Skip to contentSkip to conversation

ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

PATCH Superseded Geert Uytterhoeven 4 participants 22 messages Sep 25, 2014 – Oct 6, 2014

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Geert Uytterhoeven

author maintainer
#0 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Simon, Magnus,

This series improves PM domain support for r8a7740/armadillo800eva legacy, mimicking (parts of) the existing PM domain support for sh7372/mackerel.

More specifically, they
- Add missing devices to existing PM domains,
- Add missing PM domains, hooking up devices and subdomains,

Changes in v2:
- Dropped FIXMes on behavior that turned out to be expected,
- Hook up the HDMI-Link hardware block,
- Really add the A4MP and D4 PM domains, as fixes are available (see dependencies below),
- Move special CPU domain handling from A4S to A3SM,

There are a few dependencies to make resume from s2ram work:
- "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume",
- "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution.

Apart from that, I'm not aware of any regressions (except perhaps for USB, which I was unable to get working anyway).

Please note that DT/multi-platform PM domain support is added in an (almost) independent series.

Thanks!

Geert Uytterhoeven (11):
ARM: shmobile: r8a7740: Add missing A3SP pm domain devices
ARM: shmobile: r8a7740: Add missing A4S pm domain devices
ARM: shmobile: armadillo800eva legacy: Add missing A3SP pm domain
devices
ARM: shmobile: armadillo800eva legacy: Add missing A4S pm domain
devices
ARM: shmobile: r8a7740: Add A3RV pm domain support
ARM: shmobile: r8a7740: Add A3SG pm domain support
ARM: shmobile: r8a7740/armadillo legacy: Add A4MP pm domain support
ARM: shmobile: r8a7740: Add D4 pm domain support
ARM: shmobile: r8a7740/armadillo legacy: Add A4R pm domain support
ARM: shmobile: r8a7740: Add A4SU pm domain support
ARM: shmobile: r8a7740: Add A3SM pm domain support

 arch/arm/mach-shmobile/board-armadillo800eva.c | 12 +++++++++-
 arch/arm/mach-shmobile/pm-r8a7740.c            | 32 ++++++++++++++++++++++----
 arch/arm/mach-shmobile/setup-r8a7740.c         | 11 +++++++++
 3 files changed, 50 insertions(+), 5 deletions(-)
Signature · 12 lines
1.9.1

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds
author maintainer
#1 lore raw

PATCH v2 1/11 ARM: shmobile: r8a7740: Add missing A3SP pm domain devices

Commit 802a5639aa7041b2 ("ARM: shmobile: r8a7740: add A3SP pm domain support") added the A3SP power domain, but forgot to hook up the IPPMU, DMAC0/1/2, and USBDMAC hardware blocks.

---
v2:
- No changes

 arch/arm/mach-shmobile/setup-r8a7740.c | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm/mach-shmobile/setup-r8a7740.c +5 -0
index 8894e1b7ab0e65bb..6992d4013ef013da 100644
@@ -756,6 +756,11 @@ void __init r8a7740_add_standard_devices(void)
 		{ "A3SP", &scif7_device },
 		{ "A3SP", &scif8_device },
 		{ "A3SP", &i2c1_device },
+		{ "A3SP", &ipmmu_device },
+		{ "A3SP", &dma0_device },
+		{ "A3SP", &dma1_device },
+		{ "A3SP", &dma2_device },
+		{ "A3SP", &usb_dma_device },
 	};
 
 	/* I2C work-around */
Signature · 1 line
1.9.1
author maintainer
#2 lore raw

PATCH v2 2/11 ARM: shmobile: r8a7740: Add missing A4S pm domain devices

In reply to Geert Uytterhoeven

Commit 8459293c27bcd13a ("ARM: shmobile: r8a7740: add A4S pm domain support") added the A4S power domain, but forgot to hook up the INTCA hardware block.

---
v2:
- No changes

 arch/arm/mach-shmobile/setup-r8a7740.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/arm/mach-shmobile/setup-r8a7740.c +4 -0
index 6992d4013ef013da..dbf8a93ccc4f412d 100644
@@ -746,6 +746,10 @@ static void r8a7740_i2c_workaround(struct platform_device *pdev)
 void __init r8a7740_add_standard_devices(void)
 {
 	static struct pm_domain_device domain_devices[] __initdata = {
+		{ "A4S",  &irqpin0_device },
+		{ "A4S",  &irqpin1_device },
+		{ "A4S",  &irqpin2_device },
+		{ "A4S",  &irqpin3_device },
 		{ "A3SP", &scif0_device },
 		{ "A3SP", &scif1_device },
 		{ "A3SP", &scif2_device },
Signature · 1 line
1.9.1
author maintainer
#3 lore raw

PATCH v2 3/11 ARM: shmobile: armadillo800eva legacy: Add missing A3SP pm domain devices

In reply to Geert Uytterhoeven

Commit 802a5639aa7041b2 ("ARM: shmobile: r8a7740: add A3SP pm domain support") added the A3SP power domain, but forgot to hook up the TPU, SDHI0/1, and MMCIF hardware blocks.

Note: As the default PM QoS latency constraint for SDHI is only 100 ?s (cfr. commit c419e611c3c59c0e ("tmio_mmc / PM: Use PM QoS latency constraint"), while DEFAULT_DEV_LATENCY_NS is 250000, suspend fails with -EBUSY, unless the constraint is increased first to more than 500 ?s using e.g.

echo 501 > /sys/devices/platform/sh_mobile_sdhi.0/power/pm_qos_resume_latency_us

---
v2:
- Converted FIXME to Note, as this is expected behavior.

 arch/arm/mach-shmobile/board-armadillo800eva.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
arch/arm/mach-shmobile/board-armadillo800eva.c +7 -1
index f19459a8c3edbb77..64a68450c28a072c 100644
@@ -1234,8 +1234,11 @@ static void __init eva_init(void)
 	static struct pm_domain_device domain_devices[] __initdata = {
 		{ "A4LC", &lcdc0_device },
 		{ "A4LC", &hdmi_lcdc_device },
+		{ "A3SP", &pwm_device },
+		{ "A3SP", &sdhi0_device },
+		{ "A3SP", &sh_mmcif_device },
 	};
-	struct platform_device *usb = NULL;
+	struct platform_device *usb = NULL, *sdhi1 = NULL;
 
 	regulator_register_always_on(0, "fixed-3.3V", fixed3v3_power_consumers,
 				     ARRAY_SIZE(fixed3v3_power_consumers), 3300000);
@@ -1304,6 +1307,7 @@ static void __init eva_init(void)
 
 		platform_device_register(&vcc_sdhi1);
 		platform_device_register(&sdhi1_device);
+		sdhi1 = &sdhi1_device;
 	}
 
 
@@ -1324,6 +1328,8 @@ static void __init eva_init(void)
 				       ARRAY_SIZE(domain_devices));
 	if (usb)
 		rmobile_add_device_to_domain("A3SP", usb);
+	if (sdhi1)
+		rmobile_add_device_to_domain("A3SP", sdhi1);
 
 	r8a7740_pm_init();
 }
Signature · 1 line
1.9.1
author maintainer
#4 lore raw

PATCH v2 4/11 ARM: shmobile: armadillo800eva legacy: Add missing A4S pm domain devices

In reply to Geert Uytterhoeven

Commit 8459293c27bcd13a ("ARM: shmobile: r8a7740: add A4S pm domain support") added the A4S power domain, but forgot to hook up the GbEther hardware block.

---
v2:
- No changes

 arch/arm/mach-shmobile/board-armadillo800eva.c | 1 +
 1 file changed, 1 insertion(+)
arch/arm/mach-shmobile/board-armadillo800eva.c +1 -0
index 64a68450c28a072c..735036038951868d 100644
@@ -1234,6 +1234,7 @@ static void __init eva_init(void)
 	static struct pm_domain_device domain_devices[] __initdata = {
 		{ "A4LC", &lcdc0_device },
 		{ "A4LC", &hdmi_lcdc_device },
+		{ "A4S",  &sh_eth_device },
 		{ "A3SP", &pwm_device },
 		{ "A3SP", &sdhi0_device },
 		{ "A3SP", &sh_mmcif_device },
Signature · 1 line
1.9.1
author maintainer
#5 lore raw

PATCH v2 5/11 ARM: shmobile: r8a7740: Add A3RV pm domain support

In reply to Geert Uytterhoeven

Add support for the A3RV power domain.
This domain contains the VPU5F and VCP1 hardware blocks, which are currently not used by any driver.

---
v2:
- No changes

 arch/arm/mach-shmobile/pm-r8a7740.c | 3 +++
 1 file changed, 3 insertions(+)
arch/arm/mach-shmobile/pm-r8a7740.c +3 -0
index e3f1464482371bb4..1d916038660d4b8a 100644
@@ -37,6 +37,9 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 		.genpd.name	= "A4LC",
 		.bit_shift	= 1,
 	}, {
+		.genpd.name	= "A3RV",
+		.bit_shift	= 6,
+	}, {
 		.genpd.name	= "A4S",
 		.bit_shift	= 10,
 		.gov		= &pm_domain_always_on_gov,
Signature · 1 line
1.9.1
author maintainer
#6 lore raw

PATCH v2 6/11 ARM: shmobile: r8a7740: Add A3SG pm domain support

In reply to Geert Uytterhoeven

Add support for the A3SG power domain, and hook it up as a subdomain of A4S.
This domain contains the SGX540 hardware block, which is currently not used by any driver.

---
v2:
- No changes

 arch/arm/mach-shmobile/pm-r8a7740.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/arm/mach-shmobile/pm-r8a7740.c +4 -0
index 1d916038660d4b8a..932c2a621e81ac26 100644
@@ -51,6 +51,9 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 		.gov		= &pm_domain_always_on_gov,
 		.no_debug	= true,
 		.suspend	= r8a7740_pd_a3sp_suspend,
+	}, {
+		.genpd.name	= "A3SG",
+		.bit_shift	= 13,
 	},
 };
 
@@ -58,6 +61,7 @@ void __init r8a7740_init_pm_domains(void)
 {
 	rmobile_init_domains(r8a7740_pm_domains, ARRAY_SIZE(r8a7740_pm_domains));
 	pm_genpd_add_subdomain_names("A4S", "A3SP");
+	pm_genpd_add_subdomain_names("A4S", "A3SG");
 }
 #endif /* CONFIG_PM && !CONFIG_ARCH_MULTIPLATFORM */
 
Signature · 1 line
1.9.1
author maintainer
#7 lore raw

PATCH v2 7/11 ARM: shmobile: r8a7740/armadillo legacy: Add A4MP pm domain support

In reply to Geert Uytterhoeven

Add support for the A4MP power domain, and hook up the HDMI-Link and FSI hardware blocks.
This domain also contains the SPU2, FMSI, and BBIF2 hardware blocks, but these are currently not used by any driver.

---
This depends on "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume". Else resuming from s2ram will cause an interrupt storm and "irq 163: nobody cared".

v2:
- Also hook up HDMI-Link, which was not documented in the datasheet as belonging to A4MP,
- Really add A4MP and drop the FIXME, as a fix for the interrupt storm is available.

---
 arch/arm/mach-shmobile/board-armadillo800eva.c | 2 ++
 arch/arm/mach-shmobile/pm-r8a7740.c            | 3 +++
 2 files changed, 5 insertions(+)
arch/arm/mach-shmobile/board-armadillo800eva.c +2 -0
index 735036038951868d..c37d7b697a35322d 100644
@@ -1234,6 +1234,8 @@ static void __init eva_init(void)
 	static struct pm_domain_device domain_devices[] __initdata = {
 		{ "A4LC", &lcdc0_device },
 		{ "A4LC", &hdmi_lcdc_device },
+		{ "A4MP", &hdmi_device },
+		{ "A4MP", &fsi_device },
 		{ "A4S",  &sh_eth_device },
 		{ "A3SP", &pwm_device },
 		{ "A3SP", &sdhi0_device },
arch/arm/mach-shmobile/pm-r8a7740.c +3 -0
index 932c2a621e81ac26..081a4f9726a0c909 100644
@@ -37,6 +37,9 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 		.genpd.name	= "A4LC",
 		.bit_shift	= 1,
 	}, {
+		.genpd.name	= "A4MP",
+		.bit_shift	= 2,
+	}, {
 		.genpd.name	= "A3RV",
 		.bit_shift	= 6,
 	}, {
Signature · 1 line
1.9.1
author maintainer
#8 lore raw

PATCH v2 8/11 ARM: shmobile: r8a7740: Add D4 pm domain support

In reply to Geert Uytterhoeven

Add support for the D4 power domain.
This domain contains the Coresight-ETM hardware block.

---
This depends on "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution. Else resuming from s2ram will cause a crash.

v2:
- Really add D4 and drop the FIXME, as a fix for the crash is available.

---
 arch/arm/mach-shmobile/pm-r8a7740.c | 3 +++
 1 file changed, 3 insertions(+)
arch/arm/mach-shmobile/pm-r8a7740.c +3 -0
index 081a4f9726a0c909..6697ac00eff72957 100644
@@ -40,6 +40,9 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 		.genpd.name	= "A4MP",
 		.bit_shift	= 2,
 	}, {
+		.genpd.name	= "D4",
+		.bit_shift	= 3,
+	}, {
 		.genpd.name	= "A3RV",
 		.bit_shift	= 6,
 	}, {
Signature · 1 line
1.9.1
author maintainer
#9 lore raw

PATCH v2 9/11 ARM: shmobile: r8a7740/armadillo legacy: Add A4R pm domain support

In reply to Geert Uytterhoeven

Add support for the A4R power domain, and hook up the A3RV subdomain, and the CEU0, TMU0, and IIC0 hardware blocks.
This domain also contains the Realtime CPU (SH-4A), Realtime CPU debug modules, H-UDI, RT-SHwy, INTCS, RT-HPB, VIO6C, JPU, RTDMAC1/2, SSP, MSIOF0, CMT0, ICB, DREQPAK (RT), 2DDMAC, IPMMUI, and 2DG hardware blocks, but these are currently not used by any driver.

---
v2:
- No changes

 arch/arm/mach-shmobile/board-armadillo800eva.c | 1 +
 arch/arm/mach-shmobile/pm-r8a7740.c            | 4 ++++
 arch/arm/mach-shmobile/setup-r8a7740.c         | 2 ++
 3 files changed, 7 insertions(+)
arch/arm/mach-shmobile/board-armadillo800eva.c +1 -0
index c37d7b697a35322d..2632beaad6c838c2 100644
@@ -1236,6 +1236,7 @@ static void __init eva_init(void)
 		{ "A4LC", &hdmi_lcdc_device },
 		{ "A4MP", &hdmi_device },
 		{ "A4MP", &fsi_device },
+		{ "A4R",  &ceu0_device },
 		{ "A4S",  &sh_eth_device },
 		{ "A3SP", &pwm_device },
 		{ "A3SP", &sdhi0_device },
arch/arm/mach-shmobile/pm-r8a7740.c +4 -0
index 6697ac00eff72957..7226c3543ecdc0ea 100644
@@ -43,6 +43,9 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 		.genpd.name	= "D4",
 		.bit_shift	= 3,
 	}, {
+		.genpd.name	= "A4R",
+		.bit_shift	= 5,
+	}, {
 		.genpd.name	= "A3RV",
 		.bit_shift	= 6,
 	}, {
@@ -66,6 +69,7 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 void __init r8a7740_init_pm_domains(void)
 {
 	rmobile_init_domains(r8a7740_pm_domains, ARRAY_SIZE(r8a7740_pm_domains));
+	pm_genpd_add_subdomain_names("A4R", "A3RV");
 	pm_genpd_add_subdomain_names("A4S", "A3SP");
 	pm_genpd_add_subdomain_names("A4S", "A3SG");
 }
arch/arm/mach-shmobile/setup-r8a7740.c +2 -0
index dbf8a93ccc4f412d..b77c226ab891a5dd 100644
@@ -746,6 +746,8 @@ static void r8a7740_i2c_workaround(struct platform_device *pdev)
 void __init r8a7740_add_standard_devices(void)
 {
 	static struct pm_domain_device domain_devices[] __initdata = {
+		{ "A4R",  &tmu0_device },
+		{ "A4R",  &i2c0_device },
 		{ "A4S",  &irqpin0_device },
 		{ "A4S",  &irqpin1_device },
 		{ "A4S",  &irqpin2_device },
Signature · 1 line
1.9.1
author maintainer
#10 lore raw

PATCH v2 10/11 ARM: shmobile: r8a7740: Add A4SU pm domain support

In reply to Geert Uytterhoeven

Add support for the A4SU power domain.
This domain contains the USBPHY hardware block, which is currently not used by any driver.

---
v2:
- Drop FIXME about hooking up USB platform devices, as the Armadillo platform code doesn't register a USB host device anyway?

---
 arch/arm/mach-shmobile/pm-r8a7740.c | 3 +++
 1 file changed, 3 insertions(+)
arch/arm/mach-shmobile/pm-r8a7740.c +3 -0
index 7226c3543ecdc0ea..f3a1c92f5de6177d 100644
@@ -63,6 +63,9 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 	}, {
 		.genpd.name	= "A3SG",
 		.bit_shift	= 13,
+	}, {
+		.genpd.name	= "A4SU",
+		.bit_shift	= 20,
 	},
 };
 
Signature · 1 line
1.9.1
author maintainer
#11 lore raw

PATCH v2 11/11 ARM: shmobile: r8a7740: Add A3SM pm domain support

In reply to Geert Uytterhoeven

Add support for the A3SM power domain, and hook it up as a subdomain of A4S.

This domain contains the System CPU (Cortex-A9) hardware block. Hence move the special CPU handling from A4S to A3SM.

---
v2:
- Move special CPU domain handling from A4S to A3SM,
- Drop FIXME about not powering down A4S, as this is expected behavior.

---
 arch/arm/mach-shmobile/pm-r8a7740.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)
arch/arm/mach-shmobile/pm-r8a7740.c +8 -4
index f3a1c92f5de6177d..41e61f1b5d44f00d 100644
@@ -14,10 +14,10 @@
 #include "pm-rmobile.h"
 
 #if defined(CONFIG_PM) && !defined(CONFIG_ARCH_MULTIPLATFORM)
-static int r8a7740_pd_a4s_suspend(void)
+static int r8a7740_pd_a3sm_suspend(void)
 {
 	/*
-	 * The A4S domain contains the CPU core and therefore it should
+	 * The A3SM domain contains the CPU core and therefore it should
 	 * only be turned off if the CPU is not in use.
 	 */
 	return -EBUSY;
@@ -51,9 +51,7 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 	}, {
 		.genpd.name	= "A4S",
 		.bit_shift	= 10,
-		.gov		= &pm_domain_always_on_gov,
 		.no_debug	= true,
-		.suspend	= r8a7740_pd_a4s_suspend,
 	}, {
 		.genpd.name	= "A3SP",
 		.bit_shift	= 11,
@@ -61,6 +59,11 @@ static struct rmobile_pm_domain r8a7740_pm_domains[] = {
 		.no_debug	= true,
 		.suspend	= r8a7740_pd_a3sp_suspend,
 	}, {
+		.genpd.name	= "A3SM",
+		.bit_shift	= 12,
+		.gov		= &pm_domain_always_on_gov,
+		.suspend	= r8a7740_pd_a3sm_suspend,
+	}, {
 		.genpd.name	= "A3SG",
 		.bit_shift	= 13,
 	}, {
@@ -74,6 +77,7 @@ void __init r8a7740_init_pm_domains(void)
 	rmobile_init_domains(r8a7740_pm_domains, ARRAY_SIZE(r8a7740_pm_domains));
 	pm_genpd_add_subdomain_names("A4R", "A3RV");
 	pm_genpd_add_subdomain_names("A4S", "A3SP");
+	pm_genpd_add_subdomain_names("A4S", "A3SM");
 	pm_genpd_add_subdomain_names("A4S", "A3SG");
 }
 #endif /* CONFIG_PM && !CONFIG_ARCH_MULTIPLATFORM */
Signature · 1 line
1.9.1
#12 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

In reply to Geert Uytterhoeven
Quoted 16 lines · Geert Uytterhoeven

Hi Simon, Magnus,

This series improves PM domain support for r8a7740/armadillo800eva legacy, mimicking (parts of) the existing PM domain support for sh7372/mackerel.

More specifically, they
- Add missing devices to existing PM domains,
- Add missing PM domains, hooking up devices and subdomains,

Changes in v2:
- Dropped FIXMes on behavior that turned out to be expected,
- Hook up the HDMI-Link hardware block,
- Really add the A4MP and D4 PM domains, as fixes are available (see dependencies below),
- Move special CPU domain handling from A4S to A3SM,

Quoted 8 lines

There are a few dependencies to make resume from s2ram work:
- "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume",
- "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution.

Apart from that, I'm not aware of any regressions (except perhaps for USB, which I was unable to get working anyway).

Hi Geert,

what is your feeling about (me) queueing up part of all of this series now?

Quoted 40 lines

Please note that DT/multi-platform PM domain support is added in an (almost) independent series.

Thanks!

Geert Uytterhoeven (11):
ARM: shmobile: r8a7740: Add missing A3SP pm domain devices
ARM: shmobile: r8a7740: Add missing A4S pm domain devices
ARM: shmobile: armadillo800eva legacy: Add missing A3SP pm domain
devices
ARM: shmobile: armadillo800eva legacy: Add missing A4S pm domain
devices
ARM: shmobile: r8a7740: Add A3RV pm domain support
ARM: shmobile: r8a7740: Add A3SG pm domain support
ARM: shmobile: r8a7740/armadillo legacy: Add A4MP pm domain support
ARM: shmobile: r8a7740: Add D4 pm domain support
ARM: shmobile: r8a7740/armadillo legacy: Add A4R pm domain support
ARM: shmobile: r8a7740: Add A4SU pm domain support
ARM: shmobile: r8a7740: Add A3SM pm domain support

 arch/arm/mach-shmobile/board-armadillo800eva.c | 12 +++++++++-
 arch/arm/mach-shmobile/pm-r8a7740.c            | 32 ++++++++++++++++++++++----
 arch/arm/mach-shmobile/setup-r8a7740.c         | 11 +++++++++
 3 files changed, 50 insertions(+), 5 deletions(-)
Signature · 12 lines
1.9.1

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds
#13 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Simon,

Quoted 21 lines · Simon Horman
Quoted 20 lines · Geert Uytterhoeven

This series improves PM domain support for r8a7740/armadillo800eva legacy, mimicking (parts of) the existing PM domain support for sh7372/mackerel.

More specifically, they
- Add missing devices to existing PM domains,
- Add missing PM domains, hooking up devices and subdomains,

Changes in v2:
- Dropped FIXMes on behavior that turned out to be expected,
- Hook up the HDMI-Link hardware block,
- Really add the A4MP and D4 PM domains, as fixes are available (see dependencies below),
- Move special CPU domain handling from A4S to A3SM,

There are a few dependencies to make resume from s2ram work:
- "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume",
- "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution.

Quoted 6 lines
Quoted 2 lines

Apart from that, I'm not aware of any regressions (except perhaps for USB, which I was unable to get working anyway).

Hi Geert,

what is your feeling about (me) queueing up part of all of this series now?

If you (Magnus?) can live with non-working resume from s2ram for now, it can be queued for v3.19.

I don't think I can improve on the HDMI fix without getting docs and spending (much) time to get HDMI working on Armadillo.

The hw_breakpoint fix has to be replaced by something better/acceptable, I'm working on that.

Quoted 20 lines
Quoted 20 lines

Please note that DT/multi-platform PM domain support is added in an (almost) independent series.

Thanks!

Geert Uytterhoeven (11):
ARM: shmobile: r8a7740: Add missing A3SP pm domain devices
ARM: shmobile: r8a7740: Add missing A4S pm domain devices
ARM: shmobile: armadillo800eva legacy: Add missing A3SP pm domain
devices
ARM: shmobile: armadillo800eva legacy: Add missing A4S pm domain
devices
ARM: shmobile: r8a7740: Add A3RV pm domain support
ARM: shmobile: r8a7740: Add A3SG pm domain support
ARM: shmobile: r8a7740/armadillo legacy: Add A4MP pm domain support
ARM: shmobile: r8a7740: Add D4 pm domain support
ARM: shmobile: r8a7740/armadillo legacy: Add A4R pm domain support
ARM: shmobile: r8a7740: Add A4SU pm domain support
ARM: shmobile: r8a7740: Add A3SM pm domain support

Quoted 7 lines
Quoted 7 lines
 arch/arm/mach-shmobile/board-armadillo800eva.c | 12 +++++++++-
 arch/arm/mach-shmobile/pm-r8a7740.c            | 32 ++++++++++++++++++++++----
 arch/arm/mach-shmobile/setup-r8a7740.c         | 11 +++++++++
 3 files changed, 50 insertions(+), 5 deletions(-)

--
1.9.1

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

maintainer
#14 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Geert,

Quoted 27 lines · Geert Uytterhoeven

Hi Simon,

Quoted 24 lines · Simon Horman
Quoted 22 lines · Geert Uytterhoeven

This series improves PM domain support for r8a7740/armadillo800eva legacy, mimicking (parts of) the existing PM domain support for sh7372/mackerel.

More specifically, they
- Add missing devices to existing PM domains,
- Add missing PM domains, hooking up devices and subdomains,

Changes in v2:
- Dropped FIXMes on behavior that turned out to be expected,
- Hook up the HDMI-Link hardware block,
- Really add the A4MP and D4 PM domains, as fixes are available (see dependencies below),
- Move special CPU domain handling from A4S to A3SM,

There are a few dependencies to make resume from s2ram work:
- "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume",
- "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution.

Apart from that, I'm not aware of any regressions (except perhaps for USB, which I was unable to get working anyway).

Quoted 6 lines
Quoted 3 lines

Hi Geert,

what is your feeling about (me) queueing up part of all of this series now?

If you (Magnus?) can live with non-working resume from s2ram for now, it can be queued for v3.19.

If the series as a whole introduces a known regression then I think we should only merge the portions that work without introducing some issue.

Quoted 2 lines

I don't think I can improve on the HDMI fix without getting docs and spending (much) time to get HDMI working on Armadillo.

Ok, and I suspect it is quite unlikely we will receive any improved docs for r8a7740.

Quoted 2 lines

The hw_breakpoint fix has to be replaced by something better/acceptable, I'm working on that.

Thanks!

So what is a good way to proceed with this series I wonder?

Cheers,

/ magnus

#15 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Magnus,

Quoted 20 lines · Magnus Damm
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines

There are a few dependencies to make resume from s2ram work:
- "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume",
- "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution.

Apart from that, I'm not aware of any regressions (except perhaps for USB, which I was unable to get working anyway).

Quoted 6 lines
Quoted 3 lines

Hi Geert,

what is your feeling about (me) queueing up part of all of this series now?

If you (Magnus?) can live with non-working resume from s2ram for now, it can be queued for v3.19.

If the series as a whole introduces a known regression then I think we should only merge the portions that work without introducing some issue.

Quoted 5 lines
Quoted 2 lines

I don't think I can improve on the HDMI fix without getting docs and spending (much) time to get HDMI working on Armadillo.

Ok, and I suspect it is quite unlikely we will receive any improved docs for r8a7740.

So I will resubmit the HDMI fix.

Quoted 6 lines
Quoted 2 lines

The hw_breakpoint fix has to be replaced by something better/acceptable, I'm working on that.

Thanks!

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

maintainer
#16 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Geert,

Quoted 44 lines · Geert Uytterhoeven

Hi Magnus,

Quoted 20 lines · Magnus Damm
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines

There are a few dependencies to make resume from s2ram work:
- "[PATCH/RFC] fbdev: sh_mobile_hdmi: Re-init regs before irq re-enable on resume",
- "[PATCH] ARM: hw_breakpoint: Trap undef instruction exceptions on wake-up", or an alternative solution.

Apart from that, I'm not aware of any regressions (except perhaps for USB, which I was unable to get working anyway).

Quoted 6 lines
Quoted 3 lines

Hi Geert,

what is your feeling about (me) queueing up part of all of this series now?

If you (Magnus?) can live with non-working resume from s2ram for now, it can be queued for v3.19.

If the series as a whole introduces a known regression then I think we should only merge the portions that work without introducing some issue.

Quoted 5 lines
Quoted 2 lines

I don't think I can improve on the HDMI fix without getting docs and spending (much) time to get HDMI working on Armadillo.

Ok, and I suspect it is quite unlikely we will receive any improved docs for r8a7740.

So I will resubmit the HDMI fix.

Quoted 6 lines
Quoted 2 lines

The hw_breakpoint fix has to be replaced by something better/acceptable, I'm working on that.

Thanks!

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Quoted 4 lines

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Thanks for leading the way! From my side I don't have any special feeling, so why don't you and Simon decide?

If it was up to me only then I think option #1 that simply reduces the feature set seems rather natural, but perhaps it requires a lot of rework of the patches. If we go with option #1 we can always incrementally add D4 and A4MP later I assume.

Cheers,

/ magnus

#17 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Magnus,

Quoted 8 lines · Magnus Damm
Quoted 8 lines
Quoted 1 line

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Quoted 7 lines
Quoted 4 lines

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Thanks for leading the way! From my side I don't have any special feeling, so why don't you and Simon decide?

OK.

Quoted 4 lines

If it was up to me only then I think option #1 that simply reduces the feature set seems rather natural, but perhaps it requires a lot of rework of the patches. If we go with option #1 we can always incrementally add D4 and A4MP later I assume.

Yes we can.

Please wait a bit, I'm looking into the !CONFIG_PM_RUNTIME case. Mostly multi-platform, but it may affect legacy too, where there's interaction with the drivers/sh/pm_runtime.c hack.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

maintainer
#18 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Geert,

Quoted 19 lines · Geert Uytterhoeven

Hi Magnus,

Quoted 8 lines · Magnus Damm
Quoted 8 lines
Quoted 1 line

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Quoted 7 lines
Quoted 4 lines

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Thanks for leading the way! From my side I don't have any special feeling, so why don't you and Simon decide?

Quoted 8 lines

OK.

Quoted 4 lines

If it was up to me only then I think option #1 that simply reduces the feature set seems rather natural, but perhaps it requires a lot of rework of the patches. If we go with option #1 we can always incrementally add D4 and A4MP later I assume.

Yes we can.

Thanks for the clarification.

Quoted 3 lines

Please wait a bit, I'm looking into the !CONFIG_PM_RUNTIME case. Mostly multi-platform, but it may affect legacy too, where there's interaction with the drivers/sh/pm_runtime.c hack.

I see, thanks for looking into that. No need for stop-and-go development rush though...

Cheers,

/ magnus

#19 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

In reply to Geert Uytterhoeven
Quoted 26 lines · Geert Uytterhoeven

Hi Magnus,

Quoted 8 lines · Magnus Damm
Quoted 8 lines
Quoted 1 line

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Quoted 7 lines
Quoted 4 lines

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Thanks for leading the way! From my side I don't have any special feeling, so why don't you and Simon decide?

OK.

Quoted 4 lines

If it was up to me only then I think option #1 that simply reduces the feature set seems rather natural, but perhaps it requires a lot of rework of the patches. If we go with option #1 we can always incrementally add D4 and A4MP later I assume.

Quoted 5 lines

Yes we can.

Please wait a bit, I'm looking into the !CONFIG_PM_RUNTIME case. Mostly multi-platform, but it may affect legacy too, where there's interaction with the drivers/sh/pm_runtime.c hack.

I also feel that option 1 is slightly nicer. But I'm happy to consider option 2 if its a better fit for you.

As always I'm happy to wait and will now do so.

#20 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Simon, Magnus,

Quoted 25 lines · Simon Horman
Quoted 24 lines · Geert Uytterhoeven
Quoted 8 lines · Magnus Damm
Quoted 8 lines
Quoted 1 line

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Quoted 7 lines
Quoted 4 lines

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Thanks for leading the way! From my side I don't have any special feeling, so why don't you and Simon decide?

OK.

Quoted 4 lines

If it was up to me only then I think option #1 that simply reduces the feature set seems rather natural, but perhaps it requires a lot of rework of the patches. If we go with option #1 we can always incrementally add D4 and A4MP later I assume.

Quoted 5 lines
Quoted 5 lines

Yes we can.

Please wait a bit, I'm looking into the !CONFIG_PM_RUNTIME case. Mostly multi-platform, but it may affect legacy too, where there's interaction with the drivers/sh/pm_runtime.c hack.

I posted a fix for the !CONFIG_PM_RUNTIME problem
"[PATCH] PM / clock_ops: Fix crash in clocks management code if !CONFIG_PM_RUNTIME"
(https://lkml.org/lkml/2014/10/1/506). I hope to see that fix in v3.18-rc1.

Quoted 4 lines

I also feel that option 1 is slightly nicer. But I'm happy to consider option 2 if its a better fit for you.

As always I'm happy to wait and will now do so.

The fix for the HDMI-related interrupt storm, also affecting A4MP, is in fbdev's next, so I expect to see that in v3.18-rc1.

For D4, a final solution seems to be a long-term plan...
I'll disable powering down the D4 power area, and will repost the patches after v3.18-rc1, when the other prerequisites are in.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

maintainer
#21 lore raw

v2 cover ARM: shmobile: r8a7740/armadillo800eva legacy PM domain support

Hi Geert,

Quoted 42 lines · Geert Uytterhoeven

Hi Simon, Magnus,

Quoted 25 lines · Simon Horman
Quoted 24 lines · Geert Uytterhoeven
Quoted 8 lines · Magnus Damm
Quoted 8 lines
Quoted 1 line

So what is a good way to proceed with this series I wonder?

There are two options:
1. Drop the two patches that introduce the D4 and A4MP PM domains,
2. Amend the patches that introduce the D4 and A4MP PM domains to comment out the entries in the table in pm-r8a7740.c, like was done in the first submission.

Quoted 7 lines
Quoted 4 lines

Both will cause conflicts when applying the successive patches.

If you don't care about bisectability, you can implement option 2 as a separate change, after applying the whole series.

Thanks for leading the way! From my side I don't have any special feeling, so why don't you and Simon decide?

OK.

Quoted 4 lines

If it was up to me only then I think option #1 that simply reduces the feature set seems rather natural, but perhaps it requires a lot of rework of the patches. If we go with option #1 we can always incrementally add D4 and A4MP later I assume.

Quoted 5 lines
Quoted 5 lines

Yes we can.

Please wait a bit, I'm looking into the !CONFIG_PM_RUNTIME case. Mostly multi-platform, but it may affect legacy too, where there's interaction with the drivers/sh/pm_runtime.c hack.

I posted a fix for the !CONFIG_PM_RUNTIME problem
"[PATCH] PM / clock_ops: Fix crash in clocks management code if !CONFIG_PM_RUNTIME"
(https://lkml.org/lkml/2014/10/1/506). I hope to see that fix in v3.18-rc1.

Quoted 3 lines

I also feel that option 1 is slightly nicer. But I'm happy to consider option 2 if its a better fit for you.

Quoted 8 lines
Quoted 1 line

As always I'm happy to wait and will now do so.

The fix for the HDMI-related interrupt storm, also affecting A4MP, is in fbdev's next, so I expect to see that in v3.18-rc1.

For D4, a final solution seems to be a long-term plan...
I'll disable powering down the D4 power area, and will repost the patches after v3.18-rc1, when the other prerequisites are in.

Sounds good. Thanks for your efforts!

Cheers,

/ magnus