Prior to this patch, it would attempt to optimize and allocate registers
for the program even if it failed to compile. This seems wasteful.
More importantly, the "message length > 11" failure seems to choke the
instruction scheduler, making it somehow use an undefined value and
segmentation fault.
Signed-off-by: Kenneth Graunke <kenn...@whitecape.org>
---
src/mesa/drivers/dri/i965/brw_fs.cpp | 2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
Eric,
Yeah. Somehow I wasn't thinking straight and thought that'd check for
failure after every instruction. But it doesn't; those are only the
top-level variable declarations and functions. No sense in that at all.
This version moves it below the loop---after the bulk of codegen, but
prior to optimizing and register allocation. Seems reasonable.
The 16-element stack array theory sounds about right. Presumably we
should never run into that limit for good code, and this ought to reject
the bad code.
diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp
b/src/mesa/drivers/dri/i965/brw_fs.cpp
index 7c73a8f..3a16afe 100644
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -1533,6 +1533,8 @@ fs_visitor::run()
this->result = reg_undef;
ir->accept(this);
}
+ if (failed)
+ return false;
emit_fb_writes();
--
1.7.5.4
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
