Bug 10340

Summary: suspend order - again
Product: Power Management Reporter: Rafael J. Wysocki (rjw)
Component: Hibernation/SuspendAssignee: power-management_other
Status: CLOSED CODE_FIX    
Severity: normal CC: acpi-bugzilla, carlos, mjg59-kernel, shaohua.li
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.5.25-rc Subsystem:
Regression: Yes Bisected commit-id:
Bug Depends on:    
Bug Blocks: 7216, 9832    

Description Rafael J. Wysocki 2008-03-27 10:07:12 UTC 
Subject    : suspend order - again
Submitter  : Shaohua Li <shaohua.li@intel.com>
Date       : 2008-03-14 09:22
References : http://marc.info/?l=linux-acpi&m=120549574009914&w=4
Handled-By : Shaohua Li <shaohua.li@intel.com>
Patch      : http://marc.info/?l=linux-acpi&m=120649617317396&w=4

This entry is being used for tracking a regression from 2.6.24.  Please don't
close it until the problem is fixed in the mainline.
Comment 1 Rafael J. Wysocki 2008-03-27 10:11:49 UTC 
The regression is that the default suspend ordering, which is the "old" ACPI 1.0 now, doesn't work with Asus A6B00VC.

Now, the question is how many boxes like this there are in the wild and whether or not we're going to blacklist them all.  I guess if there are more than 3 machine models, we should go back to the ACPI 2.0 suspend ordering, no? ;-)
Comment 2 Carlos Corbacho 2008-03-27 10:36:44 UTC 
I suspect we may have is a problem endemic to nVidia chipsets & bad BIOSes (at least, I'm hoping from all the machines that exhibit the problem that this is the case).

However, we really need more information either way before we can decide how or what to blacklist, order, etc.

In the short term, the original regression the re-order fixed had been around for 3 releases, and we have a fix for the one regression found so far. So I'd vote to apply the DMI quirk patch for .25, and review things in .26 depending on what feedback we get.
Comment 3 Rafael J. Wysocki 2008-03-29 11:50:37 UTC 
Here's another example, and quite a serious one, I'd say:
https://bugzilla.novell.com/show_bug.cgi?id=374217#c16

I'm seriously considering to post a patch that will bring the previous default suspend ordering back.
Comment 4 Carlos Corbacho 2008-03-29 12:12:48 UTC 
I agree, MacBook is a far more serious example than the three year old Asus laptop.

At two to three, and given Macbook has probably sold far more than the affected nVidia boards combined, you've convinced me that a wholesale switch to the 'old' order is a bad idea.

So: let's revert the ordering change patch, and we'll switch to just blacklisting the bad nVidia systems that need the 'old' ordering in 2.6.26
Comment 5 Rafael J. Wysocki 2008-03-30 05:02:57 UTC 
Patch posted: http://marc.info/?l=linux-acpi&m=120683999519118&w=4