The nvidia driver currently uses these callbacks to work around problems where
RAC will disable access to the hardware at unexpected times. This change
restores these hooks until we can come up with a better API for working around
RAC.
This reverts commit d7c0ba2e9e.
Conflicts:
hw/xfree86/loader/xf86sym.c
Some BIOSes (hi XGI!) will attempt to enumerate the PCI bus by asking
for the config space of every possible device number. This despite
perfectly functional BIOS methods to enumerate the bus exactly.
We can get a case with gnome-terminal + links, where we get two arrays
of glyphs all with 0 width and 0 heights in them. If this happens
we manage to get to this case without any buffer setup and segfault.
- By adding a small hack to the xserver i was able to easily test the performance of the normally rare direct case (using cairo).
- It turned out to be 70% slower for me (large test on an otherwise idle computer), which seems enough of a reason to remove it.
- AddTraps could also use a 2nd look, but since noone is using that it's a bit hard and less useful to test.
If anyone can come up with an example of a bus where:
- both i/o and memory resources are addressable
- access to them can be controlled
- but they can't be controlled independently
then by all means, reinstate this logic.
instead of calling CFRunLoopRun() directly. The leak wasn't reproducible on
this machine, but someone was able to produce a leak trace with Instruments
that indicates it was leaking in the CFRunLoopRun() path.
x-input.m: dequeue and ignore events when pbproxy_active is false.
x-selection.h: add an is_active method that is used by x-input.m to ignore
events.
x-selection.m: Handle nearly every preference, except for primary_on_grab,
which I don't really understand yet.
(cherry picked from commit 4d51ad851e)