Re: 5.10.1: UBSAN: shift-out-of-bounds in ./include/linux/log2.h:57:1
From: Toralf Förster
Date: Fri Dec 18 2020 - 05:22:36 EST
On 12/18/20 7:54 AM, Randy Dunlap wrote:
Hi,
[adding linux-mm]
On 12/16/20 1:54 AM, Toralf Förster wrote:
Hi,
I got this recently at this hardened Gentoo Linux server:
Linux mr-fox 5.10.1 #1 SMP Tue Dec 15 22:09:42 CET 2020 x86_64 Intel(R)
Xeon(R) CPU E5-1650 v3 @ 3.50GHz GenuineIntel GNU/Linux
Dec 15 23:31:51 mr-fox kernel: [ 1974.206972]
================================================================================
Dec 15 23:31:51 mr-fox kernel: [ 1974.206977] UBSAN: shift-out-of-bounds
in ./include/linux/log2.h:57:13
Dec 15 23:31:51 mr-fox kernel: [ 1974.206980] shift exponent 64 is too
large for 64-bit type 'long unsigned int'
Dec 15 23:31:51 mr-fox kernel: [ 1974.206982] CPU: 11 PID: 21051 Comm:
cc1 Tainted: G T 5.10.1 #1
Dec 15 23:31:51 mr-fox kernel: [ 1974.206984] Hardware name: ASUSTeK
COMPUTER INC. Z10PA-U8 Series/Z10PA-U8 Series, BIOS 3703 08/02/2018
Dec 15 23:31:51 mr-fox kernel: [ 1974.206985] Call Trace:
Dec 15 23:31:51 mr-fox kernel: [ 1974.206993] dump_stack+0x57/0x6a
Dec 15 23:31:51 mr-fox kernel: [ 1974.206996] ubsan_epilogue+0x5/0x40
Dec 15 23:31:51 mr-fox kernel: [ 1974.206999]
__ubsan_handle_shift_out_of_bounds.cold+0x61/0x10e
Dec 15 23:31:51 mr-fox kernel: [ 1974.207002]
ondemand_readahead.cold+0x16/0x21
Dec 15 23:31:51 mr-fox kernel: [ 1974.207007]
generic_file_buffered_read+0x452/0x890
Dec 15 23:31:51 mr-fox kernel: [ 1974.207011] new_sync_read+0x156/0x200
Dec 15 23:31:51 mr-fox kernel: [ 1974.207014] vfs_read+0xf8/0x190
Dec 15 23:31:51 mr-fox kernel: [ 1974.207016] ksys_read+0x65/0xe0
Dec 15 23:31:51 mr-fox kernel: [ 1974.207018] do_syscall_64+0x33/0x40
Dec 15 23:31:51 mr-fox kernel: [ 1974.207021]
entry_SYSCALL_64_after_hwframe+0x44/0xa9
Dec 15 23:31:51 mr-fox kernel: [ 1974.207024] RIP: 0033:0x7f01b2df198e
Dec 15 23:31:51 mr-fox kernel: [ 1974.207026] Code: c0 e9 b6 fe ff ff 50
48 8d 3d 66 c3 09 00 e8 59 e2 01 00 66 0f 1f 84 00 00 00 00 00 64 8b 04
25 18 00 00 00 85 c0 75 14 0f 05 <48> 3d 00 f0 ff ff 77 5a c3 66 0f 1f
84 00 00 00 00 00 48 83 ec 28
Dec 15 23:31:51 mr-fox kernel: [ 1974.207028] RSP: 002b:00007fff2167e998
EFLAGS: 00000246 ORIG_RAX: 0000000000000000
Dec 15 23:31:51 mr-fox kernel: [ 1974.207030] RAX: ffffffffffffffda RBX:
0000000000000000 RCX: 00007f01b2df198e
Dec 15 23:31:51 mr-fox kernel: [ 1974.207032] RDX: 0000000000000000 RSI:
00000000054dcc50 RDI: 0000000000000004
Dec 15 23:31:51 mr-fox kernel: [ 1974.207033] RBP: 00000000054dcc50 R08:
00000000054dcc50 R09: 0000000000000000
Dec 15 23:31:51 mr-fox kernel: [ 1974.207034] R10: 0000000000000000 R11:
0000000000000246 R12: 00000000054dc3b0
Dec 15 23:31:51 mr-fox kernel: [ 1974.207035] R13: 0000000000008000 R14:
00000000054c9800 R15: 0000000000000000
Dec 15 23:31:51 mr-fox kernel: [ 1974.207037]
================================================================================
Known issue ?
Not that I have heard about, but that's not conclusive.
Looks to me like this is in mm/readahead.c:
static unsigned long get_init_ra_size(unsigned long size, unsigned long max)
{
unsigned long newsize = roundup_pow_of_two(size);
What filesystem? What workload?
/ is a 32 GB ext4 filesystem.
Data are at 3 BTRFS filesystems, 1x 500 GB and 2x 1.6TB.
2 Tor relays run at 100% each and utilizes the 1 GBit/s by 50%-60% [1]
7 build bots are running over the Gentoo software repostory [2]
1 AFL bot fuzzies the Tor sources.
Those 8 jobs are contained by a cgroup of 9 CPUs and 120 GB RAM [3],
each job is contained further by an own sub cgroup of 1.5 CPU and 20 GB
RAM [4]
The host is monitored using sysstat, the load is about 11.8, CPU[all] at
80%, proc/s at 1800, cswchs/s at 20000 and so on.
[1] https://metrics.torproject.org/rs.html#search/zwiebeltoralf
[2] https://zwiebeltoralf.de/tinderbox.html
[3] https://github.com/toralf/tinderbox/blob/master/bin/cgroup.sh
[4] https://github.com/toralf/tinderbox/blob/master/bin/bwrap.sh#L15
--
Toralf