blk-cgroup: skip dying blkg in blkcg_activate_policy()

When switching IO schedulers on a block device, blkcg_activate_policy()
can race with concurrent blkcg deletion, leading to a use-after-free of
the blkg.

T1:				  T2:
elv_iosched_store		  blkg_destroy
elevator_switch			  kill(&blkg->refcnt) // blkg->refcnt=0
...				  blkg_release // call_rcu
blkcg_activate_policy		  __blkg_release
list for blkg			  blkg_free
				  blkg_free_workfn
				  ->pd_free_fn(pd)
blkg_get(blkg) // blkg->refcnt=0->1
				  list_del_init(&blkg->q_node)
				  kfree(blkg)
blkg_put(pinned_blkg) // blkg->refcnt=1->0
blkg_release // call_rcu again
call_rcu(..., __blkg_release)

Fix this by replacing blkg_get() with blkg_tryget(), which fails if
the blkg's refcount has already reached zero. If blkg_tryget() fails,
skip processing this blkg since it's already being destroyed.

The uaf call trace is as follows:

==================================================================
 BUG: KASAN: slab-use-after-free in rcu_accelerate_cbs+0x114/0x120
 Read of size 8 at addr ffff88815a20b5d8 by task bash/1068
 CPU: 0 PID: 1068 Comm: bash Not tainted 6.6.0-g6918ead378dc-dirty #31
 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.1-2.fc37 04/01/2014
Call Trace:
 <IRQ>
 rcu_accelerate_cbs+0x114/0x120
 rcu_report_qs_rdp+0x1fb/0x3e0
 rcu_core+0x4d7/0x6f0
 handle_softirqs+0x198/0x550
 irq_exit_rcu+0x130/0x190
 sysvec_apic_timer_interrupt+0x6e/0x90
 </IRQ>
 <TASK>
 asm_sysvec_apic_timer_interrupt+0x16/0x20

Allocated by task 1031:
 kasan_save_stack+0x1c/0x40
 kasan_set_track+0x21/0x30
 __kasan_kmalloc+0x8b/0x90
 blkg_alloc+0xb6/0x9c0
 blkg_create+0x8c6/0x1010
 blkg_lookup_create+0x2ca/0x660
 bio_associate_blkg_from_css+0xfb/0x4e0
 bio_associate_blkg+0x62/0xf0
 bio_init+0x272/0x8d0
 bio_alloc_bioset+0x45a/0x760
 ext4_bio_write_folio+0x68e/0x10d0
 mpage_submit_folio+0x14a/0x2b0
 mpage_process_page_bufs+0x1b1/0x390
 mpage_prepare_extent_to_map+0xa91/0x1060
 ext4_do_writepages+0x948/0x1c50
 ext4_writepages+0x23f/0x4a0
 do_writepages+0x162/0x5e0
 filemap_fdatawrite_wbc+0x11a/0x180
 __filemap_fdatawrite_range+0x9d/0xd0
 file_write_and_wait_range+0x91/0x110
 ext4_sync_file+0x1c1/0xaa0
 __x64_sys_fsync+0x55/0x90
 do_syscall_64+0x55/0x100
 entry_SYSCALL_64_after_hwframe+0x78/0xe2

Freed by task 24:
 kasan_save_stack+0x1c/0x40
 kasan_set_track+0x21/0x30
 kasan_save_free_info+0x27/0x40
 __kasan_slab_free+0x106/0x180
 __kmem_cache_free+0x162/0x350
 process_one_work+0x573/0xd30
 worker_thread+0x67f/0xc30
 kthread+0x28b/0x350
 ret_from_fork+0x30/0x70
 ret_from_fork_asm+0x1b/0x30

Fixes: f1c006f ("blk-cgroup: synchronize pd_free_fn() from blkg_free_workfn() and blkcg_deactivate_policy()")
Signed-off-by: Zheng Qixing <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>

Author: Zheng Qixing

block/blk-cgroup.c

@@ -1645,9 +1645,10 @@ int blkcg_activate_policy(struct gendisk *disk, const struct blkcg_policy *pol)
 			 * GFP_NOWAIT failed.  Free the existing one and
 			 * prealloc for @blkg w/ GFP_KERNEL.
 			 */
+			if (!blkg_tryget(blkg))
+				continue;
 			if (pinned_blkg)
 				blkg_put(pinned_blkg);
-			blkg_get(blkg);
 			pinned_blkg = blkg;
 
 			spin_unlock_irq(&q->queue_lock);