Skip to contentSkip to conversation

ARM: aspeed: romulus: Add reset tolerance for power GPIOs

PATCH Applied Lei YU 2 participants 4 messages Oct 30, 2017 – Oct 31, 2017

Subsystems (3)
lore
Share & tools
lore

Lei YU

author
#0 lore raw

PATCH 1/1 ARM: aspeed: romulus: Add reset tolerance for power GPIOs

This is a similar change as Witherspoon, that enables the host to stay up across BMC reboots.

---
 arch/arm/mach-aspeed/aspeed.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
arch/arm/mach-aspeed/aspeed.c +10 -0
index c84c871..5965bbf 100644
@@ -196,7 +196,17 @@ static void __init do_witherspoon_setup(void)
 
 static void __init do_romulus_setup(void)
 {
+	unsigned long reg;
+
 	do_common_setup();
+
+	/* Reset tolerance for BMC_POWER_UP (GPIOD1) */
+	reg = readl(AST_IO(AST_BASE_GPIO | 0x01c));
+	writel(reg | BIT(25), AST_IO(AST_BASE_GPIO | 0x01c));
+
+	/* Reset tolerance for SOFTWARE_PGOOD (GPIOR1) and SEQ_CONT (GPIOS7) */
+	reg = readl(AST_IO(AST_BASE_GPIO | 0x12c));
+	writel(reg | BIT(9) | BIT(23), AST_IO(AST_BASE_GPIO | 0x12c));
 }
 
 static void __init do_lanyang_setup(void)
Signature · 1 line
1.9.1
#1 lore raw
Quoted 19 lines · Lei YU

This is a similar change as Witherspoon, that enables the host to stay up across BMC reboots.

---
 arch/arm/mach-aspeed/aspeed.c | 10 ++++++++++

 1 file changed, 10 insertions(+)

arch/arm/mach-aspeed/aspeed.c +0 -0
index c84c871..5965bbf 100644
@@ -196,7 +196,17 @@ static void __init do_witherspoon_setup(void)


 static void __init do_romulus_setup(void)
 {

arch/arm/mach-aspeed/aspeed.c +2 -0
+	unsigned long reg;
+

  do_common_setup();

arch/arm/mach-aspeed/aspeed.c static void __init do_witherspoon_setup(void) · line 207

Quoted 6 lines
arch/arm/mach-aspeed/aspeed.c +6 -0
+
+	/* Reset tolerance for BMC_POWER_UP (GPIOD1) */
+	reg = readl(AST_IO(AST_BASE_GPIO | 0x01c));
+	writel(reg | BIT(25), AST_IO(AST_BASE_GPIO | 0x01c));
+
+	/* Reset tolerance for SOFTWARE_PGOOD (GPIOR1) and SEQ_CONT (GPIOS7) */

What is SEQ_CONT? I'd guess at "power sequence continue". GPIOS7 doesn't have an associated net name in the schematic I have for Romulus, though it's pulled up to 3.3V. Maybe my schematic is out of date?

Cheers,

Andrew

author
#2 lore raw
Quoted 15 lines · Andrew Jeffery
Quoted 14 lines · Lei YU

This is a similar change as Witherspoon, that enables the host to stay up across BMC reboots.

---
 arch/arm/mach-aspeed/aspeed.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
arch/arm/mach-aspeed/aspeed.c +0 -0
index c84c871..5965bbf 100644
@@ -196,7 +196,17 @@ static void __init do_witherspoon_setup(void)
 
Quoted 15 lines

arch/arm/mach-aspeed/aspeed.c static void __init do_witherspoon_setup(void) · line 207

Quoted 11 lines
arch/arm/mach-aspeed/aspeed.c +8 -0
 static void __init do_romulus_setup(void)
 {
+     unsigned long reg;
+
      do_common_setup();
+
+     /* Reset tolerance for BMC_POWER_UP (GPIOD1) */
+     reg = readl(AST_IO(AST_BASE_GPIO | 0x01c));
+     writel(reg | BIT(25), AST_IO(AST_BASE_GPIO | 0x01c));
+
+     /* Reset tolerance for SOFTWARE_PGOOD (GPIOR1) and SEQ_CONT (GPIOS7) */

What is SEQ_CONT? I'd guess at "power sequence continue". GPIOS7 doesn't have an associated net name in the schematic I have for Romulus, though it's pulled up to 3.3V. Maybe my schematic is out of date?

Exactly, it is "power sequence continue" that Romulus CPLD handles this:
* When it's high, CPLD pauses power sequence and when it's low it continue the sequence.
This behavior is used for vcs workaround for DD1.x
* When system is powered on it shall be low, and if it becomes high, CPLD results in PGOOD fault.

Pass1 schematic does not have this and it's done by Nick's EC; Pass2 schematic has this GPIO named SEQ_CONT.

Thanks.

Quoted 4 lines

Cheers,

Andrew