CVE-2024-42266

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> btrfs: make cow_file_range_inline() honor locked_page on error<br /> <br /> The btrfs buffered write path runs through __extent_writepage() which<br /> has some tricky return value handling for writepage_delalloc().<br /> Specifically, when that returns 1, we exit, but for other return values<br /> we continue and end up calling btrfs_folio_end_all_writers(). If the<br /> folio has been unlocked (note that we check the PageLocked bit at the<br /> start of __extent_writepage()), this results in an assert panic like<br /> this one from syzbot:<br /> <br /> BTRFS: error (device loop0 state EAL) in free_log_tree:3267: errno=-5 IO failure<br /> BTRFS warning (device loop0 state EAL): Skipping commit of aborted transaction.<br /> BTRFS: error (device loop0 state EAL) in cleanup_transaction:2018: errno=-5 IO failure<br /> assertion failed: folio_test_locked(folio), in fs/btrfs/subpage.c:871<br /> ------------[ cut here ]------------<br /> kernel BUG at fs/btrfs/subpage.c:871!<br /> Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI<br /> CPU: 1 PID: 5090 Comm: syz-executor225 Not tainted<br /> 6.10.0-syzkaller-05505-gb1bc554e009e #0<br /> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS<br /> Google 06/27/2024<br /> RIP: 0010:btrfs_folio_end_all_writers+0x55b/0x610 fs/btrfs/subpage.c:871<br /> Code: e9 d3 fb ff ff e8 25 22 c2 fd 48 c7 c7 c0 3c 0e 8c 48 c7 c6 80 3d<br /> 0e 8c 48 c7 c2 60 3c 0e 8c b9 67 03 00 00 e8 66 47 ad 07 90 0b e8<br /> 6e 45 b0 07 4c 89 ff be 08 00 00 00 e8 21 12 25 fe 4c 89<br /> RSP: 0018:ffffc900033d72e0 EFLAGS: 00010246<br /> RAX: 0000000000000045 RBX: 00fff0000000402c RCX: 663b7a08c50a0a00<br /> RDX: 0000000000000000 RSI: 0000000080000000 RDI: 0000000000000000<br /> RBP: ffffc900033d73b0 R08: ffffffff8176b98c R09: 1ffff9200067adfc<br /> R10: dffffc0000000000 R11: fffff5200067adfd R12: 0000000000000001<br /> R13: dffffc0000000000 R14: 0000000000000000 R15: ffffea0001cbee80<br /> FS: 0000000000000000(0000) GS:ffff8880b9500000(0000)<br /> knlGS:0000000000000000<br /> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> CR2: 00007f5f076012f8 CR3: 000000000e134000 CR4: 00000000003506f0<br /> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000<br /> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400<br /> Call Trace:<br /> <br /> __extent_writepage fs/btrfs/extent_io.c:1597 [inline]<br /> extent_write_cache_pages fs/btrfs/extent_io.c:2251 [inline]<br /> btrfs_writepages+0x14d7/0x2760 fs/btrfs/extent_io.c:2373<br /> do_writepages+0x359/0x870 mm/page-writeback.c:2656<br /> filemap_fdatawrite_wbc+0x125/0x180 mm/filemap.c:397<br /> __filemap_fdatawrite_range mm/filemap.c:430 [inline]<br /> __filemap_fdatawrite mm/filemap.c:436 [inline]<br /> filemap_flush+0xdf/0x130 mm/filemap.c:463<br /> btrfs_release_file+0x117/0x130 fs/btrfs/file.c:1547<br /> __fput+0x24a/0x8a0 fs/file_table.c:422<br /> task_work_run+0x24f/0x310 kernel/task_work.c:222<br /> exit_task_work include/linux/task_work.h:40 [inline]<br /> do_exit+0xa2f/0x27f0 kernel/exit.c:877<br /> do_group_exit+0x207/0x2c0 kernel/exit.c:1026<br /> __do_sys_exit_group kernel/exit.c:1037 [inline]<br /> __se_sys_exit_group kernel/exit.c:1035 [inline]<br /> __x64_sys_exit_group+0x3f/0x40 kernel/exit.c:1035<br /> x64_sys_call+0x2634/0x2640<br /> arch/x86/include/generated/asm/syscalls_64.h:232<br /> do_syscall_x64 arch/x86/entry/common.c:52 [inline]<br /> do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83<br /> entry_SYSCALL_64_after_hwframe+0x77/0x7f<br /> RIP: 0033:0x7f5f075b70c9<br /> Code: Unable to access opcode bytes at<br /> 0x7f5f075b709f.<br /> <br /> I was hitting the same issue by doing hundreds of accelerated runs of<br /> generic/475, which also hits IO errors by design.<br /> <br /> I instrumented that reproducer with bpftrace and found that the<br /> undesirable folio_unlock was coming from the following callstack:<br /> <br /> folio_unlock+5<br /> __process_pages_contig+475<br /> cow_file_range_inline.constprop.0+230<br /> cow_file_range+803<br /> btrfs_run_delalloc_range+566<br /> writepage_delalloc+332<br /> __extent_writepage # inlined in my stacktrace, but I added it here<br /> extent_write_cache_pages+622<br /> <br /> Looking at the bisected-to pa<br /> ---truncated---