* Re: [4.18 rc7] BUG: sleeping function called from invalid context at mm/slab.h:421
[not found] ` <bd7f3ea4-d9a8-e437-9936-ee4513b47ac1@suse.cz>
@ 2018-08-08 10:40 ` Sergey Senozhatsky
0 siblings, 0 replies; only message in thread
From: Sergey Senozhatsky @ 2018-08-08 10:40 UTC (permalink / raw)
To: Vlastimil Babka, Jiri Slaby, Greg Kroah-Hartman, Daniel Vetter,
Bartlomiej Zolnierkiewicz
Cc: Mikhail Gavrilov, linux-mm, Petr Mladek, Steven Rostedt,
Sergey Senozhatsky, Peter Zijlstra, linux-fbdev, linux-kernel
Hello,
On (08/08/18 11:01), Vlastimil Babka wrote:
> On 08/08/2018 05:50 AM, Mikhail Gavrilov wrote:
> > Hi guys.
> > I am catched new bug.
> > Can anyone look?
>
> fbcon_startup() calls kzalloc(sizeof(struct fbcon_ops), GFP_KERNEL) so
> it tells slab it can sleep. The problem must be higher in the stack,
> CCing printk people.
Cc-ing fbcon/vt people. I'm not sure I know how exactly console
takeover is expected to work.
printk must be atomic, we can't sleep in console drivers [e.g. printk
from IRQs, etc.]
> > [226995.988988] BUG: sleeping function called from invalid context at
> > mm/slab.h:421
> > [226995.988988] in_atomic(): 1, irqs_disabled(): 1, pid: 22658, name: gsd-rfkill
> > [226995.988989] INFO: lockdep is turned off.
> > [226995.988989] irq event stamp: 0
> > [226995.988990] hardirqs last enabled at (0): [<0000000000000000>]
> > (null)
> > [226995.988991] hardirqs last disabled at (0): [<ffffffffa00b6b4a>]
> > copy_process.part.32+0x72a/0x1e60
> > [226995.988991] softirqs last enabled at (0): [<ffffffffa00b6b4a>]
> > copy_process.part.32+0x72a/0x1e60
> > [226995.988992] softirqs last disabled at (0): [<0000000000000000>]
> > (null)
> > [226995.988993] CPU: 6 PID: 22658 Comm: gsd-rfkill Tainted: G W
> > 4.18.0-0.rc7.git1.1.fc29.x86_64 #1
> > [226995.988993] Hardware name: Gigabyte Technology Co., Ltd.
> > Z87M-D3H/Z87M-D3H, BIOS F11 08/12/2014
> > [226995.988994] Call Trace:
> > [226995.988994] dump_stack+0x85/0xc0
> > [226995.988995] ___might_sleep.cold.72+0xac/0xbc
> > [226995.988995] kmem_cache_alloc_trace+0x202/0x2f0
> > [226995.988996] ? fbcon_startup+0xae/0x300
> > [226995.988996] fbcon_startup+0xae/0x300
> > [226995.988997] do_take_over_console+0x6d/0x180
> > [226995.988997] do_fbcon_takeover+0x58/0xb0
> > [226995.988997] fbcon_output_notifier.cold.35+0x5/0x23
> > [226995.988998] notifier_call_chain+0x39/0x90
> > [226995.988999] vt_console_print+0x363/0x420
> > [226995.988999] console_unlock+0x422/0x610
> > [226995.988999] vprintk_emit+0x268/0x540
> > [226995.989000] printk+0x58/0x6f
> > [226995.989000] rfkill_fop_release.cold.16+0xc/0x11 [rfkill]
> > [226995.989001] __fput+0xc7/0x250
> > [226995.989001] task_work_run+0xa1/0xd0
> > [226995.989002] exit_to_usermode_loop+0xd8/0xe0
> > [226995.989002] do_syscall_64+0x1df/0x1f0
> > [226995.989003] entry_SYSCALL_64_after_hwframe+0x49/0xbe
-ss
^ permalink raw reply [flat|nested] only message in thread