"value":"In the Linux kernel, the following vulnerability has been resolved:\n\ndm-raid456, md/raid456: fix a deadlock for dm-raid456 while io concurrent with reshape\n\nFor raid456, if reshape is still in progress, then IO across reshape\nposition will wait for reshape to make progress. However, for dm-raid,\nin following cases reshape will never make progress hence IO will hang:\n\n1) the array is read-only;\n2) MD_RECOVERY_WAIT is set;\n3) MD_RECOVERY_FROZEN is set;\n\nAfter commit c467e97f079f (\"md/raid6: use valid sector values to determine\nif an I/O should wait on the reshape\") fix the problem that IO across\nreshape position doesn't wait for reshape, the dm-raid test\nshell/lvconvert-raid-reshape.sh start to hang:\n\n[root@fedora ~]# cat /proc/979/stack\n[<0>] wait_woken+0x7d/0x90\n[<0>] raid5_make_request+0x929/0x1d70 [raid456]\n[<0>] md_handle_request+0xc2/0x3b0 [md_mod]\n[<0>] raid_map+0x2c/0x50 [dm_raid]\n[<0>] __map_bio+0x251/0x380 [dm_mod]\n[<0>] dm_submit_bio+0x1f0/0x760 [dm_mod]\n[<0>] __submit_bio+0xc2/0x1c0\n[<0>] submit_bio_noacct_nocheck+0x17f/0x450\n[<0>] submit_bio_noacct+0x2bc/0x780\n[<0>] submit_bio+0x70/0xc0\n[<0>] mpage_readahead+0x169/0x1f0\n[<0>] blkdev_readahead+0x18/0x30\n[<0>] read_pages+0x7c/0x3b0\n[<0>] page_cache_ra_unbounded+0x1ab/0x280\n[<0>] force_page_cache_ra+0x9e/0x130\n[<0>] page_cache_sync_ra+0x3b/0x110\n[<0>] filemap_get_pages+0x143/0xa30\n[<0>] filemap_read+0xdc/0x4b0\n[<0>] blkdev_read_iter+0x75/0x200\n[<0>] vfs_read+0x272/0x460\n[<0>] ksys_read+0x7a/0x170\n[<0>] __x64_sys_read+0x1c/0x30\n[<0>] do_syscall_64+0xc6/0x230\n[<0>] entry_SYSCALL_64_after_hwframe+0x6c/0x74\n\nThis is because reshape can't make progress.\n\nFor md/raid, the problem doesn't exist because register new sync_thread\ndoesn't rely on the IO to be done any more:\n\n1) If array is read-only, it can switch to read-write by ioctl/sysfs;\n2) md/raid never set MD_RECOVERY_WAIT;\n3) If MD_RECOVERY_FROZEN is set, mddev_suspend() doesn't hold\n 'reconfig_mutex', hence it can be cleared and reshape can continue by\n sysfs api 'sync_action'.\n\nHowever, I'm not sure yet how to avoid the problem in dm-raid yet. This\npatch on the one hand make sure raid_message() can't change\nsync_thread() through raid_message() after presuspend(), on the other\nhand detect the above 3 cases before wait for IO do be done in\ndm_suspend(), and let dm-raid requeue those IO."