we were only running the llvm paths when the input elts were linear,
now we can handle abritrary fetch elts arrays. we do this by generating
two paths - linear and fetch_elts one and just selecting the right one
at run time.
max_index must be observed to prevent crashes due to bad index data.
I've been using this patch for some time without regressions.
Some places, where we use internal vertex buffer, it is not entirely
clear what max_index should be, so passing just ~0 to avoid regressions
for now.
doc additions: shader export ARRAY_BASE for EXPORT_POS: 60 is position,
61 is misc vec(VS_OUT_MISC_VEC - used here),
62, 63 are clip distance vectors(VS_OUT_CCDIST#)
sorry for formating - there seem to be so many different styles in r600
Certain headers, such as GL/glew.h, are in both the Mesa include and the
default installed include directories. On recent distros the needed
symbols can be found in both places. On older distros the installed
headers could be lacking symbols, so for a header that exists in both
places, the local one should be found first.
The POSIX function pthread_cond_wait can have spurious wakeups when
waiting on a condition variable.
Add a 64-bit counter that is incremented whenever the barrier becomes
full. A woken thread checks the counter. If the counter has not changed
then it has been spuriously woken and goes back to sleep. If the counter
has changed then it was properly signaled and exits the barrier.
Tested on Mac OS X.
This patch was based on ideas from Luca Barbieri.
The struct st_module isn't needed as it is the same thing as the st_api
struct. That is they both represent the API. Instead just use a single
function entry point to the the API.
Everybody should respect max_index, specially llvm generated code, which
likes to eat vertices 4 at a time, so it may end up chew a bit a bit more
than actually exists.
Transfer, being now a context operation, should happen in order with
all other contexts operations. If there is rendering pending on the
resource then the driver must flush and potentially wait itself
internally.
Instead of avoiding using transfers internally (as done in llvmpipe) I've
opted to simply pass PIPE_TRANSFER_UNSYNCHRONIZED in all internal
transfers, to avoid infinite recursion.
