clk_must_disable() implementation on AT91

PATCH In review David Brownell 2 participants 2 messages Mar 22, 2007 – Mar 23, 2007

Subsystems (5)

Arm (32-bit) USB Arm sub-architectures Arm/Microchip (AT91) SoC support +1 more subsystems

lore

Share & tools

lore

David Brownell

author

2007-03-22 18:13 UTC #0 lore raw

PATCH 2/2 clk_must_disable() implementation on AT91

This supports the new clk_must_disable() interface for AT91 systems:

- Implement the call, replacing at91_suspend_entering_slow_clock() and eliminating various "that's not exported" build warnings;

- Use it in three drivers: USB Host, USB Peripheral, and RS232 serial.

Briefly, those are three of the drivers that need to act differently when going into deeper sleep states (suspend-to-RAM, vs standby), since among other things they can't act as wakeup event sources without using the clocks which are disabled in those deeper sleep states.

(The at91_ethernet driver would be another such driver, but it doesn't currently implement wake-on-LAN even in the "standby" mode.)

Signed-off-by David Brownell <[email protected]>

---
 arch/arm/mach-at91/clock.c    |   18 ++++++++++++++++++
 arch/arm/mach-at91/generic.h  |    1 +
 arch/arm/mach-at91/pm.c       |    7 +------
 drivers/serial/atmel_serial.c |    3 ++-
 drivers/usb/gadget/at91_udc.c |    2 +-
 drivers/usb/host/ohci-at91.c  |    2 +-
 6 files changed, 24 insertions(+), 9 deletions(-)

arch/arm/mach-at91/clock.c +18 -0

Index: at91/arch/arm/mach-at91/clock.c
===================================================================
@@ -33,6 +33,7 @@
 #include <asm/arch/cpu.h>
 
 #include "clock.h"
+#include "generic.h"
 
 
 /*

@@ -255,6 +256,23 @@ EXPORT_SYMBOL(clk_get_rate);
 
 /*------------------------------------------------------------------------*/
 
+#ifdef CONFIG_PM
+
+int clk_must_disable(struct clk *clk)
+{
+	if (!at91_suspend_entering_slow_clock())
+		return 0;
+
+	while (clk->parent)
+		clk = clk->parent;
+	return clk != &clk32k;
+}
+EXPORT_SYMBOL(clk_must_disable);
+
+#endif
+
+/*------------------------------------------------------------------------*/
+
 #ifdef CONFIG_AT91_PROGRAMMABLE_CLOCKS
 
 /*

arch/arm/mach-at91/generic.h +1 -0

Index: at91/arch/arm/mach-at91/generic.h
===================================================================
@@ -38,6 +38,7 @@ extern void __init at91_clock_associate(
  /* Power Management */
 extern void at91_irq_suspend(void);
 extern void at91_irq_resume(void);
+extern int at91_suspend_entering_slow_clock(void);
 
  /* GPIO */
 #define AT91RM9200_PQFP		3	/* AT91RM9200 PQFP package has 3 banks */

arch/arm/mach-at91/pm.c +1 -6

Index: at91/arch/arm/mach-at91/pm.c
===================================================================
@@ -104,20 +104,15 @@ static int at91_pm_verify_clocks(void)
 }
 
 /*
- * Call this from platform driver suspend() to see how deeply to suspend.
+ * This is called from clk_must_disable(), to see how deeply to suspend.
  * For example, some controllers (like OHCI) need one of the PLL clocks
  * in order to act as a wakeup source, and those are not available when
  * going into slow clock mode.
- *
- * REVISIT: generalize as clk_will_be_available(clk)?  Other platforms have
- * the very same problem (but not using at91 main_clk), and it'd be better
- * to add one generic API rather than lots of platform-specific ones.
  */
 int at91_suspend_entering_slow_clock(void)
 {
 	return (target_state == PM_SUSPEND_MEM);
 }
-EXPORT_SYMBOL(at91_suspend_entering_slow_clock);
 
 
 static void (*slow_clock)(void);

drivers/serial/atmel_serial.c +2 -1

Index: at91/drivers/serial/atmel_serial.c
===================================================================
@@ -888,7 +888,8 @@ static int atmel_serial_suspend(struct p
 	struct uart_port *port = platform_get_drvdata(pdev);
 	struct atmel_uart_port *atmel_port = (struct atmel_uart_port *) port;
 
-	if (device_may_wakeup(&pdev->dev) && !at91_suspend_entering_slow_clock())
+	if (device_may_wakeup(&pdev->dev)
+			&& !clk_must_disable(atmel_port->clk))
 		enable_irq_wake(port->irq);
 	else {
 		uart_suspend_port(&atmel_uart, port);

drivers/usb/gadget/at91_udc.c +1 -1

Index: at91/drivers/usb/gadget/at91_udc.c
===================================================================
@@ -1804,7 +1804,7 @@ static int at91udc_suspend(struct platfo
 	 */
 	if ((!udc->suspended && udc->addr)
 			|| !wake
-			|| at91_suspend_entering_slow_clock()) {
+			|| clk_must_disable(udc->fclk)) {
 		pullup(udc, 0);
 		wake = 0;
 	} else

drivers/usb/host/ohci-at91.c +1 -1

Index: at91/drivers/usb/host/ohci-at91.c
===================================================================
@@ -299,7 +299,7 @@ ohci_hcd_at91_drv_suspend(struct platfor
 	 *
 	 * REVISIT: some boards will be able to turn VBUS off...
 	 */
-	if (at91_suspend_entering_slow_clock()) {
+	if (clk_must_disable(fclk)) {
 		ohci_usb_reset (ohci);
 		at91_stop_clock();
 	}

Pavel Machek

2007-03-23 12:21 UTC #1 lore raw

Hi!

Quoted 12 lines

This supports the new clk_must_disable() interface for AT91 systems:
- Implement the call, replacing at91_suspend_entering_slow_clock() and eliminating various "that's not exported" build warnings;
- Use it in three drivers: USB Host, USB Peripheral, and RS232 serial.
Briefly, those are three of the drivers that need to act differently when going into deeper sleep states (suspend-to-RAM, vs standby), since among other things they can't act as wakeup event sources without using the clocks which are disabled in those deeper sleep states.

Quoted 4 lines

(The at91_ethernet driver would be another such driver, but it doesn't currently implement wake-on-LAN even in the "standby" mode.)
Signed-off-by David Brownell <[email protected]>

FWIW, it looks okay to me. But you probably need to cc akpm if you want it merged...

Signature · 2 lines

(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html