Bug 5056

Summary: IBM ACPI doesn't seem to work with ACPI patches 2005-07-28 for kernel 2.6.12.4
Product: Drivers Reporter: Martin Steigerwald (Martin)
Component: PlatformAssignee: Borislav Deianov (borislav)
Status: CLOSED CODE_FIX    
Severity: normal CC: acpi-bugzilla
Priority: P2    
Hardware: i386   
OS: Linux   
Kernel Version: 2.6.12.4 Subsystem:
Regression: --- Bisected commit-id:
Attachments: config file I used for the kernel
ibm-acpi-0.12 (replaces drivers/acpi/ibm_acpi.c)

Description Martin Steigerwald 2005-08-13 06:28:36 UTC
Distribution: Debian Etch with some parts of unstable 
Hardware Environment: IBM ThinkPad T23, P3 1.13 GHz, 384 MB RAM 
Software Environment:  
2.6.12.4 + ACPI patches 2005-07-28 + some hibernate scripts and ACPI events 
 
Problem Description: 
 
Kernel module "ibm-acpi" doesn't seem to work with ACPI patches 2005-07-28. 
 
root@deepdance: -> modprobe ibm-acpi 
FATAL: Error inserting ibm_acpi  
(/lib/modules/2.6.12.4-tp23-acpi-2005-07-29/kernel/drivers/acpi/ibm_acpi.ko):  
No such device 
 
This is with 2.6.12.4 compiled with GCC 4.0.1: 
 
root@deepdance: -> cat /proc/version 
Linux version 2.6.12.4-tp23-acpi-2005-07-29 (root@deepdance) (gcc version  
4.0.1 (Debian 4.0.1-2)) #1 Sat Aug 13 13:18:05 CEST 2005 
 
It used to work before on 2.6.12.2 without ACPI patches.  
 
It also works okay with 2.6.12.4 without ACPI patches (which I am now using) 
 
Steps to reproduce: 
 
1) Get plain kernel 2.6.12 sources 
2) apply 2.6.12.4 patch 
3) apply ACPI patches 2005-07-28 for 2.6.12.4 to it 
4) I had a T23 
5) I used a hand crafted kernel config file (I will attach this) 
6) I used make-kpkg 
7) reboot 
8) do "modprobe ibm-acpi" 
 
Maybe step 4 to 6 are not required to reproduce the problem. 
 
I will be keeping the Debian kernel packages I produced for a while.  
 
If you like me to try something out, want those kernel packages or need further 
information drop me a note. 
 
Regards,  
Martin Steigerwald
Comment 1 Martin Steigerwald 2005-08-13 06:30:07 UTC
I tried with IBM ACPI which is contained in kernel source 2.6.12, I did not use 
extra sources. 
Comment 2 Martin Steigerwald 2005-08-13 06:31:06 UTC
Created attachment 5624 [details]
config file I used for the kernel
Comment 3 Sanjoy Mahajan 2005-08-13 16:54:51 UTC
See the patch in Bug #5035, additional comment #2.  It improved matters for me,
though didn't solve them.  I think the rest of the solution for me involves
getting 0.11 of the ibm-acpi drive to compile, since 0.8, which comes with the
kernels, doesn't work on the TP 600X.
Comment 4 Martin Steigerwald 2005-08-14 01:49:18 UTC
Hello Sanjoy, thanks for the message. Strange I didn't find your bug. Maybe I 
forgot to search for resolved ones. Bug #5035 is marked resolved although from 
the last comment I believe it isn't yet (probably you should reopen it?). I am 
not in a need for a quick fix as without latest ACPI patches at least those 
features I use with IBM ACPI work okay here on 2.6.12.4, but it would be nice 
when that is fixed in the 2.6.13 release kernel then! ;-). IMHO preferably by 
integrating latest stable IBM ACPI into it. Regards, Martin 
Comment 5 Len Brown 2005-08-17 09:13:21 UTC
Note that RESOLVED means that there is a patch proposed for
testing or integration, not that the fix has shipped in
Linus' tree, the CLOSED state is reserved for that.

Martin, if you have the time it would be very useful to know
if the following kernels work or fail for you:
1. 2.6.12 + latest ACPI patch (ACPI patch has changed in this area
         since you last tested it)
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/2.6.12/
2. 2.6.13 (vanilla kernel.org)
3. the latest ibm_acpi driver on top of #1 or #2.
http://bugzilla.kernel.org/show_bug.cgi?id=4947#c5

Note that I'll soon be updating the ACPI patch to include
the latest ibm_acpi driver, so depending on when you pull
the acpi patch you might do #1 and #3 at the same time:-)

thanks,
-Len
Comment 6 Borislav Deianov 2005-08-17 10:00:25 UTC
Created attachment 5659 [details]
ibm-acpi-0.12 (replaces drivers/acpi/ibm_acpi.c)

Martin, if you choose to try the latest vanilla 2.6.13 (you'll need
2.6.13-rc6 or later) you can just replace ibm_acpi.c with the attached
version. Same code as included in
http://bugzilla.kernel.org/show_bug.cgi?id=4947#c5 but that patch
won't directly apply as it is generated on top of the ACPI patch.
Comment 7 Len Brown 2005-09-02 22:32:13 UTC
so does ibm_acpi 0.12 resolve this issue?
Comment 8 Martin Steigerwald 2005-09-03 08:39:40 UTC
Hello - Hmmm, today I compiled a Linux 2.6.13 vanilla kernel without additional 
ACPI patches as the ones from acpi.sf.net are against 2.6.13-rc7.   
   
IBM ACPI works, as well as suspend to RAM and suspend to disk (which displays   
some new messages that seem to indicate some improvements;-).   
   
Linux 2.6.13 still seems to have IBM ACPI 0.8 tough. I thought your updated  
ACPI patches would contain 0.12 already.   
  
martin@deepdance:/proc/acpi/ibm -> cat driver  
driver:         IBM ThinkPad ACPI Extras  
version:        0.8  
  
I will do another compile with IBM ACPI 0.12 patch added in the next days.  
 
I am already pleased as everything works and it seems 2.6.13 has improved on 
latency issues quite a bit again ;-) (Preemptible kernel with 1000Hz timer 
frequency. Gets closer to my old Amiga again with but with more feature-rich 
software.) - Regards, Martin 
Comment 9 Martin Steigerwald 2005-09-06 13:47:33 UTC
Hello again,  
 
I now compiled vanilla 2.6.13 with IBM ACPI 0.12a from  
 
http://bugzilla.kernel.org/show_bug.cgi?id=4947#c5  
 
as there is no 0.12 on ibmacpi.sourceforge.net.  
 
So I have now the following: 
 
martin@deepdance:~ -> cat /proc/version 
Linux version 2.6.13-tp23-ibmacpi-0.12 (root@deepdance) (gcc version 4.0.1 
(Debian 4.0.1-2)) #1 Tue Sep 6 21:34:48 CEST 2005 
martin@deepdance:~ -> cat /proc/acpi/ibm/driver 
driver:         IBM ThinkPad ACPI Extras 
version:        0.12a 
 
IBM ACPI works as well as suspend to disk and suspend to RAM. 
 
As ACPI in Linux Kernel 2.6.13 works with IBM ACPI 0.8 (as in kernel code) and 
IBM ACPI 0.12a I think its safe to close this bug, which I do now. 
 
Regards, Martin Steigerwald