APM suspend/resume ceased to work with 2.4.28

APM suspend/resume ceased to work with 2.4.28

[Philippe Troin]

Seen on a Dell Inspiron 3800 with BIOS revision A17.

APM suspend/resume works perfectly with 2.4.27 (or at least, as well
as APM can).

Since I did not see any differences in arch/i386/kernel/apm.c between
.27 and .28, I'm at loss to explain the problem.

Phil.

Re: APM suspend/resume ceased to work with 2.4.28

[Marcelo Tosatti]

On Sun, Nov 21, 2004 at 07:25:20PM -0800, Philippe Troin wrote:
> Seen on a Dell Inspiron 3800 with BIOS revision A17.
> 
> APM suspend/resume works perfectly with 2.4.27 (or at least, as well
> as APM can).
> 
> Since I did not see any differences in arch/i386/kernel/apm.c between
> .27 and .28, I'm at loss to explain the problem.

Guess: Are you using ACPI ? 

:(

Re: APM suspend/resume ceased to work with 2.4.28

[Philippe Troin]

Marcelo Tosatti <marcelo.tosatti@cyclades.com> writes:

> On Sun, Nov 21, 2004 at 07:25:20PM -0800, Philippe Troin wrote:
> > Seen on a Dell Inspiron 3800 with BIOS revision A17.
> > 
> > APM suspend/resume works perfectly with 2.4.27 (or at least, as well
> > as APM can).
> > 
> > Since I did not see any differences in arch/i386/kernel/apm.c between
> > .27 and .28, I'm at loss to explain the problem.
> 
> Guess: Are you using ACPI ? 

ACPI is compiled in, but the kernel is booted with noacpi...

Phil.

Re: APM suspend/resume ceased to work with 2.4.28

[Marcelo Tosatti]

On Mon, Nov 22, 2004 at 04:03:05PM -0800, Philippe Troin wrote:
> Marcelo Tosatti <marcelo.tosatti@cyclades.com> writes:
> 
> > On Sun, Nov 21, 2004 at 07:25:20PM -0800, Philippe Troin wrote:
> > > Seen on a Dell Inspiron 3800 with BIOS revision A17.
> > > 
> > > APM suspend/resume works perfectly with 2.4.27 (or at least, as well
> > > as APM can).
> > > 
> > > Since I did not see any differences in arch/i386/kernel/apm.c between
> > > .27 and .28, I'm at loss to explain the problem.
> > 
> > Guess: Are you using ACPI ? 
> 
> ACPI is compiled in, but the kernel is booted with noacpi...
> 
> Phil.

Phil, 

Can you please try 2.4.28-pre3 ? 

Lets try to find out where it started to happen. I've got no clue, 
there are no indeed no APM related changes in 2.4.28.

Re: APM suspend/resume ceased to work with 2.4.28

[Philippe Troin]

Marcelo Tosatti <marcelo.tosatti@cyclades.com> writes:

> On Mon, Nov 22, 2004 at 04:03:05PM -0800, Philippe Troin wrote:
> > Marcelo Tosatti <marcelo.tosatti@cyclades.com> writes:
> > 
> > > On Sun, Nov 21, 2004 at 07:25:20PM -0800, Philippe Troin wrote:
> > > > Seen on a Dell Inspiron 3800 with BIOS revision A17.
> > > > 
> > > > APM suspend/resume works perfectly with 2.4.27 (or at least, as well
> > > > as APM can).
> > > > 
> > > > Since I did not see any differences in arch/i386/kernel/apm.c between
> > > > .27 and .28, I'm at loss to explain the problem.
> > > 
> > > Guess: Are you using ACPI ? 
> > 
> > ACPI is compiled in, but the kernel is booted with noacpi...
> > 
> > Phil.
> 
> Phil, 
> 
> Can you please try 2.4.28-pre3 ? 
> 
> Lets try to find out where it started to happen. I've got no clue, 
> there are no indeed no APM related changes in 2.4.28.

The bug was introduced between rc1 and rc2.

I'm still working on it to narrow it down to which delta introduced
the bug.

I suspect either the ACPI or the APIC/DMI changes.

Phil.

Re: APM suspend/resume ceased to work with 2.4.28

[Philippe Troin]

[-- Attachment #1: Type: text/plain, Size: 617 bytes --]

Philippe Troin <phil@fifi.org> writes:

> Where do we go from here?

I just diffed out the boot log between 2.4.27 and 2.4.28.  And I found
this new message appearing in 2.4.28:

        ACPI: IRQ9 SCI: Edge set to Level Trigger.

Weird since IRQ 9 is not in use in this system.  IRQ9 is typically
used by ACPI.

After some digging in the sources, I found that the newly introduced
(2.4.28-rc2) function acpi_early_init() runs wether or not acpi=off
was specified on the command line.

This trivial patch makes my laptop suspend-happy.  Len, please confirm
that this is ok.  Marcelo, please include in 29-pre.

Phil.

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: linux-2.4.28-8-acpi.patch --]
[-- Type: text/x-patch, Size: 440 bytes --]

diff -ruN linux-2.4.28.orig/drivers/acpi/bus.c linux-2.4.28/drivers/acpi/bus.c
--- linux-2.4.28.orig/drivers/acpi/bus.c	Wed Nov 17 03:54:21 2004
+++ linux-2.4.28/drivers/acpi/bus.c	Thu Dec  2 16:49:54 2004
@@ -1850,6 +1850,9 @@
 	acpi_status		status = AE_OK;
 	struct acpi_buffer	buffer = {sizeof(acpi_fadt), &acpi_fadt};
 
+	if (acpi_disabled)
+		return;
+
 	ACPI_FUNCTION_TRACE("acpi_bus_init");
 
 	status = acpi_initialize_subsystem();

Re: APM suspend/resume ceased to work with 2.4.28

[Len Brown]

On Thu, 2004-12-02 at 20:50, Philippe Troin wrote:
> Philippe Troin <phil@fifi.org> writes:

> This trivial patch makes my laptop suspend-happy.

Good catch Phil!

My fault -- we fixed this in 2.6 and I failed to backport it to 2.4:-(

I've included your patch w/ minor syntax change to the ACPI patch.

Marcelo, please do a 

	bk pull bk://linux-acpi.bkbits.net/24-latest-release

thanks,
-Len

ps. a plain patch is also available here:
ftp://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/24-latest-release/acpi-20040326-24-latest-release.diff.gz

This will update the following files:

 drivers/acpi/bus.c |    9 ++++++---
 1 files changed, 6 insertions(+), 3 deletions(-)

through these ChangeSets:

<len.brown@intel.com> (04/12/02 1.1458.1.9)
   [ACPI] acpi=off must disable acpi_early_init()
   
   Signed-off-by: Philippe Troin <phil@fifi.org>
   Signed-off-by: Len Brown <len.brown@intel.com>