Bug 9662

Summary:	boot hang unless button press - EC: GPE = 0x10, I/O: command/status = 0x66, data = 0x62 - Acer 4520
Product:	ACPI	Reporter:	Dan O'Connell (Belboz99)
Component:	Other	Assignee:	acpi_other
Status:	CLOSED PATCH_ALREADY_AVAILABLE
Severity:	normal	CC:	acpi-bugzilla
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.23.12	Subsystem:
Regression:	---	Bisected commit-id:

Description Dan O'Connell 2007-12-29 13:43:08 UTC

Most recent kernel where this bug did not occur: NA on this hardware
Distribution: LinuxMint 4.0
Hardware Environment: Acer 4520 (all stock)
Software Environment: Dual-Boot with Vista
Problem Description:  Kernel hangs on boot at:

ACPI: EC: GPE = 0x10, I/O: command/status = 0x66, data = 0x62

Steps to reproduce:

I custom compiled the kernel to solve another ACPI issue which I may file a bug report for later.   I will try with apci=off momentarily.

Oddly enough, when the power button is pressed, the kernel resumes booting and runs as normal! 


acpi=off results in a kernel panic telling me to run with noapic, running with noapic results in it telling me that I need to run with pnpbios=off.   

Finally, if I boot with "acpi=off noapic pnpbios=off" it hangs at loading my DVD-ROM, after about 5 minutes it will give an error dealing with a lack of a root file system, (no /dev/sda5).

This is the latest kernel, so this should be fixed, unless I did something stupid compiling it, then I should be fixed. :-P

Regardless, there have been similar errors on the two stock kernels that shipped with LinuxMint, the 2.6.22.14-generic fails to boot entirely (power button trick doesn't work) and 2.6.22.14-386 works with power button trick, albeit at a different point in the boot sequence.

Thanks,
Dan O.

Comment 1 Len Brown 2007-12-30 22:36:00 UTC

Has any version of Linux every booted properly w/o workarounds on this box?

What is the difference between 2.6.22.14-generic which "fails entirely"
(and how does it fail, does it hang in a different place)
vs. 2.6.22.14-386, which you report like 2.6.23.12,
hangs at the ACPI EC message and continues upon a power button press?

Can you try 2.6.24-rc?  There have been some EC issues fixed on Acers
in the latest release.

Comment 2 Dan O'Connell 2007-12-31 20:50:16 UTC

Hey Len,

2.6.22.14-generic is a LinuxMint stock kernel.   It would start booting, then fail to detect a /dev/root.   Apparently, it wouldn't even recognize the hard drive as a device, thus it failed to boot entirely.

I am compiling 2.6.24-rc right now.  So far everything has gone as normal.


Of note is that I attempted to install Fedora 8 and Fedora 8 Re-Spin (a 12-18-07 update to Fedora 8) and while both installed 100% normal, both failed to boot, both hanging on "/dev/root not found" or similar.


Mind you, I have Linux running on 3 other desktops here, one of which runs a web-server, ftp server, acts as a router and HTPC.  I've been using Linux for about 5 years now and this is the roughest Linux experience I've had in at least 3  years.

Thanks for the reply,

Dan

Comment 3 Dan O'Connell 2008-01-01 06:05:33 UTC

Success!!

I have successfully compiled the latest kernel from the Linus git tree:
git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git linux26

I used make oldconfig, holding down enter to accept all defaults.

I then made about 3 changes.

1.  Change the CPU architecture to AMD Athlon 64 / Hammer from Generic.
2.  Enable SATA (Prod) support as "built-in" rather than as a module.
3.  Enabled ext3 journaling fs support.   Believe it or not, this was NOT enabled before, and yes, my root fs is ext3.   I also enabled it as built-in.

I also enabled a few SATA drivers as built-in where-as before they were listed as module.

I then ran:
make -j2
make install
reboot

And what do you know?  It runs absolutely fine!


Unfortunately, it still hangs on resume from Suspend mode, but I can live with that as long as Hibernate works OK.  (Didn't before).



Just for the record, the current kernel is now:  2.6.24-rc6 


The kernel boots as normal, save for a few PCI warnings.


I just wonder whether it was the newer kernel or the changes I made that did the trick.

Thanks Len, that was a great idea to try the rc kernel. ;)

Dan

Comment 4 Len Brown 2008-01-01 10:16:48 UTC

Thanks for testing, Dan.
As the upstream 2.6.24-rc6 kernel works, we're done here.

Re: distros not working on this box.  I'm not familiar
with the (non-Intel:-) chipset, it might be that upstream
has a driver update for it the distros don't have yet.
I expect 2.6.24 will ship shortly, and you'll be able to
get it included with test releases of FC9 etc.