Skip to contentSkip to conversation

OLPC: Add XO-1 poweroff support

PATCH Superseded Daniel Drake 2 participants 4 messages Oct 7, 2010 x86

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 OLPC: Add XO-1 poweroff support

Add a pm_power_off handler for the OLPC XO-1 laptop.

---
 arch/x86/Kconfig           |    6 ++
 arch/x86/kernel/Makefile   |    1 +
 arch/x86/kernel/olpc-xo1.c |  112 ++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 119 insertions(+), 0 deletions(-)
 create mode 100644 arch/x86/kernel/olpc-xo1.c

The new olpc-xo1.c file will also be used for further functionality (suspend/resume, lid switch device, etc); patches to be submitted shortly after the review/merge of this one.

arch/x86/Kconfig +6 -0
index cea0cd9..19e6439 100644
@@ -2065,6 +2065,12 @@ config OLPC
 	  Add support for detecting the unique features of the OLPC
 	  XO hardware.
 
+config OLPC_XO1
+	bool "OLPC XO-1 support"
+	depends on OLPC
+	---help---
+	  Add support for non-essential features of the OLPC XO-1 laptop.
+
 config OLPC_OPENFIRMWARE
 	bool "Support for OLPC's Open Firmware"
 	depends on !X86_64 && !X86_PAE
arch/x86/kernel/Makefile +1 -0
index fedf32a..247588f 100644
@@ -106,6 +106,7 @@ obj-$(CONFIG_SCx200)		+= scx200.o
 scx200-y			+= scx200_32.o
 
 obj-$(CONFIG_OLPC)		+= olpc.o
+obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o
 obj-$(CONFIG_OLPC_OPENFIRMWARE)	+= olpc_ofw.o
 obj-$(CONFIG_X86_MRST)		+= mrst.o
 
arch/x86/kernel/olpc-xo1.c new file +112 -0

112 lines not on this page · raw message

Signature · 1 line
1.7.2.3
#1 lore raw
Quoted 20 lines · On Thu, 7 Oct 2010 19:59:52 +0100 (BST) Daniel Drake <[email protected]> wrote:

Add a pm_power_off handler for the OLPC XO-1 laptop.

---
 arch/x86/Kconfig           |    6 ++
 arch/x86/kernel/Makefile   |    1 +
 arch/x86/kernel/olpc-xo1.c |  112

++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 119 insertions(+), 0 deletions(-) create mode 100644 arch/x86/kernel/olpc-xo1.c

The new olpc-xo1.c file will also be used for further functionality (suspend/resume, lid switch device, etc); patches to be submitted shortly after the review/merge of this one.

arch/x86/Kconfig +0 -0
index cea0cd9..19e6439 100644

arch/x86/Kconfig config OLPC · line 2069

Quoted 5 lines
arch/x86/Kconfig +2 -0
@@ -2065,6 +2065,12 @@ config OLPC
 	  Add support for detecting the unique features of the OLPC
 	  XO hardware.
 
+config OLPC_XO1
+	bool "OLPC XO-1 support"

Any particular reason why this can't be modular?

[...]

arch/x86/Kconfig config OLPC · line 2072

Quoted 3 lines
arch/x86/Kconfig +3 -0
+
+	pm_power_off = xo1_power_off;
+

If this were to be modular, I'm not sure what the best option is for pm_power_off. If the old value was saved upon module load, and restored upon module unload, I could imagine races like the following:

module A load:
saved_ppo = pm_power_off; /* NULL */
pm_power_off = frob;
module B load:
old_ppo = pm_power_off; /* frob */
pm_power_off = foo;
module A unload:
pm_power_off = saved_ppo; /* NULL */

So I guess just clobbering whatever's in pm_power_off and setting back to NULL upon unload, with the assumption that driver's only really going to clobber it in the case of actual OLPC hardware, and the callbacks being clobbered wouldn't really have done the correct thing anyways?

Quoted 5 lines
arch/x86/Kconfig +5 -0
+	printk(KERN_INFO "OLPC XO-1 support registered\n");
+	return 0;
+}
+device_initcall(olpc_xo1_init);
+
#2 lore raw
Quoted 1 line · Andres Salomon

Any particular reason why this can't be modular?

Because of the pm_power_off thing.
Also, a similar situation would be encountered with set_suspend_ops().

I don't really understand your suggestion. If we set it to NULL on module unload then the power off would crash on shutdown, right?

Daniel

#3 lore raw
Quoted 10 lines · On Thu, 7 Oct 2010 21:42:59 +0100 Daniel Drake <[email protected]> wrote:
Quoted 1 line · Andres Salomon

Any particular reason why this can't be modular?

Because of the pm_power_off thing.
Also, a similar situation would be encountered with set_suspend_ops().

I don't really understand your suggestion. If we set it to NULL on module unload then the power off would crash on shutdown, right?

Daniel

See arch/x86/kernel/apm_32.c for an example of modular pm_power_off handling.

Power off shouldn't crash; at least, native_machine_power_off first check if pm_power_off is !NULL before calling it.