| Age | Commit message (Collapse) | Author |
|---|
|
This relies on a new kernel ioctl to get the available aperture size.
In order to provide reasonable performance from dri_bufmgr_check_aperture, we
now require that once a buffer has been used as the target of a relocation,
it gets no further relocations added to it. This cuts the cost of
check_aperture from 10% to 1% in the 3D driver with no code changes, but
slightly complicates our plans for the 2D driver.
|
|
|
|
Don't count on ioctl returning -errno; use errno directly.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
|
|
|
|
|
|
|
|
libdrm_intel needs symbols from libdrm, so link against it.
|
|
Bug #18035. Fixes deadlock in glean texCube testcase.
|
|
When using bufmgr_fake without DRM, the X server idles the ring whenever it
wants to wait for something to complete (brutal, but effective). In this
case, bufmgr_fake must treat the pending fence as having passed. However, it
wasn't recording the fences as it emitted them, nor cleaning buffers as they
passed.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
|
We want to be able to use the bufmgr from multiple threads for GL, and thus
we need to protect the internal structures.
The pthread-stubs package is used so that programs not linked against
pthreads get weak symbols to stubs and don't eat most of the cost.
|
|
http://bugs.freedesktop.org/show_bug.cgi?id=17705
|
|
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=17705
|
|
The gltestperf demo in some cases took over seven seconds to make it through
one batchbuffer on a GM965.
Bug #17004.
|
|
I'd swapped the operands, so if we weren't in lockstep with the hardware we
said the sequence was always passed. Additionally, a race was available that
we might have failed at recovering from. Instead, I've replaced the logic
with new stuff that should be more robust and not rely on all the parties in
userland following the same IRQ_EMIT() == 1 protocol. Also, in a radical
departure from past efforts, include a long comment describing the failure
modes and how we're working around them.
Thanks to haihao for catching the original issue.
|
|
|
|
|
|
Thanks to airlied for catching this.
|
|
This avoids duplicating the effort in 3 places. Also, added emit/wait fence
callbacks back in bufmgr_fake since we need it for non-drm 2d. Sigh.
|
|
In the process, work around the glaring bugs of the kernel irq wait function.
|
|
dri_bufmgr.h is replaced by intel_bufmgr.h, and several functions are renamed,
though the structures and many functions remain dri_bufmgr_* and dri_bo_*
|
|
|
|
Store the global name in global_name, don't overwrite the gem_handle.
|
|
|
|
|
|
|
|
While the bufmgr isn't thread-safe at the moment, we need it to be for shared
objects between contexts.
|
|
another name
|
|
|
|
I915_GEM_DOMAIN_CPU is very expensive to wait for -- it generally requires
clflushing the frame buffer.
|
|
When a software fallback has completed, usermode must notify the kernel so
that any scanout buffers can be synchronized. This ioctl should be called
whenever a fallback completes to flush CPU and chipset caches.
|
|
Thanks to Thomas Hellstrom for catching the issue, no thanks to the kernel
developer who authoritatively told me that they would get restarted on their
own.
|
|
This reduces the diff from Mesa and reduces the illegibility of what I did.
|
|
This is the create (may want location flags), pread/pwrite/mmap
(performance tuning hints), and set_domain (will 32 bits be enough for
everyone?) ioctls. Left in the generic set are just flink/open/close.
The 2D driver must be updated for this change, and API but not ABI is broken
for 3D. The driver version is bumped to mark this.
|
|
|
|
Place the buffer reuse links right into the dri_bo_gem object.
|
|
|
|
The code was discarding the dri_bo_gem structure and saving only the kernel
handle. This lost the mmap address, causing pain when the next buffer user
wanted to map the buffer.
|
|
This will be used by the X Server for VT switch.
|
|
|
|
|
