Explicitly fail read/writePixels in invalid color space scenarios
It's not well defined what to do when moving from a nullptr color space to
a tagged destination (drawing, reading, writing, etc...). In these
scenarios, at least, we can choose to disallow the operation (rather than
produce an unexpected or inconsistent result).
BUG=skia:
Change-Id: I033b23c6f2bb00664efc8fdab1b3f52053d77695
Reviewed-on: https://skia-review.googlesource.com/6600
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Matt Sarett <msarett@google.com>
diff --git a/src/gpu/GrContext.cpp b/src/gpu/GrContext.cpp
index 30f596f..310f455 100644
--- a/src/gpu/GrContext.cpp
+++ b/src/gpu/GrContext.cpp
@@ -278,6 +278,11 @@
return false;
}
+ // We don't allow writing to a color space tagged destination if the source isn't tagged
+ if (dstColorSpace && !srcColorSpace) {
+ return false;
+ }
+
GrGpu::DrawPreference drawPreference = GrGpu::kNoDraw_DrawPreference;
// Don't prefer to draw for the conversion (and thereby access a texture from the cache) when
// we've already determined that there isn't a roundtrip preserving conversion processor pair.
@@ -427,6 +432,11 @@
return false;
}
+ // We don't allow reading to a color space tagged destination if the source isn't tagged
+ if (dstColorSpace && !srcColorSpace) {
+ return false;
+ }
+
GrGpu::DrawPreference drawPreference = GrGpu::kNoDraw_DrawPreference;
// Don't prefer to draw for the conversion (and thereby access a texture from the cache) when
// we've already determined that there isn't a roundtrip preserving conversion processor pair.
diff --git a/tests/ImageTest.cpp b/tests/ImageTest.cpp
index 161dff8..b2ac36d 100644
--- a/tests/ImageTest.cpp
+++ b/tests/ImageTest.cpp
@@ -45,8 +45,7 @@
// see https://bug.skia.org/3965
//REPORTER_ASSERT(reporter, a->isOpaque() == b->isOpaque());
- // The codecs may have given us back F16, we can't read from F16 raster to N32, only S32.
- SkImageInfo info = SkImageInfo::MakeS32(widthA, heightA, a->alphaType());
+ SkImageInfo info = SkImageInfo::MakeN32(widthA, heightA, a->alphaType());
SkAutoPixmapStorage pmapA, pmapB;
pmapA.alloc(info);
pmapB.alloc(info);
diff --git a/tests/ReadPixelsTest.cpp b/tests/ReadPixelsTest.cpp
index 71cd8f5..487152c 100644
--- a/tests/ReadPixelsTest.cpp
+++ b/tests/ReadPixelsTest.cpp
@@ -487,3 +487,40 @@
}
}
#endif
+
+#if SK_SUPPORT_GPU
+DEF_GPUTEST_FOR_RENDERING_CONTEXTS(ReadPixelsColorSpaceVariants_Gpu, reporter, ctxInfo) {
+ // Create surfaces with and without an attached color space
+ sk_sp<SkColorSpace> srgbColorSpace = SkColorSpace::MakeNamed(SkColorSpace::kSRGB_Named);
+ SkImageInfo srgbInfo = SkImageInfo::MakeS32(DEV_W, DEV_H, kPremul_SkAlphaType);
+ SkImageInfo legacyInfo = srgbInfo.makeColorSpace(nullptr);
+
+ sk_sp<SkSurface> srgbSurface = SkSurface::MakeRenderTarget(ctxInfo.grContext(), SkBudgeted::kNo,
+ srgbInfo);
+ sk_sp<SkSurface> legacySurface = SkSurface::MakeRenderTarget(ctxInfo.grContext(),
+ SkBudgeted::kNo, legacyInfo);
+ SkCanvas* srgbCanvas = srgbSurface->getCanvas();
+ SkCanvas* legacyCanvas = legacySurface->getCanvas();
+
+ struct {
+ SkCanvas* fCanvas;
+ const SkImageInfo& fBmpInfo;
+ bool fExpectSuccess;
+ } kTestConfigs[] ={
+ // Both kinds of surface should be able to read into a legacy destination
+ { srgbCanvas, legacyInfo, true },
+ { legacyCanvas, legacyInfo, true },
+ // Tagged surface should be able to read into tagged destination
+ { srgbCanvas, srgbInfo, true },
+ // Legacy surface shouldn't read into tagged destination
+ { legacyCanvas, srgbInfo, false },
+ };
+
+ for (auto testConfig : kTestConfigs) {
+ SkBitmap bmp;
+ bmp.setInfo(testConfig.fBmpInfo);
+ bool result = testConfig.fCanvas->readPixels(&bmp, 0, 0);
+ REPORTER_ASSERT(reporter, result == testConfig.fExpectSuccess);
+ }
+}
+#endif