Mesa (main): panfrost: Make pan_merge macro more robust

[GitLab Mirror]

Module: Mesa
Branch: main
Commit: 311077c483200e3594a755bba3a4df0c5d561df7
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=311077c483200e3594a755bba3a4df0c5d561df7

Author: Alyssa Rosenzweig <alyssa at collabora.com>
Date:   Mon Dec  6 15:06:23 2021 -0500

panfrost: Make pan_merge macro more robust

Consider the following innocuous-looking code:

   pan_merge(packed, vtx->attributes[i], ATTRIBUTE);

Under the current implementation, this code is completely broken. Why?
The current implemention is a macro which hardcodes the loop index i,
which shadows the i used to index attributes. Pull out a helper method
so we do the right thing without resulting to further preprocessor abuse
(__COUNTER__).

While making things more robust, assert the crucial pan_merge
invariant that the total size is a multiple of 4; if this fails, the
routine risks memory corruption.

Signed-off-by: Alyssa Rosenzweig <alyssa at collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14119>

---

 src/panfrost/lib/genxml/gen_pack.py | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/src/panfrost/lib/genxml/gen_pack.py b/src/panfrost/lib/genxml/gen_pack.py
index 93464525fb3..2a729f64ae0 100644
--- a/src/panfrost/lib/genxml/gen_pack.py
+++ b/src/panfrost/lib/genxml/gen_pack.py
@@ -176,11 +176,16 @@ __gen_unpack_padded(const uint8_t *restrict cl, uint32_t start, uint32_t end)
 #define pan_section_print(fp, A, S, var, indent)                          \\
         PREFIX4(A, SECTION, S, print)(fp, &(var), indent)
 
+static inline void pan_merge_helper(uint32_t *dst, const uint32_t *src, size_t bytes)
+{
+        assert((bytes & 3) == 0);
+
+        for (unsigned i = 0; i < (bytes / 4); ++i)
+                dst[i] |= src[i];
+}
+
 #define pan_merge(packed1, packed2, type) \
-        do { \
-                for (unsigned i = 0; i < (PREFIX2(type, LENGTH) / 4); ++i) \
-                        (packed1).opaque[i] |= (packed2).opaque[i]; \
-        } while(0)
+        pan_merge_helper((packed1).opaque, (packed2).opaque, pan_size(type))
 
 /* From presentations, 16x16 tiles externally. Use shift for fast computation
  * of tile numbers. */