On Wed, Nov 18, 2015 at 11:00:53PM +0800, Rui Wang wrote: > The machine hang completely with the following message on the console: > > [ 487.777538] BUG: unable to handle kernel NULL pointer dereference at 0000000000000060 > [ 487.777554] IP: [<ffffffff8158aaee>] _raw_spin_lock+0xe/0x30 > [ 487.777557] PGD 42e9f7067 PUD 42f2fa067 PMD 0 > [ 487.777560] Oops: 0002 [#1] SMP > ... > [ 487.777618] CPU: 21 PID: 3190 Comm: Xorg Tainted: G E 4.4.0-rc1-3-default+ #6 > [ 487.777620] Hardware name: Intel Corporation BRICKLAND/BRICKLAND, BIOS BRHSXSD1.86B.0059.R00.1501081238 01/08/2015 > [ 487.777621] task: ffff880853ae4680 ti: ffff8808696d4000 task.ti: ffff8808696d4000 > [ 487.777625] RIP: 0010:[<ffffffff8158aaee>] [<ffffffff8158aaee>] _raw_spin_lock+0xe/0x30 > [ 487.777627] RSP: 0018:ffff8808696d79c0 EFLAGS: 00010246 > [ 487.777628] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000 > [ 487.777629] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 0000000000000060 > [ 487.777630] RBP: ffff8808696d79e0 R08: 0000000000000000 R09: ffff88086924a780 > [ 487.777631] R10: 000000000001bb40 R11: 0000000000003246 R12: 0000000000000000 > [ 487.777632] R13: ffff880463a27360 R14: ffff88046ca50218 R15: 0000000000000080 > [ 487.777634] FS: 00007f3f81c5a8c0(0000) GS:ffff88086f060000(0000) knlGS:0000000000000000 > [ 487.777635] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 487.777636] CR2: 0000000000000060 CR3: 000000042e678000 CR4: 00000000001406e0 > [ 487.777638] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 > [ 487.777639] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 > [ 487.777639] Stack: > [ 487.777642] ffffffffa00eb5fa ffff8808696d7b60 ffff88086b87d800 0000000000000000 > [ 487.777644] ffff8808696d7ac8 ffffffffa01694b6 ffff8808696d7ae8 ffffffff8109c8d5 > [ 487.777647] ffff880469158740 ffff880463a27000 ffff88086b87d800 ffff88086b87d800 > [ 487.777647] Call Trace: > [ 487.777674] [<ffffffffa00eb5fa>] ? drm_gem_object_lookup+0x1a/0xa0 [drm] > [ 487.777681] [<ffffffffa01694b6>] mga_crtc_cursor_set+0xc6/0xb60 [mgag200] > [ 487.777691] [<ffffffff8109c8d5>] ? find_busiest_group+0x35/0x4a0 > [ 487.777696] [<ffffffff81086294>] ? __might_sleep+0x44/0x80 > [ 487.777699] [<ffffffff815888c2>] ? __ww_mutex_lock+0x22/0x9c > [ 487.777722] [<ffffffffa0104f64>] ? drm_modeset_lock+0x34/0xf0 [drm] > [ 487.777733] [<ffffffffa0148d9e>] restore_fbdev_mode+0xee/0x2a0 [drm_kms_helper] > [ 487.777742] [<ffffffffa014afce>] drm_fb_helper_restore_fbdev_mode_unlocked+0x2e/0x70 [drm_kms_helper] > [ 487.777748] [<ffffffffa014b037>] drm_fb_helper_set_par+0x27/0x50 [drm_kms_helper] > [ 487.777752] [<ffffffff8134560c>] fb_set_var+0x18c/0x3f0 > [ 487.777777] [<ffffffffa02a9b0a>] ? __ext4_handle_dirty_metadata+0x8a/0x210 [ext4] > [ 487.777783] [<ffffffff8133cb97>] fbcon_blank+0x1b7/0x2b0 > [ 487.777790] [<ffffffff813be2a3>] do_unblank_screen+0xb3/0x1c0 > [ 487.777795] [<ffffffff813b5aba>] vt_ioctl+0x118a/0x1210 > [ 487.777801] [<ffffffff813a8fe0>] tty_ioctl+0x3f0/0xc90 > [ 487.777808] [<ffffffff81172018>] ? kzfree+0x28/0x30 > [ 487.777813] [<ffffffff811e053f>] ? mntput+0x1f/0x30 > [ 487.777817] [<ffffffff811d3f5d>] do_vfs_ioctl+0x30d/0x570 > [ 487.777822] [<ffffffff8107ed3a>] ? task_work_run+0x8a/0xa0 > [ 487.777825] [<ffffffff811d4234>] SyS_ioctl+0x74/0x80 > [ 487.777829] [<ffffffff8158aeae>] entry_SYSCALL_64_fastpath+0x12/0x71 > [ 487.777851] Code: 65 ff 0d ce 02 a8 7e 5d c3 ba 01 00 00 00 f0 0f b1 17 85 c0 75 e8 b0 01 5d c3 0f 1f 00 65 ff 05 b1 02 a8 7e 31 c0 ba 01 00 00 00 <f0> 0f b1 17 85 c0 75 01 c3 55 89 c6 48 89 e5 e8 4e f5 b1 ff 5d > [ 487.777854] RIP [<ffffffff8158aaee>] _raw_spin_lock+0xe/0x30 > [ 487.777855] RSP <ffff8808696d79c0> > [ 487.777856] CR2: 0000000000000060 > [ 487.777860] ---[ end trace 672a2cd555e0ebd3 ]--- > > The cursor code may be entered with file_priv == NULL && handle == NULL. > The problem was introduced by: > > "bf89209 drm/mga200g: Hold a proper reference for cursor_set" > > which calls drm_gem_object_lookup(dev, file_priv...). Previously this wasn't > a problem because we checked the handle. Move the check early in the function > can fix the problem. > > Signed-off-by: Rui Wang <rui.y.wang@xxxxxxxxx> Oh dear, thanks for catching this. Reviewed-by: Daniel Vetter <daniel.vetter@xxxxxxxx> Dave, please pick this up. -Daniel > --- > drivers/gpu/drm/mgag200/mgag200_cursor.c | 11 +++++------ > 1 file changed, 5 insertions(+), 6 deletions(-) > > diff --git a/drivers/gpu/drm/mgag200/mgag200_cursor.c b/drivers/gpu/drm/mgag200/mgag200_cursor.c > index 4f2068f..a7bf6a9 100644 > --- a/drivers/gpu/drm/mgag200/mgag200_cursor.c > +++ b/drivers/gpu/drm/mgag200/mgag200_cursor.c > @@ -70,6 +70,11 @@ int mga_crtc_cursor_set(struct drm_crtc *crtc, > BUG_ON(pixels_2 != pixels_current && pixels_2 != pixels_prev); > BUG_ON(pixels_current == pixels_prev); > > + if (!handle || !file_priv) { > + mga_hide_cursor(mdev); > + return 0; > + } > + > obj = drm_gem_object_lookup(dev, file_priv, handle); > if (!obj) > return -ENOENT; > @@ -88,12 +93,6 @@ int mga_crtc_cursor_set(struct drm_crtc *crtc, > goto out_unreserve1; > } > > - if (!handle) { > - mga_hide_cursor(mdev); > - ret = 0; > - goto out1; > - } > - > /* Move cursor buffers into VRAM if they aren't already */ > if (!pixels_1->pin_count) { > ret = mgag200_bo_pin(pixels_1, TTM_PL_FLAG_VRAM, > -- > 1.8.5.2 > -- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/dri-devel