diff options
| author | Frank Barchard <fbarchard@google.com> | 2023-03-24 02:18:55 -0700 |
|---|---|---|
| committer | libyuv LUCI CQ <libyuv-scoped@luci-project-accounts.iam.gserviceaccount.com> | 2023-03-24 10:59:40 +0000 |
| commit | 1a971f8cc3513766f3497ed73e613217b860935d (patch) | |
| tree | c4f32278899be535c7ca48baf20e8cd0c97f0ffb /source/row_neon.cc | |
| parent | 3f219a3501e555ffef7aeaa90abbaf90a90d2258 (diff) | |
| download | libyuv-1a971f8cc3513766f3497ed73e613217b860935d.tar.gz | |
clang 17 -flto-thin bug fix for Neon YUVtoRGB and ARGBToRGB565Dither
- YUV to RGB AArch32 kRGBCoeffBias rewind pointer
- ARGBToRGB565Dither declare width and source pointers as modified
Bug: chromium:1424089
Change-Id: I987180652331bab16ce27d8d166399a687ee890e
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/4370099
Reviewed-by: Mirko Bonadei <mbonadei@chromium.org>
Commit-Queue: Frank Barchard <fbarchard@chromium.org>
Diffstat (limited to 'source/row_neon.cc')
| -rw-r--r-- | source/row_neon.cc | 25 |
1 files changed, 13 insertions, 12 deletions
diff --git a/source/row_neon.cc b/source/row_neon.cc index b3e800d9..bd45082f 100644 --- a/source/row_neon.cc +++ b/source/row_neon.cc @@ -91,10 +91,11 @@ extern "C" { #define YUVTORGB_SETUP \ "vld4.8 {d26[], d27[], d28[], d29[]}, [%[kUVCoeff]] \n" \ - "vld1.16 {d31[]}, [%[kRGBCoeffBias]]! \n" \ - "vld1.16 {d20[], d21[]}, [%[kRGBCoeffBias]]! \n" \ - "vld1.16 {d22[], d23[]}, [%[kRGBCoeffBias]]! \n" \ - "vld1.16 {d24[], d25[]}, [%[kRGBCoeffBias]] \n" + "vld1.16 {d31[]}, [%[kRGBCoeffBias]]! \n" \ + "vld1.16 {d20[], d21[]}, [%[kRGBCoeffBias]]! \n" \ + "vld1.16 {d22[], d23[]}, [%[kRGBCoeffBias]]! \n" \ + "vld1.16 {d24[], d25[]}, [%[kRGBCoeffBias]] \n" \ + "sub %[kRGBCoeffBias], %[kRGBCoeffBias], #10 \n" // q0: B uint16x8_t // q1: G uint16x8_t @@ -1754,20 +1755,20 @@ void ARGBToRGB565DitherRow_NEON(const uint8_t* src_argb, const uint32_t dither4, int width) { asm volatile( - "vdup.32 d7, %2 \n" // dither4 + "vdup.32 d7, %3 \n" // dither4 "1: \n" - "vld4.8 {d0, d2, d4, d6}, [%1]! \n" // load 8 pixels of ARGB. - "subs %3, %3, #8 \n" // 8 processed per loop. + "vld4.8 {d0, d2, d4, d6}, [%0]! \n" // load 8 pixels of ARGB. + "subs %2, %2, #8 \n" // 8 processed per loop. "vqadd.u8 d0, d0, d7 \n" "vqadd.u8 d2, d2, d7 \n" "vqadd.u8 d4, d4, d7 \n" // add for dither ARGBTORGB565 - "vst1.8 {q2}, [%0]! \n" // store 8 RGB565. + "vst1.8 {q2}, [%1]! \n" // store 8 RGB565. "bgt 1b \n" - : "+r"(dst_rgb) // %0 - : "r"(src_argb), // %1 - "r"(dither4), // %2 - "r"(width) // %3 + : "+r"(src_argb), // %0 + "+r"(dst_rgb), // %1 + "+r"(width) // %2 + : "r"(dither4) // %3 : "cc", "memory", "q0", "q1", "q2", "q3"); } |