This is a resend, the earlier set seem to have gone into void. This set was collected with intentionally failing at context creation and then fixing all the places where segfault resulted. Some of the places where null is being checked are a bit funny but its only because during first time context creation things have not yet settled properly. After this set Waffle is able to correctly report if there was failure during context creation, before this set only possible result is segfault.
/Juha-Pekka Juha-Pekka Heikkila (13): mesa/meta: Don't free meta if it was never initialized i965: If cache has no BO don't try to unmap BO mesa/main: Don't go freeing texture data which was never allocated mesa/main: Don't go freeing pipeline data which is not reserved mesa/program: Don't clear caches which are not allocated mesa/main: Don't free QueryObjects which were never allocated mesa/main: Don't free unallocated vertex array state mesa/main: Don't go deleting unallocated TransformFeedbackObjects mesa/main: Don't free unallocated performance monitor mesa/main: Check context pointer in _mesa_error before using it i965: Add missing return on error path mesa/main: Don't proceed on deleting null hash table mesa/main: Verify context creation on progress src/mesa/drivers/common/meta.c | 3 ++ src/mesa/drivers/dri/i965/brw_state_cache.c | 11 ++-- src/mesa/drivers/dri/i965/intel_buffer_objects.c | 1 + src/mesa/main/errors.c | 26 ++++++---- src/mesa/main/hash.c | 3 +- src/mesa/main/performance_monitor.c | 8 +-- src/mesa/main/pipelineobj.c | 9 ++-- src/mesa/main/queryobj.c | 6 ++- src/mesa/main/shared.c | 66 ++++++++++++++++++++++-- src/mesa/main/texstate.c | 6 ++- src/mesa/main/transformfeedback.c | 12 +++-- src/mesa/main/varray.c | 6 ++- src/mesa/program/prog_cache.c | 16 +++--- 13 files changed, 132 insertions(+), 41 deletions(-) -- 1.8.5.1 _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
