewed by, if it helps anything...

Thanks!

Actually you fixed another problems there with following patches
(physical blocks sizes), we discussed this with Karel
that it need some changes because the original patch caused
that reported blocks differs between old and new kernel (lsblk -t),
even if block size was not used.

Yes there was definitely some funkiness there, let me know if I covered
everything or if I need to make more changes.