[PATCH 0/2] ARM: DRA7: hwmod: fix gpmc hwmod for core-retention

[PATCH 0/2] ARM: DRA7: hwmod: fix gpmc hwmod for core-retention

[Roger Quadros]

Hi,

Disabling GPMC node in device tree prevents core retention on dra7-evm.
Fix that by getting rid of HWMOD_INIT_NO_IDLE & NO_RESET.

Fixup gpmc hwmod as it doesn't support SIDLE_SMART_WKUP.

cheers,
-roger

Roger Quadros (2):
  ARM: DRA7: hwmod: fix gpmc hwmod
  ARM: DRA7: hwmod: Fix GPMC from preventing core suspend

 arch/arm/mach-omap2/omap_hwmod_7xx_data.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

-- 
2.1.4

[PATCH 1/2] ARM: DRA7: hwmod: fix gpmc hwmod

[Roger Quadros]

GPMC smart idle is not really broken but it does not support
smart idle with wakeup.

Fixes: 556708fe8718 ("ARM: OMAP: DRA7: hwmod: Make gpmc software supervised as the smart idle is broken")

Signed-off-by: Roger Quadros <rogerq@ti.com>
---
 arch/arm/mach-omap2/omap_hwmod_7xx_data.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/arch/arm/mach-omap2/omap_hwmod_7xx_data.c b/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
index 0e64c2f..8c2fad5 100644
--- a/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
+++ b/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
@@ -819,8 +819,7 @@ static struct omap_hwmod dra7xx_gpmc_hwmod = {
 	.name		= "gpmc",
 	.class		= &dra7xx_gpmc_hwmod_class,
 	.clkdm_name	= "l3main1_clkdm",
-	.flags		= (HWMOD_INIT_NO_IDLE | HWMOD_INIT_NO_RESET |
-			   HWMOD_SWSUP_SIDLE),
+	.flags		= HWMOD_INIT_NO_IDLE | HWMOD_INIT_NO_RESET,
 	.main_clk	= "l3_iclk_div",
 	.prcm = {
 		.omap4 = {
@@ -1421,8 +1420,7 @@ static struct omap_hwmod_class_sysconfig dra7xx_ocp2scp_sysc = {
 	.syss_offs	= 0x0014,
 	.sysc_flags	= (SYSC_HAS_AUTOIDLE | SYSC_HAS_SIDLEMODE |
 			   SYSC_HAS_SOFTRESET | SYSS_HAS_RESET_STATUS),
-	.idlemodes	= (SIDLE_FORCE | SIDLE_NO | SIDLE_SMART |
-			   SIDLE_SMART_WKUP),
+	.idlemodes	= (SIDLE_FORCE | SIDLE_NO | SIDLE_SMART),
 	.sysc_fields	= &omap_hwmod_sysc_type1,
 };
 
-- 
2.1.4

Re: [PATCH 1/2] ARM: DRA7: hwmod: fix gpmc hwmod

[Paul Walmsley]

On Tue, 2 Jun 2015, Roger Quadros wrote:

> GPMC smart idle is not really broken but it does not support
> smart idle with wakeup.
> 
> Fixes: 556708fe8718 ("ARM: OMAP: DRA7: hwmod: Make gpmc software supervised as the smart idle is broken")
> 
> Signed-off-by: Roger Quadros <rogerq@ti.com>

Thanks, queued for v4.2-rc fixes.

- Paul

Re: [PATCH 1/2] ARM: DRA7: hwmod: fix gpmc hwmod

[Roger Quadros]

Paul,

On Tue, 2 Jun 2015 21:34:58 +0000
Paul Walmsley <paul@pwsan.com> wrote:

> On Tue, 2 Jun 2015, Roger Quadros wrote:
> 
> > GPMC smart idle is not really broken but it does not support
> > smart idle with wakeup.
> > 
> > Fixes: 556708fe8718 ("ARM: OMAP: DRA7: hwmod: Make gpmc software supervised as the smart idle is broken")
> > 
> > Signed-off-by: Roger Quadros <rogerq@ti.com>
> 
> Thanks, queued for v4.2-rc fixes.

you might want to take v2 of the patch [1] or did you fix
the conflict yourself?

cheers,
-roger

[1] http://article.gmane.org/gmane.linux.ports.arm.omap/126730

[PATCH v2 1/1] ARM: DRA7: hwmod: fix gpmc hwmod

[Roger Quadros]

GPMC smart idle is not really broken but it does not support
smart idle with wakeup.

Fixes: 556708fe8718 ("ARM: OMAP: DRA7: hwmod: Make gpmc software supervised as the smart idle is broken")

Signed-off-by: Roger Quadros <rogerq@ti.com>
---
v2 - Rebased on top of for-v4.2/omap-hwmod-a
Patch 2/2 is no longer needed

  arch/arm/mach-omap2/omap_hwmod_7xx_data.c | 5 ++---
  1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/arch/arm/mach-omap2/omap_hwmod_7xx_data.c b/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
index a0411f3..152526a 100644
--- a/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
+++ b/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
@@ -820,7 +820,7 @@ static struct omap_hwmod dra7xx_gpmc_hwmod = {
  	.class		= &dra7xx_gpmc_hwmod_class,
  	.clkdm_name	= "l3main1_clkdm",
  	/* Skip reset for CONFIG_OMAP_GPMC_DEBUG for bootloader timings */
-	.flags		= HWMOD_SWSUP_SIDLE | DEBUG_OMAP_GPMC_HWMOD_FLAGS,
+	.flags		= DEBUG_OMAP_GPMC_HWMOD_FLAGS,
  	.main_clk	= "l3_iclk_div",
  	.prcm = {
  		.omap4 = {
@@ -1421,8 +1421,7 @@ static struct omap_hwmod_class_sysconfig dra7xx_ocp2scp_sysc = {
  	.syss_offs	= 0x0014,
  	.sysc_flags	= (SYSC_HAS_AUTOIDLE | SYSC_HAS_SIDLEMODE |
  			   SYSC_HAS_SOFTRESET | SYSS_HAS_RESET_STATUS),
-	.idlemodes	= (SIDLE_FORCE | SIDLE_NO | SIDLE_SMART |
-			   SIDLE_SMART_WKUP),
+	.idlemodes	= (SIDLE_FORCE | SIDLE_NO | SIDLE_SMART),
  	.sysc_fields	= &omap_hwmod_sysc_type1,
  };
  
-- 
2.1.4

[PATCH 2/2] ARM: DRA7: hwmod: Fix GPMC from preventing core suspend

[Roger Quadros]

GPMC hwmod is flagged as HWMOD_INIT_NO_IDLE so it is kept
enabled at boot. If the GPMC driver is not loaded then
GPMC will not be idled thus preventing CORE from going idle
during suspend.

Disable HWMOD_INIT_NO_IDLE and HWMOD_INIT_NO_RESET.

The only reason HWMOD_INIT_NO_RESET was there was to retain
GPMC timings/settings configured by bootloader. We no longer
need that as we're configuring the timins in the kernel.

There is no reasoning as to why HWMOD_INIT_NO_IDLE was there.
Seems to have beein blindly copied from omap3/4 hwmod code.

Signed-off-by: Roger Quadros <rogerq@ti.com>
---
 arch/arm/mach-omap2/omap_hwmod_7xx_data.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/arch/arm/mach-omap2/omap_hwmod_7xx_data.c b/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
index 8c2fad5..52c6565 100644
--- a/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
+++ b/arch/arm/mach-omap2/omap_hwmod_7xx_data.c
@@ -819,7 +819,6 @@ static struct omap_hwmod dra7xx_gpmc_hwmod = {
 	.name		= "gpmc",
 	.class		= &dra7xx_gpmc_hwmod_class,
 	.clkdm_name	= "l3main1_clkdm",
-	.flags		= HWMOD_INIT_NO_IDLE | HWMOD_INIT_NO_RESET,
 	.main_clk	= "l3_iclk_div",
 	.prcm = {
 		.omap4 = {
-- 
2.1.4

Re: [PATCH 2/2] ARM: DRA7: hwmod: Fix GPMC from preventing core suspend

[Paul Walmsley]

On Tue, 2 Jun 2015, Roger Quadros wrote:

> GPMC hwmod is flagged as HWMOD_INIT_NO_IDLE so it is kept
> enabled at boot. If the GPMC driver is not loaded then
> GPMC will not be idled thus preventing CORE from going idle
> during suspend.
> 
> Disable HWMOD_INIT_NO_IDLE and HWMOD_INIT_NO_RESET.
> 
> The only reason HWMOD_INIT_NO_RESET was there was to retain
> GPMC timings/settings configured by bootloader. We no longer
> need that as we're configuring the timins in the kernel.
> 
> There is no reasoning as to why HWMOD_INIT_NO_IDLE was there.
> Seems to have beein blindly copied from omap3/4 hwmod code.
> 
> Signed-off-by: Roger Quadros <rogerq@ti.com>

Hi Roger, could you take a look at Tony's patch "memory: omap-gpmc: Add 
Kconfig option for debug" and see if this needs to be changed in light of 
that patch?

regards,

- Paul

Re: [PATCH 2/2] ARM: DRA7: hwmod: Fix GPMC from preventing core suspend

[Roger Quadros]

Paul,

On 03/06/15 00:36, Paul Walmsley wrote:
> On Tue, 2 Jun 2015, Roger Quadros wrote:
>
>> GPMC hwmod is flagged as HWMOD_INIT_NO_IDLE so it is kept
>> enabled at boot. If the GPMC driver is not loaded then
>> GPMC will not be idled thus preventing CORE from going idle
>> during suspend.
>>
>> Disable HWMOD_INIT_NO_IDLE and HWMOD_INIT_NO_RESET.
>>
>> The only reason HWMOD_INIT_NO_RESET was there was to retain
>> GPMC timings/settings configured by bootloader. We no longer
>> need that as we're configuring the timins in the kernel.
>>
>> There is no reasoning as to why HWMOD_INIT_NO_IDLE was there.
>> Seems to have beein blindly copied from omap3/4 hwmod code.
>>
>> Signed-off-by: Roger Quadros <rogerq@ti.com>
>
> Hi Roger, could you take a look at Tony's patch "memory: omap-gpmc: Add
> Kconfig option for debug" and see if this needs to be changed in light of
> that patch?

Yes. We don't need this patch if we can just rid of HWMOD_SWSUP_SIDLE
as done in the first patch.

cheers,
-roger