gmbr3/chromium-beta
SHA256
1
0
Fork 0
forked from chromium/chromium-beta
Code Pull Requests Activity
Files
main
chromium-beta/ppc-skia-revert-2.patch
Ruediger Oertel 7fbe7a4cf1 add ppc-skia-revert*
2025-05-16 13:41:17 +02:00

273 lines
11 KiB
Diff
Raw Permalink Blame History

From c759c1a4717afadfedab300b6314f428b05741ac Mon Sep 17 00:00:00 2001
From: Kaylee Lubick <kjlubick@google.com>
Date: Mon, 21 Apr 2025 14:51:16 -0400
Subject: [PATCH] Remove alignment assumptions in highp gather
Follow-up to http://review.skia.org/981638
We didn't have any reports that this was causing a problem
but I didn't want to leave the same root cause lurking.
Change-Id: I0d42c4abf4d642f5a33e094f88bde7bf387d2f2f
Bug: b/409859319
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/983176
Reviewed-by: Daniel Dilan <danieldilan@google.com>
diff --git a/src/opts/SkRasterPipeline_opts.h b/src/opts/SkRasterPipeline_opts.h
index d14df3677c..fb644ba5e3 100644
--- a/src/opts/SkRasterPipeline_opts.h
+++ b/src/opts/SkRasterPipeline_opts.h
@@ -161,6 +161,8 @@ namespace SK_OPTS_NS {
template <typename T>
SI T gather(const T* p, U32 ix) { return p[ix]; }
+ template <typename T>
+ SI T gather_unaligned(const T* p, U32 ix) { return gather<T>(p, ix); }
SI void scatter_masked(I32 src, int* dst, U32 ix, I32 mask) {
dst[ix] = mask ? src : dst[ix];
@@ -274,9 +276,23 @@ namespace SK_OPTS_NS {
#endif
template <typename T>
- SI V<T> gather(const T* p, U32 ix) {
- return V<T>{p[ix[0]], p[ix[1]], p[ix[2]], p[ix[3]]};
+ SI V<T> gather(const T* ptr, U32 ix) {
+ // The compiler assumes ptr is aligned, which caused crashes on some
+ // arm32 chips because a register was marked as "aligned to 32 bits"
+ // incorrectly. https://crbug.com/skia/409859319
+ SkASSERTF(reinterpret_cast<uintptr_t>(ptr) % alignof(T) == 0,
+ "Should use gather_unaligned");
+ return V<T>{ptr[ix[0]], ptr[ix[1]], ptr[ix[2]], ptr[ix[3]]};
+ }
+ template <typename T>
+ SI V<T> gather_unaligned(const T* ptr, U32 ix) {
+ // This tells the compiler ptr might not be aligned appropriately, so
+ // it generates better assembly.
+ typedef T __attribute__ ((aligned (1))) unaligned_ptr;
+ const unaligned_ptr* uptr = static_cast<const unaligned_ptr*>(ptr);
+ return V<T>{uptr[ix[0]], uptr[ix[1]], uptr[ix[2]], uptr[ix[3]]};
}
+
SI void scatter_masked(I32 src, int* dst, U32 ix, I32 mask) {
I32 before = gather(dst, ix);
I32 after = if_then_else(mask, src, before);
@@ -389,6 +405,11 @@ namespace SK_OPTS_NS {
};
return sk_bit_cast<U64>(parts);
}
+ template <typename T>
+ SI V<T> gather_unaligned(const T* p, U32 ix) {
+ return gather(p, ix);
+ }
+
template <typename V, typename S>
SI void scatter_masked(V src, S* dst, U32 ix, I32 mask) {
V before = gather(dst, ix);
@@ -627,6 +648,11 @@ namespace SK_OPTS_NS {
};
return sk_bit_cast<U64>(parts);
}
+ template <typename T>
+ SI V<T> gather_unaligned(const T* p, U32 ix) {
+ return gather(p, ix);
+ }
+
SI void scatter_masked(I32 src, int* dst, U32 ix, I32 mask) {
I32 before = gather(dst, ix);
I32 after = if_then_else(mask, src, before);
@@ -827,6 +853,10 @@ namespace SK_OPTS_NS {
SI V<T> gather(const T* p, U32 ix) {
return V<T>{p[ix[0]], p[ix[1]], p[ix[2]], p[ix[3]]};
}
+ template <typename T>
+ SI V<T> gather_unaligned(const T* p, U32 ix) {
+ return gather(p, ix);
+ }
SI void scatter_masked(I32 src, int* dst, U32 ix, I32 mask) {
I32 before = gather(dst, ix);
I32 after = if_then_else(mask, src, before);
@@ -988,6 +1018,10 @@ namespace SK_OPTS_NS {
return V<T>{ p[ix[0]], p[ix[1]], p[ix[2]], p[ix[3]],
p[ix[4]], p[ix[5]], p[ix[6]], p[ix[7]], };
}
+ template <typename T>
+ SI V<T> gather_unaligned(const T* p, U32 ix) {
+ return gather(p, ix);
+ }
template <typename V, typename S>
SI void scatter_masked(V src, S* dst, U32 ix, I32 mask) {
@@ -1195,6 +1229,10 @@ namespace SK_OPTS_NS {
ret = (F)__lsx_vinsgr2vr_w(ret, p[ix3], 3);
return ret;
}
+ template <typename T>
+ SI V<T> gather_unaligned(const T* p, U32 ix) {
+ return gather(p, ix);
+ }
template <typename V, typename S>
SI void scatter_masked(V src, S* dst, U32 ix, I32 mask) {
@@ -2854,7 +2892,7 @@ HIGHP_STAGE(load_565_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_565, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint16_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r,g);
- from_565(gather(ptr, ix), &r,&g,&b);
+ from_565(gather_unaligned(ptr, ix), &r,&g,&b);
a = F1;
}
HIGHP_STAGE(store_565, const SkRasterPipelineContexts::MemoryCtx* ctx) {
@@ -2877,7 +2915,7 @@ HIGHP_STAGE(load_4444_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_4444, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint16_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r,g);
- from_4444(gather(ptr, ix), &r,&g,&b,&a);
+ from_4444(gather_unaligned(ptr, ix), &r,&g,&b,&a);
}
HIGHP_STAGE(store_4444, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<uint16_t>(ctx, dx,dy);
@@ -2899,7 +2937,7 @@ HIGHP_STAGE(load_8888_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_8888, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint32_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r,g);
- from_8888(gather(ptr, ix), &r,&g,&b,&a);
+ from_8888(gather_unaligned(ptr, ix), &r,&g,&b,&a);
}
HIGHP_STAGE(store_8888, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<uint32_t>(ctx, dx,dy);
@@ -2926,7 +2964,7 @@ HIGHP_STAGE(load_rg88_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_rg88, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint16_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- from_88(gather(ptr, ix), &r, &g);
+ from_88(gather_unaligned(ptr, ix), &r, &g);
b = F0;
a = F1;
}
@@ -2950,7 +2988,7 @@ HIGHP_STAGE(gather_a16, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint16_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
r = g = b = F0;
- a = from_short(gather(ptr, ix));
+ a = from_short(gather_unaligned(ptr, ix));
}
HIGHP_STAGE(store_a16, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<uint16_t>(ctx, dx,dy);
@@ -2974,7 +3012,7 @@ HIGHP_STAGE(load_rg1616_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_rg1616, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint32_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- from_1616(gather(ptr, ix), &r, &g);
+ from_1616(gather_unaligned(ptr, ix), &r, &g);
b = F0;
a = F1;
}
@@ -2997,7 +3035,7 @@ HIGHP_STAGE(load_16161616_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_16161616, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint64_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- from_16161616(gather(ptr, ix), &r, &g, &b, &a);
+ from_16161616(gather_unaligned(ptr, ix), &r, &g, &b, &a);
}
HIGHP_STAGE(store_16161616, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<uint16_t>(ctx, 4*dx,4*dy);
@@ -3021,7 +3059,7 @@ HIGHP_STAGE(load_10x6_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_10x6, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint64_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- from_10x6(gather(ptr, ix), &r, &g, &b, &a);
+ from_10x6(gather_unaligned(ptr, ix), &r, &g, &b, &a);
}
HIGHP_STAGE(store_10x6, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<uint16_t>(ctx, 4*dx,4*dy);
@@ -3053,17 +3091,17 @@ HIGHP_STAGE(load_1010102_xr_dst, const SkRasterPipelineContexts::MemoryCtx* ctx)
HIGHP_STAGE(gather_1010102, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint32_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r,g);
- from_1010102(gather(ptr, ix), &r,&g,&b,&a);
+ from_1010102(gather_unaligned(ptr, ix), &r,&g,&b,&a);
}
HIGHP_STAGE(gather_1010102_xr, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint32_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- from_1010102_xr(gather(ptr, ix), &r,&g,&b,&a);
+ from_1010102_xr(gather_unaligned(ptr, ix), &r,&g,&b,&a);
}
HIGHP_STAGE(gather_10101010_xr, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint64_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- from_10101010_xr(gather(ptr, ix), &r, &g, &b, &a);
+ from_10101010_xr(gather_unaligned(ptr, ix), &r, &g, &b, &a);
}
HIGHP_STAGE(load_10101010_xr, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<const uint64_t>(ctx, dx, dy);
@@ -3127,7 +3165,7 @@ HIGHP_STAGE(load_f16_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_f16, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint64_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r,g);
- auto px = gather(ptr, ix);
+ auto px = gather_unaligned(ptr, ix);
U16 R,G,B,A;
load4((const uint16_t*)&px, &R,&G,&B,&A);
@@ -3164,7 +3202,7 @@ HIGHP_STAGE(gather_af16, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint16_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
r = g = b = F0;
- a = from_half(gather(ptr, ix));
+ a = from_half(gather_unaligned(ptr, ix));
}
HIGHP_STAGE(store_af16, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<uint16_t>(ctx, dx,dy);
@@ -3194,7 +3232,7 @@ HIGHP_STAGE(load_rgf16_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_rgf16, const SkRasterPipelineContexts::GatherCtx* ctx) {
const uint32_t* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r, g);
- auto px = gather(ptr, ix);
+ auto px = gather_unaligned(ptr, ix);
U16 R,G;
load2((const uint16_t*)&px, &R, &G);
@@ -3220,10 +3258,10 @@ HIGHP_STAGE(load_f32_dst, const SkRasterPipelineContexts::MemoryCtx* ctx) {
HIGHP_STAGE(gather_f32, const SkRasterPipelineContexts::GatherCtx* ctx) {
const float* ptr;
U32 ix = ix_and_ptr(&ptr, ctx, r,g);
- r = gather(ptr, 4*ix + 0);
- g = gather(ptr, 4*ix + 1);
- b = gather(ptr, 4*ix + 2);
- a = gather(ptr, 4*ix + 3);
+ r = gather_unaligned(ptr, 4*ix + 0);
+ g = gather_unaligned(ptr, 4*ix + 1);
+ b = gather_unaligned(ptr, 4*ix + 2);
+ a = gather_unaligned(ptr, 4*ix + 3);
}
HIGHP_STAGE(store_f32, const SkRasterPipelineContexts::MemoryCtx* ctx) {
auto ptr = ptr_at_xy<float>(ctx, 4*dx,4*dy);
@@ -5009,7 +5047,7 @@ HIGHP_STAGE(bilerp_clamp_8888, const SkRasterPipelineContexts::GatherCtx* ctx) {
U32 ix = ix_and_ptr(&ptr, ctx, x,y);
F sr,sg,sb,sa;
- from_8888(gather(ptr, ix), &sr,&sg,&sb,&sa);
+ from_8888(gather_unaligned(ptr, ix), &sr,&sg,&sb,&sa);
// In bilinear interpolation, the 4 pixels at +/- 0.5 offsets from the sample pixel center
// are combined in direct proportion to their area overlapping that logical query pixel.
@@ -5061,7 +5099,7 @@ HIGHP_STAGE(bicubic_clamp_8888, const SkRasterPipelineContexts::GatherCtx* ctx)
U32 ix = ix_and_ptr(&ptr, ctx, sample_x, sample_y);
F sr,sg,sb,sa;
- from_8888(gather(ptr, ix), &sr,&sg,&sb,&sa);
+ from_8888(gather_unaligned(ptr, ix), &sr,&sg,&sb,&sa);
r = mad(scale, sr, r);
g = mad(scale, sg, g);
@@ -5959,7 +5997,6 @@ SI void store(T* ptr, V v) {
}
#endif
-
// ~~~~~~ 32-bit memory loads and stores ~~~~~~ //
SI void from_8888(U32 rgba, U16* r, U16* g, U16* b, U16* a) {
View Git Blame Copy Permalink