Re: WARNING in drm_modeset_lock_all

From: Chris Wilson
Date: Mon Oct 30 2017 - 10:24:54 EST


Quoting syzbot (2017-10-27 09:09:50)
> Hello,
>
> syzkaller hit the following crash on
> 6f20b7a58cb9c0fe00badcdfd65b1f4a8f28dfc6
> git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/master
> compiler: gcc (GCC) 7.1.1 20170620
> .config is attached
> Raw console output is attached.
>
>
>
>
> ------------[ cut here ]------------
> WARNING: CPU: 2 PID: 11675 at drivers/gpu/drm/drm_modeset_lock.c:92
> drm_modeset_lock_all+0x1fc/0x2d0 drivers/gpu/drm/drm_modeset_lock.c:92
> Kernel panic - not syncing: panic_on_warn set ...
>
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
> Call Trace:
> __dump_stack lib/dump_stack.c:16 [inline]
> dump_stack+0x194/0x257 lib/dump_stack.c:52
> fail_dump lib/fault-inject.c:51 [inline]
> should_fail+0x8c0/0xa40 lib/fault-inject.c:149
> should_failslab+0xec/0x120 mm/failslab.c:31
> slab_pre_alloc_hook mm/slab.h:422 [inline]
> slab_alloc mm/slab.c:3383 [inline]
> kmem_cache_alloc_trace+0x4b/0x750 mm/slab.c:3625
> kmalloc include/linux/slab.h:493 [inline]
> kzalloc include/linux/slab.h:666 [inline]
> drm_modeset_lock_all+0x49/0x2d0 drivers/gpu/drm/drm_modeset_lock.c:91
> drm_mode_obj_get_properties_ioctl+0x87/0x2b0
> drivers/gpu/drm/drm_mode_object.c:359
> drm_ioctl_kernel+0x1e7/0x2e0 drivers/gpu/drm/drm_ioctl.c:735
> drm_ioctl+0x72e/0xa50 drivers/gpu/drm/drm_ioctl.c:831
> vfs_ioctl fs/ioctl.c:45 [inline]
> do_vfs_ioctl+0x1b1/0x1530 fs/ioctl.c:685
> SYSC_ioctl fs/ioctl.c:700 [inline]
> SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691
> entry_SYSCALL_64_fastpath+0x1f/0xbe

This bug is for the unexpected allocation failure inside
drm_modeset_lock_all() (in this case from should_fail). To properly
document this behaviour, we should probably use

diff --git a/drivers/gpu/drm/drm_modeset_lock.c b/drivers/gpu/drm/drm_modeset_lock.c
index e123497da0ca..963e23db0fe7 100644
--- a/drivers/gpu/drm/drm_modeset_lock.c
+++ b/drivers/gpu/drm/drm_modeset_lock.c
@@ -93,7 +93,7 @@ void drm_modeset_lock_all(struct drm_device *dev)
struct drm_modeset_acquire_ctx *ctx;
int ret;

- ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
+ ctx = kzalloc(sizeof(*ctx), GFP_KERNEL | __GFP_NOFAIL);
if (WARN_ON(!ctx))
return;

Then it will turn up on somebody'ss too-small-to-fail fix list.
-Chris