[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: acpi vs apm

Cameron Patrick <cameron@patrick.wattle.id.au> wrote:

> What kind of userland support do you see as being missing?  I use the
> hibernate package for ACPI sleep and it works pretty well.  Most of
> the problems that I've seen with ACPI have been kernel or BIOS issues
> (e.g. the screen not being switched on when resuming unless you give
> particular kernel options).

The ACPI spec makes it the OS's responsibility to reinitialise the video
hardware, not the BIOS's. The kernel is (for the most part) incapable of
doing so - there are some cases where you can get it to bring it back,
but there's a huge range of hardware where those options crash the
machine on resume.

The main things that need doing are:

1) Dealing with network interfaces and the like sensibly - at the
moment, this will often require unloading and reloading modules pre/post
2) Working with video state. The vbetool package makes it possible to
save and restore the graphics card state from userland, which tends to
work much better than the kernel fudges. In the long run, either X or
the framebuffer drivers need to get much better at programming the
3) Decent integration into user configuration. In a lot of cases, you
want PM policy to be definable by the person carrying the computer
around, even if they don't have administrative access.

I've got some amount of code for doing most of these things, but it's
very heavily a post-Sarge thing - it involves touching a lot of
Matthew Garrett | mjg59-chiark.mail.debian.devel@srcf.ucam.org

Reply to: