| Age | Commit message (Collapse) | Author |
|---|
|
just do it per blit. No noticable performance difference (or behavior
difference, in my tests), and a significant difference in sanity.
|
|
|
|
the register header was meaningful. It turns out those were the proper values
for the plain INCR/DECR ops, while what we were using as INCR/DECR were the
_WRAP versions. Tested with stencil_wrap (didn't expose normal/_WRAP swapping)
and stencilwrap (exposed it nicely) tests. Props to idr for poking me about
adding this.
|
|
|
|
* Don't dispatch vertices directly to MMIO; queue them up in dma-like buffers
first. This makes things more uniform between AGP and MMIO paths, cleans up
some locking ugliness, and makes the driver look more like other drivers.
* Don't use the AGP Cmd buffer provided by the server. Instead allocate one in
the client, which avoids the need for lots of synchronization stuff.
* Mark some MMIO accesses volatile that should have been.
* Disable the AGP submission path by default (agp_disable=true) due to
unresolved issues in the new code. The old code had its own (serious) errors
with AGP, so this is not really a step backwards.
|
|
|
|
|
|
(build tested, but not physically tested)
|
