On 9/21/21 2:08 AM, Christoph Hellwig wrote:
On Mon, Sep 20, 2021 at 08:38:31PM -0700, Bart Van Assche wrote:
On 9/20/21 04:24, Christoph Hellwig wrote:
Ming reported that for SCSI we have a lifetime problem now that
the BDI moved from the request_queue to the disk as del_gendisk
doesn't finish all outstanding file system I/O. It turns out
this actually is an older problem, although the case where it could
be hit before was very unusual (unbinding of a SCSI upper driver
while the scsi_device stays around). This series fixes this by
draining all I/O in del_gendisk.
Several failures are reported when running blktests against Jens' for-next
branch if KASAN and lockdep are enabled. Is this patch series sufficient
to make blktests pass again?
I don't see any new failures (I have a few consistent ones due to the
fact that blktests is completly fucked up and wants to load modules
everywhere which doesn't exactly work with builtin drivers). Care
to post your issues?
This is the first complaint I run into with KASAN enabled (this failure prevents
running more tests) for Jens' for-next branch merged with Linus' master branch:
root[4270]: run blktests block/010
null_blk: module loaded
==================================================================
BUG: KASAN: null-ptr-deref in null_map_queues+0x131/0x1a0 [null_blk]
Read of size 8 at addr 0000000000000000 by task modprobe/4320
CPU: 9 PID: 4320 Comm: modprobe Tainted: G E 5.15.0-rc2-dbg+ #2
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.14.0-2 04/01/2014
Call Trace:
show_stack+0x52/0x58
dump_stack_lvl+0x49/0x5e
kasan_report.cold+0x64/0xdb
__asan_load8+0x69/0x90
null_map_queues+0x131/0x1a0 [null_blk]
blk_mq_update_queue_map+0x122/0x1a0
blk_mq_alloc_tag_set+0x1e8/0x570
null_init_tag_set+0x197/0x220 [null_blk]
null_init+0x1dc/0x1000 [null_blk]
do_one_initcall+0xc7/0x440
do_init_module+0x10a/0x3d0
load_module+0x115c/0x1220
__do_sys_finit_module+0x124/0x1a0
__x64_sys_finit_module+0x42/0x50
do_syscall_64+0x35/0xb0
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7fa873062d6d
Code: c9 0c 00 0f 05 eb a9 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d d3 80 0c 00 f7 d8 64 89 01 48
RSP: 002b:00007fff2aa058b8 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fa873062d6d
RDX: 0000000000000000 RSI: 00005613ba37f750 RDI: 0000000000000003
RBP: 00005613ba37f540 R08: 0000000000000000 R09: 0000000000000050
R10: 0000000000000003 R11: 0000000000000246 R12: 00005613ba37f750
R13: 00005613ba37f7a0 R14: 00005613ba37f750 R15: 00005613ba37f460
==================================================================
This regression may have been introduced by commit 5f7acddf706c ("null_blk: poll
queue support").
Thanks,
Bart.
