[Chromium-dev] glGenPrograms segfault

[Mike H. <mho...@gr...>]

Arbfraglight works happily locally running with the conf file listed at 
the bottom, but completely borks when the appfaker is run remotely.  The 
bug can be demonstrated with arbfraglight.

There seems to be something funky with extension loading on remote 
nodes.  In particular, glGenProgramsARB is failing.  crGetProcAdress as 
well as the wgl/glx version returns an address successfully, but 
actually calling the function causes a segfault or occasionally an abort 
(SIGABRT).  I've tried multiple ATI driver revisions including ones 
known to work with previous Chromium releases.  Somewhere, we must be 
stepping on memory or getting a bogus entry in the function table.

Can someone with Nvidia systems see if they can reproduce the bug?

Thanks!
-Mike

import sys
sys.path.append( "../server" )
from mothership import *

Demo = 'arbfraglight'

Demo = os.path.join(crbindir, Demo)

mothershipPort = 10000
localHostname = os.uname()[1]

# Set up the server node
servernode1 = CRNetworkNode( )
# Note: each client has the -swap flag and we tell the server to only
# do one SwapBuffers here.
servernode1.Conf( 'only_swap_once', 1 )
servernode1.Conf( 'shared_windows', 1 )
renderspu = SPU( 'render' )
renderspu.Conf( 'window_geometry', [500, 500, 400, 400] )
servernode1.AddSPU( renderspu )

# Set up first app/client node
appnode1 = CRApplicationNode(  )
appnode1.SetApplication( '%s' % (Demo) )
appnode1.StartDir( crbindir )
spu = SPU('readback')
spu.Conf('window_geometry', [0, 0, 400, 400])
appnode1.AddSPU( spu )
spu = SPU('pack')
appnode1.AddSPU( spu )
spu.AddServer( servernode1, 'tcpip' )

cr = CR()
cr.MTU( 1024*1024 )
# Note: adding nodes in the order in which they must be started!
cr.AddNode( appnode1 )
cr.AddNode( servernode1 )
cr.Go()