[PATCH v1 0/3] Add LoongarchV1.1 instructions support.

[PATCH v1 0/3] Add LoongarchV1.1 instructions support.

[Lulu Cheng]

Lulu Cheng (3):
  LoongArch: Add LA664 support.
  LoongArch: Implement atomic operations using LoongArch1.1
    instructions.
  LoongArch: atomic_load and atomic_store are implemented using dbar
    grading.

 gcc/config.gcc                                |  10 +-
 .../loongarch/genopts/loongarch-strings       |   1 +
 gcc/config/loongarch/genopts/loongarch.opt.in |   3 +
 gcc/config/loongarch/loongarch-cpu.cc         |   4 +
 gcc/config/loongarch/loongarch-def.c          |  21 ++
 gcc/config/loongarch/loongarch-def.h          |  10 +-
 gcc/config/loongarch/loongarch-opts.cc        |   8 +-
 gcc/config/loongarch/loongarch-opts.h         |   6 +-
 gcc/config/loongarch/loongarch-str.h          |   1 +
 gcc/config/loongarch/loongarch.cc             |   7 +-
 gcc/config/loongarch/loongarch.opt            |   3 +
 gcc/config/loongarch/sync.md                  | 256 ++++++++++++++----
 12 files changed, 263 insertions(+), 67 deletions(-)

-- 
2.31.1

[PATCH v1 1/3] LoongArch: Add LA664 support.

[Lulu Cheng]

Define ISA_BASE_LA64V110, which represents the base instruction set defined in LoongArch1.1.
Support the configure setting --with-arch =la664, and support -march=la664,-mtune=la664.

gcc/ChangeLog:

	* config.gcc: Support LA664.
	* config/loongarch/genopts/loongarch-strings: Likewise.
	* config/loongarch/genopts/loongarch.opt.in: Likewise.
	* config/loongarch/loongarch-cpu.cc (fill_native_cpu_config): Likewise.
	* config/loongarch/loongarch-def.c: Likewise.
	* config/loongarch/loongarch-def.h (N_ISA_BASE_TYPES): Likewise.
	(ISA_BASE_LA64V110): Define macro.
	(N_ARCH_TYPES): Update value.
	(N_TUNE_TYPES): Update value.
	(CPU_LA664): New macro.
	* config/loongarch/loongarch-opts.cc (isa_default_abi): Likewise.
	(isa_base_compat_p): Likewise.
	* config/loongarch/loongarch-opts.h (TARGET_64BIT): This parameter is enabled
	when la_target.isa.base is equal to ISA_BASE_LA64V100 or ISA_BASE_LA64V110.
	(TARGET_uARCH_LA664): Define macro.
	* config/loongarch/loongarch-str.h (STR_CPU_LA664): Likewise.
	* config/loongarch/loongarch.cc (loongarch_cpu_sched_reassociation_width):
	Add LA664 support.
	* config/loongarch/loongarch.opt: Regenerate.
---
 gcc/config.gcc                                | 10 ++++-----
 .../loongarch/genopts/loongarch-strings       |  1 +
 gcc/config/loongarch/genopts/loongarch.opt.in |  3 +++
 gcc/config/loongarch/loongarch-cpu.cc         |  4 ++++
 gcc/config/loongarch/loongarch-def.c          | 21 +++++++++++++++++++
 gcc/config/loongarch/loongarch-def.h          |  8 ++++---
 gcc/config/loongarch/loongarch-opts.cc        |  8 +++----
 gcc/config/loongarch/loongarch-opts.h         |  4 +++-
 gcc/config/loongarch/loongarch-str.h          |  1 +
 gcc/config/loongarch/loongarch.cc             |  1 +
 gcc/config/loongarch/loongarch.opt            |  3 +++
 11 files changed, 51 insertions(+), 13 deletions(-)

diff --git a/gcc/config.gcc b/gcc/config.gcc
index 6d51bd93f3f..b88591b6fd8 100644
--- a/gcc/config.gcc
+++ b/gcc/config.gcc
@@ -5039,7 +5039,7 @@ case "${target}" in
 
 		# Perform initial sanity checks on --with-* options.
 		case ${with_arch} in
-		"" | abi-default | loongarch64 | la464) ;; # OK, append here.
+		"" | abi-default | loongarch64 | la[46]64) ;; # OK, append here.
 		native)
 			if test x${host} != x${target}; then
 				echo "--with-arch=native is illegal for cross-compiler." 1>&2
@@ -5088,7 +5088,7 @@ case "${target}" in
 		case ${abi_base}/${abi_ext} in
 		lp64*/base)
 			# architectures that support lp64* ABI
-			arch_pattern="native|abi-default|loongarch64|la464"
+			arch_pattern="native|abi-default|loongarch64|la[46]64"
 			# default architecture for lp64* ABI
 			arch_default="abi-default"
 			;;
@@ -5163,7 +5163,7 @@ case "${target}" in
 		# Check default with_tune configuration using with_arch.
 		case ${with_arch} in
 		loongarch64)
-			tune_pattern="native|abi-default|loongarch64|la464"
+			tune_pattern="native|abi-default|loongarch64|la[46]64"
 			;;
 		*)
 			# By default, $with_tune == $with_arch
@@ -5219,7 +5219,7 @@ case "${target}" in
 					# Fixed: use the default gcc configuration for all multilib
 					# builds by default.
 					with_multilib_default="" ;;
-				arch,native|arch,loongarch64|arch,la464) # OK, append here.
+				arch,native|arch,loongarch64|arch,la[46]64) # OK, append here.
 					with_multilib_default="/march=${component}" ;;
 				arch,*)
 					with_multilib_default="/march=abi-default"
@@ -5307,7 +5307,7 @@ case "${target}" in
 				if test x${parse_state} = x"arch"; then
 					# -march option
 					case ${component} in
-					native | abi-default | loongarch64 | la464) # OK, append here.
+					native | abi-default | loongarch64 | la[46]64) # OK, append here.
 						# Append -march spec for each multilib variant.
 						loongarch_multilib_list_make="${loongarch_multilib_list_make}/march=${component}"
 						parse_state="opts"
diff --git a/gcc/config/loongarch/genopts/loongarch-strings b/gcc/config/loongarch/genopts/loongarch-strings
index 8e412f7536e..7bc4824007e 100644
--- a/gcc/config/loongarch/genopts/loongarch-strings
+++ b/gcc/config/loongarch/genopts/loongarch-strings
@@ -26,6 +26,7 @@ STR_CPU_NATIVE	      native
 STR_CPU_ABI_DEFAULT   abi-default
 STR_CPU_LOONGARCH64   loongarch64
 STR_CPU_LA464	      la464
+STR_CPU_LA664	      la664
 
 # Base architecture
 STR_ISA_BASE_LA64V100 la64
diff --git a/gcc/config/loongarch/genopts/loongarch.opt.in b/gcc/config/loongarch/genopts/loongarch.opt.in
index 158701d327a..00b4733d75b 100644
--- a/gcc/config/loongarch/genopts/loongarch.opt.in
+++ b/gcc/config/loongarch/genopts/loongarch.opt.in
@@ -107,6 +107,9 @@ Enum(cpu_type) String(@@STR_CPU_LOONGARCH64@@) Value(CPU_LOONGARCH64)
 EnumValue
 Enum(cpu_type) String(@@STR_CPU_LA464@@) Value(CPU_LA464)
 
+EnumValue
+Enum(cpu_type) String(@@STR_CPU_LA664@@) Value(CPU_LA664)
+
 m@@OPTSTR_ARCH@@=
 Target RejectNegative Joined Enum(cpu_type) Var(la_opt_cpu_arch) Init(M_OPT_UNSET)
 -m@@OPTSTR_ARCH@@=PROCESSOR	Generate code for the given PROCESSOR ISA.
diff --git a/gcc/config/loongarch/loongarch-cpu.cc b/gcc/config/loongarch/loongarch-cpu.cc
index 7a2866f60f9..f3a13414143 100644
--- a/gcc/config/loongarch/loongarch-cpu.cc
+++ b/gcc/config/loongarch/loongarch-cpu.cc
@@ -106,6 +106,10 @@ fill_native_cpu_config (struct loongarch_target *tgt)
       native_cpu_type = CPU_LA464;
       break;
 
+    case 0x0014d000:   /* LA664 */
+      native_cpu_type = CPU_LA664;
+      break;
+
     default:
       /* Unknown PRID.  */
       if (tune_native_p)
diff --git a/gcc/config/loongarch/loongarch-def.c b/gcc/config/loongarch/loongarch-def.c
index 430ef8b2d95..067629141b6 100644
--- a/gcc/config/loongarch/loongarch-def.c
+++ b/gcc/config/loongarch/loongarch-def.c
@@ -28,6 +28,7 @@ loongarch_cpu_strings[N_TUNE_TYPES] = {
   [CPU_ABI_DEFAULT]	  = STR_CPU_ABI_DEFAULT,
   [CPU_LOONGARCH64]	  = STR_CPU_LOONGARCH64,
   [CPU_LA464]		  = STR_CPU_LA464,
+  [CPU_LA664]		  = STR_CPU_LA664,
 };
 
 struct loongarch_isa
@@ -42,6 +43,11 @@ loongarch_cpu_default_isa[N_ARCH_TYPES] = {
       .fpu = ISA_EXT_FPU64,
       .simd = ISA_EXT_SIMD_LASX,
   },
+  [CPU_LA664] = {
+      .base = ISA_BASE_LA64V110,
+      .fpu = ISA_EXT_FPU64,
+      .simd = ISA_EXT_SIMD_LASX,
+  },
 };
 
 struct loongarch_cache
@@ -58,6 +64,12 @@ loongarch_cpu_cache[N_TUNE_TYPES] = {
       .l2d_size = 256,
       .simultaneous_prefetches = 4,
   },
+  [CPU_LA664] = {
+      .l1d_line_size = 64,
+      .l1d_size = 64,
+      .l2d_size = 256,
+      .simultaneous_prefetches = 4,
+  },
 };
 
 struct loongarch_align
@@ -70,6 +82,10 @@ loongarch_cpu_align[N_TUNE_TYPES] = {
     .function = "32",
     .label = "16",
   },
+  [CPU_LA664] = {
+    .function = "32",
+    .label = "16",
+  },
 };
 
 
@@ -104,6 +120,9 @@ loongarch_cpu_rtx_cost_data[N_TUNE_TYPES] = {
   [CPU_LA464] = {
       DEFAULT_COSTS
   },
+  [CPU_LA664] = {
+      DEFAULT_COSTS
+  },
 };
 
 /* RTX costs to use when optimizing for size.  */
@@ -127,6 +146,7 @@ loongarch_cpu_issue_rate[N_TUNE_TYPES] = {
   [CPU_NATIVE]	      = 4,
   [CPU_LOONGARCH64]   = 4,
   [CPU_LA464]	      = 4,
+  [CPU_LA664]	      = 6,
 };
 
 int
@@ -134,6 +154,7 @@ loongarch_cpu_multipass_dfa_lookahead[N_TUNE_TYPES] = {
   [CPU_NATIVE]	      = 4,
   [CPU_LOONGARCH64]   = 4,
   [CPU_LA464]	      = 4,
+  [CPU_LA664]	      = 6,
 };
 
 /* Wiring string definitions from loongarch-str.h to global arrays
diff --git a/gcc/config/loongarch/loongarch-def.h b/gcc/config/loongarch/loongarch-def.h
index 6e2a6987910..db497f3ffe2 100644
--- a/gcc/config/loongarch/loongarch-def.h
+++ b/gcc/config/loongarch/loongarch-def.h
@@ -55,7 +55,8 @@ extern "C" {
 /* enum isa_base */
 extern const char* loongarch_isa_base_strings[];
 #define ISA_BASE_LA64V100     0
-#define N_ISA_BASE_TYPES      1
+#define ISA_BASE_LA64V110     1
+#define N_ISA_BASE_TYPES      2
 
 /* enum isa_ext_* */
 extern const char* loongarch_isa_ext_strings[];
@@ -141,8 +142,9 @@ struct loongarch_target
 #define CPU_ABI_DEFAULT   1
 #define CPU_LOONGARCH64	  2
 #define CPU_LA464	  3
-#define N_ARCH_TYPES	  4
-#define N_TUNE_TYPES	  4
+#define CPU_LA664	  4
+#define N_ARCH_TYPES	  5
+#define N_TUNE_TYPES	  5
 
 /* parallel tables.  */
 extern const char* loongarch_cpu_strings[];
diff --git a/gcc/config/loongarch/loongarch-opts.cc b/gcc/config/loongarch/loongarch-opts.cc
index e5921189a06..67a59152a01 100644
--- a/gcc/config/loongarch/loongarch-opts.cc
+++ b/gcc/config/loongarch/loongarch-opts.cc
@@ -552,17 +552,17 @@ isa_default_abi (const struct loongarch_isa *isa)
   switch (isa->fpu)
     {
       case ISA_EXT_FPU64:
-	if (isa->base == ISA_BASE_LA64V100)
+	if (isa->base >= ISA_BASE_LA64V100)
 	  abi.base = ABI_BASE_LP64D;
 	break;
 
       case ISA_EXT_FPU32:
-	if (isa->base == ISA_BASE_LA64V100)
+	if (isa->base >= ISA_BASE_LA64V100)
 	  abi.base = ABI_BASE_LP64F;
 	break;
 
       case ISA_EXT_NONE:
-	if (isa->base == ISA_BASE_LA64V100)
+	if (isa->base >= ISA_BASE_LA64V100)
 	  abi.base = ABI_BASE_LP64S;
 	break;
 
@@ -582,7 +582,7 @@ isa_base_compat_p (const struct loongarch_isa *set1,
   switch (set2->base)
     {
       case ISA_BASE_LA64V100:
-	return (set1->base == ISA_BASE_LA64V100);
+	return (set1->base >= ISA_BASE_LA64V100);
 
       default:
 	gcc_unreachable ();
diff --git a/gcc/config/loongarch/loongarch-opts.h b/gcc/config/loongarch/loongarch-opts.h
index 8de41bbc4f7..bd2e86a5aa7 100644
--- a/gcc/config/loongarch/loongarch-opts.h
+++ b/gcc/config/loongarch/loongarch-opts.h
@@ -76,7 +76,8 @@ loongarch_update_gcc_opt_status (struct loongarch_target *target,
 #define TARGET_DOUBLE_FLOAT	  (la_target.isa.fpu == ISA_EXT_FPU64)
 #define TARGET_DOUBLE_FLOAT_ABI	  (la_target.abi.base == ABI_BASE_LP64D)
 
-#define TARGET_64BIT		  (la_target.isa.base == ISA_BASE_LA64V100)
+#define TARGET_64BIT		  (la_target.isa.base == ISA_BASE_LA64V100 \
+				   || la_target.isa.base == ISA_BASE_LA64V110)
 #define TARGET_ABI_LP64		  (la_target.abi.base == ABI_BASE_LP64D	\
 				   || la_target.abi.base == ABI_BASE_LP64F \
 				   || la_target.abi.base == ABI_BASE_LP64S)
@@ -88,6 +89,7 @@ loongarch_update_gcc_opt_status (struct loongarch_target *target,
 
 /* TARGET_ macros for use in *.md template conditionals */
 #define TARGET_uARCH_LA464	  (la_target.cpu_tune == CPU_LA464)
+#define TARGET_uARCH_LA664	  (la_target.cpu_tune == CPU_LA664)
 
 /* Note: optimize_size may vary across functions,
    while -m[no]-memcpy imposes a global constraint.  */
diff --git a/gcc/config/loongarch/loongarch-str.h b/gcc/config/loongarch/loongarch-str.h
index 072558c28f1..fc4f41bfc1e 100644
--- a/gcc/config/loongarch/loongarch-str.h
+++ b/gcc/config/loongarch/loongarch-str.h
@@ -30,6 +30,7 @@ along with GCC; see the file COPYING3.  If not see
 #define STR_CPU_ABI_DEFAULT "abi-default"
 #define STR_CPU_LOONGARCH64 "loongarch64"
 #define STR_CPU_LA464 "la464"
+#define STR_CPU_LA664 "la664"
 
 #define STR_ISA_BASE_LA64V100 "la64"
 
diff --git a/gcc/config/loongarch/loongarch.cc b/gcc/config/loongarch/loongarch.cc
index d05743bec87..464f6c4dd63 100644
--- a/gcc/config/loongarch/loongarch.cc
+++ b/gcc/config/loongarch/loongarch.cc
@@ -10173,6 +10173,7 @@ loongarch_cpu_sched_reassociation_width (struct loongarch_target *target,
     {
     case CPU_LOONGARCH64:
     case CPU_LA464:
+    case CPU_LA664:
       /* Vector part.  */
       if (LSX_SUPPORTED_MODE_P (mode) || LASX_SUPPORTED_MODE_P (mode))
 	{
diff --git a/gcc/config/loongarch/loongarch.opt b/gcc/config/loongarch/loongarch.opt
index a5988411fbb..7f129e53ba5 100644
--- a/gcc/config/loongarch/loongarch.opt
+++ b/gcc/config/loongarch/loongarch.opt
@@ -114,6 +114,9 @@ Enum(cpu_type) String(loongarch64) Value(CPU_LOONGARCH64)
 EnumValue
 Enum(cpu_type) String(la464) Value(CPU_LA464)
 
+EnumValue
+Enum(cpu_type) String(la664) Value(CPU_LA664)
+
 march=
 Target RejectNegative Joined Enum(cpu_type) Var(la_opt_cpu_arch) Init(M_OPT_UNSET)
 -march=PROCESSOR	Generate code for the given PROCESSOR ISA.
-- 
2.31.1

[PATCH v1 2/3] LoongArch: Implement atomic operations using LoongArch1.1 instructions.

[Lulu Cheng]

1. short and char type calls for atomic_add_fetch and __atomic_fetch_add are
   implemented using amadd{_db}.{b/h}.
2. Use amcas{_db}.{b/h/w/d} to implement __atomic_compare_exchange_n and __atomic_compare_exchange.
3. The short and char types of the functions __atomic_exchange and __atomic_exchange_n are
   implemented using amswap{_db}.{b/h}.

gcc/ChangeLog:

	* config/loongarch/loongarch-def.h: Add comments.
	* config/loongarch/loongarch-opts.h (ISA_BASE_IS_LA64V110): Define macro.
	* config/loongarch/loongarch.cc (loongarch_memmodel_needs_rel_acq_fence):
	Remove redundant code implementations.
	* config/loongarch/sync.md (d): Added QI, HI support.
	(atomic_add<mode>): New template.
	(atomic_exchange<mode>_short): Likewise.
	(atomic_cas_value_strong<mode>_amcas): Likewise..
	(atomic_fetch_add<mode>_short): Likewise.
---
 gcc/config/loongarch/loongarch-def.h  |   2 +
 gcc/config/loongarch/loongarch-opts.h |   2 +-
 gcc/config/loongarch/loongarch.cc     |   6 +-
 gcc/config/loongarch/sync.md          | 186 ++++++++++++++++++++------
 4 files changed, 147 insertions(+), 49 deletions(-)

diff --git a/gcc/config/loongarch/loongarch-def.h b/gcc/config/loongarch/loongarch-def.h
index db497f3ffe2..b319cded456 100644
--- a/gcc/config/loongarch/loongarch-def.h
+++ b/gcc/config/loongarch/loongarch-def.h
@@ -54,7 +54,9 @@ extern "C" {
 
 /* enum isa_base */
 extern const char* loongarch_isa_base_strings[];
+/* LoongArch V1.00.  */
 #define ISA_BASE_LA64V100     0
+/* LoongArch V1.10.  */
 #define ISA_BASE_LA64V110     1
 #define N_ISA_BASE_TYPES      2
 
diff --git a/gcc/config/loongarch/loongarch-opts.h b/gcc/config/loongarch/loongarch-opts.h
index bd2e86a5aa7..e8645d9ad5c 100644
--- a/gcc/config/loongarch/loongarch-opts.h
+++ b/gcc/config/loongarch/loongarch-opts.h
@@ -86,10 +86,10 @@ loongarch_update_gcc_opt_status (struct loongarch_target *target,
 				   || la_target.isa.simd == ISA_EXT_SIMD_LASX)
 #define ISA_HAS_LASX		  (la_target.isa.simd == ISA_EXT_SIMD_LASX)
 
-
 /* TARGET_ macros for use in *.md template conditionals */
 #define TARGET_uARCH_LA464	  (la_target.cpu_tune == CPU_LA464)
 #define TARGET_uARCH_LA664	  (la_target.cpu_tune == CPU_LA664)
+#define ISA_BASE_IS_LA64V110	  (la_target.isa.base == ISA_BASE_LA64V110)
 
 /* Note: optimize_size may vary across functions,
    while -m[no]-memcpy imposes a global constraint.  */
diff --git a/gcc/config/loongarch/loongarch.cc b/gcc/config/loongarch/loongarch.cc
index 464f6c4dd63..5bec10d7418 100644
--- a/gcc/config/loongarch/loongarch.cc
+++ b/gcc/config/loongarch/loongarch.cc
@@ -5813,16 +5813,12 @@ loongarch_print_operand_punct_valid_p (unsigned char code)
 static bool
 loongarch_memmodel_needs_rel_acq_fence (enum memmodel model)
 {
-  switch (model)
+  switch (memmodel_base (model))
     {
       case MEMMODEL_ACQ_REL:
       case MEMMODEL_SEQ_CST:
-      case MEMMODEL_SYNC_SEQ_CST:
       case MEMMODEL_RELEASE:
-      case MEMMODEL_SYNC_RELEASE:
       case MEMMODEL_ACQUIRE:
-      case MEMMODEL_CONSUME:
-      case MEMMODEL_SYNC_ACQUIRE:
 	return true;
 
       case MEMMODEL_RELAXED:
diff --git a/gcc/config/loongarch/sync.md b/gcc/config/loongarch/sync.md
index c112091a60f..66e316d80f5 100644
--- a/gcc/config/loongarch/sync.md
+++ b/gcc/config/loongarch/sync.md
@@ -38,7 +38,7 @@ (define_code_attr atomic_optab
   [(plus "add") (ior "or") (xor "xor") (and "and")])
 
 ;; This attribute gives the format suffix for atomic memory operations.
-(define_mode_attr amo [(SI "w") (DI "d")])
+(define_mode_attr amo [(QI "b") (HI "h") (SI "w") (DI "d")])
 
 ;; <amop> expands to the name of the atomic operand that implements a
 ;; particular code.
@@ -123,7 +123,18 @@ (define_insn "atomic_<atomic_optab><mode>"
 	 UNSPEC_SYNC_OLD_OP))]
   ""
   "am<amop>%A2.<amo>\t$zero,%z1,%0"
-  [(set (attr "length") (const_int 8))])
+  [(set (attr "length") (const_int 4))])
+
+(define_insn "atomic_add<mode>"
+  [(set (match_operand:SHORT 0 "memory_operand" "+ZB")
+	(unspec_volatile:SHORT
+	  [(plus:SHORT (match_dup 0)
+		       (match_operand:SHORT 1 "reg_or_0_operand" "rJ"))
+	   (match_operand:SI 2 "const_int_operand")] ;; model
+	 UNSPEC_SYNC_OLD_OP))]
+  "ISA_BASE_IS_LA64V110"
+  "amadd%A2.<amo>\t$zero,%z1,%0"
+  [(set (attr "length") (const_int 4))])
 
 (define_insn "atomic_fetch_<atomic_optab><mode>"
   [(set (match_operand:GPR 0 "register_operand" "=&r")
@@ -131,12 +142,12 @@ (define_insn "atomic_fetch_<atomic_optab><mode>"
    (set (match_dup 1)
 	(unspec_volatile:GPR
 	  [(any_atomic:GPR (match_dup 1)
-		     (match_operand:GPR 2 "reg_or_0_operand" "rJ"))
+			   (match_operand:GPR 2 "reg_or_0_operand" "rJ"))
 	   (match_operand:SI 3 "const_int_operand")] ;; model
 	 UNSPEC_SYNC_OLD_OP))]
   ""
   "am<amop>%A3.<amo>\t%0,%z2,%1"
-  [(set (attr "length") (const_int 8))])
+  [(set (attr "length") (const_int 4))])
 
 (define_insn "atomic_exchange<mode>"
   [(set (match_operand:GPR 0 "register_operand" "=&r")
@@ -148,7 +159,19 @@ (define_insn "atomic_exchange<mode>"
 	(match_operand:GPR 2 "register_operand" "r"))]
   ""
   "amswap%A3.<amo>\t%0,%z2,%1"
-  [(set (attr "length") (const_int 8))])
+  [(set (attr "length") (const_int 4))])
+
+(define_insn "atomic_exchange<mode>_short"
+  [(set (match_operand:SHORT 0 "register_operand" "=&r")
+	(unspec_volatile:SHORT
+	  [(match_operand:SHORT 1 "memory_operand" "+ZB")
+	   (match_operand:SI 3 "const_int_operand")] ;; model
+	  UNSPEC_SYNC_EXCHANGE))
+   (set (match_dup 1)
+	(match_operand:SHORT 2 "register_operand" "r"))]
+  "ISA_BASE_IS_LA64V110"
+  "amswap%A3.<amo>\t%0,%z2,%1"
+  [(set (attr "length") (const_int 4))])
 
 (define_insn "atomic_cas_value_strong<mode>"
   [(set (match_operand:GPR 0 "register_operand" "=&r")
@@ -156,25 +179,36 @@ (define_insn "atomic_cas_value_strong<mode>"
    (set (match_dup 1)
 	(unspec_volatile:GPR [(match_operand:GPR 2 "reg_or_0_operand" "rJ")
 			      (match_operand:GPR 3 "reg_or_0_operand" "rJ")
-			      (match_operand:SI 4 "const_int_operand")  ;; mod_s
-			      (match_operand:SI 5 "const_int_operand")] ;; mod_f
+			      (match_operand:SI 4 "const_int_operand")]  ;; mod_s
 	 UNSPEC_COMPARE_AND_SWAP))
-   (clobber (match_scratch:GPR 6 "=&r"))]
+   (clobber (match_scratch:GPR 5 "=&r"))]
   ""
 {
   return "1:\\n\\t"
 	 "ll.<amo>\\t%0,%1\\n\\t"
 	 "bne\\t%0,%z2,2f\\n\\t"
-	 "or%i3\\t%6,$zero,%3\\n\\t"
-	 "sc.<amo>\\t%6,%1\\n\\t"
-	 "beqz\\t%6,1b\\n\\t"
+	 "or%i3\\t%5,$zero,%3\\n\\t"
+	 "sc.<amo>\\t%5,%1\\n\\t"
+	 "beqz\\t%5,1b\\n\\t"
 	 "b\\t3f\\n\\t"
 	 "2:\\n\\t"
-	 "%G5\\n\\t"
+	 "%G4\\n\\t"
 	 "3:\\n\\t";
 }
   [(set (attr "length") (const_int 28))])
 
+(define_insn "atomic_cas_value_strong<mode>_amcas"
+  [(set (match_operand:QHWD 0 "register_operand" "=&r")
+	(match_operand:QHWD 1 "memory_operand" "+ZB"))
+   (set (match_dup 1)
+	(unspec_volatile:QHWD [(match_operand:QHWD 2 "reg_or_0_operand" "rJ")
+			       (match_operand:QHWD 3 "reg_or_0_operand" "rJ")
+			       (match_operand:SI 4 "const_int_operand")]  ;; mod_s
+	 UNSPEC_COMPARE_AND_SWAP))]
+  "ISA_BASE_IS_LA64V110"
+  "ori\t%0,%z2,0\n\tamcas%A4.<amo>\t%0,%z3,%1"
+  [(set (attr "length") (const_int 8))])
+
 (define_expand "atomic_compare_and_swap<mode>"
   [(match_operand:SI 0 "register_operand" "")   ;; bool output
    (match_operand:GPR 1 "register_operand" "")  ;; val output
@@ -186,9 +220,29 @@ (define_expand "atomic_compare_and_swap<mode>"
    (match_operand:SI 7 "const_int_operand" "")] ;; mod_f
   ""
 {
-  emit_insn (gen_atomic_cas_value_strong<mode> (operands[1], operands[2],
-						operands[3], operands[4],
-						operands[6], operands[7]));
+  rtx mod_s, mod_f;
+
+  mod_s = operands[6];
+  mod_f = operands[7];
+
+  /* Normally the succ memory model must be stronger than fail, but in the
+     unlikely event of fail being ACQUIRE and succ being RELEASE we need to
+     promote succ to ACQ_REL so that we don't lose the acquire semantics.  */
+
+  if (is_mm_acquire (memmodel_base (INTVAL (mod_f)))
+      && is_mm_release (memmodel_base (INTVAL (mod_s))))
+    mod_s = GEN_INT (MEMMODEL_ACQ_REL);
+
+  operands[6] = mod_s;
+
+  if (ISA_BASE_IS_LA64V110)
+    emit_insn (gen_atomic_cas_value_strong<mode>_amcas (operands[1], operands[2],
+							 operands[3], operands[4],
+							 operands[6]));
+  else
+    emit_insn (gen_atomic_cas_value_strong<mode> (operands[1], operands[2],
+						  operands[3], operands[4],
+						  operands[6]));
 
   rtx compare = operands[1];
   if (operands[3] != const0_rtx)
@@ -292,31 +346,53 @@ (define_expand "atomic_compare_and_swap<mode>"
    (match_operand:SI 7 "const_int_operand" "")] ;; mod_f
   ""
 {
-  union loongarch_gen_fn_ptrs generator;
-  generator.fn_7 = gen_atomic_cas_value_cmp_and_7_si;
-  loongarch_expand_atomic_qihi (generator, operands[1], operands[2],
-				operands[3], operands[4], operands[7]);
+  rtx mod_s, mod_f;
 
-  rtx compare = operands[1];
-  if (operands[3] != const0_rtx)
-    {
-      machine_mode mode = GET_MODE (operands[3]);
-      rtx op1 = convert_modes (SImode, mode, operands[1], true);
-      rtx op3 = convert_modes (SImode, mode, operands[3], true);
-      rtx difference = gen_rtx_MINUS (SImode, op1, op3);
-      compare = gen_reg_rtx (SImode);
-      emit_insn (gen_rtx_SET (compare, difference));
-    }
+  mod_s = operands[6];
+  mod_f = operands[7];
 
-  if (word_mode != <MODE>mode)
+  /* Normally the succ memory model must be stronger than fail, but in the
+     unlikely event of fail being ACQUIRE and succ being RELEASE we need to
+     promote succ to ACQ_REL so that we don't lose the acquire semantics.  */
+
+  if (is_mm_acquire (memmodel_base (INTVAL (mod_f)))
+      && is_mm_release (memmodel_base (INTVAL (mod_s))))
+    mod_s = GEN_INT (MEMMODEL_ACQ_REL);
+
+  operands[6] = mod_s;
+
+  if (ISA_BASE_IS_LA64V110)
+    emit_insn (gen_atomic_cas_value_strong<mode>_amcas (operands[1], operands[2],
+						       operands[3], operands[4],
+						       operands[6]));
+  else
     {
-      rtx reg = gen_reg_rtx (word_mode);
-      emit_insn (gen_rtx_SET (reg, gen_rtx_SIGN_EXTEND (word_mode, compare)));
-      compare = reg;
+      union loongarch_gen_fn_ptrs generator;
+      generator.fn_7 = gen_atomic_cas_value_cmp_and_7_si;
+      loongarch_expand_atomic_qihi (generator, operands[1], operands[2],
+				    operands[3], operands[4], operands[6]);
     }
 
-  emit_insn (gen_rtx_SET (operands[0],
-			  gen_rtx_EQ (SImode, compare, const0_rtx)));
+      rtx compare = operands[1];
+      if (operands[3] != const0_rtx)
+	{
+	  machine_mode mode = GET_MODE (operands[3]);
+	  rtx op1 = convert_modes (SImode, mode, operands[1], true);
+	  rtx op3 = convert_modes (SImode, mode, operands[3], true);
+	  rtx difference = gen_rtx_MINUS (SImode, op1, op3);
+	  compare = gen_reg_rtx (SImode);
+	  emit_insn (gen_rtx_SET (compare, difference));
+	}
+
+      if (word_mode != <MODE>mode)
+	{
+	  rtx reg = gen_reg_rtx (word_mode);
+	  emit_insn (gen_rtx_SET (reg, gen_rtx_SIGN_EXTEND (word_mode, compare)));
+	  compare = reg;
+	}
+
+      emit_insn (gen_rtx_SET (operands[0],
+			      gen_rtx_EQ (SImode, compare, const0_rtx)));
   DONE;
 })
 
@@ -505,13 +581,31 @@ (define_expand "atomic_exchange<mode>"
 	(match_operand:SHORT 2 "register_operand"))]
   ""
 {
-  union loongarch_gen_fn_ptrs generator;
-  generator.fn_7 = gen_atomic_cas_value_exchange_7_si;
-  loongarch_expand_atomic_qihi (generator, operands[0], operands[1],
-				const0_rtx, operands[2], operands[3]);
+  if (ISA_BASE_IS_LA64V110)
+    emit_insn (gen_atomic_exchange<mode>_short (operands[0], operands[1], operands[2], operands[3]));
+  else
+    {
+      union loongarch_gen_fn_ptrs generator;
+      generator.fn_7 = gen_atomic_cas_value_exchange_7_si;
+      loongarch_expand_atomic_qihi (generator, operands[0], operands[1],
+				    const0_rtx, operands[2], operands[3]);
+    }
   DONE;
 })
 
+(define_insn "atomic_fetch_add<mode>_short"
+  [(set (match_operand:SHORT 0 "register_operand" "=&r")
+	(match_operand:SHORT 1 "memory_operand" "+ZB"))
+   (set (match_dup 1)
+	(unspec_volatile:SHORT
+	  [(plus:SHORT (match_dup 1)
+		     (match_operand:SHORT 2 "reg_or_0_operand" "rJ"))
+	   (match_operand:SI 3 "const_int_operand")] ;; model
+	 UNSPEC_SYNC_OLD_OP))]
+  "ISA_BASE_IS_LA64V110"
+  "amadd%A3.<amo>\t%0,%z2,%1"
+  [(set (attr "length") (const_int 4))])
+
 (define_expand "atomic_fetch_add<mode>"
   [(set (match_operand:SHORT 0 "register_operand" "=&r")
 	(match_operand:SHORT 1 "memory_operand" "+ZB"))
@@ -523,10 +617,16 @@ (define_expand "atomic_fetch_add<mode>"
 	 UNSPEC_SYNC_OLD_OP))]
   ""
 {
-  union loongarch_gen_fn_ptrs generator;
-  generator.fn_7 = gen_atomic_cas_value_add_7_si;
-  loongarch_expand_atomic_qihi (generator, operands[0], operands[1],
-				operands[1], operands[2], operands[3]);
+  if (ISA_BASE_IS_LA64V110)
+    emit_insn (gen_atomic_fetch_add<mode>_short (operands[0], operands[1],
+					     operands[2], operands[3]));
+  else
+    {
+      union loongarch_gen_fn_ptrs generator;
+      generator.fn_7 = gen_atomic_cas_value_add_7_si;
+      loongarch_expand_atomic_qihi (generator, operands[0], operands[1],
+				    operands[1], operands[2], operands[3]);
+    }
   DONE;
 })
 
-- 
2.31.1

[PATCH v1 3/3] LoongArch: atomic_load and atomic_store are implemented using dbar grading.

[Lulu Cheng]

Because the la464 memory model design allows the same address load out of order,
so in the following test example, the Load of 23 lines may be executed first over
the load of 21 lines, resulting in an error.
So when memmodel is MEMMODEL_RELAXED, the load instruction will be followed by
"dbar 0x700" when implementing _atomic_load.

  1 void *
  2 gomp_ptrlock_get_slow (gomp_ptrlock_t *ptrlock)
  3 {
  4   int *intptr;
  5   uintptr_t oldval = 1;
  6
  7   __atomic_compare_exchange_n (ptrlock, &oldval, 2, false,
  8                                MEMMODEL_RELAXED, MEMMODEL_RELAXED);
  9
 10   /* futex works on ints, not pointers.
 11      But a valid work share pointer will be at least
 12      8 byte aligned, so it is safe to assume the low
 13      32-bits of the pointer won't contain values 1 or 2.  */
 14   __asm volatile ("" : "=r" (intptr) : "0" (ptrlock));
 15 #if __BYTE_ORDER == __BIG_ENDIAN
 16   if (sizeof (*ptrlock) > sizeof (int))
 17     intptr += (sizeof (*ptrlock) / sizeof (int)) - 1;
 18 #endif
 19   do
 20     do_wait (intptr, 2);
 21   while (__atomic_load_n (intptr, MEMMODEL_RELAXED) == 2);
 22   __asm volatile ("" : : : "memory");
 23   return (void *) __atomic_load_n (ptrlock, MEMMODEL_ACQUIRE);
 24 }

gcc/ChangeLog:

	* config/loongarch/sync.md (atomic_load<mode>): New template.
---
 gcc/config/loongarch/sync.md | 70 +++++++++++++++++++++++++++++++++---
 1 file changed, 65 insertions(+), 5 deletions(-)

diff --git a/gcc/config/loongarch/sync.md b/gcc/config/loongarch/sync.md
index 66e316d80f5..67848d72b87 100644
--- a/gcc/config/loongarch/sync.md
+++ b/gcc/config/loongarch/sync.md
@@ -30,6 +30,7 @@ (define_c_enum "unspec" [
   UNSPEC_SYNC_OLD_OP
   UNSPEC_SYNC_EXCHANGE
   UNSPEC_ATOMIC_STORE
+  UNSPEC_ATOMIC_LOAD
   UNSPEC_MEMORY_BARRIER
 ])
 
@@ -103,16 +104,75 @@ (define_insn "mem_thread_fence_1"
 
 ;; Atomic memory operations.
 
+(define_insn "atomic_load<mode>"
+  [(set (match_operand:QHWD 0 "register_operand" "=r")
+    (unspec_volatile:QHWD
+      [(match_operand:QHWD 1 "memory_operand" "+m")
+       (match_operand:SI 2 "const_int_operand")]                        ;; model
+      UNSPEC_ATOMIC_LOAD))]
+  ""
+{
+  enum memmodel model = memmodel_base (INTVAL (operands[2]));
+
+  switch (model)
+    {
+    case MEMMODEL_SEQ_CST:
+      return "dbar\t0x11\\n\\t"
+	     "ld.<size>\t%0,%1\\n\\t"
+	     "dbar\t0x14\\n\\t";
+    case MEMMODEL_ACQUIRE:
+      return "ld.<size>\t%0,%1\\n\\t"
+	     "dbar\t0x14\\n\\t";
+    case MEMMODEL_RELAXED:
+      return "ld.<size>\t%0,%1\\n\\t"
+	     "dbar\t0x700\\n\\t";
+
+    default:
+      /* The valid memory order variants are __ATOMIC_RELAXED, __ATOMIC_SEQ_CST,
+	 __ATOMIC_CONSUME and __ATOMIC_ACQUIRE.
+	 The expand_builtin_atomic_store function converts all invalid memmodels
+	 to MEMMODEL_SEQ_CST.
+
+	 __atomic builtins doc: "Consume is implemented using the
+	 stronger acquire memory order because of a deficiency in C++11's
+	 semantics."  See PR 59448 and get_memmodel in builtins.cc.  */
+      gcc_unreachable ();
+    }
+}
+  [(set (attr "length") (const_int 12))])
+
 ;; Implement atomic stores with amoswap.  Fall back to fences for atomic loads.
 (define_insn "atomic_store<mode>"
-  [(set (match_operand:GPR 0 "memory_operand" "+ZB")
-    (unspec_volatile:GPR
-      [(match_operand:GPR 1 "reg_or_0_operand" "rJ")
+  [(set (match_operand:QHWD 0 "memory_operand" "+m")
+    (unspec_volatile:QHWD
+      [(match_operand:QHWD 1 "reg_or_0_operand" "rJ")
        (match_operand:SI 2 "const_int_operand")]      ;; model
       UNSPEC_ATOMIC_STORE))]
   ""
-  "amswap%A2.<amo>\t$zero,%z1,%0"
-  [(set (attr "length") (const_int 8))])
+{
+  enum memmodel model = memmodel_base (INTVAL (operands[2]));
+
+  switch (model)
+    {
+    case MEMMODEL_SEQ_CST:
+      return "dbar\t0x12\\n\\t"
+	     "st.<size>\t%z1,%0\\n\\t"
+	     "dbar\t0x18\\n\\t";
+    case MEMMODEL_RELEASE:
+      return "dbar\t0x12\\n\\t"
+	     "st.<size>\t%z1,%0\\n\\t";
+    case MEMMODEL_RELAXED:
+      return "st.<size>\t%z1,%0";
+
+    default:
+      /* The valid memory order variants are __ATOMIC_RELAXED, __ATOMIC_SEQ_CST,
+	 and __ATOMIC_RELEASE.
+	 The expand_builtin_atomic_store function converts all invalid memmodels
+	 to MEMMODEL_SEQ_CST.  */
+      gcc_unreachable ();
+    }
+}
+  [(set (attr "length") (const_int 12))])
 
 (define_insn "atomic_<atomic_optab><mode>"
   [(set (match_operand:GPR 0 "memory_operand" "+ZB")
-- 
2.31.1

Re: [PATCH v1 0/3] Add LoongarchV1.1 instructions support.

[Xi Ruoyao]

On Fri, 2023-11-17 at 16:33 +0800, Lulu Cheng wrote:
> 
> Lulu Cheng (3):
>   LoongArch: Add LA664 support.
>   LoongArch: Implement atomic operations using LoongArch1.1
>     instructions.
>   LoongArch: atomic_load and atomic_store are implemented using dbar
>     grading.
> 
>  gcc/config.gcc                                |  10 +-
>  .../loongarch/genopts/loongarch-strings       |   1 +
>  gcc/config/loongarch/genopts/loongarch.opt.in |   3 +
>  gcc/config/loongarch/loongarch-cpu.cc         |   4 +
>  gcc/config/loongarch/loongarch-def.c          |  21 ++
>  gcc/config/loongarch/loongarch-def.h          |  10 +-
>  gcc/config/loongarch/loongarch-opts.cc        |   8 +-
>  gcc/config/loongarch/loongarch-opts.h         |   6 +-
>  gcc/config/loongarch/loongarch-str.h          |   1 +
>  gcc/config/loongarch/loongarch.cc             |   7 +-
>  gcc/config/loongarch/loongarch.opt            |   3 +
>  gcc/config/loongarch/sync.md                  | 256 ++++++++++++++---
> -
>  12 files changed, 263 insertions(+), 67 deletions(-)

I'll rebase my patches for div32 and ld-seq-sa on top of this.

-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University

Re: [PATCH v1 1/3] LoongArch: Add LA664 support.

[Xi Ruoyao]

On Fri, 2023-11-17 at 16:33 +0800, Lulu Cheng wrote:
> Define ISA_BASE_LA64V110, which represents the base instruction set defined in LoongArch1.1.
> Support the configure setting --with-arch =la664, and support -march=la664,-mtune=la664.
> 
> gcc/ChangeLog:
> 
> 	* config.gcc: Support LA664.
> 	* config/loongarch/genopts/loongarch-strings: Likewise.
> 	* config/loongarch/genopts/loongarch.opt.in: Likewise.
> 	* config/loongarch/loongarch-cpu.cc (fill_native_cpu_config): Likewise.
> 	* config/loongarch/loongarch-def.c: Likewise.

There should be

diff --git a/gcc/config/loongarch/loongarch-def.c
b/gcc/config/loongarch/loongarch-def.c
index 067629141b6..f22d488acb2 100644
--- a/gcc/config/loongarch/loongarch-def.c
+++ b/gcc/config/loongarch/loongarch-def.c
@@ -165,6 +165,7 @@ loongarch_cpu_multipass_dfa_lookahead[N_TUNE_TYPES]
= {
 const char*
 loongarch_isa_base_strings[N_ISA_BASE_TYPES] = {
   [ISA_BASE_LA64V100] = STR_ISA_BASE_LA64V100,
+  [ISA_BASE_LA64V110] = STR_ISA_BASE_LA64V110,
 };
 
 const char*


-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University

Re: [PATCH v1 1/3] LoongArch: Add LA664 support.

[chenglulu]

[-- Attachment #1: Type: text/plain, Size: 1172 bytes --]


在 2023/11/17 下午8:31, Xi Ruoyao 写道:
> On Fri, 2023-11-17 at 16:33 +0800, Lulu Cheng wrote:
>> Define ISA_BASE_LA64V110, which represents the base instruction set defined in LoongArch1.1.
>> Support the configure setting --with-arch =la664, and support -march=la664,-mtune=la664.
>>
>> gcc/ChangeLog:
>>
>> 	* config.gcc: Support LA664.
>> 	* config/loongarch/genopts/loongarch-strings: Likewise.
>> 	* config/loongarch/genopts/loongarch.opt.in: Likewise.
>> 	* config/loongarch/loongarch-cpu.cc (fill_native_cpu_config): Likewise.
>> 	* config/loongarch/loongarch-def.c: Likewise.
> There should be
>
> diff --git a/gcc/config/loongarch/loongarch-def.c
> b/gcc/config/loongarch/loongarch-def.c
> index 067629141b6..f22d488acb2 100644
> --- a/gcc/config/loongarch/loongarch-def.c
> +++ b/gcc/config/loongarch/loongarch-def.c
> @@ -165,6 +165,7 @@ loongarch_cpu_multipass_dfa_lookahead[N_TUNE_TYPES]
> = {
>   const char*
>   loongarch_isa_base_strings[N_ISA_BASE_TYPES] = {
>     [ISA_BASE_LA64V100] = STR_ISA_BASE_LA64V100,
> +  [ISA_BASE_LA64V110] = STR_ISA_BASE_LA64V110,
>   };
>   
>   const char*
>
Ok, I will revise it as soon as possible.

Thanks.

  *

Re: [pushed ][PATCH v1 0/3] Add LoongarchV1.1 instructions support.

[chenglulu]

Pushed to r14-5568.

在 2023/11/17 下午7:09, Xi Ruoyao 写道:
> On Fri, 2023-11-17 at 16:33 +0800, Lulu Cheng wrote:
>> Lulu Cheng (3):
>>    LoongArch: Add LA664 support.
>>    LoongArch: Implement atomic operations using LoongArch1.1
>>      instructions.
>>    LoongArch: atomic_load and atomic_store are implemented using dbar
>>      grading.
>>
>>   gcc/config.gcc                                |  10 +-
>>   .../loongarch/genopts/loongarch-strings       |   1 +
>>   gcc/config/loongarch/genopts/loongarch.opt.in |   3 +
>>   gcc/config/loongarch/loongarch-cpu.cc         |   4 +
>>   gcc/config/loongarch/loongarch-def.c          |  21 ++
>>   gcc/config/loongarch/loongarch-def.h          |  10 +-
>>   gcc/config/loongarch/loongarch-opts.cc        |   8 +-
>>   gcc/config/loongarch/loongarch-opts.h         |   6 +-
>>   gcc/config/loongarch/loongarch-str.h          |   1 +
>>   gcc/config/loongarch/loongarch.cc             |   7 +-
>>   gcc/config/loongarch/loongarch.opt            |   3 +
>>   gcc/config/loongarch/sync.md                  | 256 ++++++++++++++---
>> -
>>   12 files changed, 263 insertions(+), 67 deletions(-)
> I'll rebase my patches for div32 and ld-seq-sa on top of this.
>