sna/gen4: Tweak compilation flags to avoid mixed settings across functions

Confusing gcc with different flags for supposedly inlined functions is
not a good idea.

References: https://bugs.freedesktop.org/show_bug.cgi?id=62198
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

3 files changed, 13 insertions, 28 deletions

diff --git a/src/sna/compiler.h b/src/sna/compiler.h
index 985844df..b5c9ac27 100644
--- a/src/sna/compiler.h
+++ b/src/sna/compiler.h
@@ -43,7 +43,7 @@
 #define likely(expr) (expr)
 #define unlikely(expr) (expr)
 #define noinline
-#define force_inline
+#define force_inline inline
 #define fastcall
 #define must_check
 #define constant
diff --git a/src/sna/gen4_vertex.c b/src/sna/gen4_vertex.c
index 35f772e9..d8425859 100644
--- a/src/sna/gen4_vertex.c
+++ b/src/sna/gen4_vertex.c
@@ -232,7 +232,7 @@ void gen4_vertex_close(struct sna *sna)
 #define OUT_VERTEX(x,y) vertex_emit_2s(sna, x,y) /* XXX assert(!too_large(x, y)); */
 #define OUT_VERTEX_F(v) vertex_emit(sna, v)
 
-inline static float
+force_inline static float
 compute_linear(const struct sna_composite_channel *channel,
 	       int16_t x, int16_t y)
 {
@@ -241,7 +241,7 @@ compute_linear(const struct sna_composite_channel *channel,
 		channel->u.linear.offset);
 }
 
-inline static void
+sse2 inline static void
 emit_texcoord(struct sna *sna,
 	      const struct sna_composite_channel *channel,
 	      int16_t x, int16_t y)
@@ -274,7 +274,7 @@ emit_texcoord(struct sna *sna,
 	}
 }
 
-sse2 inline static void
+sse2 force_inline static void
 emit_vertex(struct sna *sna,
 	    const struct sna_composite_op *op,
 	    int16_t srcX, int16_t srcY,
@@ -304,7 +304,7 @@ emit_primitive(struct sna *sna,
 		    r->dst.x,  r->dst.y);
 }
 
-sse2 inline static void
+sse2 force_inline static void
 emit_vertex_mask(struct sna *sna,
 		 const struct sna_composite_op *op,
 		 int16_t srcX, int16_t srcY,
@@ -969,21 +969,6 @@ emit_primitive_affine_source_identity(struct sna *sna,
 	v[14] = msk_y * op->mask.scale[1];
 }
 
-inline static void
-emit_composite_texcoord_affine(struct sna *sna,
-			       const struct sna_composite_channel *channel,
-			       int16_t x, int16_t y)
-{
-	float t[2];
-
-	sna_get_transformed_coordinates(x + channel->offset[0],
-					y + channel->offset[1],
-					channel->transform,
-					&t[0], &t[1]);
-	OUT_VERTEX_F(t[0] * channel->scale[0]);
-	OUT_VERTEX_F(t[1] * channel->scale[1]);
-}
-
 /* SSE4_2 */
 #if defined(sse4_2)
 
@@ -1548,7 +1533,7 @@ unsigned gen4_choose_composite_emitter(struct sna *sna, struct sna_composite_op
 	return vb;
 }
 
-inline static void
+sse2 force_inline static void
 emit_span_vertex(struct sna *sna,
 		  const struct sna_composite_spans_op *op,
 		  int16_t x, int16_t y)
diff --git a/src/sna/sna_render_inline.h b/src/sna/sna_render_inline.h
index 7d9f2cac..1a3be2cb 100644
--- a/src/sna/sna_render_inline.h
+++ b/src/sna/sna_render_inline.h
@@ -17,7 +17,7 @@ static inline bool need_redirect(struct sna *sna, PixmapPtr dst)
 		dst->drawable.height > sna->render.max_3d_size);
 }
 
-static inline float pack_2s(int16_t x, int16_t y)
+static force_inline float pack_2s(int16_t x, int16_t y)
 {
 	union {
 		struct sna_coordinate p;
@@ -28,35 +28,35 @@ static inline float pack_2s(int16_t x, int16_t y)
 	return u.f;
 }
 
-static inline int vertex_space(struct sna *sna)
+static force_inline int vertex_space(struct sna *sna)
 {
 	return sna->render.vertex_size - sna->render.vertex_used;
 }
-static inline void vertex_emit(struct sna *sna, float v)
+static force_inline void vertex_emit(struct sna *sna, float v)
 {
 	assert(sna->render.vertex_used < sna->render.vertex_size);
 	sna->render.vertices[sna->render.vertex_used++] = v;
 }
-static inline void vertex_emit_2s(struct sna *sna, int16_t x, int16_t y)
+static force_inline void vertex_emit_2s(struct sna *sna, int16_t x, int16_t y)
 {
 	vertex_emit(sna, pack_2s(x, y));
 }
 
-static inline int batch_space(struct sna *sna)
+static force_inline int batch_space(struct sna *sna)
 {
 	assert(sna->kgem.nbatch <= KGEM_BATCH_SIZE(&sna->kgem));
 	assert(sna->kgem.nbatch + KGEM_BATCH_RESERVED <= sna->kgem.surface);
 	return sna->kgem.surface - sna->kgem.nbatch - KGEM_BATCH_RESERVED;
 }
 
-static inline void batch_emit(struct sna *sna, uint32_t dword)
+static force_inline void batch_emit(struct sna *sna, uint32_t dword)
 {
 	assert(sna->kgem.mode != KGEM_NONE);
 	assert(sna->kgem.nbatch + KGEM_BATCH_RESERVED < sna->kgem.surface);
 	sna->kgem.batch[sna->kgem.nbatch++] = dword;
 }
 
-static inline void batch_emit_float(struct sna *sna, float f)
+static force_inline void batch_emit_float(struct sna *sna, float f)
 {
 	union {
 		uint32_t dw;