diff options
| author | Craig Topper <craig.topper@intel.com> | 2018-06-06 00:24:55 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2018-06-06 00:24:55 +0000 |
| commit | f3914b74c1825f4deece3317542743109dabbf56 (patch) | |
| tree | c6f2e80811bbed0f4fb57c2fd9f38ff3058ff11c /clang/lib/Sema | |
| parent | c797bb23b857e4521562073d408be2de8d15a270 (diff) | |
| download | bcm5719-llvm-f3914b74c1825f4deece3317542743109dabbf56.tar.gz bcm5719-llvm-f3914b74c1825f4deece3317542743109dabbf56.zip | |
[X86] Add builtins for vector element insert and extract for different 128 and 256 bit vector types. Use them to implement the extract and insert intrinsics.
Previously we were just using extended vector operations in the header file.
This unfortunately allowed non-constant indices to be used with the intrinsics. This is incompatible with gcc, icc, and MSVC. It also introduces a different performance characteristic because non-constant index gets lowered to a vector store and an element sized load.
By adding the builtins we can check for the index to be a constant and ensure its in range of the vector element count.
User code still has the option to use extended vector operations themselves if they need non-constant indexing.
llvm-svn: 334057
Diffstat (limited to 'clang/lib/Sema')
| -rw-r--r-- | clang/lib/Sema/SemaChecking.cpp | 27 |
1 files changed, 24 insertions, 3 deletions
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp index 643c103c8a3..b4fdcc0f7e9 100644 --- a/clang/lib/Sema/SemaChecking.cpp +++ b/clang/lib/Sema/SemaChecking.cpp @@ -2552,17 +2552,28 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) { switch (BuiltinID) { default: return false; - case X86::BI_mm_prefetch: - i = 1; l = 0; u = 7; - break; case X86::BI__builtin_ia32_vec_ext_v2si: + case X86::BI__builtin_ia32_vec_ext_v2di: i = 1; l = 0; u = 1; break; + case X86::BI__builtin_ia32_vec_set_v2di: + i = 2; l = 0; u = 1; + break; case X86::BI__builtin_ia32_vec_ext_v4hi: + case X86::BI__builtin_ia32_vec_ext_v4si: + case X86::BI__builtin_ia32_vec_ext_v4sf: + case X86::BI__builtin_ia32_vec_ext_v4di: i = 1; l = 0; u = 3; break; + case X86::BI_mm_prefetch: + case X86::BI__builtin_ia32_vec_ext_v8hi: + case X86::BI__builtin_ia32_vec_ext_v8si: + i = 1; l = 0; u = 7; + break; case X86::BI__builtin_ia32_sha1rnds4: case X86::BI__builtin_ia32_vec_set_v4hi: + case X86::BI__builtin_ia32_vec_set_v4si: + case X86::BI__builtin_ia32_vec_set_v4di: i = 2; l = 0; u = 3; break; case X86::BI__builtin_ia32_vpermil2pd: @@ -2603,12 +2614,16 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) { case X86::BI__builtin_ia32_vpcomw: case X86::BI__builtin_ia32_vpcomd: case X86::BI__builtin_ia32_vpcomq: + case X86::BI__builtin_ia32_vec_set_v8hi: + case X86::BI__builtin_ia32_vec_set_v8si: i = 2; l = 0; u = 7; break; case X86::BI__builtin_ia32_roundps: case X86::BI__builtin_ia32_roundpd: case X86::BI__builtin_ia32_roundps256: case X86::BI__builtin_ia32_roundpd256: + case X86::BI__builtin_ia32_vec_ext_v16qi: + case X86::BI__builtin_ia32_vec_ext_v16hi: i = 1; l = 0; u = 15; break; case X86::BI__builtin_ia32_roundss: @@ -2621,8 +2636,13 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) { case X86::BI__builtin_ia32_rangeps512_mask: case X86::BI__builtin_ia32_getmantsd_round_mask: case X86::BI__builtin_ia32_getmantss_round_mask: + case X86::BI__builtin_ia32_vec_set_v16qi: + case X86::BI__builtin_ia32_vec_set_v16hi: i = 2; l = 0; u = 15; break; + case X86::BI__builtin_ia32_vec_ext_v32qi: + i = 1; l = 0; u = 31; + break; case X86::BI__builtin_ia32_cmpps: case X86::BI__builtin_ia32_cmpss: case X86::BI__builtin_ia32_cmppd: @@ -2637,6 +2657,7 @@ bool Sema::CheckX86BuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) { case X86::BI__builtin_ia32_cmppd512_mask: case X86::BI__builtin_ia32_cmpsd_mask: case X86::BI__builtin_ia32_cmpss_mask: + case X86::BI__builtin_ia32_vec_set_v32qi: i = 2; l = 0; u = 31; break; case X86::BI__builtin_ia32_vcvtps2ph: |
