Re: [Chromium-dev] multicontext branch and context creation problem

[Karl R. <rk...@vr...>]

ok... i spoke too soon..

first, my previous conf, while it did display correctly, i got tons of
spewing on the console about:

CR Warning(hrchekkalfas:2794): Unhandled target in 
renderspuChromiumParametervCR(0x9904)

which appears to be coming from some debugging stuff 
in crserver/server_papi.c, and the renderspu not handling the
GL_PRINT_STRING_CR parameter.. 

then, feeling adventurous, i changed the pack spu to a tilesort spu, and
added a second readback spu & node, with each readback packing and sending
to a single render spu/node for display.

now, the crservers will start up correctly, but i seem to get into an
infinate loop while crappfaker starts up. i do get 3 windows (should have
4, i think.. the faker window is missing, afaik). the last few lines from
the debugging output are:

CR Debug(hrchekkalfas:4195): Buffer pool was empty, so I allocated 1048596
bytes
Creating
Creating 1.5
CR Debug(hrchekkalfas:4193): Buffer pool was empty, so I allocated 1048596
bytes
CR Debug(hrchekkalfas:4193): First call to MakeCurrent, window visual is
OK.
CR Debug(hrchekkalfas:4193): Creating a context: 1 (0x2) ctx = 0x810d0b8
Creating
Creating 1.5


after some poking around, it would appear that we never come back  
from the

while (writeback) {
	crNetRecv();
}

loop in spu/tilesort/tilesortspu_context.c, on the 2nd (of 2) iterations
of the loop surrounding the above chunk of code.

is there a writeback somewhere that is not taking place, in one of the
readback spu's maybe? im not exactly sure where the reply it is waiting
for is dispatched.. (or how that loop terminates, for that matter...).

thanks.

karl

  [k a r l  r a s c h e]    the math department makes me feel normal
                                                   -- campus grafiti
   rk...@vr...