[PATCH 0/2] mmc: tmio: two bug-fixes

[PATCH 0/2] mmc: tmio: two bug-fixes

[Guennadi Liakhovetski]

So, Chris, I lied:-)

There go two more bug-fixes for tmio on top of my earlier patches from 
today, so, should still be easy enough to trace. So, to pull them 
together, the previous ones were

[PATCH v3] mmc: tmio: fix recursive spinlock, don't schedule with interrupts disabled
[PATCH v3] mmc: tmio: maximize power saving

And now follow

[PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code
[PATCH 2/2] mmc: tmio: fix a deadlock

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/

[PATCH 0/2] mmc: tmio: two bug-fixes

[Guennadi Liakhovetski]

So, Chris, I lied:-)

There go two more bug-fixes for tmio on top of my earlier patches from 
today, so, should still be easy enough to trace. So, to pull them 
together, the previous ones were

[PATCH v3] mmc: tmio: fix recursive spinlock, don't schedule with interrupts disabled
[PATCH v3] mmc: tmio: maximize power saving

And now follow

[PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code
[PATCH 2/2] mmc: tmio: fix a deadlock

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/

[PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code

[Guennadi Liakhovetski]

A recent commit "mmc: tmio: Share register access functions" has swapped
arguments of a macro and broken DMA with TMIO MMC. This patch fixes the
arguments back.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
---
 drivers/mmc/host/tmio_mmc_dma.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/mmc/host/tmio_mmc_dma.c b/drivers/mmc/host/tmio_mmc_dma.c
index f24a029..7e86662 100644
--- a/drivers/mmc/host/tmio_mmc_dma.c
+++ b/drivers/mmc/host/tmio_mmc_dma.c
@@ -26,7 +26,7 @@ static void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
 {
 #if defined(CONFIG_SUPERH) || defined(CONFIG_ARCH_SHMOBILE)
 	/* Switch DMA mode on or off - SuperH specific? */
-	sd_ctrl_write16(host, enable ? 2 : 0, CTL_DMA_ENABLE);
+	sd_ctrl_write16(host, CTL_DMA_ENABLE, enable ? 2 : 0);
 #endif
 }
 
-- 
1.7.2.5

[PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code

[Guennadi Liakhovetski]

A recent commit "mmc: tmio: Share register access functions" has swapped
arguments of a macro and broken DMA with TMIO MMC. This patch fixes the
arguments back.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
---
 drivers/mmc/host/tmio_mmc_dma.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/mmc/host/tmio_mmc_dma.c b/drivers/mmc/host/tmio_mmc_dma.c
index f24a029..7e86662 100644
--- a/drivers/mmc/host/tmio_mmc_dma.c
+++ b/drivers/mmc/host/tmio_mmc_dma.c
@@ -26,7 +26,7 @@ static void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
 {
 #if defined(CONFIG_SUPERH) || defined(CONFIG_ARCH_SHMOBILE)
 	/* Switch DMA mode on or off - SuperH specific? */
-	sd_ctrl_write16(host, enable ? 2 : 0, CTL_DMA_ENABLE);
+	sd_ctrl_write16(host, CTL_DMA_ENABLE, enable ? 2 : 0);
 #endif
 }
 
-- 
1.7.2.5

[PATCH 2/2] mmc: tmio: fix a deadlock

[Guennadi Liakhovetski]

Currently the tmio-mmc driver contains a recursive runtime PM method
invocation, which leads to a deadlock on a mutex. Avoid it by taking
care not to request DMA too early.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
---
 drivers/mmc/host/tmio_mmc.h     |    5 +++++
 drivers/mmc/host/tmio_mmc_dma.c |    5 ++++-
 drivers/mmc/host/tmio_mmc_pio.c |    4 ++--
 3 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/drivers/mmc/host/tmio_mmc.h b/drivers/mmc/host/tmio_mmc.h
index 551d1ef..60ba335 100644
--- a/drivers/mmc/host/tmio_mmc.h
+++ b/drivers/mmc/host/tmio_mmc.h
@@ -109,6 +109,7 @@ static inline void tmio_mmc_kunmap_atomic(struct scatterlist *sg,
 
 #if defined(CONFIG_MMC_SDHI) || defined(CONFIG_MMC_SDHI_MODULE)
 void tmio_mmc_start_dma(struct tmio_mmc_host *host, struct mmc_data *data);
+void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable);
 void tmio_mmc_request_dma(struct tmio_mmc_host *host, struct tmio_mmc_data *pdata);
 void tmio_mmc_release_dma(struct tmio_mmc_host *host);
 #else
@@ -117,6 +118,10 @@ static inline void tmio_mmc_start_dma(struct tmio_mmc_host *host,
 {
 }
 
+static inline void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
+{
+}
+
 static inline void tmio_mmc_request_dma(struct tmio_mmc_host *host,
 				 struct tmio_mmc_data *pdata)
 {
diff --git a/drivers/mmc/host/tmio_mmc_dma.c b/drivers/mmc/host/tmio_mmc_dma.c
index 7e86662..2aa616d 100644
--- a/drivers/mmc/host/tmio_mmc_dma.c
+++ b/drivers/mmc/host/tmio_mmc_dma.c
@@ -22,8 +22,11 @@
 
 #define TMIO_MMC_MIN_DMA_LEN 8
 
-static void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
+void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
 {
+	if (!host->chan_tx || !host->chan_rx)
+		return;
+
 #if defined(CONFIG_SUPERH) || defined(CONFIG_ARCH_SHMOBILE)
 	/* Switch DMA mode on or off - SuperH specific? */
 	sd_ctrl_write16(host, CTL_DMA_ENABLE, enable ? 2 : 0);
diff --git a/drivers/mmc/host/tmio_mmc_pio.c b/drivers/mmc/host/tmio_mmc_pio.c
index 221ffb7..1f16357 100644
--- a/drivers/mmc/host/tmio_mmc_pio.c
+++ b/drivers/mmc/host/tmio_mmc_pio.c
@@ -984,7 +984,7 @@ int tmio_mmc_host_resume(struct device *dev)
 	if (host->pm_global) {
 		/* Runtime PM resume callback didn't run */
 		tmio_mmc_reset(host);
-		tmio_mmc_request_dma(host, host->pdata);
+		tmio_mmc_enable_dma(host, true);
 		host->pm_global = false;
 	}
 
@@ -1007,7 +1007,7 @@ int tmio_mmc_host_runtime_resume(struct device *dev)
 	struct tmio_mmc_data *pdata = host->pdata;
 
 	tmio_mmc_reset(host);
-	tmio_mmc_request_dma(host, host->pdata);
+	tmio_mmc_enable_dma(host, true);
 
 	if (pdata->power) {
 		/* Only entered after a card-insert interrupt */
-- 
1.7.2.5

[PATCH 2/2] mmc: tmio: fix a deadlock

[Guennadi Liakhovetski]

Currently the tmio-mmc driver contains a recursive runtime PM method
invocation, which leads to a deadlock on a mutex. Avoid it by taking
care not to request DMA too early.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
---
 drivers/mmc/host/tmio_mmc.h     |    5 +++++
 drivers/mmc/host/tmio_mmc_dma.c |    5 ++++-
 drivers/mmc/host/tmio_mmc_pio.c |    4 ++--
 3 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/drivers/mmc/host/tmio_mmc.h b/drivers/mmc/host/tmio_mmc.h
index 551d1ef..60ba335 100644
--- a/drivers/mmc/host/tmio_mmc.h
+++ b/drivers/mmc/host/tmio_mmc.h
@@ -109,6 +109,7 @@ static inline void tmio_mmc_kunmap_atomic(struct scatterlist *sg,
 
 #if defined(CONFIG_MMC_SDHI) || defined(CONFIG_MMC_SDHI_MODULE)
 void tmio_mmc_start_dma(struct tmio_mmc_host *host, struct mmc_data *data);
+void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable);
 void tmio_mmc_request_dma(struct tmio_mmc_host *host, struct tmio_mmc_data *pdata);
 void tmio_mmc_release_dma(struct tmio_mmc_host *host);
 #else
@@ -117,6 +118,10 @@ static inline void tmio_mmc_start_dma(struct tmio_mmc_host *host,
 {
 }
 
+static inline void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
+{
+}
+
 static inline void tmio_mmc_request_dma(struct tmio_mmc_host *host,
 				 struct tmio_mmc_data *pdata)
 {
diff --git a/drivers/mmc/host/tmio_mmc_dma.c b/drivers/mmc/host/tmio_mmc_dma.c
index 7e86662..2aa616d 100644
--- a/drivers/mmc/host/tmio_mmc_dma.c
+++ b/drivers/mmc/host/tmio_mmc_dma.c
@@ -22,8 +22,11 @@
 
 #define TMIO_MMC_MIN_DMA_LEN 8
 
-static void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
+void tmio_mmc_enable_dma(struct tmio_mmc_host *host, bool enable)
 {
+	if (!host->chan_tx || !host->chan_rx)
+		return;
+
 #if defined(CONFIG_SUPERH) || defined(CONFIG_ARCH_SHMOBILE)
 	/* Switch DMA mode on or off - SuperH specific? */
 	sd_ctrl_write16(host, CTL_DMA_ENABLE, enable ? 2 : 0);
diff --git a/drivers/mmc/host/tmio_mmc_pio.c b/drivers/mmc/host/tmio_mmc_pio.c
index 221ffb7..1f16357 100644
--- a/drivers/mmc/host/tmio_mmc_pio.c
+++ b/drivers/mmc/host/tmio_mmc_pio.c
@@ -984,7 +984,7 @@ int tmio_mmc_host_resume(struct device *dev)
 	if (host->pm_global) {
 		/* Runtime PM resume callback didn't run */
 		tmio_mmc_reset(host);
-		tmio_mmc_request_dma(host, host->pdata);
+		tmio_mmc_enable_dma(host, true);
 		host->pm_global = false;
 	}
 
@@ -1007,7 +1007,7 @@ int tmio_mmc_host_runtime_resume(struct device *dev)
 	struct tmio_mmc_data *pdata = host->pdata;
 
 	tmio_mmc_reset(host);
-	tmio_mmc_request_dma(host, host->pdata);
+	tmio_mmc_enable_dma(host, true);
 
 	if (pdata->power) {
 		/* Only entered after a card-insert interrupt */
-- 
1.7.2.5

Re: [PATCH 0/2] mmc: tmio: two bug-fixes

[Chris Ball]

Hi Guennadi,

On Thu, Jul 14 2011, Guennadi Liakhovetski wrote:
> There go two more bug-fixes for tmio on top of my earlier patches from 
> today, so, should still be easy enough to trace. So, to pull them 
> together, the previous ones were
>
> [PATCH v3] mmc: tmio: fix recursive spinlock, don't schedule with
> interrupts disabled
> [PATCH v3] mmc: tmio: maximize power saving
>
> And now follow
>
> [PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code
> [PATCH 2/2] mmc: tmio: fix a deadlock

Pushed these four plus "mmc: sh_mmcif: maximize power saving" to
mmc-next for 3.1.  Thanks!

By the way, feel free to send me a MAINTAINERS update taking over
TMIO if you'd like to; you've been doing the work.  :-)

- Chris.
-- 
Chris Ball   <cjb@laptop.org>   <http://printf.net/>
One Laptop Per Child

Re: [PATCH 0/2] mmc: tmio: two bug-fixes

[Chris Ball]

Hi Guennadi,

On Thu, Jul 14 2011, Guennadi Liakhovetski wrote:
> There go two more bug-fixes for tmio on top of my earlier patches from 
> today, so, should still be easy enough to trace. So, to pull them 
> together, the previous ones were
>
> [PATCH v3] mmc: tmio: fix recursive spinlock, don't schedule with
> interrupts disabled
> [PATCH v3] mmc: tmio: maximize power saving
>
> And now follow
>
> [PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code
> [PATCH 2/2] mmc: tmio: fix a deadlock

Pushed these four plus "mmc: sh_mmcif: maximize power saving" to
mmc-next for 3.1.  Thanks!

By the way, feel free to send me a MAINTAINERS update taking over
TMIO if you'd like to; you've been doing the work.  :-)

- Chris.
-- 
Chris Ball   <cjb@laptop.org>   <http://printf.net/>
One Laptop Per Child

Re: [PATCH 0/2] mmc: tmio: two bug-fixes

[Guennadi Liakhovetski]

Hi Chris

On Thu, 14 Jul 2011, Chris Ball wrote:

> Hi Guennadi,
> 
> On Thu, Jul 14 2011, Guennadi Liakhovetski wrote:
> > There go two more bug-fixes for tmio on top of my earlier patches from 
> > today, so, should still be easy enough to trace. So, to pull them 
> > together, the previous ones were
> >
> > [PATCH v3] mmc: tmio: fix recursive spinlock, don't schedule with
> > interrupts disabled
> > [PATCH v3] mmc: tmio: maximize power saving
> >
> > And now follow
> >
> > [PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code
> > [PATCH 2/2] mmc: tmio: fix a deadlock
> 
> Pushed these four plus "mmc: sh_mmcif: maximize power saving" to
> mmc-next for 3.1.  Thanks!

Thanks for a fast reaction.

> By the way, feel free to send me a MAINTAINERS update taking over
> TMIO if you'd like to; you've been doing the work.  :-)

Thanks for the offer, I appreciate it! I am not the only active tmio mmc 
contributor, but yes, I could take that role upon me, if noone objects. 
Would be good to hear, what Ian thinks? Is he maybe still planning to 
actively working on the driver? Would he like to stay listed as a 
co-maintainer? If I am to maintain the driver alone, it would be good to 
have access to some non-sh-mobile hardware, using it too, at least for 
regression testing. Any idea where to get anything like that?

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/

Re: [PATCH 0/2] mmc: tmio: two bug-fixes

[Guennadi Liakhovetski]

Hi Chris

On Thu, 14 Jul 2011, Chris Ball wrote:

> Hi Guennadi,
> 
> On Thu, Jul 14 2011, Guennadi Liakhovetski wrote:
> > There go two more bug-fixes for tmio on top of my earlier patches from 
> > today, so, should still be easy enough to trace. So, to pull them 
> > together, the previous ones were
> >
> > [PATCH v3] mmc: tmio: fix recursive spinlock, don't schedule with
> > interrupts disabled
> > [PATCH v3] mmc: tmio: maximize power saving
> >
> > And now follow
> >
> > [PATCH 1/2] mmc: tmio: fix a recently introduced bug in DMA code
> > [PATCH 2/2] mmc: tmio: fix a deadlock
> 
> Pushed these four plus "mmc: sh_mmcif: maximize power saving" to
> mmc-next for 3.1.  Thanks!

Thanks for a fast reaction.

> By the way, feel free to send me a MAINTAINERS update taking over
> TMIO if you'd like to; you've been doing the work.  :-)

Thanks for the offer, I appreciate it! I am not the only active tmio mmc 
contributor, but yes, I could take that role upon me, if noone objects. 
Would be good to hear, what Ian thinks? Is he maybe still planning to 
actively working on the driver? Would he like to stay listed as a 
co-maintainer? If I am to maintain the driver alone, it would be good to 
have access to some non-sh-mobile hardware, using it too, at least for 
regression testing. Any idea where to get anything like that?

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/

Re: [PATCH 0/2] mmc: tmio: two bug-fixes

[Ian Molton]

On Fri, 2011-07-15 at 10:46 +0200, Guennadi Liakhovetski wrote:
> Thanks for the offer, I appreciate it! I am not the only active tmio
> mmc contributor, but yes, I could take that role upon me, if noone
> objects. 
> Would be good to hear, what Ian thinks? Is he maybe still planning to 
> actively working on the driver? Would he like to stay listed as a 
> co-maintainer? If I am to maintain the driver alone, it would be good
> to have access to some non-sh-mobile hardware, using it too, at least
> for regression testing. Any idea where to get anything like that? 

Seem to remember dealing with this already - but I'm fine with it. I'd
like to keep on testing the TMIO chipsets in e-series (need to build
that again sometime soon).

-Ian

Re: [PATCH 0/2] mmc: tmio: two bug-fixes

[Ian Molton]

On Fri, 2011-07-15 at 10:46 +0200, Guennadi Liakhovetski wrote:
> Thanks for the offer, I appreciate it! I am not the only active tmio
> mmc contributor, but yes, I could take that role upon me, if noone
> objects. 
> Would be good to hear, what Ian thinks? Is he maybe still planning to 
> actively working on the driver? Would he like to stay listed as a 
> co-maintainer? If I am to maintain the driver alone, it would be good
> to have access to some non-sh-mobile hardware, using it too, at least
> for regression testing. Any idea where to get anything like that? 

Seem to remember dealing with this already - but I'm fine with it. I'd
like to keep on testing the TMIO chipsets in e-series (need to build
that again sometime soon).

-Ian