[PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Laurent Pinchart]

Hello,

This patch set implements a busy loop to ensure that enabling an MSTP clock
will not return before the clock is actually enabled. It has successfully been
tested on the Marzen board and on the Lager board where they fix random crashes
with the VSP1.

Simon, could you please apply this for v3.14 ?

Changes since v3:

- Rebased on top of renesas-devel-v3.13-rc3-20131214v2

Changes since v2:

- Don't remove unused MSTPCR registers from 2/3
- Rebased on top of renesas-devel-v3.13-rc1-20131125

Guennadi Liakhovetski (1):
  ARM: shmobile: wait for MSTP clock status to toggle, when enabling it

Laurent Pinchart (1):
  ARM: shmobile: r8a7779: Wait for status on selected MSTP clocks

Shinya Kuribayashi (1):
  ARM: shmobile: r8a7790: Wait for status on all MSTP clocks

 arch/arm/mach-shmobile/clock-r8a7779.c |  20 +++---
 arch/arm/mach-shmobile/clock-r8a7790.c | 115 ++++++++++++++++++---------------
 drivers/sh/clk/cpg.c                   |  38 +++++++++++
 include/linux/sh_clk.h                 |  19 ++++--
 4 files changed, 122 insertions(+), 70 deletions(-)

-- 
Regards,

Laurent Pinchart

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Laurent Pinchart]

Hi Simon,

On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
> Hello,
> 
> This patch set implements a busy loop to ensure that enabling an MSTP clock
> will not return before the clock is actually enabled. It has successfully
> been tested on the Marzen board and on the Lager board where they fix
> random crashes with the VSP1.
> 
> Simon, could you please apply this for v3.14 ?

Ping ?

This patch series is required to fix crashes with the VSP1 on R8A7790.

> Changes since v3:
> 
> - Rebased on top of renesas-devel-v3.13-rc3-20131214v2
> 
> Changes since v2:
> 
> - Don't remove unused MSTPCR registers from 2/3
> - Rebased on top of renesas-devel-v3.13-rc1-20131125
> 
> Guennadi Liakhovetski (1):
>   ARM: shmobile: wait for MSTP clock status to toggle, when enabling it
> 
> Laurent Pinchart (1):
>   ARM: shmobile: r8a7779: Wait for status on selected MSTP clocks
> 
> Shinya Kuribayashi (1):
>   ARM: shmobile: r8a7790: Wait for status on all MSTP clocks
> 
>  arch/arm/mach-shmobile/clock-r8a7779.c |  20 +++---
>  arch/arm/mach-shmobile/clock-r8a7790.c | 115 ++++++++++++++++--------------
>  drivers/sh/clk/cpg.c                   |  38 +++++++++++
>  include/linux/sh_clk.h                 |  19 ++++--
>  4 files changed, 122 insertions(+), 70 deletions(-)

-- 
Regards,

Laurent Pinchart

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Simon Horman]

On Tue, Jan 07, 2014 at 09:34:12AM +0100, Laurent Pinchart wrote:
> Hi Simon,
> 
> On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
> > Hello,
> > 
> > This patch set implements a busy loop to ensure that enabling an MSTP clock
> > will not return before the clock is actually enabled. It has successfully
> > been tested on the Marzen board and on the Lager board where they fix
> > random crashes with the VSP1.
> > 
> > Simon, could you please apply this for v3.14 ?
> 
> Ping ?
> 
> This patch series is required to fix crashes with the VSP1 on R8A7790.

Thanks. I haven't heard back from Paul so I'll look at queuing
this up tomorrow.

> > Changes since v3:
> > 
> > - Rebased on top of renesas-devel-v3.13-rc3-20131214v2
> > 
> > Changes since v2:
> > 
> > - Don't remove unused MSTPCR registers from 2/3
> > - Rebased on top of renesas-devel-v3.13-rc1-20131125
> > 
> > Guennadi Liakhovetski (1):
> >   ARM: shmobile: wait for MSTP clock status to toggle, when enabling it
> > 
> > Laurent Pinchart (1):
> >   ARM: shmobile: r8a7779: Wait for status on selected MSTP clocks
> > 
> > Shinya Kuribayashi (1):
> >   ARM: shmobile: r8a7790: Wait for status on all MSTP clocks
> > 
> >  arch/arm/mach-shmobile/clock-r8a7779.c |  20 +++---
> >  arch/arm/mach-shmobile/clock-r8a7790.c | 115 ++++++++++++++++--------------
> >  drivers/sh/clk/cpg.c                   |  38 +++++++++++
> >  include/linux/sh_clk.h                 |  19 ++++--
> >  4 files changed, 122 insertions(+), 70 deletions(-)
> 
> -- 
> Regards,
> 
> Laurent Pinchart
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-sh" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Simon Horman]

On Tue, Jan 07, 2014 at 05:53:38PM +0900, Simon Horman wrote:
> On Tue, Jan 07, 2014 at 09:34:12AM +0100, Laurent Pinchart wrote:
> > Hi Simon,
> > 
> > On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
> > > Hello,
> > > 
> > > This patch set implements a busy loop to ensure that enabling an MSTP clock
> > > will not return before the clock is actually enabled. It has successfully
> > > been tested on the Marzen board and on the Lager board where they fix
> > > random crashes with the VSP1.
> > > 
> > > Simon, could you please apply this for v3.14 ?
> > 
> > Ping ?
> > 
> > This patch series is required to fix crashes with the VSP1 on R8A7790.
> 
> Thanks. I haven't heard back from Paul so I'll look at queuing
> this up tomorrow.

Hi Laurent,

I have queued up this series locally in the soc-fixes-for-v3.14 which
I plan to include in the devel and next branches and push later today.

It would help me when sending a pull-request for these changes if
you could give me a description of the problem they solve and if
possible when they problem was introduced.

e.g. (I am completely guessing about the cause)

This resolves crashing when using VSP1 which has been observed on the
r8a7779 (R-Car H1) based Marzen board. This problem has been present since
VSP support was added to Marzen by deadbeef ("ARM: shmobile: marzen: Add
VSP1") in v3.W-rcX.

This also resolves the same problem on the r8a7790 (R-Car H2) based Lager
board. The problem was introduced there by by dead1234 ("ARM: shmobile:
lager: Add VSP1") in v3.Y-rcZ.

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Laurent Pinchart]

Hi Simon,

On Wednesday 08 January 2014 10:03:54 Simon Horman wrote:
> On Tue, Jan 07, 2014 at 05:53:38PM +0900, Simon Horman wrote:
> > On Tue, Jan 07, 2014 at 09:34:12AM +0100, Laurent Pinchart wrote:
> > > On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
> > > > Hello,
> > > > 
> > > > This patch set implements a busy loop to ensure that enabling an MSTP
> > > > clock will not return before the clock is actually enabled. It has
> > > > successfully been tested on the Marzen board and on the Lager board
> > > > where they fix random crashes with the VSP1.
> > > > 
> > > > Simon, could you please apply this for v3.14 ?
> > > 
> > > Ping ?
> > > 
> > > This patch series is required to fix crashes with the VSP1 on R8A7790.
> > 
> > Thanks. I haven't heard back from Paul so I'll look at queuing
> > this up tomorrow.
> 
> Hi Laurent,
> 
> I have queued up this series locally in the soc-fixes-for-v3.14 which
> I plan to include in the devel and next branches and push later today.
> 
> It would help me when sending a pull-request for these changes if
> you could give me a description of the problem they solve and if
> possible when they problem was introduced.
> 
> e.g. (I am completely guessing about the cause)
> 
> This resolves crashing when using VSP1 which has been observed on the
> r8a7779 (R-Car H1) based Marzen board. This problem has been present since
> VSP support was added to Marzen by deadbeef ("ARM: shmobile: marzen: Add
> VSP1") in v3.W-rcX.
> 
> This also resolves the same problem on the r8a7790 (R-Car H2) based Lager
> board. The problem was introduced there by by dead1234 ("ARM: shmobile:
> lager: Add VSP1") in v3.Y-rcZ.

---- 8< ----
This resolves unhandled faults caused by accessing devices with their clock 
disabled. The issue has been noticed with the VSP1 on r8a7790 and has been 
present since the VSP1 driver has been added by commit 26e0ca22c3b8 ("[media] 
v4l: Renesas R-Car VSP1 driver") in v3.12-rc1.
---- 8< ----

(Note that VSP1 support hasn't been added to board files in mainline yet.)

-- 
Regards,

Laurent Pinchart

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Simon Horman]

On Wed, Jan 08, 2014 at 02:12:54AM +0100, Laurent Pinchart wrote:
> Hi Simon,
> 
> On Wednesday 08 January 2014 10:03:54 Simon Horman wrote:
> > On Tue, Jan 07, 2014 at 05:53:38PM +0900, Simon Horman wrote:
> > > On Tue, Jan 07, 2014 at 09:34:12AM +0100, Laurent Pinchart wrote:
> > > > On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
> > > > > Hello,
> > > > > 
> > > > > This patch set implements a busy loop to ensure that enabling an MSTP
> > > > > clock will not return before the clock is actually enabled. It has
> > > > > successfully been tested on the Marzen board and on the Lager board
> > > > > where they fix random crashes with the VSP1.
> > > > > 
> > > > > Simon, could you please apply this for v3.14 ?
> > > > 
> > > > Ping ?
> > > > 
> > > > This patch series is required to fix crashes with the VSP1 on R8A7790.
> > > 
> > > Thanks. I haven't heard back from Paul so I'll look at queuing
> > > this up tomorrow.
> > 
> > Hi Laurent,
> > 
> > I have queued up this series locally in the soc-fixes-for-v3.14 which
> > I plan to include in the devel and next branches and push later today.
> > 
> > It would help me when sending a pull-request for these changes if
> > you could give me a description of the problem they solve and if
> > possible when they problem was introduced.
> > 
> > e.g. (I am completely guessing about the cause)
> > 
> > This resolves crashing when using VSP1 which has been observed on the
> > r8a7779 (R-Car H1) based Marzen board. This problem has been present since
> > VSP support was added to Marzen by deadbeef ("ARM: shmobile: marzen: Add
> > VSP1") in v3.W-rcX.
> > 
> > This also resolves the same problem on the r8a7790 (R-Car H2) based Lager
> > board. The problem was introduced there by by dead1234 ("ARM: shmobile:
> > lager: Add VSP1") in v3.Y-rcZ.
> 
> ---- 8< ----
> This resolves unhandled faults caused by accessing devices with their clock 
> disabled. The issue has been noticed with the VSP1 on r8a7790 and has been 
> present since the VSP1 driver has been added by commit 26e0ca22c3b8 ("[media] 
> v4l: Renesas R-Car VSP1 driver") in v3.12-rc1.
> ---- 8< ----

Thanks!

> (Note that VSP1 support hasn't been added to board files in mainline yet.)

Noted.

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Magnus Damm]

Hi Laurent,

On Tue, Jan 7, 2014 at 5:34 PM, Laurent Pinchart
<laurent.pinchart@ideasonboard.com> wrote:
> Hi Simon,
>
> On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
>> Hello,
>>
>> This patch set implements a busy loop to ensure that enabling an MSTP clock
>> will not return before the clock is actually enabled. It has successfully
>> been tested on the Marzen board and on the Lager board where they fix
>> random crashes with the VSP1.
>>
>> Simon, could you please apply this for v3.14 ?
>
> Ping ?
>
> This patch series is required to fix crashes with the VSP1 on R8A7790.

Thanks for fixing this issue, the missing status bit check has been
around for ages.

I suspect a similar fix is needed for r8a7791 as well, any chance that
would be needed for DU / VSP1 on Koelsch?

Cheers,

/ magnus

Re: [PATCH v4 0/3] ARM: shmobile: r8a7779/90: Poll MSTP clocks status

[Laurent Pinchart]

Hi Magnus,

On Wednesday 08 January 2014 16:04:44 Magnus Damm wrote:
> On Tue, Jan 7, 2014 at 5:34 PM, Laurent Pinchart wrote:
> > On Saturday 14 December 2013 16:23:50 Laurent Pinchart wrote:
> >> Hello,
> >> 
> >> This patch set implements a busy loop to ensure that enabling an MSTP
> >> clock will not return before the clock is actually enabled. It has
> >> successfully been tested on the Marzen board and on the Lager board where
> >> they fix random crashes with the VSP1.
> >> 
> >> Simon, could you please apply this for v3.14 ?
> > 
> > Ping ?
> > 
> > This patch series is required to fix crashes with the VSP1 on R8A7790.
> 
> Thanks for fixing this issue, the missing status bit check has been
> around for ages.

You're welcome.

> I suspect a similar fix is needed for r8a7791 as well, any chance that
> would be needed for DU / VSP1 on Koelsch?

Indeed. I'll submit a patch.

-- 
Regards,

Laurent Pinchart