diff mbox

drm: hibmc: Use set_busid function from drm core

Message ID 20170609065757.9736-1-dja@axtens.net (mailing list archive)
State New, archived
Headers show

Commit Message

Daniel Axtens June 9, 2017, 6:57 a.m. UTC
Currently, calling drmGetBusid from libdrm on a hibmc VGA
card returns a string like "0007:a1:00.0".

The busid reported by most cards begins with "pci:". For
example, on an amd64 system, a VGA card reported
"pci:0000:00:02.0".

The missing "pci:" prefix confuses Xorg and leads to crashes
and other misbehaviour.[0]

Use the standard helper from the drm core to set the busid to
include the "pci:" prefix. This is done by many other drivers.

With this patch, Xorg can be run successfully.

[0]: https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/1691991

Cc: Xinliang Liu <z.liuxinliang@hisilicon.com>
Cc: Rongrong Zou <zourongrong@gmail.com>
Signed-off-by: Daniel Axtens <dja@axtens.net>
---
 drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Emil Velikov June 13, 2017, 9:23 a.m. UTC | #1
Hi Daniel,

On 9 June 2017 at 07:57, Daniel Axtens <dja@axtens.net> wrote:
> Currently, calling drmGetBusid from libdrm on a hibmc VGA
> card returns a string like "0007:a1:00.0".
>
> The busid reported by most cards begins with "pci:". For
> example, on an amd64 system, a VGA card reported
> "pci:0000:00:02.0".
>
> The missing "pci:" prefix confuses Xorg and leads to crashes
> and other misbehaviour.[0]
>
> Use the standard helper from the drm core to set the busid to
> include the "pci:" prefix. This is done by many other drivers.
>
> With this patch, Xorg can be run successfully.
>
> [0]: https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/1691991
>
> Cc: Xinliang Liu <z.liuxinliang@hisilicon.com>
> Cc: Rongrong Zou <zourongrong@gmail.com>
> Signed-off-by: Daniel Axtens <dja@axtens.net>
> ---
>  drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
> index 2ffdbf9801bd..9663a633897a 100644
> --- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
> +++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
> @@ -69,6 +69,7 @@ static struct drm_driver hibmc_driver = {
>         .dumb_map_offset        = hibmc_dumb_mmap_offset,
>         .dumb_destroy           = drm_gem_dumb_destroy,
>         .irq_handler            = hibmc_drm_interrupt,
> +       .set_busid              = drm_pci_set_busid,

Daniel V has been working very hard on cleaning all the set_busid
"chaos" and even has some patches which remove it completely [1].
While your patch is correct, it would be nicer if we can remove the
code so that nobody else gets bitten ;-)

Can you be so kind and re-spin your patch with the cleanups (2-4) on top?
IIRC your fix should be CC kernel-stable.

Thanks
Emil

[1] https://lists.freedesktop.org/archives/dri-devel/2017-May/142670.html
Daniel Axtens June 19, 2017, 8:02 a.m. UTC | #2
Hi Emil,

> Daniel V has been working very hard on cleaning all the set_busid
> "chaos" and even has some patches which remove it completely [1].
> While your patch is correct, it would be nicer if we can remove the
> code so that nobody else gets bitten ;-)
>
> Can you be so kind and re-spin your patch with the cleanups (2-4) on top?
> IIRC your fix should be CC kernel-stable.

Thanks for pointing me towards that series.

That series (in particular patch 4) actually fixes the bug I was trying
to fix, so it renders my patch unnecessary. However, I notice that
series has attracted a lot of discussion: any ideas when is it likely to
land?

I just want to avoid the situation where we don't merge my fix because
of Daniel's series and Daniel's series also doesn't land for another few
releases!

Regards,
Daniel

>
> Thanks
> Emil
>
> [1] https://lists.freedesktop.org/archives/dri-devel/2017-May/142670.html
Daniel Vetter June 20, 2017, 9:20 a.m. UTC | #3
On Tue, Jun 13, 2017 at 10:23:34AM +0100, Emil Velikov wrote:
> Hi Daniel,
> 
> On 9 June 2017 at 07:57, Daniel Axtens <dja@axtens.net> wrote:
> > Currently, calling drmGetBusid from libdrm on a hibmc VGA
> > card returns a string like "0007:a1:00.0".
> >
> > The busid reported by most cards begins with "pci:". For
> > example, on an amd64 system, a VGA card reported
> > "pci:0000:00:02.0".
> >
> > The missing "pci:" prefix confuses Xorg and leads to crashes
> > and other misbehaviour.[0]
> >
> > Use the standard helper from the drm core to set the busid to
> > include the "pci:" prefix. This is done by many other drivers.
> >
> > With this patch, Xorg can be run successfully.
> >
> > [0]: https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/1691991
> >
> > Cc: Xinliang Liu <z.liuxinliang@hisilicon.com>
> > Cc: Rongrong Zou <zourongrong@gmail.com>
> > Signed-off-by: Daniel Axtens <dja@axtens.net>
> > ---
> >  drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
> > index 2ffdbf9801bd..9663a633897a 100644
> > --- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
> > +++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
> > @@ -69,6 +69,7 @@ static struct drm_driver hibmc_driver = {
> >         .dumb_map_offset        = hibmc_dumb_mmap_offset,
> >         .dumb_destroy           = drm_gem_dumb_destroy,
> >         .irq_handler            = hibmc_drm_interrupt,
> > +       .set_busid              = drm_pci_set_busid,
> 
> Daniel V has been working very hard on cleaning all the set_busid
> "chaos" and even has some patches which remove it completely [1].
> While your patch is correct, it would be nicer if we can remove the
> code so that nobody else gets bitten ;-)

Yeah I just applied my patches to nuke the set_busid callback. Pls drop
this one. If you want this backported, I guess we could cherry-pick it
over with a cc: stable (it's likely only going to reach 4.14 as-is).
-Daniel

> 
> Can you be so kind and re-spin your patch with the cleanups (2-4) on top?
> IIRC your fix should be CC kernel-stable.
> 
> Thanks
> Emil
> 
> [1] https://lists.freedesktop.org/archives/dri-devel/2017-May/142670.html
diff mbox

Patch

diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
index 2ffdbf9801bd..9663a633897a 100644
--- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
+++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.c
@@ -69,6 +69,7 @@  static struct drm_driver hibmc_driver = {
 	.dumb_map_offset        = hibmc_dumb_mmap_offset,
 	.dumb_destroy           = drm_gem_dumb_destroy,
 	.irq_handler		= hibmc_drm_interrupt,
+	.set_busid		= drm_pci_set_busid,
 };
 
 static int __maybe_unused hibmc_pm_suspend(struct device *dev)