[CentOS] CentOS 5 LiveCD better than the real one?

Patrick centos-list at puzzled.xs4all.nl
Thu Oct 11 14:46:37 UTC 2007

On Thu, 2007-10-11 at 05:33 -0400, Radu-Cristian FOTESCU wrote:
> Folks, 
> I am terribly puzzled by an issue reported as bug 2381
> [http://bugs.centos.org/view.php?id=2381] — definitely an upstream
> bug, as it does the same under X/OS 5 and StartCom 5.
> Simply put: Hibernation fails with horrendous I/O errors after swsusp
> starts dumping to swap.

If you are using the 8.41 ATI driver don't as it will break your
suspend/resume. I don't know if the nvidia proprietary one behaves. Not
sure about this one but worth looking into: iirc if your swap
partition/available swap is too small hibernate can fail. So perhaps try
without any propietary gfx driver and with a swap partition that's say
twice the size of your memory. Don't use vga=... in your grub.conf as it
may break suspend/resume. At least it does on my Acer TM6465 laptop.

Also make sure that the kvm, kvm_intel, iwl3945 modules are unloaded
before suspending. This is done in /etc/pm/config.d/unload_modules
Make sure you # chmod 755 unload_modules

File contains something like:
SUSPEND_MODULES="iwl3945 mac80211 cfg80211 kvm_intel kvm"

See this "quirk" webpage with tons of help:

Answer the questions on this page:
and try out the suggestions.

If that does not help get the latest hal-info source (git tree) and see
if there are any solutions in there for your laptop. Specifically look
in this file:

The hal-info rpm that is included in F8T3 does not have any entries for
the Acer TM5xxx laptops that you mentioned so check out the latest
hal-info git tree.

> What bugs me is that hibernating from the CentOS 5 LiveCD *works*!!!
> (on the same hardware)

At least you know that it can work. That's better than not working at
all :)

> How could I investigate to see what is the relevant difference between
> the LiveCD and the installed system, WRT suspend-to-disk?

Beats me. Start with the suggestions I gave you.

> It has been said that "CentOS 5 i386 Live CD is based on the ADIOS
> Live CD Project", but also that "CentOS Project would like to thank
> Patrice Guay for the creation of this CD". Who is supposed to know
> better about this issue?

Don't know.

> Also, how could I check if upstream has a similar bug report? I don't
> have any RHN acct.
> It's a shame that RHEL5 && clones fail to work properly with a
> specific laptop (Acer Travelmate 5310/5320/5710/5720 have the same
> architectural design), but an ad-hoc LiveCD works!

Suspend/resume/hibernate stuff is complicated and the zillions of
different laptops out there all with their (buggy) bios and acpi
implementations isn't helping either. So with that in mind imho I think
it could be much worse. There is a lot of work going on to ensure that
laptops just work. Laptop quirks are gradually added to hal-info which
should make more and more laptops work correctly. If you find the quirk
fix for your laptop please send the relevant info to the hal-info
mailing list (see that "quirk" web page I gave above). 


More information about the CentOS mailing list