CVE-2025-68823

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ublk: fix deadlock when reading partition table<br /> <br /> When one process(such as udev) opens ublk block device (e.g., to read<br /> the partition table via bdev_open()), a deadlock[1] can occur:<br /> <br /> 1. bdev_open() grabs disk-&gt;open_mutex<br /> 2. The process issues read I/O to ublk backend to read partition table<br /> 3. In __ublk_complete_rq(), blk_update_request() or blk_mq_end_request()<br /> runs bio-&gt;bi_end_io() callbacks<br /> 4. If this triggers fput() on file descriptor of ublk block device, the<br /> work may be deferred to current task&amp;#39;s task work (see fput() implementation)<br /> 5. This eventually calls blkdev_release() from the same context<br /> 6. blkdev_release() tries to grab disk-&gt;open_mutex again<br /> 7. Deadlock: same task waiting for a mutex it already holds<br /> <br /> The fix is to run blk_update_request() and blk_mq_end_request() with bottom<br /> halves disabled. This forces blkdev_release() to run in kernel work-queue<br /> context instead of current task work context, and allows ublk server to make<br /> forward progress, and avoids the deadlock.<br /> <br /> [axboe: rewrite comment in ublk]