Re: S3 test tool (was : Re: Bizarre oops after suspend to RAM (was: Re: Resume from Suspend to RAM))

["Jason Dagit" <dagit-LP0vGzdgvNwj5TC/SZClsA@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 1628 bytes --]

Christoph Tavan <chefturner-S0/GAf8tV78@public.gmane.org> writes:

> On Friday 10 June 2005 20:49, Frank Siegert wrote:
>> 2. I tried to add the patch manually (and can't see where I missed
>> something -- the "extern void machine_real_restart(...)" line exists), but
>> kernel compilation then fails.
>>
>> |   CC      init/version.o
>> |   LD      init/built-in.o
>> |   LD      .tmp_vmlinux1
>> | ld: BFD 2.15 assertion fail ../../bfd/linker.c:619
>> |
>> | drivers/built-in.o(.text+0x203ef): In function `acpi_enter_sleep_state':
>> | : undefined reference to `machine_real_restart'
>> |
>> | make[1]: *** [.tmp_vmlinux1] Fehler 1
>> | make[1]: Verlasse Verzeichnis »/usr/src/linux-2.6.12-rc6«
>> | make: *** [stamp-build] Fehler 2
>
> I get exactly the same error. I also would like to test that patch, but I 
> cannot compile the sources with that patch applied.

What is your version of gcc?  I just got around to trying this patch
out today using 2.6.12-rc6 and everything worked great.  In fact, the
patch even does the suspend/resume correctly.  Now, it doesn't
actually go to sleep so it's not useful, but it was just for
testing/debugging right?

I'm using gcc 3.3.5 (Debian 1:3.3.5-13).

>
>> Would you mind posting the patch as attachment and the exact
>> command I have to issue in my kernel directory?
>
> That would be great!!!

the command I used from linux-2.6.12-rc6/ was:
patch drivers/acpi/hardware/hwsleep.c patch-s3.diff

that assumes patch-s3.diff is in the linux-2.6.12-rc6 directory.  I'm
also attaching the patch file again as you requested.

HTH,
Jason


[-- Attachment #2: patch-s3.diff --]
[-- Type: text/plain, Size: 1226 bytes --]

--- a/drivers/acpi/hardware/hwsleep.c	2005-06-07 13:45:04.088273424 +0800
+++ b/drivers/acpi/hardware/hwsleep.c	2005-06-07 13:49:31.858566152 +0800
@@ -242,6 +242,19 @@ acpi_enter_sleep_state_prep (
  *              THIS FUNCTION MUST BE CALLED WITH INTERRUPTS DISABLED
  *
  ******************************************************************************/
+#define S3_DEBUG
+#ifdef S3_DEBUG
+#include <asm/io.h>
+extern void machine_real_restart(unsigned char *code, int length);
+static unsigned char jump_to_pm [] =
+{
+	0xea,
+	0x00,
+	0x00,
+	0x00,
+	0x00		/*    ljmp  $0x0000,$0x0000  */
+};
+#endif
 
 acpi_status asmlinkage
 acpi_enter_sleep_state (
@@ -315,6 +328,14 @@ acpi_enter_sleep_state (
 	PM1Acontrol |= (acpi_gbl_sleep_type_a << sleep_type_reg_info->bit_position);
 	PM1Bcontrol |= (acpi_gbl_sleep_type_b << sleep_type_reg_info->bit_position);
 
+#ifdef S3_DEBUG
+	if (sleep_state == ACPI_STATE_S3) {
+		*((short *)&jump_to_pm[3]) =
+			(short)(virt_to_phys((void *)acpi_wakeup_address)) >> 4;
+		/* Directly jump to acpi_wakeup_address */
+		machine_real_restart(jump_to_pm, sizeof(jump_to_pm));
+	}
+#endif
 	/*
 	 * We split the writes of SLP_TYP and SLP_EN to workaround
 	 * poorly implemented hardware.