Re: kernel BUG in gfs2_withdraw

[Glass Su]

> On Aug 31, 2024, at 10:38, Hui Guo <guohui.study@xxxxxxxxx> wrote:
> 
> Hi Kernel Maintainers,
> we found a crash "kernel BUG in gfs2_withdraw" (it seems a bug in
> gfs2) in upstream:
> This bug seems to have been triggered before and fixed, but it can
> still be triggered now.

IIUC, there is a new “bug”. See comment below.
> 
> HEAD Commit: d5d547aa7b51467b15d9caa86b116f8c2507c72a(Merge tag
> 'random-6.11-rc6-for-linus')
> console output:
> https://github.com/androidAppGuard/KernelBugs/blob/main/d5d547aa7b51467b15d9caa86b116f8c2507c72a/0d1c927593a03040214498e5b44a9d7881a1def6/log0
> kernel config: https://github.com/androidAppGuard/KernelBugs/blob/main/6.11.config
> syz reproducer:
> https://github.com/androidAppGuard/KernelBugs/blob/main/d5d547aa7b51467b15d9caa86b116f8c2507c72a/0d1c927593a03040214498e5b44a9d7881a1def6/repro.prog
> C reproducer: https://github.com/androidAppGuard/KernelBugs/blob/main/d5d547aa7b51467b15d9caa86b116f8c2507c72a/0d1c927593a03040214498e5b44a9d7881a1def6/repro.cprog
> the state file of fs:
> https://github.com/androidAppGuard/KernelBugs/blob/main/d5d547aa7b51467b15d9caa86b116f8c2507c72a/0d1c927593a03040214498e5b44a9d7881a1def6/mount_0.gz
> 
> Best,
> Hui Guo
> 
> The following context is the crash report.
> 
> gfs2: fsid=syz:syz.0: fatal: invalid metadata block - bh =
> 1125899906854035 (bad magic number), function = gfs2_quota_init, file
> = fs/gfs2/quota.c, line = 1432
> gfs2: fsid=syz:syz.0: about to withdraw this file system
> ------------[ cut here ]------------
> kernel BUG at fs/gfs2/util.c:340!
> Oops: invalid opcode: 0000 [#1] PREEMPT SMP KASAN NOPTI
> CPU: 0 UID: 0 PID: 12970 Comm: syz.1.390 Not tainted
> 6.11.0-rc5-00081-gd5d547aa7b51 #1
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
> RIP: 0010:gfs2_withdraw+0xf32/0x11e0 data/linux_kernel/linux/fs/gfs2/util.c:340


The line triggers BUG_ON is:

BUG_ON(sdp->sd_args.ar_debug);

repro.cprog mounts the image with option “debug” which set sd_args.ar_debug. 

memcpy((void*)0x20000000, "debug", 5);

The “debug” option should not be used in  production systems. Can you silence the syzbot issue?

— 
Su


> Code: e6 e8 52 02 d2 fd 4d 85 e4 0f 84 5d ff ff ff e8 e4 fe d1 fd 4c
> 89 e7 e8 ec 3a 36 07 49 89 c4 e9 13 ff ff ff e8 cf fe d1 fd 90 <0f> 0b
> e8 c7 fe d1 fd be 08 00 00 00 48 89 ef e8 2a 3a 2d fe f0 80
> RSP: 0018:ffffc90017a8f6c0 EFLAGS: 00010246
> RAX: 0000000000040000 RBX: ffff88805ab0c000 RCX: ffffc9000f4d1000
> RDX: 0000000000040000 RSI: ffffffff83b80071 RDI: 0000000000000001
> RBP: ffff88805ab0c0a8 R08: 0000000000000001 R09: ffffed1005885179
> R10: 0000000000000004 R11: 0000000000000000 R12: 0000000000000004
> R13: 0000000000001001 R14: ffffc90017a8f750 R15: ffff88805ab0d1f8
> FS: 00007f0c2d8db640(0000) GS:ffff88802c400000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00007fc52c957a8c CR3: 00000000633d4000 CR4: 0000000000750ef0
> PKRU: 80000000
> Call Trace:
> <TASK>
> gfs2_meta_check_ii+0x65/0xa0 data/linux_kernel/linux/fs/gfs2/util.c:501
> gfs2_metatype_check_i data/linux_kernel/linux/fs/gfs2/util.h:126 [inline]
> gfs2_quota_init+0xfda/0x1360 data/linux_kernel/linux/fs/gfs2/quota.c:1432
> gfs2_make_fs_rw+0x3a2/0x5d0 data/linux_kernel/linux/fs/gfs2/super.c:159
> gfs2_fill_super+0x2887/0x2c00 data/linux_kernel/linux/fs/gfs2/ops_fstype.c:1274
> get_tree_bdev+0x378/0x600 data/linux_kernel/linux/fs/super.c:1635
> gfs2_get_tree+0x4e/0x280 data/linux_kernel/linux/fs/gfs2/ops_fstype.c:1329
> vfs_get_tree+0x94/0x380 data/linux_kernel/linux/fs/super.c:1800
> do_new_mount data/linux_kernel/linux/fs/namespace.c:3472 [inline]
> path_mount+0x6b2/0x1ea0 data/linux_kernel/linux/fs/namespace.c:3799
> do_mount data/linux_kernel/linux/fs/namespace.c:3812 [inline]
> __do_sys_mount data/linux_kernel/linux/fs/namespace.c:4020 [inline]
> __se_sys_mount data/linux_kernel/linux/fs/namespace.c:3997 [inline]
> __x64_sys_mount+0x284/0x310 data/linux_kernel/linux/fs/namespace.c:3997
> do_syscall_x64 data/linux_kernel/linux/arch/x86/entry/common.c:52 [inline]
> do_syscall_64+0xcb/0x250 data/linux_kernel/linux/arch/x86/entry/common.c:83
> entry_SYSCALL_64_after_hwframe+0x77/0x7f
> RIP: 0033:0x7f0c2cb9b45e
> Code: 48 c7 c0 ff ff ff ff eb aa e8 5e 20 00 00 66 2e 0f 1f 84 00 00
> 00 00 00 0f 1f 40 00 f3 0f 1e fa 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d
> 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48
> RSP: 002b:00007f0c2d8dada8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5
> RAX: ffffffffffffffda RBX: 000000000003b2c4 RCX: 00007f0c2cb9b45e
> RDX: 000000002003b2c0 RSI: 000000002003b300 RDI: 00007f0c2d8dae00
> RBP: 00007f0c2d8dae40 R08: 00007f0c2d8dae40 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000246 R12: 000000002003b2c0
> R13: 000000002003b300 R14: 00007f0c2d8dae00 R15: 0000000020000000
> </TASK>
> Modules linked in:
> ---[ end trace 0000000000000000 ]---
> RIP: 0010:gfs2_withdraw+0xf32/0x11e0 data/linux_kernel/linux/fs/gfs2/util.c:340
> Code: e6 e8 52 02 d2 fd 4d 85 e4 0f 84 5d ff ff ff e8 e4 fe d1 fd 4c
> 89 e7 e8 ec 3a 36 07 49 89 c4 e9 13 ff ff ff e8 cf fe d1 fd 90 <0f> 0b
> e8 c7 fe d1 fd be 08 00 00 00 48 89 ef e8 2a 3a 2d fe f0 80
> RSP: 0018:ffffc90017a8f6c0 EFLAGS: 00010246
> RAX: 0000000000040000 RBX: ffff88805ab0c000 RCX: ffffc9000f4d1000
> RDX: 0000000000040000 RSI: ffffffff83b80071 RDI: 0000000000000001
> RBP: ffff88805ab0c0a8 R08: 0000000000000001 R09: ffffed1005885179
> R10: 0000000000000004 R11: 0000000000000000 R12: 0000000000000004
> R13: 0000000000001001 R14: ffffc90017a8f750 R15: ffff88805ab0d1f8
> FS: 00007f0c2d8db640(0000) GS:ffff88802c400000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00007fc52c957a8c CR3: 00000000633d4000 CR4: 0000000000750ef0
> PKRU: 80000000
>