authorgravatar for mlugg@mlugg.co.ukMatthew Lugg <mlugg@mlugg.co.uk> 2025-09-26 16:57:23+01:00
committergravatar for jacobly@ziglang.orgJacob Young <jacobly@ziglang.org> 2025-09-27 18:30:52-04:00
log77fca1652fce295fe0dd7c97432c23b0c4483724
tree46bc9c86038a4ad039b90cf1bdbec17e4a18d4d3
parent0c476191a4ad3514b2d10ba083ba2f86fee9b223

x86_64: fix miscompilation of `mul` on vectors of large ints

1 files changed, 40 insertions(+), 40 deletions(-)

src/codegen/x86_64/CodeGen.zig+40-40
...@@ -23288,10 +23288,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -23288,10 +23288,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
23288 .dst_temps = .{ .mem, .unused },23288 .dst_temps = .{ .mem, .unused },
23289 .clobbers = .{ .eflags = true },23289 .clobbers = .{ .eflags = true },
23290 .each = .{ .once = &.{23290 .each = .{ .once = &.{
23291 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },23291 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
23292 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },23292 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
23293 .{ ._, ._, .lea, .tmp2p, .memia(.src1, .tmp0, .add_size), ._, ._ },23293 .{ ._, ._, .lea, .tmp2p, .memia(.src1, .tmp0, .add_unaligned_size), ._, ._ },
23294 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },23294 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
23295 .{ ._, ._, .mov, .tmp4d, .sia(-8, .src0, .add_elem_size), ._, ._ },23295 .{ ._, ._, .mov, .tmp4d, .sia(-8, .src0, .add_elem_size), ._, ._ },
23296 .{ .@"1:", ._, .xor, .tmp5d, .tmp5d, ._, ._ },23296 .{ .@"1:", ._, .xor, .tmp5d, .tmp5d, ._, ._ },
23297 .{ ._, ._, .@"or", .tmp5q, .leai(.tmp1q, .tmp4), ._, ._ },23297 .{ ._, ._, .@"or", .tmp5q, .leai(.tmp1q, .tmp4), ._, ._ },
...@@ -23339,10 +23339,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -23339,10 +23339,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
23339 .dst_temps = .{ .mem, .unused },23339 .dst_temps = .{ .mem, .unused },
23340 .clobbers = .{ .eflags = true },23340 .clobbers = .{ .eflags = true },
23341 .each = .{ .once = &.{23341 .each = .{ .once = &.{
23342 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },23342 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
23343 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },23343 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
23344 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },23344 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
23345 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },23345 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
23346 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },23346 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
23347 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },23347 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
23348 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },23348 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -23393,10 +23393,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -23393,10 +23393,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
23393 .dst_temps = .{ .mem, .unused },23393 .dst_temps = .{ .mem, .unused },
23394 .clobbers = .{ .eflags = true },23394 .clobbers = .{ .eflags = true },
23395 .each = .{ .once = &.{23395 .each = .{ .once = &.{
23396 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },23396 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
23397 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },23397 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
23398 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },23398 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
23399 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },23399 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
23400 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },23400 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
23401 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },23401 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
23402 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },23402 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -23447,10 +23447,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -23447,10 +23447,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
23447 .dst_temps = .{ .mem, .unused },23447 .dst_temps = .{ .mem, .unused },
23448 .clobbers = .{ .eflags = true },23448 .clobbers = .{ .eflags = true },
23449 .each = .{ .once = &.{23449 .each = .{ .once = &.{
23450 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },23450 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
23451 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },23451 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
23452 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },23452 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
23453 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },23453 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
23454 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },23454 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
23455 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },23455 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
23456 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },23456 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -23502,10 +23502,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -23502,10 +23502,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
23502 .dst_temps = .{ .mem, .unused },23502 .dst_temps = .{ .mem, .unused },
23503 .clobbers = .{ .eflags = true },23503 .clobbers = .{ .eflags = true },
23504 .each = .{ .once = &.{23504 .each = .{ .once = &.{
23505 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },23505 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
23506 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },23506 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
23507 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },23507 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
23508 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },23508 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
23509 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },23509 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
23510 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },23510 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
23511 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },23511 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -26107,10 +26107,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -26107,10 +26107,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
26107 .dst_temps = .{ .mem, .unused },26107 .dst_temps = .{ .mem, .unused },
26108 .clobbers = .{ .eflags = true },26108 .clobbers = .{ .eflags = true },
26109 .each = .{ .once = &.{26109 .each = .{ .once = &.{
26110 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },26110 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
26111 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },26111 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
26112 .{ ._, ._, .lea, .tmp2p, .memia(.src1, .tmp0, .add_size), ._, ._ },26112 .{ ._, ._, .lea, .tmp2p, .memia(.src1, .tmp0, .add_unaligned_size), ._, ._ },
26113 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },26113 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
26114 .{ ._, ._, .mov, .tmp4d, .sia(-8, .src0, .add_elem_size), ._, ._ },26114 .{ ._, ._, .mov, .tmp4d, .sia(-8, .src0, .add_elem_size), ._, ._ },
26115 .{ .@"1:", ._, .xor, .tmp5d, .tmp5d, ._, ._ },26115 .{ .@"1:", ._, .xor, .tmp5d, .tmp5d, ._, ._ },
26116 .{ ._, ._, .@"or", .tmp5q, .leai(.tmp1q, .tmp4), ._, ._ },26116 .{ ._, ._, .@"or", .tmp5q, .leai(.tmp1q, .tmp4), ._, ._ },
...@@ -26158,10 +26158,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -26158,10 +26158,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
26158 .dst_temps = .{ .mem, .unused },26158 .dst_temps = .{ .mem, .unused },
26159 .clobbers = .{ .eflags = true },26159 .clobbers = .{ .eflags = true },
26160 .each = .{ .once = &.{26160 .each = .{ .once = &.{
26161 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },26161 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
26162 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },26162 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
26163 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },26163 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
26164 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },26164 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
26165 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },26165 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
26166 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },26166 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
26167 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },26167 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -26212,10 +26212,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -26212,10 +26212,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
26212 .dst_temps = .{ .mem, .unused },26212 .dst_temps = .{ .mem, .unused },
26213 .clobbers = .{ .eflags = true },26213 .clobbers = .{ .eflags = true },
26214 .each = .{ .once = &.{26214 .each = .{ .once = &.{
26215 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },26215 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
26216 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },26216 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
26217 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },26217 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
26218 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },26218 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
26219 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },26219 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
26220 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },26220 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
26221 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },26221 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -26266,10 +26266,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -26266,10 +26266,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
26266 .dst_temps = .{ .mem, .unused },26266 .dst_temps = .{ .mem, .unused },
26267 .clobbers = .{ .eflags = true },26267 .clobbers = .{ .eflags = true },
26268 .each = .{ .once = &.{26268 .each = .{ .once = &.{
26269 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },26269 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
26270 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },26270 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
26271 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },26271 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
26272 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },26272 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
26273 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },26273 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
26274 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },26274 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
26275 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },26275 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },
...@@ -26321,10 +26321,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {...@@ -26321,10 +26321,10 @@ fn genBody(cg: *CodeGen, body: []const Air.Inst.Index) InnerError!void {
26321 .dst_temps = .{ .mem, .unused },26321 .dst_temps = .{ .mem, .unused },
26322 .clobbers = .{ .eflags = true },26322 .clobbers = .{ .eflags = true },
26323 .each = .{ .once = &.{26323 .each = .{ .once = &.{
26324 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_size), ._, ._ },26324 .{ ._, ._, .mov, .tmp0p, .sa(.src0, .sub_unaligned_size), ._, ._ },
26325 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_size), ._, ._ },26325 .{ .@"0:", ._, .lea, .tmp1p, .memia(.src0, .tmp0, .add_unaligned_size), ._, ._ },
26326 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_size, 8), ._, ._ },26326 .{ ._, ._, .lea, .tmp2p, .memiad(.src1, .tmp0, .add_unaligned_size, 8), ._, ._ },
26327 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_size), ._, ._ },26327 .{ ._, ._, .lea, .tmp3p, .memia(.dst0, .tmp0, .add_unaligned_size), ._, ._ },
26328 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },26328 .{ ._, ._, .mov, .tmp4d, .sia(-1, .src0, .add_elem_size_div_8), ._, ._ },
26329 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },26329 .{ .@"1:", ._, .lea, .tmp5p, .leaa(.tmp4, .sub_src0_elem_size_div_8), ._, ._ },
26330 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },26330 .{ ._, ._, .xor, .tmp6d, .tmp6d, ._, ._ },