Tune linear->sRGB constants to round-trip all bytes.

tests/SRGBTest.cpp

 /*
  * Copyright 2016 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */
 
 #include "SkSRGB.h"
 #include "SkTypes.h"
 #include "Test.h"
 #include <math.h>
 
 static uint8_t linear_to_srgb(float l) {
-    // Round float to int, truncate that to uint8_t.
-    return (uint8_t)Sk4f_round( sk_linear_to_srgb(Sk4f{l}) )[0];
+    return (uint8_t)sk_linear_to_srgb(Sk4f{l})[0];
 }
 
 DEF_TEST(sk_linear_to_srgb, r) {
-    // Should map 0 -> 0 and 1 -> 1.
-    REPORTER_ASSERT(r,   0 == linear_to_srgb(0.0f));
-    REPORTER_ASSERT(r, 255 == linear_to_srgb(1.0f));
+    // All bytes should round trip.
+    for (int i = 0; i < 256; i++) {
+        int actual = linear_to_srgb(sk_linear_from_srgb[i]);
+        if (i != actual) {
+            ERRORF(r, "%d -> %d\n", i, actual);
+        }
+    }
 
-    // Should be monotonic between 0 and 1.
-    // We don't bother checking denorm values.
-    int tolerated_regressions = 0;
-#if defined(SK_ARM_HAS_NEON)
-    // Values around 0.166016 are usually 72 but drop briefly (41 floats) down to 71.
-    tolerated_regressions = 1;
-#endif
+    // Should be (mostly) monotonic between 0 and 1.
+    // TODO: It'd be nice to find coefficients that satisfy both of these tests pefectly.
+    int tolerated_regressions = 1;
+
     uint8_t prev = 0;
-    for (float f = FLT_MIN; f <= 1.0f; ) {
+    for (float f = FLT_MIN; f <= 1.0f; ) {  // We don't bother checking denorm values.
         uint8_t srgb = linear_to_srgb(f);
 
         REPORTER_ASSERT(r, srgb >= prev || tolerated_regressions > 0);
         if (srgb < prev) { tolerated_regressions--; }
         prev = srgb;
 
         union { float flt; uint32_t bits; } pun = { f };
         pun.bits++;
         f = pun.flt;
     }
 }