[PATCH] D156221: [RISCV] Support overloaded version ntlh intrinsic function

[Piyou Chen via Phabricator via cfe-commits]

BeMg created this revision.
Herald added subscribers: jobnoorman, luke, VincentWu, vkmr, frasercrmck, luismarques, apazos, sameer.abuasal, s.egerton, Jim, benna, psnobl, jocewei, PkmX, the_o, brucehoult, MartinMosbeck, rogfer01, edward-jones, zzheng, jrtc27, shiva0217, kito-cheng, niosHD, sabuasal, simoncook, johnrusso, rbar, asb, arichardson.
Herald added a project: All.
BeMg requested review of this revision.
Herald added subscribers: cfe-commits, wangpc, eopXD, MaskRay.
Herald added a project: clang.

Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D156221

Files:
  clang/lib/Headers/riscv_ntlh.h
  clang/test/CodeGen/RISCV/ntlh-intrinsics/riscv32-zihintntl.c


Index: clang/test/CodeGen/RISCV/ntlh-intrinsics/riscv32-zihintntl.c
===================================================================
--- clang/test/CodeGen/RISCV/ntlh-intrinsics/riscv32-zihintntl.c
+++ clang/test/CodeGen/RISCV/ntlh-intrinsics/riscv32-zihintntl.c
@@ -100,6 +100,24 @@
   *scvs1 = __riscv_ntl_load(scvs2, __RISCV_NTLH_ALL);   // CHECK: load <vscale x 4 x i16>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
   *scvc1 = __riscv_ntl_load(scvc2, __RISCV_NTLH_ALL);   // CHECK: load <vscale x 8 x i8>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
 
+  uc = __riscv_ntl_load(&sc);   // CHECK: load i8{{.*}}align 1, !nontemporal !4, !riscv-nontemporal-domain !8
+  sc = __riscv_ntl_load(&uc);   // CHECK: load i8{{.*}}align 1, !nontemporal !4, !riscv-nontemporal-domain !8
+  us = __riscv_ntl_load(&ss);   // CHECK: load i16{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !8
+  ss = __riscv_ntl_load(&us);   // CHECK: load i16{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !8
+  ui = __riscv_ntl_load(&si);   // CHECK: load i32{{.*}}align 4, !nontemporal !4, !riscv-nontemporal-domain !8
+  si = __riscv_ntl_load(&ui);   // CHECK: load i32{{.*}}align 4, !nontemporal !4, !riscv-nontemporal-domain !8
+  ull = __riscv_ntl_load(&sll); // CHECK: load i64{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  sll = __riscv_ntl_load(&ull); // CHECK: load i64{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  h1 = __riscv_ntl_load(&h2);   // CHECK: load half{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !8
+  f1 = __riscv_ntl_load(&f2);   // CHECK: load float{{.*}}align 4, !nontemporal !4, !riscv-nontemporal-domain !8
+  d1 = __riscv_ntl_load(&d2);   // CHECK: load double{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  v4si1 = __riscv_ntl_load(&v4si2);   // CHECK: load <4 x i32>{{.*}}align 16, !nontemporal !4, !riscv-nontemporal-domain !8
+  v8ss1 = __riscv_ntl_load(&v8ss2);   // CHECK: load <8 x i16>{{.*}}align 16, !nontemporal !4, !riscv-nontemporal-domain !8
+  v16sc1 = __riscv_ntl_load(&v16sc2);   // CHECK: load <16 x i8>{{.*}}align 16, !nontemporal !4, !riscv-nontemporal-domain !8
+  *scvi1 = __riscv_ntl_load(scvi2);   // CHECK: load <vscale x 2 x i32>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  *scvs1 = __riscv_ntl_load(scvs2);   // CHECK: load <vscale x 4 x i16>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  *scvc1 = __riscv_ntl_load(scvc2);   // CHECK: load <vscale x 8 x i8>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+
   __riscv_ntl_store(&uc, 1, __RISCV_NTLH_INNERMOST_PRIVATE);    // CHECK: store i8{{.*}}align 1, !nontemporal !4, !riscv-nontemporal-domain !5
   __riscv_ntl_store(&sc, 1, __RISCV_NTLH_INNERMOST_PRIVATE);    // CHECK: store i8{{.*}}align 1, !nontemporal !4, !riscv-nontemporal-domain !5
   __riscv_ntl_store(&us, 1, __RISCV_NTLH_INNERMOST_PRIVATE);    // CHECK: store i16{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !5
@@ -172,6 +190,23 @@
   __riscv_ntl_store(scvs2, *scvs1, __RISCV_NTLH_ALL);  // CHECK: store <vscale x 4 x i16>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
   __riscv_ntl_store(scvc2, *scvc1, __RISCV_NTLH_ALL);  // CHECK: store <vscale x 8 x i8>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
 
+  __riscv_ntl_store(&uc, 1);    // CHECK: store i8{{.*}}align 1, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&sc, 1);    // CHECK: store i8{{.*}}align 1, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&us, 1);    // CHECK: store i16{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&ss, 1);    // CHECK: store i16{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&ui, 1);    // CHECK: store i32{{.*}}align 4, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&si, 1);    // CHECK: store i32{{.*}}align 4, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&ull, 1);   // CHECK: store i64{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&sll, 1);   // CHECK: store i64{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&h1, 1.0);  // CHECK: store half{{.*}}align 2, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&f1, 1.0);  // CHECK: store float{{.*}}align 4, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&d1, 1.0);  // CHECK: store double{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&v4si1, v4si2);  // CHECK: store <4 x i32>{{.*}}align 16, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&v8ss1, v8ss2);  // CHECK: store <8 x i16>{{.*}}align 16, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(&v16sc1, v16sc2);  // CHECK: store <16 x i8>{{.*}}align 16, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(scvi2, *scvi1);  // CHECK: store <vscale x 2 x i32>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(scvs2, *scvs1);  // CHECK: store <vscale x 4 x i16>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
+  __riscv_ntl_store(scvc2, *scvc1);  // CHECK: store <vscale x 8 x i8>{{.*}}align 8, !nontemporal !4, !riscv-nontemporal-domain !8
 }
 // clang-format on
 
Index: clang/lib/Headers/riscv_ntlh.h
===================================================================
--- clang/lib/Headers/riscv_ntlh.h
+++ clang/lib/Headers/riscv_ntlh.h
@@ -21,8 +21,25 @@
   __RISCV_NTLH_ALL
 };
 
-#define __riscv_ntl_load(PTR, DOMAIN) __builtin_riscv_ntl_load((PTR), (DOMAIN))
-#define __riscv_ntl_store(PTR, VAL, DOMAIN)                                    \
+#define __riscv_ntl_load_with_domain(PTR, DOMAIN)                              \
+  __builtin_riscv_ntl_load((PTR), (DOMAIN))
+#define __riscv_ntl_load_without_domain(PTR)                                   \
+  __builtin_riscv_ntl_load((PTR), __RISCV_NTLH_ALL)
+#define SELECT_NTL_LOAD(_1, _2, NAME, ...) NAME
+
+#define __riscv_ntl_load(...)                                                  \
+  SELECT_NTL_LOAD(__VA_ARGS__, __riscv_ntl_load_with_domain,                   \
+                  __riscv_ntl_load_without_domain)                             \
+  (__VA_ARGS__)
+
+#define __riscv_ntl_store_with_domain(PTR, VAL, DOMAIN)                        \
   __builtin_riscv_ntl_store((PTR), (VAL), (DOMAIN))
+#define __riscv_ntl_store_without_domain(PTR, VAL)                             \
+  __builtin_riscv_ntl_store((PTR), (VAL), __RISCV_NTLH_ALL)
+#define SELECT_NTL_STORE(_1, _2, _3, NAME, ...) NAME
 
+#define __riscv_ntl_store(...)                                                 \
+  SELECT_NTL_STORE(__VA_ARGS__, __riscv_ntl_store_with_domain,                 \
+                   __riscv_ntl_store_without_domain)                           \
+  (__VA_ARGS__)
 #endif
\ No newline at end of file


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D156221.543900.patch
Type: text/x-patch
Size: 7002 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20230725/96359696/attachment-0001.bin>