Work around a GLSL compiler bug on Pixel 8

Pixel 8 with ARM Mali-G715 throws away "x0x1 << 16 >> 16". We need this in order to sign-extend the bottom 16 bits of x0x1. Add some throwaway code that convinces the compiler not to throw this away.

Diffs=
9ad9d82e1 Work around a GLSL compiler bug on Pixel 8 (#7247)

Co-authored-by: Chris Dalton <99840794+csmartdalton@users.noreply.github.com>

.rive_head

diff --git a/.rive_head b/.rive_head
index f350668..9fdcac6 100644
--- a/.rive_head
+++ b/.rive_head
@@ -1 +1 @@
-17f604311327e786f48dfdfca37501f10dfd5c6c
+9ad9d82e1863f9c04376005534d3c1587a98b9cf

renderer/shaders/tessellate.glsl

diff --git a/renderer/shaders/tessellate.glsl b/renderer/shaders/tessellate.glsl
index 5fccc33..efbe0a8 100644
--- a/renderer/shaders/tessellate.glsl
+++ b/renderer/shaders/tessellate.glsl
@@ -78,7 +78,21 @@
     bool isFirstSpan = _vertexID < 4;
     float y = isFirstSpan ? @a_joinTan_and_ys.z : @a_joinTan_and_ys.w;
     int x0x1 = int(isFirstSpan ? @a_args.x : @a_args.y);
+#ifdef GLSL
+    int x1up = x0x1 << 16;
+    if (@a_args.z == 0xffffffffu)
+    {
+        // Pixel 8 with ARM Mali-G715 throws away "x0x1 << 16 >> 16". We need this in order to
+        // sign-extend the bottom 16 bits of x0x1.
+        // Create a branch that we know won't be taken, in order to convince the compiler not to
+        // throw this operation away.
+        // NOTE: we could use bitfieldExtract(), but it isn't available on ES 3.0.
+        --x1up;
+    }
+    float x0 = float(x1up >> 16);
+#else
     float x0 = float(x0x1 << 16 >> 16);
+#endif
     float x1 = float(x0x1 >> 16);
     float2 coord = float2((_vertexID & 1) == 0 ? x0 : x1, (_vertexID & 2) == 0 ? y + 1. : y);