On 07/26/2013 09:53 PM, Vinson Lee wrote:
On Thu, Jul 25, 2013 at 11:19 PM, Kenneth Graunke <kenn...@whitecape.org> wrote:
On 07/25/2013 10:48 PM, Vinson Lee wrote:
Fixes "Uninitialized scalar field" defect reported by Coverity.
Signed-off-by: Vinson Lee <v...@freedesktop.org>
---
src/mesa/drivers/dri/i965/brw_blorp_clear.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp
b/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp
index f26f39d..5234208 100644
--- a/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp
+++ b/src/mesa/drivers/dri/i965/brw_blorp_clear.cpp
@@ -119,7 +119,8 @@ private:
brw_blorp_const_color_program::brw_blorp_const_color_program(
struct brw_context *brw,
const brw_blorp_const_color_prog_key *key)
- : mem_ctx(ralloc_context(NULL)),
+ : prog_data(),
+ mem_ctx(ralloc_context(NULL)),
brw(brw),
key(key),
R0(),
NAK. Since brw_blorp_prog_data doesn't define a constructor (see
brw_blorp.h:196), adding prog_data() is just a call to the C++ default/empty
constructor.
So this patch doesn't actually initialize anything at all.
struct brw_blorp_prog_data is a POD class with no constructor defined.
'prog_data()' should value-initialize and, in this case,
zero-initialize all fields of prog_data.
You're right - I forgot that explicitly calling the constructor of a POD
type zero-initializes fields; it's only constructors for non-POD types
that are no-ops.
I still don't like this patch, though...it's really not obvious what's
going on.
--Ken
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
