On Tue, Jan 17, 2006 at 09:17:01AM +1100, Bojan Smojver wrote:
Quoting Dave Jones <davej redhat com>:
>Most of the diff between in-kernel suspend and suspend2 is
>the bells & whistles like compression, splash screens, etc.
If that were true, then vanilla suspend code would work on my notebook
but it would just be uglier and slower. However, it isn't just uglier
and slower, it also hangs my X consistently on resume.
The internals of suspend2 are explained in the file
Documentation/power/internals.txt, once you apply the patch. I'm no
kernel hacker, but it looks like the algorithms for doing things have
been changed - it's not just spit and polish.
Note I said 'most'. The fundamentals aren't that different.
And as you noted, with a lot of suspend2 being moved to userspace,
and upstream headed in the same direction, the delta between
the two implementations should continue to shrink.
In conclusion, I wouldn't just dismiss suspend2 to as some
redundant
patch. It makes many real systems actually suspend and resume, unlike
the vanilla code.
I never dismissed it.
But the end goal should be 'make vanilla work', not 'rely on patching
a kernel every time an update comes out' or 'make alternative kernels
available'.
Dave