I don't think this is the way to do it - we should just fix the network drivers if they are broken. Most of them seem to suspend and resume just fine on fedora rawhide.

Ohh, and ipw3945 is not going to work well for suspend resume - I'm suprised it works at all. iwl3945 suspends and resumes very well.

Richard.

Created attachment 15506 [details] [review]
ubuntu patch

Ergh, this time without syntax errors, sorry.

Hm, I am using iwl3945 (Dell Latitude D430), and until now I kept an rmmod/modprobe iwl3945 in /etc/pm/sleep.d. Otherwise that thing could never detect any ESSIDs again and it never automatically reconnects to the wifi.

I agree that "just fix the network drivers" is a great long-term goal, of course, but it does not help a lot with pending releases (Fedora will release in a few weeks as well, right?)

If that were the preferred and only way, the entire quirks set in hal-info would be "not the way to do it". :-)

What would you recommend as a better workaround instead?

Thank you for your input!

Martin

The best way to handle this for Ubuntu is to make it its own hook instead of modifying the 50modules hook.

For what it is worth, I have been running Hardy since early December and have never encountered the iwl3945 bug this report mentions -- on my box, at least, iwl3945 works fine across a suspend/resume without needing to be removed/readded.

Proper workaround for kernels with drivers that need special care over suspend/resume is for the kernel package to provide a suspend hook that does whatever is needed until the driver is updated to work correctly across suspend/hibernate with as little userspace help as possible.

For this patch in particular, writing a tiny little hook that is part of the appropriate kernel module package would be the way to go, and I am willing to help write that hook.  Patching pm-utils directly to cover this case is not something I am willing to do.