[3.14-rc1] cirrus driver problem (qemu)

[3.14-rc1] cirrus driver problem (qemu)

[Sabrina Dubroca]

When I boot 3.14-rc1 in qemu, I get the trace below. The console stops
updating and I don't get a login prompt. I can login, but I can't see
what I'm doing. I can login normally via SSH.

If I revert the last commit in drivers/gpu/drm/cirrus:

f4b4718b61d1d5a7442a4fd6863ea80c3a10e508 drm: ast,cirrus,mgag200: use drm_can_sleep

the problem is solved.


[    1.749341] ------------[ cut here ]------------
[    1.749347] WARNING: CPU: 0 PID: 0 at kernel/locking/mutex.c:856 mutex_trylock+0x1e5/0x250()
[    1.749348] DEBUG_LOCKS_WARN_ON(in_interrupt())
[    1.749360] Modules linked in: ppdev cirrus syscopyarea sysfillrect sysimgblt drm_kms_helper evdev psmouse microcode serio_raw pcspkr ttm e1000 parport_pc parport processor button intel_agp drm intel_gtt i2c_piix4 ipv6 ext4 crc16 mbcache jbd2 sd_mod sr_mod cdrom ata_generic pata_acpi ata_piix 9pnet_virtio 9pnet libata scsi_mod
[    1.749362] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.14.0-rc1-t1 #34
[    1.749364] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[    1.749366]  0000000000000009 ffff88001fc038c8 ffffffff814e8456 ffff88001fc03910
[    1.749367]  ffff88001fc03900 ffffffff8106a0dd ffff88001d3ff990 0000000000000010
[    1.749368]  0000000000000000 00000000000001e0 ffff88001cc3b000 ffff88001fc03960
[    1.749369] Call Trace:
[    1.749372]  <IRQ>  [<ffffffff814e8456>] dump_stack+0x4d/0x6f
[    1.749374]  [<ffffffff8106a0dd>] warn_slowpath_common+0x7d/0xa0
[    1.749375]  [<ffffffff8106a14c>] warn_slowpath_fmt+0x4c/0x50
[    1.749377]  [<ffffffff814ebf45>] mutex_trylock+0x1e5/0x250
[    1.749380]  [<ffffffffa037888c>] cirrus_dirty_update+0x7c/0x2f0 [cirrus]
[    1.749381]  [<ffffffffa0378b2f>] cirrus_imageblit+0x2f/0x40 [cirrus]
[    1.749388]  [<ffffffff813188f4>] soft_cursor+0x1b4/0x250
[    1.749390]  [<ffffffff813181a3>] bit_cursor+0x613/0x650
[    1.749391]  [<ffffffff81313891>] ? get_color.isra.15+0x31/0x140
[    1.749392]  [<ffffffff8131442b>] fbcon_cursor+0x13b/0x1c0
[    1.749393]  [<ffffffff81317b90>] ? update_attr.isra.2+0x90/0x90
[    1.749398]  [<ffffffff81380fd8>] hide_cursor+0x28/0xa0
[    1.749400]  [<ffffffff81382428>] vt_console_print+0x398/0x3d0
[    1.749405]  [<ffffffff810bc0af>] ? print_prefix+0x6f/0xb0
[    1.749407]  [<ffffffff810bcc03>] call_console_drivers.constprop.18+0x93/0x110
[    1.749409]  [<ffffffff810bd33f>] console_unlock+0x3cf/0x410
[    1.749410]  [<ffffffff810bd501>] vprintk_emit+0x181/0x4f0
[    1.749412]  [<ffffffff814e60dd>] printk+0x54/0x56
[    1.749414]  [<ffffffff81397bca>] credit_entropy_bits+0x2ea/0x300
[    1.749415]  [<ffffffff81398166>] ? mix_pool_bytes.constprop.30+0x56/0xc0
[    1.749416]  [<ffffffff8139866e>] add_timer_randomness+0xee/0x120
[    1.749418]  [<ffffffff81399463>] add_disk_randomness+0x33/0xb0
[    1.749424]  [<ffffffff8127d4bc>] blk_update_bidi_request+0x5c/0x80
[    1.749426]  [<ffffffff8127d79f>] blk_end_bidi_request+0x1f/0x60
[    1.749427]  [<ffffffff8127d7f0>] blk_end_request+0x10/0x20
[    1.749433]  [<ffffffffa0009db9>] scsi_io_completion+0xa9/0x640 [scsi_mod]
[    1.749436]  [<ffffffffa00014b2>] scsi_finish_command+0xa2/0xe0 [scsi_mod]
[    1.749440]  [<ffffffffa0009c1e>] scsi_softirq_done+0x10e/0x130 [scsi_mod]
[    1.749441]  [<ffffffff81283fb3>] blk_done_softirq+0x93/0xb0
[    1.749443]  [<ffffffff8106ee15>] __do_softirq+0x105/0x2f0
[    1.749444]  [<ffffffff8106f2b2>] irq_exit+0x92/0xc0
[    1.749446]  [<ffffffff814fa798>] do_IRQ+0x58/0xf0
[    1.749447]  [<ffffffff814f03ad>] common_interrupt+0x6d/0x6d
[    1.749450]  <EOI>  [<ffffffff810546d6>] ? native_safe_halt+0x6/0x10
[    1.749453]  [<ffffffff8102047d>] default_idle+0x2d/0x110
[    1.749454]  [<ffffffff81020e9e>] arch_cpu_idle+0x2e/0x40
[    1.749455]  [<ffffffff810bec05>] cpu_startup_entry+0xa5/0x2e0
[    1.749464]  [<ffffffff81ae2120>] ? early_idt_handlers+0x120/0x120
[    1.749466]  [<ffffffff814daad4>] rest_init+0x84/0x90
[    1.749467]  [<ffffffff81ae2fa9>] start_kernel+0x443/0x44e
[    1.749468]  [<ffffffff81ae296f>] ? repair_env_string+0x5c/0x5c
[    1.749469]  [<ffffffff81ae25f6>] x86_64_start_reservations+0x2a/0x2c
[    1.749470]  [<ffffffff81ae2761>] x86_64_start_kernel+0x169/0x178
[    1.749471] ---[ end trace d478ba7c30908d4d ]---


-- 
Sabrina Dubroca

Re: [3.14-rc1] cirrus driver problem (qemu)

[Dave Airlie]

[-- Attachment #1: Type: text/plain, Size: 486 bytes --]

On Tue, Feb 4, 2014 at 1:34 AM, Sabrina Dubroca <sd@queasysnail.net> wrote:
> When I boot 3.14-rc1 in qemu, I get the trace below. The console stops
> updating and I don't get a login prompt. I can login, but I can't see
> what I'm doing. I can login normally via SSH.
>
> If I revert the last commit in drivers/gpu/drm/cirrus:
>
> f4b4718b61d1d5a7442a4fd6863ea80c3a10e508 drm: ast,cirrus,mgag200: use drm_can_sleep
>
> the problem is solved.
>

Hi does the attach patch fix it?

Dave.

[-- Attachment #2: 0001-drm-add-in_interrupt-to-drm_can_sleep-list-of-things.patch --]
[-- Type: text/x-patch, Size: 2532 bytes --]

From c6feb881e00c8db9f9f73d099a08e86c5af79d50 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Tue, 4 Feb 2014 13:19:08 +1000
Subject: [PATCH] drm: add in_interrupt to drm_can_sleep list of things

This should stop us taking locks when we shouldn't in the console drivers,

[    1.749341] ------------[ cut here ]------------
[    1.749347] WARNING: CPU: 0 PID: 0 at kernel/locking/mutex.c:856 mutex_trylock+0x1e5/0x250()
[    1.749348] DEBUG_LOCKS_WARN_ON(in_interrupt())
[    1.749360] Modules linked in: ppdev cirrus syscopyarea sysfillrect sysimgblt drm_kms_helper evdev psmouse microcode serio_raw pcspkr ttm e1000 parport_pc parport processor button intel_agp drm intel_gtt i2c_piix4 ipv6 ext4 crc16 mbcache jbd2 sd_mod sr_mod cdrom ata_generic pata_acpi ata_piix 9pnet_virtio 9pnet libata scsi_mod
[    1.749362] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.14.0-rc1-t1 #34
[    1.749364] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[    1.749366]  0000000000000009 ffff88001fc038c8 ffffffff814e8456 ffff88001fc03910
[    1.749367]  ffff88001fc03900 ffffffff8106a0dd ffff88001d3ff990 0000000000000010
[    1.749368]  0000000000000000 00000000000001e0 ffff88001cc3b000 ffff88001fc03960
[    1.749369] Call Trace:
[    1.749372]  <IRQ>  [<ffffffff814e8456>] dump_stack+0x4d/0x6f
[    1.749374]  [<ffffffff8106a0dd>] warn_slowpath_common+0x7d/0xa0
[    1.749375]  [<ffffffff8106a14c>] warn_slowpath_fmt+0x4c/0x50
[    1.749377]  [<ffffffff814ebf45>] mutex_trylock+0x1e5/0x250
[    1.749380]  [<ffffffffa037888c>] cirrus_dirty_update+0x7c/0x2f0 [cirrus]
[    1.749381]  [<ffffffffa0378b2f>] cirrus_imageblit+0x2f/0x40 [cirrus]
[    1.749388]  [<ffffffff813188f4>] soft_cursor+0x1b4/0x250
[    1.749390]  [<ffffffff813181a3>] bit_cursor+0x613/0x650
[    1.749391]  [<ffffffff81313891>] ? get_color.isra.15+0x31/0x140

Reported-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 include/drm/drmP.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/drm/drmP.h b/include/drm/drmP.h
index 1d4a920..3a98664 100644
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -1665,7 +1665,7 @@ extern void drm_platform_exit(struct drm_driver *driver, struct platform_device
 /* returns true if currently okay to sleep */
 static __inline__ bool drm_can_sleep(void)
 {
-	if (in_atomic() || in_dbg_master() || irqs_disabled())
+	if (in_atomic() || in_dbg_master() || in_interrupt() || irqs_disabled())
 		return false;
 	return true;
 }
-- 
1.8.4.2

Re: [3.14-rc1] cirrus driver problem (qemu)

[Sabrina Dubroca]

2014-02-04, 13:20:54 +1000, Dave Airlie wrote:
> On Tue, Feb 4, 2014 at 1:34 AM, Sabrina Dubroca <sd@queasysnail.net> wrote:
> > When I boot 3.14-rc1 in qemu, I get the trace below. The console stops
> > updating and I don't get a login prompt. I can login, but I can't see
> > what I'm doing. I can login normally via SSH.
> >
> > If I revert the last commit in drivers/gpu/drm/cirrus:
> >
> > f4b4718b61d1d5a7442a4fd6863ea80c3a10e508 drm: ast,cirrus,mgag200: use drm_can_sleep
> >
> > the problem is solved.
> >
> 
> Hi does the attach patch fix it?
> 
> Dave.


Same problem. Didn't you reverse the logic on in_interrupt, compared
to the old "if (!in_interrupt())" ? It looks like drm_can_sleep() is
false when in_interrupt() is true.

I modified your patch as below. Display doesn't freeze, but I still
get the warning.

Thanks,
Sabrina

---
diff --git a/include/drm/drmP.h b/include/drm/drmP.h
index 04086c5..6ab14455f 100644
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -1696,7 +1696,7 @@ extern int drm_platform_init(struct drm_driver *driver, struct platform_device *
 /* returns true if currently okay to sleep */
 static __inline__ bool drm_can_sleep(void)
 {
-	if (in_atomic() || in_dbg_master() || irqs_disabled())
+	if (in_atomic() || in_dbg_master() || !in_interrupt() || irqs_disabled())
 		return false;
 	return true;
 }

Re: [3.14-rc1] cirrus driver problem (qemu)

[Dave Airlie]

[-- Attachment #1: Type: text/plain, Size: 1066 bytes --]

On Wed, Feb 5, 2014 at 8:53 AM, Sabrina Dubroca <sd@queasysnail.net> wrote:
> 2014-02-04, 13:20:54 +1000, Dave Airlie wrote:
>> On Tue, Feb 4, 2014 at 1:34 AM, Sabrina Dubroca <sd@queasysnail.net> wrote:
>> > When I boot 3.14-rc1 in qemu, I get the trace below. The console stops
>> > updating and I don't get a login prompt. I can login, but I can't see
>> > what I'm doing. I can login normally via SSH.
>> >
>> > If I revert the last commit in drivers/gpu/drm/cirrus:
>> >
>> > f4b4718b61d1d5a7442a4fd6863ea80c3a10e508 drm: ast,cirrus,mgag200: use drm_can_sleep
>> >
>> > the problem is solved.
>> >
>>
>> Hi does the attach patch fix it?
>>
>> Dave.
>
>
> Same problem. Didn't you reverse the logic on in_interrupt, compared
> to the old "if (!in_interrupt())" ? It looks like drm_can_sleep() is
> false when in_interrupt() is true.
>
> I modified your patch as below. Display doesn't freeze, but I still
> get the warning.

Oh wow I totally screwed up there, you are right, logic inversion.

Can you try the attached?

without the in_interrupt addition.

Dave.

[-- Attachment #2: 0001-drm-mgag200-ast-cirrus-fix-regression-with-drm_can_s.patch --]
[-- Type: text/x-patch, Size: 2030 bytes --]

From c39a8bdd2586c616e9b3746db128180d4e0ca8ec Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Wed, 5 Feb 2014 14:47:45 +1000
Subject: [PATCH] drm/mgag200,ast,cirrus: fix regression with drm_can_sleep
 conversion

I totally sign inverted my way out of this one.

Reported-by: "Sabrina Dubroca" <sd@queasysnail.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 drivers/gpu/drm/ast/ast_fb.c          | 2 +-
 drivers/gpu/drm/cirrus/cirrus_fbdev.c | 2 +-
 drivers/gpu/drm/mgag200/mgag200_fb.c  | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/ast/ast_fb.c b/drivers/gpu/drm/ast/ast_fb.c
index 3f65dd6..a28640f 100644
--- a/drivers/gpu/drm/ast/ast_fb.c
+++ b/drivers/gpu/drm/ast/ast_fb.c
@@ -65,7 +65,7 @@ static void ast_dirty_update(struct ast_fbdev *afbdev,
 	 * then the BO is being moved and we should
 	 * store up the damage until later.
 	 */
-	if (!drm_can_sleep())
+	if (drm_can_sleep())
 		ret = ast_bo_reserve(bo, true);
 	if (ret) {
 		if (ret != -EBUSY)
diff --git a/drivers/gpu/drm/cirrus/cirrus_fbdev.c b/drivers/gpu/drm/cirrus/cirrus_fbdev.c
index 2fd4a92..32bbba0 100644
--- a/drivers/gpu/drm/cirrus/cirrus_fbdev.c
+++ b/drivers/gpu/drm/cirrus/cirrus_fbdev.c
@@ -39,7 +39,7 @@ static void cirrus_dirty_update(struct cirrus_fbdev *afbdev,
 	 * then the BO is being moved and we should
 	 * store up the damage until later.
 	 */
-	if (!drm_can_sleep())
+	if (drm_can_sleep())
 		ret = cirrus_bo_reserve(bo, true);
 	if (ret) {
 		if (ret != -EBUSY)
diff --git a/drivers/gpu/drm/mgag200/mgag200_fb.c b/drivers/gpu/drm/mgag200/mgag200_fb.c
index f9adc27..13b7dd8 100644
--- a/drivers/gpu/drm/mgag200/mgag200_fb.c
+++ b/drivers/gpu/drm/mgag200/mgag200_fb.c
@@ -41,7 +41,7 @@ static void mga_dirty_update(struct mga_fbdev *mfbdev,
 	 * then the BO is being moved and we should
 	 * store up the damage until later.
 	 */
-	if (!drm_can_sleep())
+	if (drm_can_sleep())
 		ret = mgag200_bo_reserve(bo, true);
 	if (ret) {
 		if (ret != -EBUSY)
-- 
1.8.4.2

Re: [3.14-rc1] cirrus driver problem (qemu)

[Sabrina Dubroca]

2014-02-05, 14:50:18 +1000, Dave Airlie wrote:
> On Wed, Feb 5, 2014 at 8:53 AM, Sabrina Dubroca <sd@queasysnail.net> wrote:
> > 2014-02-04, 13:20:54 +1000, Dave Airlie wrote:
> >> On Tue, Feb 4, 2014 at 1:34 AM, Sabrina Dubroca <sd@queasysnail.net> wrote:
> >> > When I boot 3.14-rc1 in qemu, I get the trace below. The console stops
> >> > updating and I don't get a login prompt. I can login, but I can't see
> >> > what I'm doing. I can login normally via SSH.
> >> >
> >> > If I revert the last commit in drivers/gpu/drm/cirrus:
> >> >
> >> > f4b4718b61d1d5a7442a4fd6863ea80c3a10e508 drm: ast,cirrus,mgag200: use drm_can_sleep
> >> >
> >> > the problem is solved.
> >> >
> >>
> >> Hi does the attach patch fix it?
> >>
> >> Dave.
> >
> >
> > Same problem. Didn't you reverse the logic on in_interrupt, compared
> > to the old "if (!in_interrupt())" ? It looks like drm_can_sleep() is
> > false when in_interrupt() is true.
> >
> > I modified your patch as below. Display doesn't freeze, but I still
> > get the warning.
> 
> Oh wow I totally screwed up there, you are right, logic inversion.
> 
> Can you try the attached?
> 
> without the in_interrupt addition.
> 
> Dave.

It works, thanks!
No freeze, no warning.

Sabrina