Commit 3a0709928b172a41 ("drm/vkms: Add vblank events simulated by
hrtimers") introduced ret_overrun variable. And that variable was an
unused-but-set-variable until commit 09ef09b4ab95dc40 ("drm/vkms:
WARN when hrtimer_forward_now fails") added WARN_ON(ret_overrun != 1).

Now, syzbot is hitting this WARN_ON() using a simple reproducer that
does open("/dev/dri/card1") followed by ioctl(DRM_IOCTL_WAIT_VBLANK),
and a debug printk() patch says that syzbot is getting

   output->vblank_hrtimer.base->get_time()=93531904774 (which is uptime)
   output->period_ns=0
   ret_overrun=216994

. I can't understand what "verify the hrtimer_forward_now return" in
that commit wants to say. hrtimer_forward_now() must return, and the
return value of hrtimer_forward_now() is not a boolean. Why comparing
with 1 ? Anyway, this failure is not something that worth crashing the
system. Let's remove the ret_overrun variable and WARN_ON() test.

Link: 
https://syzkaller.appspot.com/bug?id=0ba17d70d062b2595e1f061231474800f076c7cb
Signed-off-by: Tetsuo Handa <penguin-ker...@i-love.sakura.ne.jp>
Reported-by: syzbot+0871b14ca2e2fb64f...@syzkaller.appspotmail.com
Fixes: 09ef09b4ab95dc40 ("drm/vkms: WARN when hrtimer_forward_now fails")
---
 drivers/gpu/drm/vkms/vkms_crtc.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/vkms/vkms_crtc.c b/drivers/gpu/drm/vkms/vkms_crtc.c
index ac85e17428f8..cc1811ce6092 100644
--- a/drivers/gpu/drm/vkms/vkms_crtc.c
+++ b/drivers/gpu/drm/vkms/vkms_crtc.c
@@ -13,12 +13,9 @@ static enum hrtimer_restart vkms_vblank_simulate(struct 
hrtimer *timer)
                                                  vblank_hrtimer);
        struct drm_crtc *crtc = &output->crtc;
        struct vkms_crtc_state *state;
-       u64 ret_overrun;
        bool ret;
 
-       ret_overrun = hrtimer_forward_now(&output->vblank_hrtimer,
-                                         output->period_ns);
-       WARN_ON(ret_overrun != 1);
+       hrtimer_forward_now(&output->vblank_hrtimer, output->period_ns);
 
        spin_lock(&output->lock);
        ret = drm_crtc_handle_vblank(crtc);
-- 
2.18.2

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Reply via email to