office-gobmx/external/skia/clang-attributes-warning.patch.1
Noel Grandin 9c9a711ac5 Update skia to m111
(*) Remove patches already upstream

(*) Remove the skia_sk_cpu_sse_level_0_by_default.patch.1 patch and rather set
that define via -D parameter, because that is how the skia BUILD.gn
seems to do it.

(*) I hand edited the PCH file, because running the update_pch script failed for me.

Change-Id: I1fd175b9f63f8d2792a1876e4ece03fe92fb5374
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/146251
Tested-by: Jenkins
Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
2023-01-30 10:52:07 +00:00

31 lines
1.3 KiB
Groff

diff --git a/include/private/base/SkFloatingPoint.h b/include/private/base/SkFloatingPoint.h
index 3c6d22c310..60500b2d2c 100644
--- a/include/private/base/SkFloatingPoint.h
+++ b/include/private/base/SkFloatingPoint.h
@@ -159,7 +159,9 @@ static inline int64_t sk_float_saturate2int64(float x) {
// Cast double to float, ignoring any warning about too-large finite values being cast to float.
// Clang thinks this is undefined, but it's actually implementation defined to return either
// the largest float or infinity (one of the two bracketing representable floats). Good enough!
+#if defined(__clang__)
SK_NO_SANITIZE("float-cast-overflow")
+#endif
static inline float sk_double_to_float(double x) {
return static_cast<float>(x);
}
@@ -242,12 +244,16 @@ static inline int sk_float_nextlog2(float x) {
// IEEE defines how float divide behaves for non-finite values and zero-denoms, but C does not
// so we have a helper that suppresses the possible undefined-behavior warnings.
+#if defined(__clang__)
SK_NO_SANITIZE("float-divide-by-zero")
+#endif
static inline float sk_ieee_float_divide(float numer, float denom) {
return numer / denom;
}
+#if defined(__clang__)
SK_NO_SANITIZE("float-divide-by-zero")
+#endif
static inline double sk_ieee_double_divide(double numer, double denom) {
return numer / denom;
}