Bug 4975

Summary:	Acer Aspire 1691WCLi hangs on boot after "booting the kernel"
Product:	Platform Specific/Hardware	Reporter:	Alexey Dobriyan (adobriyan)
Component:	Other	Assignee:	Brad Barnett (lists)
Status:	RESOLVED CODE_FIX
Severity:	normal
Priority:	P2
Hardware:	i386
OS:	Linux
Kernel Version:	2.6.13-rc4, 2.6.12.3, 2.6.12, 2.6.11, 2.4.27	Subsystem:
Regression:	---	Bisected commit-id:
Attachments:	lspci output

Description Alexey Dobriyan 2005-07-31 14:18:56 UTC

From: Brad Barnett <bahb@L8R.net>
http://marc.theaimsgroup.com/?t=112265436500006&r=1&w=2

Pentium M 725, 512M ram.

This laptop will simply not boot with any Debian precompiled kernel, with the
exception of Debian's 2.4.27-2 initrd kernel.  I have compiled my own kernels,
using a vast array of options, 2.6.11, 2.6.12, 2.6.12.3, 2.6.13-rc4 and 2.4.27,
they also all fail in exactly the same way.  I have tried with and without
initrd, acpi, 386 or other processor options, as well as very lean, stripped
down kernels.  I have tried with both lilo and grub, but both result in the
same hang.

Lilo or grub boots the kernel, and I see the classic:

boot: vmlinuz
Loading vmlinuz.................................................
BIOS data check successful
Uncompressing Linux... Ok. booting the kernel.

[photo with this text is at http://be.back.l8r.net:8000/no_boot.jpg
                                                        --adobriyan]

Debian's 2.4.27-2 boots fine, and this is what really annoys me.  I took
Debian's 2.4.27-2 initrd config from /boot, ran make oldconfig on a fresh
2.4.27 tree (some minor options were different due to Debian's
backpatching).  This image _still_ would not boot.

If needed, I could compile a few more kernels, change GCC versions (if anyone
thinks that would help) and so on.  However, my goal here is to get 2.6 working
in order to support various bits of hardware on this laptop.

There are very few bios options to change. :/

[Jesus Delgado <jdelgado@gmail.com> has the same problems with eMachines M6807
                                                                   --adobriyan]

Comment 1 Alexey Dobriyan 2005-07-31 14:21:32 UTC

Created attachment 5437 [details]
lspci output

Comment 2 Pedro Ramalhais 2005-08-23 10:05:40 UTC

AFAICS the problem is when the kernel is enabling the IOAPIC it just hangs. The
work around is to boot with the "noapic" kernel boot parameter. I also have some
problems with IRQs and with the ACPI implementation. There's a fixed ACPI DSDT
you can download at acpi.sf.net. My laptop is a 1691WLMi but it's probably very
similar, download the original DSDT, and check against your laptop's DSDT.
I would gladly help people fix these problems, just say what you need to help
fixing it.

Comment 3 Pedro Ramalhais 2005-08-23 14:22:01 UTC

I found a patch that makes the kernel boot without need for "noapic" and
"pci=routeirq". Looks like this bug is a duplicate of bug 4700.
http://bugzilla.kernel.org/show_bug.cgi?id=4700

Comment 4 Brad Barnett 2005-09-01 16:24:00 UTC

issue resolved.. it was UP_IOAPIC, not entirely sure how I missed it...

Comment 5 Brad Barnett 2005-09-01 16:26:31 UTC

issue resolved.. 


Dropped from GCC 3.3 to GCC 2.95, got rid of UP_IOAPIC (not entirely sure how I
missed UP_IOAPIC during compiles), and the issue has vanished...

I anyone wants, I can compile various kernels against various gcc versions, if a
better picture is needed/desired.