* [PATCH 01/35] accel/tcg: Move HMP info jit and info opcount code
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 02/35] tcg: Add C_N2_I1 Richard Henderson
` (35 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Philippe Mathieu-Daudé, Alex Bennée
Move all of it into accel/tcg/monitor.c. This puts everything
about tcg that is only used by the monitor in the same place.
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
accel/tcg/internal-common.h | 2 -
include/exec/cputlb.h | 1 -
include/tcg/tcg.h | 3 -
accel/tcg/cputlb.c | 15 ----
accel/tcg/monitor.c | 154 ++++++++++++++++++++++++++++++++++++
accel/tcg/translate-all.c | 127 -----------------------------
tcg/tcg.c | 10 ---
7 files changed, 154 insertions(+), 158 deletions(-)
diff --git a/accel/tcg/internal-common.h b/accel/tcg/internal-common.h
index 3b2277e6e9..edefd0dcb7 100644
--- a/accel/tcg/internal-common.h
+++ b/accel/tcg/internal-common.h
@@ -14,8 +14,6 @@
extern int64_t max_delay;
extern int64_t max_advance;
-void dump_exec_info(GString *buf);
-
/*
* Return true if CS is not running in parallel with other cpus, either
* because there are no other cpus or we are within an exclusive context.
diff --git a/include/exec/cputlb.h b/include/exec/cputlb.h
index 19b16e58f8..6da1462c4f 100644
--- a/include/exec/cputlb.h
+++ b/include/exec/cputlb.h
@@ -26,6 +26,5 @@
/* cputlb.c */
void tlb_protect_code(ram_addr_t ram_addr);
void tlb_unprotect_code(ram_addr_t ram_addr);
-void tlb_flush_counts(size_t *full, size_t *part, size_t *elide);
#endif
#endif
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index a9282cdcc6..3a4c0f124f 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -846,9 +846,6 @@ static inline TCGv_ptr tcg_temp_new_ptr(void)
return temp_tcgv_ptr(t);
}
-void tcg_dump_info(GString *buf);
-void tcg_dump_op_count(GString *buf);
-
#define TCG_CT_CONST 1 /* any constant of register size */
typedef struct TCGArgConstraint {
diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index b8c5e345b8..13986820fe 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -321,21 +321,6 @@ static void flush_all_helper(CPUState *src, run_on_cpu_func fn,
}
}
-void tlb_flush_counts(size_t *pfull, size_t *ppart, size_t *pelide)
-{
- CPUState *cpu;
- size_t full = 0, part = 0, elide = 0;
-
- CPU_FOREACH(cpu) {
- full += qatomic_read(&cpu->neg.tlb.c.full_flush_count);
- part += qatomic_read(&cpu->neg.tlb.c.part_flush_count);
- elide += qatomic_read(&cpu->neg.tlb.c.elide_flush_count);
- }
- *pfull = full;
- *ppart = part;
- *pelide = elide;
-}
-
static void tlb_flush_by_mmuidx_async_work(CPUState *cpu, run_on_cpu_data data)
{
uint16_t asked = data.host_int;
diff --git a/accel/tcg/monitor.c b/accel/tcg/monitor.c
index caf1189e0b..093efe9714 100644
--- a/accel/tcg/monitor.c
+++ b/accel/tcg/monitor.c
@@ -8,6 +8,7 @@
#include "qemu/osdep.h"
#include "qemu/accel.h"
+#include "qemu/qht.h"
#include "qapi/error.h"
#include "qapi/type-helpers.h"
#include "qapi/qapi-commands-machine.h"
@@ -17,6 +18,7 @@
#include "sysemu/tcg.h"
#include "tcg/tcg.h"
#include "internal-common.h"
+#include "tb-context.h"
static void dump_drift_info(GString *buf)
@@ -50,6 +52,153 @@ static void dump_accel_info(GString *buf)
one_insn_per_tb ? "on" : "off");
}
+static void print_qht_statistics(struct qht_stats hst, GString *buf)
+{
+ uint32_t hgram_opts;
+ size_t hgram_bins;
+ char *hgram;
+
+ if (!hst.head_buckets) {
+ return;
+ }
+ g_string_append_printf(buf, "TB hash buckets %zu/%zu "
+ "(%0.2f%% head buckets used)\n",
+ hst.used_head_buckets, hst.head_buckets,
+ (double)hst.used_head_buckets /
+ hst.head_buckets * 100);
+
+ hgram_opts = QDIST_PR_BORDER | QDIST_PR_LABELS;
+ hgram_opts |= QDIST_PR_100X | QDIST_PR_PERCENT;
+ if (qdist_xmax(&hst.occupancy) - qdist_xmin(&hst.occupancy) == 1) {
+ hgram_opts |= QDIST_PR_NODECIMAL;
+ }
+ hgram = qdist_pr(&hst.occupancy, 10, hgram_opts);
+ g_string_append_printf(buf, "TB hash occupancy %0.2f%% avg chain occ. "
+ "Histogram: %s\n",
+ qdist_avg(&hst.occupancy) * 100, hgram);
+ g_free(hgram);
+
+ hgram_opts = QDIST_PR_BORDER | QDIST_PR_LABELS;
+ hgram_bins = qdist_xmax(&hst.chain) - qdist_xmin(&hst.chain);
+ if (hgram_bins > 10) {
+ hgram_bins = 10;
+ } else {
+ hgram_bins = 0;
+ hgram_opts |= QDIST_PR_NODECIMAL | QDIST_PR_NOBINRANGE;
+ }
+ hgram = qdist_pr(&hst.chain, hgram_bins, hgram_opts);
+ g_string_append_printf(buf, "TB hash avg chain %0.3f buckets. "
+ "Histogram: %s\n",
+ qdist_avg(&hst.chain), hgram);
+ g_free(hgram);
+}
+
+struct tb_tree_stats {
+ size_t nb_tbs;
+ size_t host_size;
+ size_t target_size;
+ size_t max_target_size;
+ size_t direct_jmp_count;
+ size_t direct_jmp2_count;
+ size_t cross_page;
+};
+
+static gboolean tb_tree_stats_iter(gpointer key, gpointer value, gpointer data)
+{
+ const TranslationBlock *tb = value;
+ struct tb_tree_stats *tst = data;
+
+ tst->nb_tbs++;
+ tst->host_size += tb->tc.size;
+ tst->target_size += tb->size;
+ if (tb->size > tst->max_target_size) {
+ tst->max_target_size = tb->size;
+ }
+ if (tb->page_addr[1] != -1) {
+ tst->cross_page++;
+ }
+ if (tb->jmp_reset_offset[0] != TB_JMP_OFFSET_INVALID) {
+ tst->direct_jmp_count++;
+ if (tb->jmp_reset_offset[1] != TB_JMP_OFFSET_INVALID) {
+ tst->direct_jmp2_count++;
+ }
+ }
+ return false;
+}
+
+static void tlb_flush_counts(size_t *pfull, size_t *ppart, size_t *pelide)
+{
+ CPUState *cpu;
+ size_t full = 0, part = 0, elide = 0;
+
+ CPU_FOREACH(cpu) {
+ full += qatomic_read(&cpu->neg.tlb.c.full_flush_count);
+ part += qatomic_read(&cpu->neg.tlb.c.part_flush_count);
+ elide += qatomic_read(&cpu->neg.tlb.c.elide_flush_count);
+ }
+ *pfull = full;
+ *ppart = part;
+ *pelide = elide;
+}
+
+static void tcg_dump_info(GString *buf)
+{
+ g_string_append_printf(buf, "[TCG profiler not compiled]\n");
+}
+
+static void dump_exec_info(GString *buf)
+{
+ struct tb_tree_stats tst = {};
+ struct qht_stats hst;
+ size_t nb_tbs, flush_full, flush_part, flush_elide;
+
+ tcg_tb_foreach(tb_tree_stats_iter, &tst);
+ nb_tbs = tst.nb_tbs;
+ /* XXX: avoid using doubles ? */
+ g_string_append_printf(buf, "Translation buffer state:\n");
+ /*
+ * Report total code size including the padding and TB structs;
+ * otherwise users might think "-accel tcg,tb-size" is not honoured.
+ * For avg host size we use the precise numbers from tb_tree_stats though.
+ */
+ g_string_append_printf(buf, "gen code size %zu/%zu\n",
+ tcg_code_size(), tcg_code_capacity());
+ g_string_append_printf(buf, "TB count %zu\n", nb_tbs);
+ g_string_append_printf(buf, "TB avg target size %zu max=%zu bytes\n",
+ nb_tbs ? tst.target_size / nb_tbs : 0,
+ tst.max_target_size);
+ g_string_append_printf(buf, "TB avg host size %zu bytes "
+ "(expansion ratio: %0.1f)\n",
+ nb_tbs ? tst.host_size / nb_tbs : 0,
+ tst.target_size ?
+ (double)tst.host_size / tst.target_size : 0);
+ g_string_append_printf(buf, "cross page TB count %zu (%zu%%)\n",
+ tst.cross_page,
+ nb_tbs ? (tst.cross_page * 100) / nb_tbs : 0);
+ g_string_append_printf(buf, "direct jump count %zu (%zu%%) "
+ "(2 jumps=%zu %zu%%)\n",
+ tst.direct_jmp_count,
+ nb_tbs ? (tst.direct_jmp_count * 100) / nb_tbs : 0,
+ tst.direct_jmp2_count,
+ nb_tbs ? (tst.direct_jmp2_count * 100) / nb_tbs : 0);
+
+ qht_statistics_init(&tb_ctx.htable, &hst);
+ print_qht_statistics(hst, buf);
+ qht_statistics_destroy(&hst);
+
+ g_string_append_printf(buf, "\nStatistics:\n");
+ g_string_append_printf(buf, "TB flush count %u\n",
+ qatomic_read(&tb_ctx.tb_flush_count));
+ g_string_append_printf(buf, "TB invalidate count %u\n",
+ qatomic_read(&tb_ctx.tb_phys_invalidate_count));
+
+ tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
+ g_string_append_printf(buf, "TLB full flushes %zu\n", flush_full);
+ g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
+ g_string_append_printf(buf, "TLB elided flushes %zu\n", flush_elide);
+ tcg_dump_info(buf);
+}
+
HumanReadableText *qmp_x_query_jit(Error **errp)
{
g_autoptr(GString) buf = g_string_new("");
@@ -66,6 +215,11 @@ HumanReadableText *qmp_x_query_jit(Error **errp)
return human_readable_text_from_str(buf);
}
+static void tcg_dump_op_count(GString *buf)
+{
+ g_string_append_printf(buf, "[TCG profiler not compiled]\n");
+}
+
HumanReadableText *qmp_x_query_opcount(Error **errp)
{
g_autoptr(GString) buf = g_string_new("");
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 8cb6ad3511..e579b0891d 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -645,133 +645,6 @@ void cpu_io_recompile(CPUState *cpu, uintptr_t retaddr)
cpu_loop_exit_noexc(cpu);
}
-static void print_qht_statistics(struct qht_stats hst, GString *buf)
-{
- uint32_t hgram_opts;
- size_t hgram_bins;
- char *hgram;
-
- if (!hst.head_buckets) {
- return;
- }
- g_string_append_printf(buf, "TB hash buckets %zu/%zu "
- "(%0.2f%% head buckets used)\n",
- hst.used_head_buckets, hst.head_buckets,
- (double)hst.used_head_buckets /
- hst.head_buckets * 100);
-
- hgram_opts = QDIST_PR_BORDER | QDIST_PR_LABELS;
- hgram_opts |= QDIST_PR_100X | QDIST_PR_PERCENT;
- if (qdist_xmax(&hst.occupancy) - qdist_xmin(&hst.occupancy) == 1) {
- hgram_opts |= QDIST_PR_NODECIMAL;
- }
- hgram = qdist_pr(&hst.occupancy, 10, hgram_opts);
- g_string_append_printf(buf, "TB hash occupancy %0.2f%% avg chain occ. "
- "Histogram: %s\n",
- qdist_avg(&hst.occupancy) * 100, hgram);
- g_free(hgram);
-
- hgram_opts = QDIST_PR_BORDER | QDIST_PR_LABELS;
- hgram_bins = qdist_xmax(&hst.chain) - qdist_xmin(&hst.chain);
- if (hgram_bins > 10) {
- hgram_bins = 10;
- } else {
- hgram_bins = 0;
- hgram_opts |= QDIST_PR_NODECIMAL | QDIST_PR_NOBINRANGE;
- }
- hgram = qdist_pr(&hst.chain, hgram_bins, hgram_opts);
- g_string_append_printf(buf, "TB hash avg chain %0.3f buckets. "
- "Histogram: %s\n",
- qdist_avg(&hst.chain), hgram);
- g_free(hgram);
-}
-
-struct tb_tree_stats {
- size_t nb_tbs;
- size_t host_size;
- size_t target_size;
- size_t max_target_size;
- size_t direct_jmp_count;
- size_t direct_jmp2_count;
- size_t cross_page;
-};
-
-static gboolean tb_tree_stats_iter(gpointer key, gpointer value, gpointer data)
-{
- const TranslationBlock *tb = value;
- struct tb_tree_stats *tst = data;
-
- tst->nb_tbs++;
- tst->host_size += tb->tc.size;
- tst->target_size += tb->size;
- if (tb->size > tst->max_target_size) {
- tst->max_target_size = tb->size;
- }
- if (tb_page_addr1(tb) != -1) {
- tst->cross_page++;
- }
- if (tb->jmp_reset_offset[0] != TB_JMP_OFFSET_INVALID) {
- tst->direct_jmp_count++;
- if (tb->jmp_reset_offset[1] != TB_JMP_OFFSET_INVALID) {
- tst->direct_jmp2_count++;
- }
- }
- return false;
-}
-
-void dump_exec_info(GString *buf)
-{
- struct tb_tree_stats tst = {};
- struct qht_stats hst;
- size_t nb_tbs, flush_full, flush_part, flush_elide;
-
- tcg_tb_foreach(tb_tree_stats_iter, &tst);
- nb_tbs = tst.nb_tbs;
- /* XXX: avoid using doubles ? */
- g_string_append_printf(buf, "Translation buffer state:\n");
- /*
- * Report total code size including the padding and TB structs;
- * otherwise users might think "-accel tcg,tb-size" is not honoured.
- * For avg host size we use the precise numbers from tb_tree_stats though.
- */
- g_string_append_printf(buf, "gen code size %zu/%zu\n",
- tcg_code_size(), tcg_code_capacity());
- g_string_append_printf(buf, "TB count %zu\n", nb_tbs);
- g_string_append_printf(buf, "TB avg target size %zu max=%zu bytes\n",
- nb_tbs ? tst.target_size / nb_tbs : 0,
- tst.max_target_size);
- g_string_append_printf(buf, "TB avg host size %zu bytes "
- "(expansion ratio: %0.1f)\n",
- nb_tbs ? tst.host_size / nb_tbs : 0,
- tst.target_size ?
- (double)tst.host_size / tst.target_size : 0);
- g_string_append_printf(buf, "cross page TB count %zu (%zu%%)\n",
- tst.cross_page,
- nb_tbs ? (tst.cross_page * 100) / nb_tbs : 0);
- g_string_append_printf(buf, "direct jump count %zu (%zu%%) "
- "(2 jumps=%zu %zu%%)\n",
- tst.direct_jmp_count,
- nb_tbs ? (tst.direct_jmp_count * 100) / nb_tbs : 0,
- tst.direct_jmp2_count,
- nb_tbs ? (tst.direct_jmp2_count * 100) / nb_tbs : 0);
-
- qht_statistics_init(&tb_ctx.htable, &hst);
- print_qht_statistics(hst, buf);
- qht_statistics_destroy(&hst);
-
- g_string_append_printf(buf, "\nStatistics:\n");
- g_string_append_printf(buf, "TB flush count %u\n",
- qatomic_read(&tb_ctx.tb_flush_count));
- g_string_append_printf(buf, "TB invalidate count %u\n",
- qatomic_read(&tb_ctx.tb_phys_invalidate_count));
-
- tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
- g_string_append_printf(buf, "TLB full flushes %zu\n", flush_full);
- g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
- g_string_append_printf(buf, "TLB elided flushes %zu\n", flush_elide);
- tcg_dump_info(buf);
-}
-
#else /* CONFIG_USER_ONLY */
void cpu_interrupt(CPUState *cpu, int mask)
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 35158a0846..847d749a7e 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -5927,11 +5927,6 @@ static void tcg_out_st_helper_args(TCGContext *s, const TCGLabelQemuLdst *ldst,
tcg_out_helper_load_common_args(s, ldst, parm, info, next_arg);
}
-void tcg_dump_op_count(GString *buf)
-{
- g_string_append_printf(buf, "[TCG profiler not compiled]\n");
-}
-
int tcg_gen_code(TCGContext *s, TranslationBlock *tb, uint64_t pc_start)
{
int i, start_words, num_insns;
@@ -6128,11 +6123,6 @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb, uint64_t pc_start)
return tcg_current_code_size(s);
}
-void tcg_dump_info(GString *buf)
-{
- g_string_append_printf(buf, "[TCG profiler not compiled]\n");
-}
-
#ifdef ELF_HOST_MACHINE
/* In order to use this feature, the backend needs to do three things:
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 02/35] tcg: Add C_N2_I1
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
2023-11-07 2:48 ` [PATCH 01/35] accel/tcg: Move HMP info jit and info opcount code Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 03/35] tcg/loongarch64: Use C_N2_I1 for INDEX_op_qemu_ld_a*_i128 Richard Henderson
` (34 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Jiajie Chen
Constraint with two outputs, both in new registers.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-2-richard.henderson@linaro.org>
---
tcg/tcg.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 847d749a7e..6766b60b8a 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -653,6 +653,7 @@ static void tcg_out_movext3(TCGContext *s, const TCGMovExtend *i1,
#define C_O1_I4(O1, I1, I2, I3, I4) C_PFX5(c_o1_i4_, O1, I1, I2, I3, I4),
#define C_N1_I2(O1, I1, I2) C_PFX3(c_n1_i2_, O1, I1, I2),
+#define C_N2_I1(O1, O2, I1) C_PFX3(c_n2_i1_, O1, O2, I1),
#define C_O2_I1(O1, O2, I1) C_PFX3(c_o2_i1_, O1, O2, I1),
#define C_O2_I2(O1, O2, I1, I2) C_PFX4(c_o2_i2_, O1, O2, I1, I2),
@@ -675,6 +676,7 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode);
#undef C_O1_I3
#undef C_O1_I4
#undef C_N1_I2
+#undef C_N2_I1
#undef C_O2_I1
#undef C_O2_I2
#undef C_O2_I3
@@ -694,6 +696,7 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode);
#define C_O1_I4(O1, I1, I2, I3, I4) { .args_ct_str = { #O1, #I1, #I2, #I3, #I4 } },
#define C_N1_I2(O1, I1, I2) { .args_ct_str = { "&" #O1, #I1, #I2 } },
+#define C_N2_I1(O1, O2, I1) { .args_ct_str = { "&" #O1, "&" #O2, #I1 } },
#define C_O2_I1(O1, O2, I1) { .args_ct_str = { #O1, #O2, #I1 } },
#define C_O2_I2(O1, O2, I1, I2) { .args_ct_str = { #O1, #O2, #I1, #I2 } },
@@ -715,6 +718,7 @@ static const TCGTargetOpDef constraint_sets[] = {
#undef C_O1_I3
#undef C_O1_I4
#undef C_N1_I2
+#undef C_N2_I1
#undef C_O2_I1
#undef C_O2_I2
#undef C_O2_I3
@@ -734,6 +738,7 @@ static const TCGTargetOpDef constraint_sets[] = {
#define C_O1_I4(O1, I1, I2, I3, I4) C_PFX5(c_o1_i4_, O1, I1, I2, I3, I4)
#define C_N1_I2(O1, I1, I2) C_PFX3(c_n1_i2_, O1, I1, I2)
+#define C_N2_I1(O1, O2, I1) C_PFX3(c_n2_i1_, O1, O2, I1)
#define C_O2_I1(O1, O2, I1) C_PFX3(c_o2_i1_, O1, O2, I1)
#define C_O2_I2(O1, O2, I1, I2) C_PFX4(c_o2_i2_, O1, O2, I1, I2)
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 03/35] tcg/loongarch64: Use C_N2_I1 for INDEX_op_qemu_ld_a*_i128
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
2023-11-07 2:48 ` [PATCH 01/35] accel/tcg: Move HMP info jit and info opcount code Richard Henderson
2023-11-07 2:48 ` [PATCH 02/35] tcg: Add C_N2_I1 Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 04/35] util: Add cpuinfo for loongarch64 Richard Henderson
` (33 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Jiajie Chen
Use new registers for the output, so that we never overlap
the input address, which could happen for user-only.
This avoids a "tmp = addr + 0" in that case.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-3-richard.henderson@linaro.org>
---
tcg/loongarch64/tcg-target-con-set.h | 2 +-
tcg/loongarch64/tcg-target.c.inc | 17 +++++++++++------
2 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/tcg/loongarch64/tcg-target-con-set.h b/tcg/loongarch64/tcg-target-con-set.h
index 77d62e38e7..cae6c2aad6 100644
--- a/tcg/loongarch64/tcg-target-con-set.h
+++ b/tcg/loongarch64/tcg-target-con-set.h
@@ -38,4 +38,4 @@ C_O1_I2(w, w, wM)
C_O1_I2(w, w, wA)
C_O1_I3(w, w, w, w)
C_O1_I4(r, rZ, rJ, rZ, rZ)
-C_O2_I1(r, r, r)
+C_N2_I1(r, r, r)
diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc
index ccf133db4b..a1a387df31 100644
--- a/tcg/loongarch64/tcg-target.c.inc
+++ b/tcg/loongarch64/tcg-target.c.inc
@@ -1103,13 +1103,18 @@ static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg data_lo, TCGReg data_hi
}
} else {
/* Otherwise use a pair of LD/ST. */
- tcg_out_opc_add_d(s, TCG_REG_TMP0, h.base, h.index);
+ TCGReg base = h.base;
+ if (h.index != TCG_REG_ZERO) {
+ base = TCG_REG_TMP0;
+ tcg_out_opc_add_d(s, base, h.base, h.index);
+ }
if (is_ld) {
- tcg_out_opc_ld_d(s, data_lo, TCG_REG_TMP0, 0);
- tcg_out_opc_ld_d(s, data_hi, TCG_REG_TMP0, 8);
+ tcg_debug_assert(base != data_lo);
+ tcg_out_opc_ld_d(s, data_lo, base, 0);
+ tcg_out_opc_ld_d(s, data_hi, base, 8);
} else {
- tcg_out_opc_st_d(s, data_lo, TCG_REG_TMP0, 0);
- tcg_out_opc_st_d(s, data_hi, TCG_REG_TMP0, 8);
+ tcg_out_opc_st_d(s, data_lo, base, 0);
+ tcg_out_opc_st_d(s, data_hi, base, 8);
}
}
@@ -2049,7 +2054,7 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op)
case INDEX_op_qemu_ld_a32_i128:
case INDEX_op_qemu_ld_a64_i128:
- return C_O2_I1(r, r, r);
+ return C_N2_I1(r, r, r);
case INDEX_op_qemu_st_a32_i128:
case INDEX_op_qemu_st_a64_i128:
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 04/35] util: Add cpuinfo for loongarch64
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (2 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 03/35] tcg/loongarch64: Use C_N2_I1 for INDEX_op_qemu_ld_a*_i128 Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 05/35] tcg/loongarch64: Use cpuinfo.h Richard Henderson
` (32 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Jiajie Chen
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-4-richard.henderson@linaro.org>
---
host/include/loongarch64/host/cpuinfo.h | 21 +++++++++++++++
util/cpuinfo-loongarch.c | 35 +++++++++++++++++++++++++
util/meson.build | 2 ++
3 files changed, 58 insertions(+)
create mode 100644 host/include/loongarch64/host/cpuinfo.h
create mode 100644 util/cpuinfo-loongarch.c
diff --git a/host/include/loongarch64/host/cpuinfo.h b/host/include/loongarch64/host/cpuinfo.h
new file mode 100644
index 0000000000..fab664a10b
--- /dev/null
+++ b/host/include/loongarch64/host/cpuinfo.h
@@ -0,0 +1,21 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ * Host specific cpu identification for LoongArch
+ */
+
+#ifndef HOST_CPUINFO_H
+#define HOST_CPUINFO_H
+
+#define CPUINFO_ALWAYS (1u << 0) /* so cpuinfo is nonzero */
+#define CPUINFO_LSX (1u << 1)
+
+/* Initialized with a constructor. */
+extern unsigned cpuinfo;
+
+/*
+ * We cannot rely on constructor ordering, so other constructors must
+ * use the function interface rather than the variable above.
+ */
+unsigned cpuinfo_init(void);
+
+#endif /* HOST_CPUINFO_H */
diff --git a/util/cpuinfo-loongarch.c b/util/cpuinfo-loongarch.c
new file mode 100644
index 0000000000..08b6d7460c
--- /dev/null
+++ b/util/cpuinfo-loongarch.c
@@ -0,0 +1,35 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ * Host specific cpu identification for LoongArch.
+ */
+
+#include "qemu/osdep.h"
+#include "host/cpuinfo.h"
+
+#ifdef CONFIG_GETAUXVAL
+# include <sys/auxv.h>
+#else
+# include "elf.h"
+#endif
+#include <asm/hwcap.h>
+
+unsigned cpuinfo;
+
+/* Called both as constructor and (possibly) via other constructors. */
+unsigned __attribute__((constructor)) cpuinfo_init(void)
+{
+ unsigned info = cpuinfo;
+ unsigned long hwcap;
+
+ if (info) {
+ return info;
+ }
+
+ hwcap = qemu_getauxval(AT_HWCAP);
+
+ info = CPUINFO_ALWAYS;
+ info |= (hwcap & HWCAP_LOONGARCH_LSX ? CPUINFO_LSX : 0);
+
+ cpuinfo = info;
+ return info;
+}
diff --git a/util/meson.build b/util/meson.build
index 769b24f2e0..1bfff81087 100644
--- a/util/meson.build
+++ b/util/meson.build
@@ -113,6 +113,8 @@ if cpu == 'aarch64'
util_ss.add(files('cpuinfo-aarch64.c'))
elif cpu in ['x86', 'x86_64']
util_ss.add(files('cpuinfo-i386.c'))
+elif cpu == 'loongarch64'
+ util_ss.add(files('cpuinfo-loongarch.c'))
elif cpu in ['ppc', 'ppc64']
util_ss.add(files('cpuinfo-ppc.c'))
endif
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 05/35] tcg/loongarch64: Use cpuinfo.h
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (3 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 04/35] util: Add cpuinfo for loongarch64 Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 06/35] host/include/loongarch64: Add atomic16 load and store Richard Henderson
` (31 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Jiajie Chen
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jiajie Chen <c@jia.je>
Message-Id: <20230916220151.526140-5-richard.henderson@linaro.org>
---
tcg/loongarch64/tcg-target.h | 8 ++++----
tcg/loongarch64/tcg-target.c.inc | 8 +-------
2 files changed, 5 insertions(+), 11 deletions(-)
diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h
index 03017672f6..1bea15b02e 100644
--- a/tcg/loongarch64/tcg-target.h
+++ b/tcg/loongarch64/tcg-target.h
@@ -29,6 +29,8 @@
#ifndef LOONGARCH_TCG_TARGET_H
#define LOONGARCH_TCG_TARGET_H
+#include "host/cpuinfo.h"
+
#define TCG_TARGET_INSN_UNIT_SIZE 4
#define TCG_TARGET_NB_REGS 64
@@ -85,8 +87,6 @@ typedef enum {
TCG_VEC_TMP0 = TCG_REG_V23,
} TCGReg;
-extern bool use_lsx_instructions;
-
/* used for function call generation */
#define TCG_REG_CALL_STACK TCG_REG_SP
#define TCG_TARGET_STACK_ALIGN 16
@@ -171,10 +171,10 @@ extern bool use_lsx_instructions;
#define TCG_TARGET_HAS_muluh_i64 1
#define TCG_TARGET_HAS_mulsh_i64 1
-#define TCG_TARGET_HAS_qemu_ldst_i128 use_lsx_instructions
+#define TCG_TARGET_HAS_qemu_ldst_i128 (cpuinfo & CPUINFO_LSX)
#define TCG_TARGET_HAS_v64 0
-#define TCG_TARGET_HAS_v128 use_lsx_instructions
+#define TCG_TARGET_HAS_v128 (cpuinfo & CPUINFO_LSX)
#define TCG_TARGET_HAS_v256 0
#define TCG_TARGET_HAS_not_vec 1
diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc
index a1a387df31..ae13c1f666 100644
--- a/tcg/loongarch64/tcg-target.c.inc
+++ b/tcg/loongarch64/tcg-target.c.inc
@@ -32,8 +32,6 @@
#include "../tcg-ldst.c.inc"
#include <asm/hwcap.h>
-bool use_lsx_instructions;
-
#ifdef CONFIG_DEBUG_TCG
static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = {
"zero",
@@ -2314,10 +2312,6 @@ static void tcg_target_init(TCGContext *s)
exit(EXIT_FAILURE);
}
- if (hwcap & HWCAP_LOONGARCH_LSX) {
- use_lsx_instructions = 1;
- }
-
tcg_target_available_regs[TCG_TYPE_I32] = ALL_GENERAL_REGS;
tcg_target_available_regs[TCG_TYPE_I64] = ALL_GENERAL_REGS;
@@ -2333,7 +2327,7 @@ static void tcg_target_init(TCGContext *s)
tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S8);
tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S9);
- if (use_lsx_instructions) {
+ if (cpuinfo & CPUINFO_LSX) {
tcg_target_available_regs[TCG_TYPE_V128] = ALL_VECTOR_REGS;
tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_V24);
tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_V25);
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 06/35] host/include/loongarch64: Add atomic16 load and store
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (4 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 05/35] tcg/loongarch64: Use cpuinfo.h Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 07/35] accel/tcg: Remove redundant case in store_atom_16 Richard Henderson
` (30 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
While loongarch64 does not have a 128-bit cmpxchg, it does
have 128-bit atomic load and store via the vector unit.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230916220151.526140-6-richard.henderson@linaro.org>
---
.../include/loongarch64/host/atomic128-ldst.h | 52 +++++++++++++++++++
| 39 ++++++++++++++
.../loongarch64/host/store-insert-al16.h | 12 +++++
3 files changed, 103 insertions(+)
create mode 100644 host/include/loongarch64/host/atomic128-ldst.h
create mode 100644 host/include/loongarch64/host/load-extract-al16-al8.h
create mode 100644 host/include/loongarch64/host/store-insert-al16.h
diff --git a/host/include/loongarch64/host/atomic128-ldst.h b/host/include/loongarch64/host/atomic128-ldst.h
new file mode 100644
index 0000000000..9a4a8f8b9e
--- /dev/null
+++ b/host/include/loongarch64/host/atomic128-ldst.h
@@ -0,0 +1,52 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ * Load/store for 128-bit atomic operations, LoongArch version.
+ *
+ * See docs/devel/atomics.rst for discussion about the guarantees each
+ * atomic primitive is meant to provide.
+ */
+
+#ifndef LOONGARCH_ATOMIC128_LDST_H
+#define LOONGARCH_ATOMIC128_LDST_H
+
+#include "host/cpuinfo.h"
+#include "tcg/debug-assert.h"
+
+#define HAVE_ATOMIC128_RO likely(cpuinfo & CPUINFO_LSX)
+#define HAVE_ATOMIC128_RW HAVE_ATOMIC128_RO
+
+/*
+ * As of gcc 13 and clang 16, there is no compiler support for LSX at all.
+ * Use inline assembly throughout.
+ */
+
+static inline Int128 atomic16_read_ro(const Int128 *ptr)
+{
+ uint64_t l, h;
+
+ tcg_debug_assert(HAVE_ATOMIC128_RO);
+ asm("vld $vr0, %2, 0\n\t"
+ "vpickve2gr.d %0, $vr0, 0\n\t"
+ "vpickve2gr.d %1, $vr0, 1"
+ : "=r"(l), "=r"(h) : "r"(ptr), "m"(*ptr) : "f0");
+
+ return int128_make128(l, h);
+}
+
+static inline Int128 atomic16_read_rw(Int128 *ptr)
+{
+ return atomic16_read_ro(ptr);
+}
+
+static inline void atomic16_set(Int128 *ptr, Int128 val)
+{
+ uint64_t l = int128_getlo(val), h = int128_gethi(val);
+
+ tcg_debug_assert(HAVE_ATOMIC128_RW);
+ asm("vinsgr2vr.d $vr0, %1, 0\n\t"
+ "vinsgr2vr.d $vr0, %2, 1\n\t"
+ "vst $vr0, %3, 0"
+ : "=m"(*ptr) : "r"(l), "r"(h), "r"(ptr) : "f0");
+}
+
+#endif /* LOONGARCH_ATOMIC128_LDST_H */
--git a/host/include/loongarch64/host/load-extract-al16-al8.h b/host/include/loongarch64/host/load-extract-al16-al8.h
new file mode 100644
index 0000000000..d1fb59d8af
--- /dev/null
+++ b/host/include/loongarch64/host/load-extract-al16-al8.h
@@ -0,0 +1,39 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ * Atomic extract 64 from 128-bit, LoongArch version.
+ *
+ * Copyright (C) 2023 Linaro, Ltd.
+ */
+
+#ifndef LOONGARCH_LOAD_EXTRACT_AL16_AL8_H
+#define LOONGARCH_LOAD_EXTRACT_AL16_AL8_H
+
+#include "host/cpuinfo.h"
+#include "tcg/debug-assert.h"
+
+/**
+ * load_atom_extract_al16_or_al8:
+ * @pv: host address
+ * @s: object size in bytes, @s <= 8.
+ *
+ * Load @s bytes from @pv, when pv % s != 0. If [p, p+s-1] does not
+ * cross an 16-byte boundary then the access must be 16-byte atomic,
+ * otherwise the access must be 8-byte atomic.
+ */
+static inline uint64_t load_atom_extract_al16_or_al8(void *pv, int s)
+{
+ uintptr_t pi = (uintptr_t)pv;
+ Int128 *ptr_align = (Int128 *)(pi & ~7);
+ int shr = (pi & 7) * 8;
+ uint64_t l, h;
+
+ tcg_debug_assert(HAVE_ATOMIC128_RO);
+ asm("vld $vr0, %2, 0\n\t"
+ "vpickve2gr.d %0, $vr0, 0\n\t"
+ "vpickve2gr.d %1, $vr0, 1"
+ : "=r"(l), "=r"(h) : "r"(ptr_align), "m"(*ptr_align) : "f0");
+
+ return (l >> shr) | (h << (-shr & 63));
+}
+
+#endif /* LOONGARCH_LOAD_EXTRACT_AL16_AL8_H */
diff --git a/host/include/loongarch64/host/store-insert-al16.h b/host/include/loongarch64/host/store-insert-al16.h
new file mode 100644
index 0000000000..919fd8d744
--- /dev/null
+++ b/host/include/loongarch64/host/store-insert-al16.h
@@ -0,0 +1,12 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ * Atomic store insert into 128-bit, LoongArch version.
+ */
+
+#ifndef LOONGARCH_STORE_INSERT_AL16_H
+#define LOONGARCH_STORE_INSERT_AL16_H
+
+void store_atom_insert_al16(Int128 *ps, Int128 val, Int128 msk)
+ QEMU_ERROR("unsupported atomic");
+
+#endif /* LOONGARCH_STORE_INSERT_AL16_H */
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 07/35] accel/tcg: Remove redundant case in store_atom_16
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (5 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 06/35] host/include/loongarch64: Add atomic16 load and store Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 08/35] accel/tcg: Fix condition for store_atom_insert_al16 Richard Henderson
` (29 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
We handled the HAVE_ATOMIC128_RW case with atomic16_set at the top of
the function; the only thing left for a host without that support is
to fall through to cpu_loop_exit_atomic.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230916220151.526140-7-richard.henderson@linaro.org>
---
accel/tcg/ldst_atomicity.c.inc | 4 ----
1 file changed, 4 deletions(-)
diff --git a/accel/tcg/ldst_atomicity.c.inc b/accel/tcg/ldst_atomicity.c.inc
index 1cf5b92166..e8f97506fa 100644
--- a/accel/tcg/ldst_atomicity.c.inc
+++ b/accel/tcg/ldst_atomicity.c.inc
@@ -1103,10 +1103,6 @@ static void store_atom_16(CPUState *cpu, uintptr_t ra,
}
break;
case MO_128:
- if (HAVE_ATOMIC128_RW) {
- atomic16_set(pv, val);
- return;
- }
break;
default:
g_assert_not_reached();
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 08/35] accel/tcg: Fix condition for store_atom_insert_al16
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (6 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 07/35] accel/tcg: Remove redundant case in store_atom_16 Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 09/35] tcg: Mark tcg_gen_op* as noinline Richard Henderson
` (28 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Store bytes under a mask is fundamentally a cmpxchg, not a straight store.
Use HAVE_CMPXCHG128 instead of HAVE_ATOMIC128_RW.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230916220151.526140-8-richard.henderson@linaro.org>
---
accel/tcg/cputlb.c | 2 +-
accel/tcg/ldst_atomicity.c.inc | 10 +++++-----
2 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index 13986820fe..f35c5f359b 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -2691,7 +2691,7 @@ static uint64_t do_st16_leN(CPUState *cpu, MMULookupPageData *p,
case MO_ATOM_WITHIN16_PAIR:
/* Since size > 8, this is the half that must be atomic. */
- if (!HAVE_ATOMIC128_RW) {
+ if (!HAVE_CMPXCHG128) {
cpu_loop_exit_atomic(cpu, ra);
}
return store_whole_le16(p->haddr, p->size, val_le);
diff --git a/accel/tcg/ldst_atomicity.c.inc b/accel/tcg/ldst_atomicity.c.inc
index e8f97506fa..33a04dec52 100644
--- a/accel/tcg/ldst_atomicity.c.inc
+++ b/accel/tcg/ldst_atomicity.c.inc
@@ -825,7 +825,7 @@ static uint64_t store_whole_le16(void *pv, int size, Int128 val_le)
int sh = o * 8;
Int128 m, v;
- qemu_build_assert(HAVE_ATOMIC128_RW);
+ qemu_build_assert(HAVE_CMPXCHG128);
/* Like MAKE_64BIT_MASK(0, sz), but larger. */
if (sz <= 64) {
@@ -887,7 +887,7 @@ static void store_atom_2(CPUState *cpu, uintptr_t ra,
return;
}
} else if ((pi & 15) == 7) {
- if (HAVE_ATOMIC128_RW) {
+ if (HAVE_CMPXCHG128) {
Int128 v = int128_lshift(int128_make64(val), 56);
Int128 m = int128_lshift(int128_make64(0xffff), 56);
store_atom_insert_al16(pv - 7, v, m);
@@ -956,7 +956,7 @@ static void store_atom_4(CPUState *cpu, uintptr_t ra,
return;
}
} else {
- if (HAVE_ATOMIC128_RW) {
+ if (HAVE_CMPXCHG128) {
store_whole_le16(pv, 4, int128_make64(cpu_to_le32(val)));
return;
}
@@ -1021,7 +1021,7 @@ static void store_atom_8(CPUState *cpu, uintptr_t ra,
}
break;
case MO_64:
- if (HAVE_ATOMIC128_RW) {
+ if (HAVE_CMPXCHG128) {
store_whole_le16(pv, 8, int128_make64(cpu_to_le64(val)));
return;
}
@@ -1076,7 +1076,7 @@ static void store_atom_16(CPUState *cpu, uintptr_t ra,
}
break;
case -MO_64:
- if (HAVE_ATOMIC128_RW) {
+ if (HAVE_CMPXCHG128) {
uint64_t val_le;
int s2 = pi & 15;
int s1 = 16 - s2;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 09/35] tcg: Mark tcg_gen_op* as noinline
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (7 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 08/35] accel/tcg: Fix condition for store_atom_insert_al16 Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 10/35] tcg: Move tcg_gen_op* out of line Richard Henderson
` (27 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Encourage the compiler to tail-call rather than inline
across the dozens of opcode expanders.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-2-richard.henderson@linaro.org>
---
tcg/tcg-op.c | 22 ++++++++++++++--------
1 file changed, 14 insertions(+), 8 deletions(-)
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 828eb9ee46..9aed19e957 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -31,20 +31,26 @@
#include "tcg-internal.h"
-void tcg_gen_op1(TCGOpcode opc, TCGArg a1)
+/*
+ * Encourage the compiler to tail-call to a function, rather than inlining.
+ * Minimizes code size across 99 bottles of beer on the wall.
+ */
+#define NI __attribute__((noinline))
+
+void NI tcg_gen_op1(TCGOpcode opc, TCGArg a1)
{
TCGOp *op = tcg_emit_op(opc, 1);
op->args[0] = a1;
}
-void tcg_gen_op2(TCGOpcode opc, TCGArg a1, TCGArg a2)
+void NI tcg_gen_op2(TCGOpcode opc, TCGArg a1, TCGArg a2)
{
TCGOp *op = tcg_emit_op(opc, 2);
op->args[0] = a1;
op->args[1] = a2;
}
-void tcg_gen_op3(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3)
+void NI tcg_gen_op3(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3)
{
TCGOp *op = tcg_emit_op(opc, 3);
op->args[0] = a1;
@@ -52,7 +58,7 @@ void tcg_gen_op3(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3)
op->args[2] = a3;
}
-void tcg_gen_op4(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3, TCGArg a4)
+void NI tcg_gen_op4(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3, TCGArg a4)
{
TCGOp *op = tcg_emit_op(opc, 4);
op->args[0] = a1;
@@ -61,8 +67,8 @@ void tcg_gen_op4(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3, TCGArg a4)
op->args[3] = a4;
}
-void tcg_gen_op5(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
- TCGArg a4, TCGArg a5)
+void NI tcg_gen_op5(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
+ TCGArg a4, TCGArg a5)
{
TCGOp *op = tcg_emit_op(opc, 5);
op->args[0] = a1;
@@ -72,8 +78,8 @@ void tcg_gen_op5(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
op->args[4] = a5;
}
-void tcg_gen_op6(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
- TCGArg a4, TCGArg a5, TCGArg a6)
+void NI tcg_gen_op6(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
+ TCGArg a4, TCGArg a5, TCGArg a6)
{
TCGOp *op = tcg_emit_op(opc, 6);
op->args[0] = a1;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 10/35] tcg: Move tcg_gen_op* out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (8 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 09/35] tcg: Mark tcg_gen_op* as noinline Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 11/35] tcg: Move generic expanders " Richard Henderson
` (26 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
In addition to moving out of line, with CONFIG_DEBUG_TCG
mark them all noinline.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-3-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 252 +++++++-----------------------------
tcg/tcg-op.c | 208 +++++++++++++++++++++++++++++
2 files changed, 252 insertions(+), 208 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index 677aea6dd1..e093f541a7 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -25,214 +25,50 @@ void vec_gen_2(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg);
void vec_gen_3(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg);
void vec_gen_4(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg, TCGArg);
-static inline void tcg_gen_op1_i32(TCGOpcode opc, TCGv_i32 a1)
-{
- tcg_gen_op1(opc, tcgv_i32_arg(a1));
-}
-
-static inline void tcg_gen_op1_i64(TCGOpcode opc, TCGv_i64 a1)
-{
- tcg_gen_op1(opc, tcgv_i64_arg(a1));
-}
-
-static inline void tcg_gen_op1i(TCGOpcode opc, TCGArg a1)
-{
- tcg_gen_op1(opc, a1);
-}
-
-static inline void tcg_gen_op2_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2)
-{
- tcg_gen_op2(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2));
-}
-
-static inline void tcg_gen_op2_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2)
-{
- tcg_gen_op2(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2));
-}
-
-static inline void tcg_gen_op2i_i32(TCGOpcode opc, TCGv_i32 a1, TCGArg a2)
-{
- tcg_gen_op2(opc, tcgv_i32_arg(a1), a2);
-}
-
-static inline void tcg_gen_op2i_i64(TCGOpcode opc, TCGv_i64 a1, TCGArg a2)
-{
- tcg_gen_op2(opc, tcgv_i64_arg(a1), a2);
-}
-
-static inline void tcg_gen_op2ii(TCGOpcode opc, TCGArg a1, TCGArg a2)
-{
- tcg_gen_op2(opc, a1, a2);
-}
-
-static inline void tcg_gen_op3_i32(TCGOpcode opc, TCGv_i32 a1,
- TCGv_i32 a2, TCGv_i32 a3)
-{
- tcg_gen_op3(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), tcgv_i32_arg(a3));
-}
-
-static inline void tcg_gen_op3_i64(TCGOpcode opc, TCGv_i64 a1,
- TCGv_i64 a2, TCGv_i64 a3)
-{
- tcg_gen_op3(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), tcgv_i64_arg(a3));
-}
-
-static inline void tcg_gen_op3i_i32(TCGOpcode opc, TCGv_i32 a1,
- TCGv_i32 a2, TCGArg a3)
-{
- tcg_gen_op3(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), a3);
-}
-
-static inline void tcg_gen_op3i_i64(TCGOpcode opc, TCGv_i64 a1,
- TCGv_i64 a2, TCGArg a3)
-{
- tcg_gen_op3(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), a3);
-}
-
-static inline void tcg_gen_ldst_op_i32(TCGOpcode opc, TCGv_i32 val,
- TCGv_ptr base, TCGArg offset)
-{
- tcg_gen_op3(opc, tcgv_i32_arg(val), tcgv_ptr_arg(base), offset);
-}
-
-static inline void tcg_gen_ldst_op_i64(TCGOpcode opc, TCGv_i64 val,
- TCGv_ptr base, TCGArg offset)
-{
- tcg_gen_op3(opc, tcgv_i64_arg(val), tcgv_ptr_arg(base), offset);
-}
-
-static inline void tcg_gen_op4_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4)
-{
- tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4));
-}
-
-static inline void tcg_gen_op4_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4)
-{
- tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4));
-}
-
-static inline void tcg_gen_op4i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGArg a4)
-{
- tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), a4);
-}
-
-static inline void tcg_gen_op4i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGArg a4)
-{
- tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), a4);
-}
-
-static inline void tcg_gen_op4ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGArg a3, TCGArg a4)
-{
- tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), a3, a4);
-}
-
-static inline void tcg_gen_op4ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGArg a3, TCGArg a4)
-{
- tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), a3, a4);
-}
-
-static inline void tcg_gen_op5_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4, TCGv_i32 a5)
-{
- tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5));
-}
-
-static inline void tcg_gen_op5_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4, TCGv_i64 a5)
-{
- tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5));
-}
-
-static inline void tcg_gen_op5i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4, TCGArg a5)
-{
- tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4), a5);
-}
-
-static inline void tcg_gen_op5i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4, TCGArg a5)
-{
- tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), a5);
-}
-
-static inline void tcg_gen_op5ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGArg a4, TCGArg a5)
-{
- tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), a4, a5);
-}
-
-static inline void tcg_gen_op5ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGArg a4, TCGArg a5)
-{
- tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), a4, a5);
-}
-
-static inline void tcg_gen_op6_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4,
- TCGv_i32 a5, TCGv_i32 a6)
-{
- tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5),
- tcgv_i32_arg(a6));
-}
-
-static inline void tcg_gen_op6_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4,
- TCGv_i64 a5, TCGv_i64 a6)
-{
- tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5),
- tcgv_i64_arg(a6));
-}
-
-static inline void tcg_gen_op6i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4,
- TCGv_i32 a5, TCGArg a6)
-{
- tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5), a6);
-}
-
-static inline void tcg_gen_op6i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4,
- TCGv_i64 a5, TCGArg a6)
-{
- tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5), a6);
-}
-
-static inline void tcg_gen_op6ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4,
- TCGArg a5, TCGArg a6)
-{
- tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4), a5, a6);
-}
-
-static inline void tcg_gen_op6ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4,
- TCGArg a5, TCGArg a6)
-{
- tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), a5, a6);
-}
-
+void tcg_gen_op1_i32(TCGOpcode, TCGv_i32);
+void tcg_gen_op1_i64(TCGOpcode, TCGv_i64);
+void tcg_gen_op1i(TCGOpcode, TCGArg);
+void tcg_gen_op2_i32(TCGOpcode, TCGv_i32, TCGv_i32);
+void tcg_gen_op2_i64(TCGOpcode, TCGv_i64, TCGv_i64);
+void tcg_gen_op2i_i32(TCGOpcode, TCGv_i32, TCGArg);
+void tcg_gen_op2i_i64(TCGOpcode, TCGv_i64, TCGArg);
+void tcg_gen_op2ii(TCGOpcode, TCGArg, TCGArg);
+void tcg_gen_op3_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32);
+void tcg_gen_op3_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64);
+void tcg_gen_op3i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGArg);
+void tcg_gen_op3i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGArg);
+void tcg_gen_ldst_op_i32(TCGOpcode, TCGv_i32, TCGv_ptr, TCGArg);
+void tcg_gen_ldst_op_i64(TCGOpcode, TCGv_i64, TCGv_ptr, TCGArg);
+void tcg_gen_op4_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32, TCGv_i32);
+void tcg_gen_op4_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64, TCGv_i64);
+void tcg_gen_op4i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32, TCGArg);
+void tcg_gen_op4i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64, TCGArg);
+void tcg_gen_op4ii_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGArg, TCGArg);
+void tcg_gen_op4ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGArg, TCGArg);
+void tcg_gen_op5_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
+ TCGv_i32, TCGv_i32);
+void tcg_gen_op5_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
+ TCGv_i64, TCGv_i64);
+void tcg_gen_op5i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
+ TCGv_i32, TCGArg);
+void tcg_gen_op5i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
+ TCGv_i64, TCGArg);
+void tcg_gen_op5ii_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
+ TCGArg, TCGArg);
+void tcg_gen_op5ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
+ TCGArg, TCGArg);
+void tcg_gen_op6_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
+ TCGv_i32, TCGv_i32, TCGv_i32);
+void tcg_gen_op6_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
+ TCGv_i64, TCGv_i64, TCGv_i64);
+void tcg_gen_op6i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
+ TCGv_i32, TCGv_i32, TCGArg);
+void tcg_gen_op6i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
+ TCGv_i64, TCGv_i64, TCGArg);
+void tcg_gen_op6ii_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
+ TCGv_i32, TCGArg, TCGArg);
+void tcg_gen_op6ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
+ TCGv_i64, TCGArg, TCGArg);
/* Generic ops. */
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 9aed19e957..6c826b46b0 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -90,6 +90,214 @@ void NI tcg_gen_op6(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
op->args[5] = a6;
}
+/*
+ * With CONFIG_DEBUG_TCG, tcgv_*_tmp via tcgv_*_arg, is an out-of-line
+ * assertion check. Force tail calls to avoid too much code expansion.
+ */
+#ifdef CONFIG_DEBUG_TCG
+# define DNI NI
+#else
+# define DNI
+#endif
+
+void DNI tcg_gen_op1_i32(TCGOpcode opc, TCGv_i32 a1)
+{
+ tcg_gen_op1(opc, tcgv_i32_arg(a1));
+}
+
+void DNI tcg_gen_op1_i64(TCGOpcode opc, TCGv_i64 a1)
+{
+ tcg_gen_op1(opc, tcgv_i64_arg(a1));
+}
+
+void DNI tcg_gen_op1i(TCGOpcode opc, TCGArg a1)
+{
+ tcg_gen_op1(opc, a1);
+}
+
+void DNI tcg_gen_op2_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2)
+{
+ tcg_gen_op2(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2));
+}
+
+void DNI tcg_gen_op2_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2)
+{
+ tcg_gen_op2(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2));
+}
+
+void DNI tcg_gen_op2i_i32(TCGOpcode opc, TCGv_i32 a1, TCGArg a2)
+{
+ tcg_gen_op2(opc, tcgv_i32_arg(a1), a2);
+}
+
+void DNI tcg_gen_op2i_i64(TCGOpcode opc, TCGv_i64 a1, TCGArg a2)
+{
+ tcg_gen_op2(opc, tcgv_i64_arg(a1), a2);
+}
+
+void DNI tcg_gen_op2ii(TCGOpcode opc, TCGArg a1, TCGArg a2)
+{
+ tcg_gen_op2(opc, a1, a2);
+}
+
+void DNI tcg_gen_op3_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGv_i32 a3)
+{
+ tcg_gen_op3(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), tcgv_i32_arg(a3));
+}
+
+void DNI tcg_gen_op3_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGv_i64 a3)
+{
+ tcg_gen_op3(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), tcgv_i64_arg(a3));
+}
+
+void DNI tcg_gen_op3i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGArg a3)
+{
+ tcg_gen_op3(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), a3);
+}
+
+void DNI tcg_gen_op3i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGArg a3)
+{
+ tcg_gen_op3(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), a3);
+}
+
+void DNI tcg_gen_ldst_op_i32(TCGOpcode opc, TCGv_i32 val,
+ TCGv_ptr base, TCGArg offset)
+{
+ tcg_gen_op3(opc, tcgv_i32_arg(val), tcgv_ptr_arg(base), offset);
+}
+
+void DNI tcg_gen_ldst_op_i64(TCGOpcode opc, TCGv_i64 val,
+ TCGv_ptr base, TCGArg offset)
+{
+ tcg_gen_op3(opc, tcgv_i64_arg(val), tcgv_ptr_arg(base), offset);
+}
+
+void DNI tcg_gen_op4_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4)
+{
+ tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), tcgv_i32_arg(a4));
+}
+
+void DNI tcg_gen_op4_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4)
+{
+ tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), tcgv_i64_arg(a4));
+}
+
+void DNI tcg_gen_op4i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGArg a4)
+{
+ tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), a4);
+}
+
+void DNI tcg_gen_op4i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGArg a4)
+{
+ tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), a4);
+}
+
+void DNI tcg_gen_op4ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGArg a3, TCGArg a4)
+{
+ tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), a3, a4);
+}
+
+void DNI tcg_gen_op4ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGArg a3, TCGArg a4)
+{
+ tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), a3, a4);
+}
+
+void DNI tcg_gen_op5_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4, TCGv_i32 a5)
+{
+ tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5));
+}
+
+void DNI tcg_gen_op5_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4, TCGv_i64 a5)
+{
+ tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5));
+}
+
+void DNI tcg_gen_op5i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4, TCGArg a5)
+{
+ tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), tcgv_i32_arg(a4), a5);
+}
+
+void DNI tcg_gen_op5i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4, TCGArg a5)
+{
+ tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), tcgv_i64_arg(a4), a5);
+}
+
+void DNI tcg_gen_op5ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGArg a4, TCGArg a5)
+{
+ tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), a4, a5);
+}
+
+void DNI tcg_gen_op5ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGArg a4, TCGArg a5)
+{
+ tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), a4, a5);
+}
+
+void DNI tcg_gen_op6_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGv_i32 a3,
+ TCGv_i32 a4, TCGv_i32 a5, TCGv_i32 a6)
+{
+ tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5),
+ tcgv_i32_arg(a6));
+}
+
+void DNI tcg_gen_op6_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGv_i64 a3,
+ TCGv_i64 a4, TCGv_i64 a5, TCGv_i64 a6)
+{
+ tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5),
+ tcgv_i64_arg(a6));
+}
+
+void DNI tcg_gen_op6i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGv_i32 a3,
+ TCGv_i32 a4, TCGv_i32 a5, TCGArg a6)
+{
+ tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5), a6);
+}
+
+void DNI tcg_gen_op6i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGv_i64 a3,
+ TCGv_i64 a4, TCGv_i64 a5, TCGArg a6)
+{
+ tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5), a6);
+}
+
+void DNI tcg_gen_op6ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4, TCGArg a5, TCGArg a6)
+{
+ tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
+ tcgv_i32_arg(a3), tcgv_i32_arg(a4), a5, a6);
+}
+
+void DNI tcg_gen_op6ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4, TCGArg a5, TCGArg a6)
+{
+ tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
+ tcgv_i64_arg(a3), tcgv_i64_arg(a4), a5, a6);
+}
+
/* Generic ops. */
static void add_last_as_label_use(TCGLabel *l)
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 11/35] tcg: Move generic expanders out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (9 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 10/35] tcg: Move tcg_gen_op* out of line Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 12/35] tcg: Move 32-bit " Richard Henderson
` (25 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-4-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 19 +++----------------
tcg/tcg-op.c | 16 ++++++++++++++++
2 files changed, 19 insertions(+), 16 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index e093f541a7..2134961a98 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -72,12 +72,7 @@ void tcg_gen_op6ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
/* Generic ops. */
-static inline void gen_set_label(TCGLabel *l)
-{
- l->present = 1;
- tcg_gen_op1(INDEX_op_set_label, label_arg(l));
-}
-
+void gen_set_label(TCGLabel *l);
void tcg_gen_br(TCGLabel *l);
void tcg_gen_mb(TCGBar);
@@ -121,16 +116,8 @@ void tcg_gen_goto_tb(unsigned idx);
*/
void tcg_gen_lookup_and_goto_ptr(void);
-static inline void tcg_gen_plugin_cb_start(unsigned from, unsigned type,
- unsigned wr)
-{
- tcg_gen_op3(INDEX_op_plugin_cb_start, from, type, wr);
-}
-
-static inline void tcg_gen_plugin_cb_end(void)
-{
- tcg_emit_op(INDEX_op_plugin_cb_end, 0);
-}
+void tcg_gen_plugin_cb_start(unsigned from, unsigned type, unsigned wr);
+void tcg_gen_plugin_cb_end(void);
/* 32 bit ops */
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 6c826b46b0..a8cbad212d 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -300,6 +300,12 @@ void DNI tcg_gen_op6ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
/* Generic ops. */
+void gen_set_label(TCGLabel *l)
+{
+ l->present = 1;
+ tcg_gen_op1(INDEX_op_set_label, label_arg(l));
+}
+
static void add_last_as_label_use(TCGLabel *l)
{
TCGLabelUse *u = tcg_malloc(sizeof(TCGLabelUse));
@@ -333,6 +339,16 @@ void tcg_gen_mb(TCGBar mb_type)
}
}
+void tcg_gen_plugin_cb_start(unsigned from, unsigned type, unsigned wr)
+{
+ tcg_gen_op3(INDEX_op_plugin_cb_start, from, type, wr);
+}
+
+void tcg_gen_plugin_cb_end(void)
+{
+ tcg_emit_op(INDEX_op_plugin_cb_end, 0);
+}
+
/* 32 bit ops */
void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg)
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 12/35] tcg: Move 32-bit expanders out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (10 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 11/35] tcg: Move generic expanders " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 13/35] tcg: Move 64-bit " Richard Henderson
` (24 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-5-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 140 ++++++------------------------------
tcg/tcg-op.c | 116 ++++++++++++++++++++++++++++++
2 files changed, 137 insertions(+), 119 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index 2134961a98..cdaa1415d1 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -197,128 +197,30 @@ void tcg_gen_abs_i32(TCGv_i32, TCGv_i32);
/* Replicate a value of size @vece from @in to all the lanes in @out */
void tcg_gen_dup_i32(unsigned vece, TCGv_i32 out, TCGv_i32 in);
-static inline void tcg_gen_discard_i32(TCGv_i32 arg)
-{
- tcg_gen_op1_i32(INDEX_op_discard, arg);
-}
+void tcg_gen_discard_i32(TCGv_i32 arg);
+void tcg_gen_mov_i32(TCGv_i32 ret, TCGv_i32 arg);
-static inline void tcg_gen_mov_i32(TCGv_i32 ret, TCGv_i32 arg)
-{
- if (ret != arg) {
- tcg_gen_op2_i32(INDEX_op_mov_i32, ret, arg);
- }
-}
+void tcg_gen_ld8u_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset);
+void tcg_gen_ld8s_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset);
+void tcg_gen_ld16u_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset);
+void tcg_gen_ld16s_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset);
+void tcg_gen_ld_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset);
-static inline void tcg_gen_ld8u_i32(TCGv_i32 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_ld8u_i32, ret, arg2, offset);
-}
+void tcg_gen_st8_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset);
+void tcg_gen_st16_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset);
+void tcg_gen_st_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset);
-static inline void tcg_gen_ld8s_i32(TCGv_i32 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_ld8s_i32, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld16u_i32(TCGv_i32 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_ld16u_i32, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld16s_i32(TCGv_i32 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_ld16s_i32, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld_i32(TCGv_i32 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_ld_i32, ret, arg2, offset);
-}
-
-static inline void tcg_gen_st8_i32(TCGv_i32 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_st8_i32, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_st16_i32(TCGv_i32 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_st16_i32, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_st_i32(TCGv_i32 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i32(INDEX_op_st_i32, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_add_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_add_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_sub_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_sub_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_and_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_and_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_or_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_or_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_xor_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_xor_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_shl_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_shl_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_shr_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_shr_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_sar_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_sar_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_mul_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
-{
- tcg_gen_op3_i32(INDEX_op_mul_i32, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_neg_i32(TCGv_i32 ret, TCGv_i32 arg)
-{
- if (TCG_TARGET_HAS_neg_i32) {
- tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg);
- } else {
- tcg_gen_subfi_i32(ret, 0, arg);
- }
-}
-
-static inline void tcg_gen_not_i32(TCGv_i32 ret, TCGv_i32 arg)
-{
- if (TCG_TARGET_HAS_not_i32) {
- tcg_gen_op2_i32(INDEX_op_not_i32, ret, arg);
- } else {
- tcg_gen_xori_i32(ret, arg, -1);
- }
-}
+void tcg_gen_add_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_sub_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_and_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_or_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_xor_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_shl_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_shr_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_sar_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_mul_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2);
+void tcg_gen_neg_i32(TCGv_i32 ret, TCGv_i32 arg);
+void tcg_gen_not_i32(TCGv_i32 ret, TCGv_i32 arg);
/* 64 bit ops */
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index a8cbad212d..ab6281ebec 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -351,11 +351,28 @@ void tcg_gen_plugin_cb_end(void)
/* 32 bit ops */
+void tcg_gen_discard_i32(TCGv_i32 arg)
+{
+ tcg_gen_op1_i32(INDEX_op_discard, arg);
+}
+
+void tcg_gen_mov_i32(TCGv_i32 ret, TCGv_i32 arg)
+{
+ if (ret != arg) {
+ tcg_gen_op2_i32(INDEX_op_mov_i32, ret, arg);
+ }
+}
+
void tcg_gen_movi_i32(TCGv_i32 ret, int32_t arg)
{
tcg_gen_mov_i32(ret, tcg_constant_i32(arg));
}
+void tcg_gen_add_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_add_i32, ret, arg1, arg2);
+}
+
void tcg_gen_addi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
/* some cases can be optimized here */
@@ -366,6 +383,11 @@ void tcg_gen_addi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
}
}
+void tcg_gen_sub_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_sub_i32, ret, arg1, arg2);
+}
+
void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2)
{
if (arg1 == 0 && TCG_TARGET_HAS_neg_i32) {
@@ -386,6 +408,20 @@ void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
}
}
+void tcg_gen_neg_i32(TCGv_i32 ret, TCGv_i32 arg)
+{
+ if (TCG_TARGET_HAS_neg_i32) {
+ tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg);
+ } else {
+ tcg_gen_subfi_i32(ret, 0, arg);
+ }
+}
+
+void tcg_gen_and_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_and_i32, ret, arg1, arg2);
+}
+
void tcg_gen_andi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
/* Some cases can be optimized here. */
@@ -414,6 +450,11 @@ void tcg_gen_andi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
tcg_gen_and_i32(ret, arg1, tcg_constant_i32(arg2));
}
+void tcg_gen_or_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_or_i32, ret, arg1, arg2);
+}
+
void tcg_gen_ori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
/* Some cases can be optimized here. */
@@ -426,6 +467,11 @@ void tcg_gen_ori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
}
}
+void tcg_gen_xor_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_xor_i32, ret, arg1, arg2);
+}
+
void tcg_gen_xori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
/* Some cases can be optimized here. */
@@ -439,6 +485,20 @@ void tcg_gen_xori_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
}
}
+void tcg_gen_not_i32(TCGv_i32 ret, TCGv_i32 arg)
+{
+ if (TCG_TARGET_HAS_not_i32) {
+ tcg_gen_op2_i32(INDEX_op_not_i32, ret, arg);
+ } else {
+ tcg_gen_xori_i32(ret, arg, -1);
+ }
+}
+
+void tcg_gen_shl_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_shl_i32, ret, arg1, arg2);
+}
+
void tcg_gen_shli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
tcg_debug_assert(arg2 >= 0 && arg2 < 32);
@@ -449,6 +509,11 @@ void tcg_gen_shli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
}
}
+void tcg_gen_shr_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_shr_i32, ret, arg1, arg2);
+}
+
void tcg_gen_shri_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
tcg_debug_assert(arg2 >= 0 && arg2 < 32);
@@ -459,6 +524,11 @@ void tcg_gen_shri_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
}
}
+void tcg_gen_sar_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_sar_i32, ret, arg1, arg2);
+}
+
void tcg_gen_sari_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
tcg_debug_assert(arg2 >= 0 && arg2 < 32);
@@ -527,6 +597,11 @@ void tcg_gen_negsetcondi_i32(TCGCond cond, TCGv_i32 ret,
tcg_gen_negsetcond_i32(cond, ret, arg1, tcg_constant_i32(arg2));
}
+void tcg_gen_mul_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
+{
+ tcg_gen_op3_i32(INDEX_op_mul_i32, ret, arg1, arg2);
+}
+
void tcg_gen_muli_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
if (arg2 == 0) {
@@ -1385,6 +1460,47 @@ void tcg_gen_abs_i32(TCGv_i32 ret, TCGv_i32 a)
tcg_temp_free_i32(t);
}
+void tcg_gen_ld8u_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_ld8u_i32, ret, arg2, offset);
+}
+
+void tcg_gen_ld8s_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_ld8s_i32, ret, arg2, offset);
+}
+
+void tcg_gen_ld16u_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_ld16u_i32, ret, arg2, offset);
+}
+
+void tcg_gen_ld16s_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_ld16s_i32, ret, arg2, offset);
+}
+
+void tcg_gen_ld_i32(TCGv_i32 ret, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_ld_i32, ret, arg2, offset);
+}
+
+void tcg_gen_st8_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_st8_i32, arg1, arg2, offset);
+}
+
+void tcg_gen_st16_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_st16_i32, arg1, arg2, offset);
+}
+
+void tcg_gen_st_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset)
+{
+ tcg_gen_ldst_op_i32(INDEX_op_st_i32, arg1, arg2, offset);
+}
+
+
/* 64-bit ops */
#if TCG_TARGET_REG_BITS == 32
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 13/35] tcg: Move 64-bit expanders out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (11 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 12/35] tcg: Move 32-bit " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 14/35] tcg: Move vec_gen_* declarations to tcg-internal.h Richard Henderson
` (23 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
This one is more complicated, combining 32-bit and 64-bit
expansion with C if instead of preprocessor #if.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-6-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 144 +---------------------
tcg/tcg-op.c | 231 +++++++++++++++++++++++++-----------
2 files changed, 169 insertions(+), 206 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index cdaa1415d1..f5ec54f874 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -305,130 +305,6 @@ void tcg_gen_abs_i64(TCGv_i64, TCGv_i64);
/* Replicate a value of size @vece from @in to all the lanes in @out */
void tcg_gen_dup_i64(unsigned vece, TCGv_i64 out, TCGv_i64 in);
-#if TCG_TARGET_REG_BITS == 64
-static inline void tcg_gen_discard_i64(TCGv_i64 arg)
-{
- tcg_gen_op1_i64(INDEX_op_discard, arg);
-}
-
-static inline void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg)
-{
- if (ret != arg) {
- tcg_gen_op2_i64(INDEX_op_mov_i64, ret, arg);
- }
-}
-
-static inline void tcg_gen_ld8u_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld8u_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld8s_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld8s_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld16u_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld16u_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld16s_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld16s_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld32u_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld32u_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld32s_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld32s_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_ld_i64(TCGv_i64 ret, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_ld_i64, ret, arg2, offset);
-}
-
-static inline void tcg_gen_st8_i64(TCGv_i64 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_st8_i64, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_st16_i64(TCGv_i64 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_st16_i64, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_st32_i64(TCGv_i64 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_st32_i64, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_st_i64(TCGv_i64 arg1, TCGv_ptr arg2,
- tcg_target_long offset)
-{
- tcg_gen_ldst_op_i64(INDEX_op_st_i64, arg1, arg2, offset);
-}
-
-static inline void tcg_gen_add_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_add_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_sub_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_sub_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_and_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_and_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_or_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_or_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_xor_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_xor_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_shl_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_shl_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_shr_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_shr_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_sar_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_sar_i64, ret, arg1, arg2);
-}
-
-static inline void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
-{
- tcg_gen_op3_i64(INDEX_op_mul_i64, ret, arg1, arg2);
-}
-#else /* TCG_TARGET_REG_BITS == 32 */
void tcg_gen_st8_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset);
void tcg_gen_st16_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset);
void tcg_gen_st32_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset);
@@ -453,16 +329,8 @@ void tcg_gen_shl_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2);
void tcg_gen_shr_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2);
void tcg_gen_sar_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2);
void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2);
-#endif /* TCG_TARGET_REG_BITS */
+void tcg_gen_neg_i64(TCGv_i64 ret, TCGv_i64 arg);
-static inline void tcg_gen_neg_i64(TCGv_i64 ret, TCGv_i64 arg)
-{
- if (TCG_TARGET_HAS_neg_i64) {
- tcg_gen_op2_i64(INDEX_op_neg_i64, ret, arg);
- } else {
- tcg_gen_subfi_i64(ret, 0, arg);
- }
-}
/* Size changing operations. */
@@ -473,19 +341,17 @@ void tcg_gen_extrl_i64_i32(TCGv_i32 ret, TCGv_i64 arg);
void tcg_gen_extrh_i64_i32(TCGv_i32 ret, TCGv_i64 arg);
void tcg_gen_extr_i64_i32(TCGv_i32 lo, TCGv_i32 hi, TCGv_i64 arg);
void tcg_gen_extr32_i64(TCGv_i64 lo, TCGv_i64 hi, TCGv_i64 arg);
+void tcg_gen_concat32_i64(TCGv_i64 ret, TCGv_i64 lo, TCGv_i64 hi);
-void tcg_gen_mov_i128(TCGv_i128 dst, TCGv_i128 src);
void tcg_gen_extr_i128_i64(TCGv_i64 lo, TCGv_i64 hi, TCGv_i128 arg);
void tcg_gen_concat_i64_i128(TCGv_i128 ret, TCGv_i64 lo, TCGv_i64 hi);
+/* 128 bit ops */
+
+void tcg_gen_mov_i128(TCGv_i128 dst, TCGv_i128 src);
void tcg_gen_ld_i128(TCGv_i128 ret, TCGv_ptr base, tcg_target_long offset);
void tcg_gen_st_i128(TCGv_i128 val, TCGv_ptr base, tcg_target_long offset);
-static inline void tcg_gen_concat32_i64(TCGv_i64 ret, TCGv_i64 lo, TCGv_i64 hi)
-{
- tcg_gen_deposit_i64(ret, lo, hi, 32, 32);
-}
-
/* Local load/store bit ops */
void tcg_gen_qemu_ld_i32_chk(TCGv_i32, TCGTemp *, TCGArg, MemOp, TCGType);
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index ab6281ebec..579a2aab15 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -1503,152 +1503,238 @@ void tcg_gen_st_i32(TCGv_i32 arg1, TCGv_ptr arg2, tcg_target_long offset)
/* 64-bit ops */
-#if TCG_TARGET_REG_BITS == 32
-/* These are all inline for TCG_TARGET_REG_BITS == 64. */
-
void tcg_gen_discard_i64(TCGv_i64 arg)
{
- tcg_gen_discard_i32(TCGV_LOW(arg));
- tcg_gen_discard_i32(TCGV_HIGH(arg));
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op1_i64(INDEX_op_discard, arg);
+ } else {
+ tcg_gen_discard_i32(TCGV_LOW(arg));
+ tcg_gen_discard_i32(TCGV_HIGH(arg));
+ }
}
void tcg_gen_mov_i64(TCGv_i64 ret, TCGv_i64 arg)
{
- TCGTemp *ts = tcgv_i64_temp(arg);
-
- /* Canonicalize TCGv_i64 TEMP_CONST into TCGv_i32 TEMP_CONST. */
- if (ts->kind == TEMP_CONST) {
- tcg_gen_movi_i64(ret, ts->val);
+ if (ret == arg) {
+ return;
+ }
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op2_i64(INDEX_op_mov_i64, ret, arg);
} else {
- tcg_gen_mov_i32(TCGV_LOW(ret), TCGV_LOW(arg));
- tcg_gen_mov_i32(TCGV_HIGH(ret), TCGV_HIGH(arg));
+ TCGTemp *ts = tcgv_i64_temp(arg);
+
+ /* Canonicalize TCGv_i64 TEMP_CONST into TCGv_i32 TEMP_CONST. */
+ if (ts->kind == TEMP_CONST) {
+ tcg_gen_movi_i64(ret, ts->val);
+ } else {
+ tcg_gen_mov_i32(TCGV_LOW(ret), TCGV_LOW(arg));
+ tcg_gen_mov_i32(TCGV_HIGH(ret), TCGV_HIGH(arg));
+ }
}
}
void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg)
{
- tcg_gen_movi_i32(TCGV_LOW(ret), arg);
- tcg_gen_movi_i32(TCGV_HIGH(ret), arg >> 32);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_mov_i64(ret, tcg_constant_i64(arg));
+ } else {
+ tcg_gen_movi_i32(TCGV_LOW(ret), arg);
+ tcg_gen_movi_i32(TCGV_HIGH(ret), arg >> 32);
+ }
}
void tcg_gen_ld8u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_ld8u_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_movi_i32(TCGV_HIGH(ret), 0);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld8u_i64, ret, arg2, offset);
+ } else {
+ tcg_gen_ld8u_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_movi_i32(TCGV_HIGH(ret), 0);
+ }
}
void tcg_gen_ld8s_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_ld8s_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_sari_i32(TCGV_HIGH(ret), TCGV_LOW(ret), 31);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld8s_i64, ret, arg2, offset);
+ } else {
+ tcg_gen_ld8s_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_sari_i32(TCGV_HIGH(ret), TCGV_LOW(ret), 31);
+ }
}
void tcg_gen_ld16u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_ld16u_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_movi_i32(TCGV_HIGH(ret), 0);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld16u_i64, ret, arg2, offset);
+ } else {
+ tcg_gen_ld16u_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_movi_i32(TCGV_HIGH(ret), 0);
+ }
}
void tcg_gen_ld16s_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_ld16s_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_sari_i32(TCGV_HIGH(ret), TCGV_LOW(ret), 31);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld16s_i64, ret, arg2, offset);
+ } else {
+ tcg_gen_ld16s_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_sari_i32(TCGV_HIGH(ret), TCGV_LOW(ret), 31);
+ }
}
void tcg_gen_ld32u_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_movi_i32(TCGV_HIGH(ret), 0);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld32u_i64, ret, arg2, offset);
+ } else {
+ tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_movi_i32(TCGV_HIGH(ret), 0);
+ }
}
void tcg_gen_ld32s_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_sari_i32(TCGV_HIGH(ret), TCGV_LOW(ret), 31);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld32s_i64, ret, arg2, offset);
+ } else {
+ tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_sari_i32(TCGV_HIGH(ret), TCGV_LOW(ret), 31);
+ }
}
void tcg_gen_ld_i64(TCGv_i64 ret, TCGv_ptr arg2, tcg_target_long offset)
{
- /* Since arg2 and ret have different types,
- they cannot be the same temporary */
-#if HOST_BIG_ENDIAN
- tcg_gen_ld_i32(TCGV_HIGH(ret), arg2, offset);
- tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset + 4);
-#else
- tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset);
- tcg_gen_ld_i32(TCGV_HIGH(ret), arg2, offset + 4);
-#endif
+ /*
+ * For 32-bit host, since arg2 and ret have different types,
+ * they cannot be the same temporary -- no chance of overlap.
+ */
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_ld_i64, ret, arg2, offset);
+ } else if (HOST_BIG_ENDIAN) {
+ tcg_gen_ld_i32(TCGV_HIGH(ret), arg2, offset);
+ tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset + 4);
+ } else {
+ tcg_gen_ld_i32(TCGV_LOW(ret), arg2, offset);
+ tcg_gen_ld_i32(TCGV_HIGH(ret), arg2, offset + 4);
+ }
}
void tcg_gen_st8_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_st8_i32(TCGV_LOW(arg1), arg2, offset);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_st8_i64, arg1, arg2, offset);
+ } else {
+ tcg_gen_st8_i32(TCGV_LOW(arg1), arg2, offset);
+ }
}
void tcg_gen_st16_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_st16_i32(TCGV_LOW(arg1), arg2, offset);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_st16_i64, arg1, arg2, offset);
+ } else {
+ tcg_gen_st16_i32(TCGV_LOW(arg1), arg2, offset);
+ }
}
void tcg_gen_st32_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset)
{
- tcg_gen_st_i32(TCGV_LOW(arg1), arg2, offset);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_st32_i64, arg1, arg2, offset);
+ } else {
+ tcg_gen_st_i32(TCGV_LOW(arg1), arg2, offset);
+ }
}
void tcg_gen_st_i64(TCGv_i64 arg1, TCGv_ptr arg2, tcg_target_long offset)
{
-#if HOST_BIG_ENDIAN
- tcg_gen_st_i32(TCGV_HIGH(arg1), arg2, offset);
- tcg_gen_st_i32(TCGV_LOW(arg1), arg2, offset + 4);
-#else
- tcg_gen_st_i32(TCGV_LOW(arg1), arg2, offset);
- tcg_gen_st_i32(TCGV_HIGH(arg1), arg2, offset + 4);
-#endif
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_ldst_op_i64(INDEX_op_st_i64, arg1, arg2, offset);
+ } else if (HOST_BIG_ENDIAN) {
+ tcg_gen_st_i32(TCGV_HIGH(arg1), arg2, offset);
+ tcg_gen_st_i32(TCGV_LOW(arg1), arg2, offset + 4);
+ } else {
+ tcg_gen_st_i32(TCGV_LOW(arg1), arg2, offset);
+ tcg_gen_st_i32(TCGV_HIGH(arg1), arg2, offset + 4);
+ }
}
void tcg_gen_add_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- tcg_gen_add2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), TCGV_LOW(arg1),
- TCGV_HIGH(arg1), TCGV_LOW(arg2), TCGV_HIGH(arg2));
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_add_i64, ret, arg1, arg2);
+ } else {
+ tcg_gen_add2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), TCGV_LOW(arg1),
+ TCGV_HIGH(arg1), TCGV_LOW(arg2), TCGV_HIGH(arg2));
+ }
}
void tcg_gen_sub_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), TCGV_LOW(arg1),
- TCGV_HIGH(arg1), TCGV_LOW(arg2), TCGV_HIGH(arg2));
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_sub_i64, ret, arg1, arg2);
+ } else {
+ tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret), TCGV_LOW(arg1),
+ TCGV_HIGH(arg1), TCGV_LOW(arg2), TCGV_HIGH(arg2));
+ }
}
void tcg_gen_and_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- tcg_gen_and_i32(TCGV_LOW(ret), TCGV_LOW(arg1), TCGV_LOW(arg2));
- tcg_gen_and_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), TCGV_HIGH(arg2));
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_and_i64, ret, arg1, arg2);
+ } else {
+ tcg_gen_and_i32(TCGV_LOW(ret), TCGV_LOW(arg1), TCGV_LOW(arg2));
+ tcg_gen_and_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), TCGV_HIGH(arg2));
+ }
}
void tcg_gen_or_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- tcg_gen_or_i32(TCGV_LOW(ret), TCGV_LOW(arg1), TCGV_LOW(arg2));
- tcg_gen_or_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), TCGV_HIGH(arg2));
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_or_i64, ret, arg1, arg2);
+ } else {
+ tcg_gen_or_i32(TCGV_LOW(ret), TCGV_LOW(arg1), TCGV_LOW(arg2));
+ tcg_gen_or_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), TCGV_HIGH(arg2));
+ }
}
void tcg_gen_xor_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- tcg_gen_xor_i32(TCGV_LOW(ret), TCGV_LOW(arg1), TCGV_LOW(arg2));
- tcg_gen_xor_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), TCGV_HIGH(arg2));
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_xor_i64, ret, arg1, arg2);
+ } else {
+ tcg_gen_xor_i32(TCGV_LOW(ret), TCGV_LOW(arg1), TCGV_LOW(arg2));
+ tcg_gen_xor_i32(TCGV_HIGH(ret), TCGV_HIGH(arg1), TCGV_HIGH(arg2));
+ }
}
void tcg_gen_shl_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- gen_helper_shl_i64(ret, arg1, arg2);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_shl_i64, ret, arg1, arg2);
+ } else {
+ gen_helper_shl_i64(ret, arg1, arg2);
+ }
}
void tcg_gen_shr_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- gen_helper_shr_i64(ret, arg1, arg2);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_shr_i64, ret, arg1, arg2);
+ } else {
+ gen_helper_shr_i64(ret, arg1, arg2);
+ }
}
void tcg_gen_sar_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
{
- gen_helper_sar_i64(ret, arg1, arg2);
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_sar_i64, ret, arg1, arg2);
+ } else {
+ gen_helper_sar_i64(ret, arg1, arg2);
+ }
}
void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
@@ -1656,6 +1742,12 @@ void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
TCGv_i64 t0;
TCGv_i32 t1;
+ if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op3_i64(INDEX_op_mul_i64, ret, arg1, arg2);
+ return;
+ }
+
+
t0 = tcg_temp_ebb_new_i64();
t1 = tcg_temp_ebb_new_i32();
@@ -1672,15 +1764,6 @@ void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
tcg_temp_free_i32(t1);
}
-#else
-
-void tcg_gen_movi_i64(TCGv_i64 ret, int64_t arg)
-{
- tcg_gen_mov_i64(ret, tcg_constant_i64(arg));
-}
-
-#endif /* TCG_TARGET_REG_SIZE == 32 */
-
void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2)
{
/* some cases can be optimized here */
@@ -1723,6 +1806,15 @@ void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2)
}
}
+void tcg_gen_neg_i64(TCGv_i64 ret, TCGv_i64 arg)
+{
+ if (TCG_TARGET_HAS_neg_i64) {
+ tcg_gen_op2_i64(INDEX_op_neg_i64, ret, arg);
+ } else {
+ tcg_gen_subfi_i64(ret, 0, arg);
+ }
+}
+
void tcg_gen_andi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2)
{
if (TCG_TARGET_REG_BITS == 32) {
@@ -3218,6 +3310,11 @@ void tcg_gen_extr32_i64(TCGv_i64 lo, TCGv_i64 hi, TCGv_i64 arg)
tcg_gen_shri_i64(hi, arg, 32);
}
+void tcg_gen_concat32_i64(TCGv_i64 ret, TCGv_i64 lo, TCGv_i64 hi)
+{
+ tcg_gen_deposit_i64(ret, lo, hi, 32, 32);
+}
+
void tcg_gen_extr_i128_i64(TCGv_i64 lo, TCGv_i64 hi, TCGv_i128 arg)
{
tcg_gen_mov_i64(lo, TCGV128_LOW(arg));
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 14/35] tcg: Move vec_gen_* declarations to tcg-internal.h
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (12 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 13/35] tcg: Move 64-bit " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 15/35] tcg: Move tcg_gen_opN " Richard Henderson
` (22 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
These are used within tcg-op-vec.c and tcg/host/tcg-target.c.inc.
There are no uses outside tcg/.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-7-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 4 ----
tcg/tcg-internal.h | 4 ++++
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index f5ec54f874..3f8b214376 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -21,10 +21,6 @@ void tcg_gen_op4(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg);
void tcg_gen_op5(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg);
void tcg_gen_op6(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg);
-void vec_gen_2(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg);
-void vec_gen_3(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg);
-void vec_gen_4(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg, TCGArg);
-
void tcg_gen_op1_i32(TCGOpcode, TCGv_i32);
void tcg_gen_op1_i64(TCGOpcode, TCGv_i64);
void tcg_gen_op1i(TCGOpcode, TCGArg);
diff --git a/tcg/tcg-internal.h b/tcg/tcg-internal.h
index 40a69e6e6e..f18d282abb 100644
--- a/tcg/tcg-internal.h
+++ b/tcg/tcg-internal.h
@@ -83,4 +83,8 @@ static inline TCGv_i64 TCGV128_HIGH(TCGv_i128 t)
bool tcg_target_has_memory_bswap(MemOp memop);
+void vec_gen_2(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg);
+void vec_gen_3(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg);
+void vec_gen_4(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg, TCGArg);
+
#endif /* TCG_INTERNAL_H */
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 15/35] tcg: Move tcg_gen_opN declarations to tcg-internal.h
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (13 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 14/35] tcg: Move vec_gen_* declarations to tcg-internal.h Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 16/35] tcg: Unexport tcg_gen_op*_{i32,i64} Richard Henderson
` (21 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
These are used within tcg-op.c and tcg-op-ldst.c.
There are no uses outside tcg/.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-8-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 7 -------
tcg/tcg-internal.h | 7 +++++++
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index 3f8b214376..b922545118 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -14,13 +14,6 @@
/* Basic output routines. Not for general consumption. */
-void tcg_gen_op1(TCGOpcode, TCGArg);
-void tcg_gen_op2(TCGOpcode, TCGArg, TCGArg);
-void tcg_gen_op3(TCGOpcode, TCGArg, TCGArg, TCGArg);
-void tcg_gen_op4(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg);
-void tcg_gen_op5(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg);
-void tcg_gen_op6(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg);
-
void tcg_gen_op1_i32(TCGOpcode, TCGv_i32);
void tcg_gen_op1_i64(TCGOpcode, TCGv_i64);
void tcg_gen_op1i(TCGOpcode, TCGArg);
diff --git a/tcg/tcg-internal.h b/tcg/tcg-internal.h
index f18d282abb..c9ac34fc3d 100644
--- a/tcg/tcg-internal.h
+++ b/tcg/tcg-internal.h
@@ -83,6 +83,13 @@ static inline TCGv_i64 TCGV128_HIGH(TCGv_i128 t)
bool tcg_target_has_memory_bswap(MemOp memop);
+void tcg_gen_op1(TCGOpcode, TCGArg);
+void tcg_gen_op2(TCGOpcode, TCGArg, TCGArg);
+void tcg_gen_op3(TCGOpcode, TCGArg, TCGArg, TCGArg);
+void tcg_gen_op4(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg);
+void tcg_gen_op5(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg);
+void tcg_gen_op6(TCGOpcode, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg, TCGArg);
+
void vec_gen_2(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg);
void vec_gen_3(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg);
void vec_gen_4(TCGOpcode, TCGType, unsigned, TCGArg, TCGArg, TCGArg, TCGArg);
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 16/35] tcg: Unexport tcg_gen_op*_{i32,i64}
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (14 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 15/35] tcg: Move tcg_gen_opN " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 17/35] tcg: Move tcg_constant_* out of line Richard Henderson
` (20 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
These functions are no longer used outside tcg-op.c.
There are several that are completely unused, so remove them.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-9-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 47 -------------
tcg/tcg-op.c | 131 ++++++++++++++----------------------
2 files changed, 52 insertions(+), 126 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index b922545118..760c67683b 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -12,53 +12,6 @@
#include "exec/helper-proto-common.h"
#include "exec/helper-gen-common.h"
-/* Basic output routines. Not for general consumption. */
-
-void tcg_gen_op1_i32(TCGOpcode, TCGv_i32);
-void tcg_gen_op1_i64(TCGOpcode, TCGv_i64);
-void tcg_gen_op1i(TCGOpcode, TCGArg);
-void tcg_gen_op2_i32(TCGOpcode, TCGv_i32, TCGv_i32);
-void tcg_gen_op2_i64(TCGOpcode, TCGv_i64, TCGv_i64);
-void tcg_gen_op2i_i32(TCGOpcode, TCGv_i32, TCGArg);
-void tcg_gen_op2i_i64(TCGOpcode, TCGv_i64, TCGArg);
-void tcg_gen_op2ii(TCGOpcode, TCGArg, TCGArg);
-void tcg_gen_op3_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32);
-void tcg_gen_op3_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64);
-void tcg_gen_op3i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGArg);
-void tcg_gen_op3i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGArg);
-void tcg_gen_ldst_op_i32(TCGOpcode, TCGv_i32, TCGv_ptr, TCGArg);
-void tcg_gen_ldst_op_i64(TCGOpcode, TCGv_i64, TCGv_ptr, TCGArg);
-void tcg_gen_op4_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32, TCGv_i32);
-void tcg_gen_op4_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64, TCGv_i64);
-void tcg_gen_op4i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32, TCGArg);
-void tcg_gen_op4i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64, TCGArg);
-void tcg_gen_op4ii_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGArg, TCGArg);
-void tcg_gen_op4ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGArg, TCGArg);
-void tcg_gen_op5_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
- TCGv_i32, TCGv_i32);
-void tcg_gen_op5_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
- TCGv_i64, TCGv_i64);
-void tcg_gen_op5i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
- TCGv_i32, TCGArg);
-void tcg_gen_op5i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
- TCGv_i64, TCGArg);
-void tcg_gen_op5ii_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
- TCGArg, TCGArg);
-void tcg_gen_op5ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
- TCGArg, TCGArg);
-void tcg_gen_op6_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
- TCGv_i32, TCGv_i32, TCGv_i32);
-void tcg_gen_op6_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
- TCGv_i64, TCGv_i64, TCGv_i64);
-void tcg_gen_op6i_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
- TCGv_i32, TCGv_i32, TCGArg);
-void tcg_gen_op6i_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
- TCGv_i64, TCGv_i64, TCGArg);
-void tcg_gen_op6ii_i32(TCGOpcode, TCGv_i32, TCGv_i32, TCGv_i32,
- TCGv_i32, TCGArg, TCGArg);
-void tcg_gen_op6ii_i64(TCGOpcode, TCGv_i64, TCGv_i64, TCGv_i64,
- TCGv_i64, TCGArg, TCGArg);
-
/* Generic ops. */
void gen_set_label(TCGLabel *l);
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 579a2aab15..9aba103590 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -100,204 +100,177 @@ void NI tcg_gen_op6(TCGOpcode opc, TCGArg a1, TCGArg a2, TCGArg a3,
# define DNI
#endif
-void DNI tcg_gen_op1_i32(TCGOpcode opc, TCGv_i32 a1)
+static void DNI tcg_gen_op1_i32(TCGOpcode opc, TCGv_i32 a1)
{
tcg_gen_op1(opc, tcgv_i32_arg(a1));
}
-void DNI tcg_gen_op1_i64(TCGOpcode opc, TCGv_i64 a1)
+static void DNI tcg_gen_op1_i64(TCGOpcode opc, TCGv_i64 a1)
{
tcg_gen_op1(opc, tcgv_i64_arg(a1));
}
-void DNI tcg_gen_op1i(TCGOpcode opc, TCGArg a1)
+static void DNI tcg_gen_op1i(TCGOpcode opc, TCGArg a1)
{
tcg_gen_op1(opc, a1);
}
-void DNI tcg_gen_op2_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2)
+static void DNI tcg_gen_op2_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2)
{
tcg_gen_op2(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2));
}
-void DNI tcg_gen_op2_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2)
+static void DNI tcg_gen_op2_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2)
{
tcg_gen_op2(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2));
}
-void DNI tcg_gen_op2i_i32(TCGOpcode opc, TCGv_i32 a1, TCGArg a2)
-{
- tcg_gen_op2(opc, tcgv_i32_arg(a1), a2);
-}
-
-void DNI tcg_gen_op2i_i64(TCGOpcode opc, TCGv_i64 a1, TCGArg a2)
-{
- tcg_gen_op2(opc, tcgv_i64_arg(a1), a2);
-}
-
-void DNI tcg_gen_op2ii(TCGOpcode opc, TCGArg a1, TCGArg a2)
-{
- tcg_gen_op2(opc, a1, a2);
-}
-
-void DNI tcg_gen_op3_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGv_i32 a3)
+static void DNI tcg_gen_op3_i32(TCGOpcode opc, TCGv_i32 a1,
+ TCGv_i32 a2, TCGv_i32 a3)
{
tcg_gen_op3(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), tcgv_i32_arg(a3));
}
-void DNI tcg_gen_op3_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGv_i64 a3)
+static void DNI tcg_gen_op3_i64(TCGOpcode opc, TCGv_i64 a1,
+ TCGv_i64 a2, TCGv_i64 a3)
{
tcg_gen_op3(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), tcgv_i64_arg(a3));
}
-void DNI tcg_gen_op3i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGArg a3)
+static void DNI tcg_gen_op3i_i32(TCGOpcode opc, TCGv_i32 a1,
+ TCGv_i32 a2, TCGArg a3)
{
tcg_gen_op3(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), a3);
}
-void DNI tcg_gen_op3i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGArg a3)
+static void DNI tcg_gen_op3i_i64(TCGOpcode opc, TCGv_i64 a1,
+ TCGv_i64 a2, TCGArg a3)
{
tcg_gen_op3(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), a3);
}
-void DNI tcg_gen_ldst_op_i32(TCGOpcode opc, TCGv_i32 val,
- TCGv_ptr base, TCGArg offset)
+static void DNI tcg_gen_ldst_op_i32(TCGOpcode opc, TCGv_i32 val,
+ TCGv_ptr base, TCGArg offset)
{
tcg_gen_op3(opc, tcgv_i32_arg(val), tcgv_ptr_arg(base), offset);
}
-void DNI tcg_gen_ldst_op_i64(TCGOpcode opc, TCGv_i64 val,
- TCGv_ptr base, TCGArg offset)
+static void DNI tcg_gen_ldst_op_i64(TCGOpcode opc, TCGv_i64 val,
+ TCGv_ptr base, TCGArg offset)
{
tcg_gen_op3(opc, tcgv_i64_arg(val), tcgv_ptr_arg(base), offset);
}
-void DNI tcg_gen_op4_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4)
+static void DNI tcg_gen_op4_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4)
{
tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), tcgv_i32_arg(a4));
}
-void DNI tcg_gen_op4_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4)
+static void DNI tcg_gen_op4_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4)
{
tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
tcgv_i64_arg(a3), tcgv_i64_arg(a4));
}
-void DNI tcg_gen_op4i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGArg a4)
+static void DNI tcg_gen_op4i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGArg a4)
{
tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), a4);
}
-void DNI tcg_gen_op4i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGArg a4)
+static void DNI tcg_gen_op4i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGArg a4)
{
tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
tcgv_i64_arg(a3), a4);
}
-void DNI tcg_gen_op4ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGArg a3, TCGArg a4)
+static void DNI tcg_gen_op4ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGArg a3, TCGArg a4)
{
tcg_gen_op4(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2), a3, a4);
}
-void DNI tcg_gen_op4ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGArg a3, TCGArg a4)
+static void DNI tcg_gen_op4ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGArg a3, TCGArg a4)
{
tcg_gen_op4(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2), a3, a4);
}
-void DNI tcg_gen_op5_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4, TCGv_i32 a5)
+static void DNI tcg_gen_op5_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4, TCGv_i32 a5)
{
tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5));
}
-void DNI tcg_gen_op5_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4, TCGv_i64 a5)
+static void DNI tcg_gen_op5_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4, TCGv_i64 a5)
{
tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5));
}
-void DNI tcg_gen_op5i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4, TCGArg a5)
-{
- tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
- tcgv_i32_arg(a3), tcgv_i32_arg(a4), a5);
-}
-
-void DNI tcg_gen_op5i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4, TCGArg a5)
-{
- tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), a5);
-}
-
-void DNI tcg_gen_op5ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGArg a4, TCGArg a5)
+static void DNI tcg_gen_op5ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGArg a4, TCGArg a5)
{
tcg_gen_op5(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), a4, a5);
}
-void DNI tcg_gen_op5ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGArg a4, TCGArg a5)
+static void DNI tcg_gen_op5ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGArg a4, TCGArg a5)
{
tcg_gen_op5(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
tcgv_i64_arg(a3), a4, a5);
}
-void DNI tcg_gen_op6_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGv_i32 a3,
- TCGv_i32 a4, TCGv_i32 a5, TCGv_i32 a6)
+static void DNI tcg_gen_op6_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4,
+ TCGv_i32 a5, TCGv_i32 a6)
{
tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5),
tcgv_i32_arg(a6));
}
-void DNI tcg_gen_op6_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGv_i64 a3,
- TCGv_i64 a4, TCGv_i64 a5, TCGv_i64 a6)
+static void DNI tcg_gen_op6_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4,
+ TCGv_i64 a5, TCGv_i64 a6)
{
tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5),
tcgv_i64_arg(a6));
}
-void DNI tcg_gen_op6i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2, TCGv_i32 a3,
- TCGv_i32 a4, TCGv_i32 a5, TCGArg a6)
+static void DNI tcg_gen_op6i_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4,
+ TCGv_i32 a5, TCGArg a6)
{
tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), tcgv_i32_arg(a4), tcgv_i32_arg(a5), a6);
}
-void DNI tcg_gen_op6i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2, TCGv_i64 a3,
- TCGv_i64 a4, TCGv_i64 a5, TCGArg a6)
+static void DNI tcg_gen_op6i_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
+ TCGv_i64 a3, TCGv_i64 a4,
+ TCGv_i64 a5, TCGArg a6)
{
tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
tcgv_i64_arg(a3), tcgv_i64_arg(a4), tcgv_i64_arg(a5), a6);
}
-void DNI tcg_gen_op6ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
- TCGv_i32 a3, TCGv_i32 a4, TCGArg a5, TCGArg a6)
+static void DNI tcg_gen_op6ii_i32(TCGOpcode opc, TCGv_i32 a1, TCGv_i32 a2,
+ TCGv_i32 a3, TCGv_i32 a4,
+ TCGArg a5, TCGArg a6)
{
tcg_gen_op6(opc, tcgv_i32_arg(a1), tcgv_i32_arg(a2),
tcgv_i32_arg(a3), tcgv_i32_arg(a4), a5, a6);
}
-void DNI tcg_gen_op6ii_i64(TCGOpcode opc, TCGv_i64 a1, TCGv_i64 a2,
- TCGv_i64 a3, TCGv_i64 a4, TCGArg a5, TCGArg a6)
-{
- tcg_gen_op6(opc, tcgv_i64_arg(a1), tcgv_i64_arg(a2),
- tcgv_i64_arg(a3), tcgv_i64_arg(a4), a5, a6);
-}
-
/* Generic ops. */
void gen_set_label(TCGLabel *l)
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 17/35] tcg: Move tcg_constant_* out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (15 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 16/35] tcg: Unexport tcg_gen_op*_{i32,i64} Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 18/35] tcg: Move tcg_temp_new_*, tcg_global_mem_new_* " Richard Henderson
` (19 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-10-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 8 ++++++++
include/tcg/tcg.h | 26 --------------------------
tcg/tcg-internal.h | 7 +++++++
tcg/tcg.c | 15 +++++++++++++++
4 files changed, 30 insertions(+), 26 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index 760c67683b..dddf93067e 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -12,6 +12,11 @@
#include "exec/helper-proto-common.h"
#include "exec/helper-gen-common.h"
+TCGv_i32 tcg_constant_i32(int32_t val);
+TCGv_i64 tcg_constant_i64(int64_t val);
+TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val);
+TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val);
+
/* Generic ops. */
void gen_set_label(TCGLabel *l);
@@ -459,6 +464,9 @@ void tcg_gen_stl_vec(TCGv_vec r, TCGv_ptr base, TCGArg offset, TCGType t);
# define NAT TCGv_i64
#endif
+TCGv_ptr tcg_constant_ptr_int(intptr_t x);
+#define tcg_constant_ptr(X) tcg_constant_ptr_int((intptr_t)(X))
+
static inline void tcg_gen_ld_ptr(TCGv_ptr r, TCGv_ptr a, intptr_t o)
{
glue(tcg_gen_ld_,PTR)((NAT)r, a, o);
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 3a4c0f124f..1ae131c242 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -934,32 +934,6 @@ void tcg_remove_ops_after(TCGOp *op);
void tcg_optimize(TCGContext *s);
-/*
- * Locate or create a read-only temporary that is a constant.
- * This kind of temporary need not be freed, but for convenience
- * will be silently ignored by tcg_temp_free_*.
- */
-TCGTemp *tcg_constant_internal(TCGType type, int64_t val);
-
-static inline TCGv_i32 tcg_constant_i32(int32_t val)
-{
- return temp_tcgv_i32(tcg_constant_internal(TCG_TYPE_I32, val));
-}
-
-static inline TCGv_i64 tcg_constant_i64(int64_t val)
-{
- return temp_tcgv_i64(tcg_constant_internal(TCG_TYPE_I64, val));
-}
-
-TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val);
-TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val);
-
-#if UINTPTR_MAX == UINT32_MAX
-# define tcg_constant_ptr(x) ((TCGv_ptr)tcg_constant_i32((intptr_t)(x)))
-#else
-# define tcg_constant_ptr(x) ((TCGv_ptr)tcg_constant_i64((intptr_t)(x)))
-#endif
-
TCGLabel *gen_new_label(void);
/**
diff --git a/tcg/tcg-internal.h b/tcg/tcg-internal.h
index c9ac34fc3d..6c9d9e48db 100644
--- a/tcg/tcg-internal.h
+++ b/tcg/tcg-internal.h
@@ -83,6 +83,13 @@ static inline TCGv_i64 TCGV128_HIGH(TCGv_i128 t)
bool tcg_target_has_memory_bswap(MemOp memop);
+/*
+ * Locate or create a read-only temporary that is a constant.
+ * This kind of temporary need not be freed, but for convenience
+ * will be silently ignored by tcg_temp_free_*.
+ */
+TCGTemp *tcg_constant_internal(TCGType type, int64_t val);
+
void tcg_gen_op1(TCGOpcode, TCGArg);
void tcg_gen_op2(TCGOpcode, TCGArg, TCGArg);
void tcg_gen_op3(TCGOpcode, TCGArg, TCGArg, TCGArg);
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 6766b60b8a..ab0d227c00 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1806,6 +1806,21 @@ TCGTemp *tcg_constant_internal(TCGType type, int64_t val)
return ts;
}
+TCGv_i32 tcg_constant_i32(int32_t val)
+{
+ return temp_tcgv_i32(tcg_constant_internal(TCG_TYPE_I32, val));
+}
+
+TCGv_i64 tcg_constant_i64(int64_t val)
+{
+ return temp_tcgv_i64(tcg_constant_internal(TCG_TYPE_I64, val));
+}
+
+TCGv_ptr tcg_constant_ptr_int(intptr_t val)
+{
+ return temp_tcgv_ptr(tcg_constant_internal(TCG_TYPE_PTR, val));
+}
+
TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val)
{
val = dup_const(vece, val);
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 18/35] tcg: Move tcg_temp_new_*, tcg_global_mem_new_* out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (16 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 17/35] tcg: Move tcg_constant_* out of line Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 19/35] tcg: Move tcg_temp_free_* " Richard Henderson
` (18 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-11-richard.henderson@linaro.org>
---
include/tcg/tcg-op-common.h | 11 ++++++
include/tcg/tcg-temp-internal.h | 27 +++-----------
include/tcg/tcg.h | 51 --------------------------
tcg/tcg.c | 64 +++++++++++++++++++++++++++++++--
4 files changed, 76 insertions(+), 77 deletions(-)
diff --git a/include/tcg/tcg-op-common.h b/include/tcg/tcg-op-common.h
index dddf93067e..2d932a515e 100644
--- a/include/tcg/tcg-op-common.h
+++ b/include/tcg/tcg-op-common.h
@@ -17,6 +17,17 @@ TCGv_i64 tcg_constant_i64(int64_t val);
TCGv_vec tcg_constant_vec(TCGType type, unsigned vece, int64_t val);
TCGv_vec tcg_constant_vec_matching(TCGv_vec match, unsigned vece, int64_t val);
+TCGv_i32 tcg_temp_new_i32(void);
+TCGv_i64 tcg_temp_new_i64(void);
+TCGv_ptr tcg_temp_new_ptr(void);
+TCGv_i128 tcg_temp_new_i128(void);
+TCGv_vec tcg_temp_new_vec(TCGType type);
+TCGv_vec tcg_temp_new_vec_matching(TCGv_vec match);
+
+TCGv_i32 tcg_global_mem_new_i32(TCGv_ptr reg, intptr_t off, const char *name);
+TCGv_i64 tcg_global_mem_new_i64(TCGv_ptr reg, intptr_t off, const char *name);
+TCGv_ptr tcg_global_mem_new_ptr(TCGv_ptr reg, intptr_t off, const char *name);
+
/* Generic ops. */
void gen_set_label(TCGLabel *l);
diff --git a/include/tcg/tcg-temp-internal.h b/include/tcg/tcg-temp-internal.h
index dded2917e5..2d45cc45d2 100644
--- a/include/tcg/tcg-temp-internal.h
+++ b/include/tcg/tcg-temp-internal.h
@@ -56,28 +56,9 @@ static inline void tcg_temp_free_vec(TCGv_vec arg)
tcg_temp_free_internal(tcgv_vec_temp(arg));
}
-static inline TCGv_i32 tcg_temp_ebb_new_i32(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_I32, TEMP_EBB);
- return temp_tcgv_i32(t);
-}
-
-static inline TCGv_i64 tcg_temp_ebb_new_i64(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_I64, TEMP_EBB);
- return temp_tcgv_i64(t);
-}
-
-static inline TCGv_i128 tcg_temp_ebb_new_i128(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_I128, TEMP_EBB);
- return temp_tcgv_i128(t);
-}
-
-static inline TCGv_ptr tcg_temp_ebb_new_ptr(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_PTR, TEMP_EBB);
- return temp_tcgv_ptr(t);
-}
+TCGv_i32 tcg_temp_ebb_new_i32(void);
+TCGv_i64 tcg_temp_ebb_new_i64(void);
+TCGv_ptr tcg_temp_ebb_new_ptr(void);
+TCGv_i128 tcg_temp_ebb_new_i128(void);
#endif /* TCG_TEMP_FREE_H */
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 1ae131c242..fc218fd381 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -795,57 +795,6 @@ void tb_target_set_jmp_target(const TranslationBlock *, int,
void tcg_set_frame(TCGContext *s, TCGReg reg, intptr_t start, intptr_t size);
-TCGTemp *tcg_global_mem_new_internal(TCGType, TCGv_ptr,
- intptr_t, const char *);
-TCGTemp *tcg_temp_new_internal(TCGType, TCGTempKind);
-TCGv_vec tcg_temp_new_vec(TCGType type);
-TCGv_vec tcg_temp_new_vec_matching(TCGv_vec match);
-
-static inline TCGv_i32 tcg_global_mem_new_i32(TCGv_ptr reg, intptr_t offset,
- const char *name)
-{
- TCGTemp *t = tcg_global_mem_new_internal(TCG_TYPE_I32, reg, offset, name);
- return temp_tcgv_i32(t);
-}
-
-static inline TCGv_i32 tcg_temp_new_i32(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_I32, TEMP_TB);
- return temp_tcgv_i32(t);
-}
-
-static inline TCGv_i64 tcg_global_mem_new_i64(TCGv_ptr reg, intptr_t offset,
- const char *name)
-{
- TCGTemp *t = tcg_global_mem_new_internal(TCG_TYPE_I64, reg, offset, name);
- return temp_tcgv_i64(t);
-}
-
-static inline TCGv_i64 tcg_temp_new_i64(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_I64, TEMP_TB);
- return temp_tcgv_i64(t);
-}
-
-static inline TCGv_i128 tcg_temp_new_i128(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_I128, TEMP_TB);
- return temp_tcgv_i128(t);
-}
-
-static inline TCGv_ptr tcg_global_mem_new_ptr(TCGv_ptr reg, intptr_t offset,
- const char *name)
-{
- TCGTemp *t = tcg_global_mem_new_internal(TCG_TYPE_PTR, reg, offset, name);
- return temp_tcgv_ptr(t);
-}
-
-static inline TCGv_ptr tcg_temp_new_ptr(void)
-{
- TCGTemp *t = tcg_temp_new_internal(TCG_TYPE_PTR, TEMP_TB);
- return temp_tcgv_ptr(t);
-}
-
#define TCG_CT_CONST 1 /* any constant of register size */
typedef struct TCGArgConstraint {
diff --git a/tcg/tcg.c b/tcg/tcg.c
index ab0d227c00..ec358ce5c0 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1572,8 +1572,8 @@ void tcg_set_frame(TCGContext *s, TCGReg reg, intptr_t start, intptr_t size)
= tcg_global_reg_new_internal(s, TCG_TYPE_PTR, reg, "_frame");
}
-TCGTemp *tcg_global_mem_new_internal(TCGType type, TCGv_ptr base,
- intptr_t offset, const char *name)
+static TCGTemp *tcg_global_mem_new_internal(TCGv_ptr base, intptr_t offset,
+ const char *name, TCGType type)
{
TCGContext *s = tcg_ctx;
TCGTemp *base_ts = tcgv_ptr_temp(base);
@@ -1632,7 +1632,25 @@ TCGTemp *tcg_global_mem_new_internal(TCGType type, TCGv_ptr base,
return ts;
}
-TCGTemp *tcg_temp_new_internal(TCGType type, TCGTempKind kind)
+TCGv_i32 tcg_global_mem_new_i32(TCGv_ptr reg, intptr_t off, const char *name)
+{
+ TCGTemp *ts = tcg_global_mem_new_internal(reg, off, name, TCG_TYPE_I32);
+ return temp_tcgv_i32(ts);
+}
+
+TCGv_i64 tcg_global_mem_new_i64(TCGv_ptr reg, intptr_t off, const char *name)
+{
+ TCGTemp *ts = tcg_global_mem_new_internal(reg, off, name, TCG_TYPE_I64);
+ return temp_tcgv_i64(ts);
+}
+
+TCGv_ptr tcg_global_mem_new_ptr(TCGv_ptr reg, intptr_t off, const char *name)
+{
+ TCGTemp *ts = tcg_global_mem_new_internal(reg, off, name, TCG_TYPE_PTR);
+ return temp_tcgv_ptr(ts);
+}
+
+static TCGTemp *tcg_temp_new_internal(TCGType type, TCGTempKind kind)
{
TCGContext *s = tcg_ctx;
TCGTemp *ts;
@@ -1696,6 +1714,46 @@ TCGTemp *tcg_temp_new_internal(TCGType type, TCGTempKind kind)
return ts;
}
+TCGv_i32 tcg_temp_new_i32(void)
+{
+ return temp_tcgv_i32(tcg_temp_new_internal(TCG_TYPE_I32, TEMP_TB));
+}
+
+TCGv_i32 tcg_temp_ebb_new_i32(void)
+{
+ return temp_tcgv_i32(tcg_temp_new_internal(TCG_TYPE_I32, TEMP_EBB));
+}
+
+TCGv_i64 tcg_temp_new_i64(void)
+{
+ return temp_tcgv_i64(tcg_temp_new_internal(TCG_TYPE_I64, TEMP_TB));
+}
+
+TCGv_i64 tcg_temp_ebb_new_i64(void)
+{
+ return temp_tcgv_i64(tcg_temp_new_internal(TCG_TYPE_I64, TEMP_EBB));
+}
+
+TCGv_ptr tcg_temp_new_ptr(void)
+{
+ return temp_tcgv_ptr(tcg_temp_new_internal(TCG_TYPE_PTR, TEMP_TB));
+}
+
+TCGv_ptr tcg_temp_ebb_new_ptr(void)
+{
+ return temp_tcgv_ptr(tcg_temp_new_internal(TCG_TYPE_PTR, TEMP_EBB));
+}
+
+TCGv_i128 tcg_temp_new_i128(void)
+{
+ return temp_tcgv_i128(tcg_temp_new_internal(TCG_TYPE_I128, TEMP_TB));
+}
+
+TCGv_i128 tcg_temp_ebb_new_i128(void)
+{
+ return temp_tcgv_i128(tcg_temp_new_internal(TCG_TYPE_I128, TEMP_EBB));
+}
+
TCGv_vec tcg_temp_new_vec(TCGType type)
{
TCGTemp *t;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 19/35] tcg: Move tcg_temp_free_* out of line
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (17 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 18/35] tcg: Move tcg_temp_new_*, tcg_global_mem_new_* " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 20/35] tcg/mips: Split out tcg_out_setcond_int Richard Henderson
` (17 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231029210848.78234-12-richard.henderson@linaro.org>
---
include/tcg/tcg-temp-internal.h | 29 +++++------------------------
tcg/tcg.c | 25 +++++++++++++++++++++++++
2 files changed, 30 insertions(+), 24 deletions(-)
diff --git a/include/tcg/tcg-temp-internal.h b/include/tcg/tcg-temp-internal.h
index 2d45cc45d2..44192c55a9 100644
--- a/include/tcg/tcg-temp-internal.h
+++ b/include/tcg/tcg-temp-internal.h
@@ -31,30 +31,11 @@
void tcg_temp_free_internal(TCGTemp *);
-static inline void tcg_temp_free_i32(TCGv_i32 arg)
-{
- tcg_temp_free_internal(tcgv_i32_temp(arg));
-}
-
-static inline void tcg_temp_free_i64(TCGv_i64 arg)
-{
- tcg_temp_free_internal(tcgv_i64_temp(arg));
-}
-
-static inline void tcg_temp_free_i128(TCGv_i128 arg)
-{
- tcg_temp_free_internal(tcgv_i128_temp(arg));
-}
-
-static inline void tcg_temp_free_ptr(TCGv_ptr arg)
-{
- tcg_temp_free_internal(tcgv_ptr_temp(arg));
-}
-
-static inline void tcg_temp_free_vec(TCGv_vec arg)
-{
- tcg_temp_free_internal(tcgv_vec_temp(arg));
-}
+void tcg_temp_free_i32(TCGv_i32 arg);
+void tcg_temp_free_i64(TCGv_i64 arg);
+void tcg_temp_free_i128(TCGv_i128 arg);
+void tcg_temp_free_ptr(TCGv_ptr arg);
+void tcg_temp_free_vec(TCGv_vec arg);
TCGv_i32 tcg_temp_ebb_new_i32(void);
TCGv_i64 tcg_temp_ebb_new_i64(void);
diff --git a/tcg/tcg.c b/tcg/tcg.c
index ec358ce5c0..258bd1c10b 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1809,6 +1809,31 @@ void tcg_temp_free_internal(TCGTemp *ts)
}
}
+void tcg_temp_free_i32(TCGv_i32 arg)
+{
+ tcg_temp_free_internal(tcgv_i32_temp(arg));
+}
+
+void tcg_temp_free_i64(TCGv_i64 arg)
+{
+ tcg_temp_free_internal(tcgv_i64_temp(arg));
+}
+
+void tcg_temp_free_i128(TCGv_i128 arg)
+{
+ tcg_temp_free_internal(tcgv_i128_temp(arg));
+}
+
+void tcg_temp_free_ptr(TCGv_ptr arg)
+{
+ tcg_temp_free_internal(tcgv_ptr_temp(arg));
+}
+
+void tcg_temp_free_vec(TCGv_vec arg)
+{
+ tcg_temp_free_internal(tcgv_vec_temp(arg));
+}
+
TCGTemp *tcg_constant_internal(TCGType type, int64_t val)
{
TCGContext *s = tcg_ctx;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 20/35] tcg/mips: Split out tcg_out_setcond_int
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (18 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 19/35] tcg: Move tcg_temp_free_* " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 21/35] tcg/mips: Always implement movcond Richard Henderson
` (16 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Return the temp and a set of flags, to be used as a
primitive for setcond, brcond, movcond.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-2-richard.henderson@linaro.org>
---
tcg/mips/tcg-target.c.inc | 302 +++++++++++++++-----------------------
1 file changed, 118 insertions(+), 184 deletions(-)
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 328984ccff..89681f00fe 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -871,71 +871,83 @@ static void tcg_out_addsub2(TCGContext *s, TCGReg rl, TCGReg rh, TCGReg al,
}
}
-/* Bit 0 set if inversion required; bit 1 set if swapping required. */
-#define MIPS_CMP_INV 1
-#define MIPS_CMP_SWAP 2
+#define SETCOND_INV TCG_TARGET_NB_REGS
+#define SETCOND_NEZ (SETCOND_INV << 1)
+#define SETCOND_FLAGS (SETCOND_INV | SETCOND_NEZ)
-static const uint8_t mips_cmp_map[16] = {
- [TCG_COND_LT] = 0,
- [TCG_COND_LTU] = 0,
- [TCG_COND_GE] = MIPS_CMP_INV,
- [TCG_COND_GEU] = MIPS_CMP_INV,
- [TCG_COND_LE] = MIPS_CMP_INV | MIPS_CMP_SWAP,
- [TCG_COND_LEU] = MIPS_CMP_INV | MIPS_CMP_SWAP,
- [TCG_COND_GT] = MIPS_CMP_SWAP,
- [TCG_COND_GTU] = MIPS_CMP_SWAP,
-};
+static int tcg_out_setcond_int(TCGContext *s, TCGCond cond, TCGReg ret,
+ TCGReg arg1, TCGReg arg2)
+{
+ int flags = 0;
+
+ switch (cond) {
+ case TCG_COND_EQ: /* -> NE */
+ case TCG_COND_GE: /* -> LT */
+ case TCG_COND_GEU: /* -> LTU */
+ case TCG_COND_LE: /* -> GT */
+ case TCG_COND_LEU: /* -> GTU */
+ cond = tcg_invert_cond(cond);
+ flags ^= SETCOND_INV;
+ break;
+ default:
+ break;
+ }
+
+ switch (cond) {
+ case TCG_COND_NE:
+ flags |= SETCOND_NEZ;
+ if (arg2 == 0) {
+ return arg1 | flags;
+ }
+ tcg_out_opc_reg(s, OPC_XOR, ret, arg1, arg2);
+ break;
+ case TCG_COND_LT:
+ tcg_out_opc_reg(s, OPC_SLT, ret, arg1, arg2);
+ break;
+ case TCG_COND_LTU:
+ tcg_out_opc_reg(s, OPC_SLTU, ret, arg1, arg2);
+ break;
+ case TCG_COND_GT:
+ tcg_out_opc_reg(s, OPC_SLT, ret, arg2, arg1);
+ break;
+ case TCG_COND_GTU:
+ tcg_out_opc_reg(s, OPC_SLTU, ret, arg2, arg1);
+ break;
+ default:
+ g_assert_not_reached();
+ }
+ return ret | flags;
+}
+
+static void tcg_out_setcond_end(TCGContext *s, TCGReg ret, int tmpflags)
+{
+ if (tmpflags != ret) {
+ TCGReg tmp = tmpflags & ~SETCOND_FLAGS;
+
+ switch (tmpflags & SETCOND_FLAGS) {
+ case SETCOND_INV:
+ /* Intermediate result is boolean: simply invert. */
+ tcg_out_opc_imm(s, OPC_XORI, ret, tmp, 1);
+ break;
+ case SETCOND_NEZ:
+ /* Intermediate result is zero/non-zero: test != 0. */
+ tcg_out_opc_reg(s, OPC_SLTU, ret, TCG_REG_ZERO, tmp);
+ break;
+ case SETCOND_NEZ | SETCOND_INV:
+ /* Intermediate result is zero/non-zero: test == 0. */
+ tcg_out_opc_imm(s, OPC_SLTIU, ret, tmp, 1);
+ break;
+ default:
+ g_assert_not_reached();
+ }
+ }
+}
static void tcg_out_setcond(TCGContext *s, TCGCond cond, TCGReg ret,
TCGReg arg1, TCGReg arg2)
{
- MIPSInsn s_opc = OPC_SLTU;
- int cmp_map;
-
- switch (cond) {
- case TCG_COND_EQ:
- if (arg2 != 0) {
- tcg_out_opc_reg(s, OPC_XOR, ret, arg1, arg2);
- arg1 = ret;
- }
- tcg_out_opc_imm(s, OPC_SLTIU, ret, arg1, 1);
- break;
-
- case TCG_COND_NE:
- if (arg2 != 0) {
- tcg_out_opc_reg(s, OPC_XOR, ret, arg1, arg2);
- arg1 = ret;
- }
- tcg_out_opc_reg(s, OPC_SLTU, ret, TCG_REG_ZERO, arg1);
- break;
-
- case TCG_COND_LT:
- case TCG_COND_GE:
- case TCG_COND_LE:
- case TCG_COND_GT:
- s_opc = OPC_SLT;
- /* FALLTHRU */
-
- case TCG_COND_LTU:
- case TCG_COND_GEU:
- case TCG_COND_LEU:
- case TCG_COND_GTU:
- cmp_map = mips_cmp_map[cond];
- if (cmp_map & MIPS_CMP_SWAP) {
- TCGReg t = arg1;
- arg1 = arg2;
- arg2 = t;
- }
- tcg_out_opc_reg(s, s_opc, ret, arg1, arg2);
- if (cmp_map & MIPS_CMP_INV) {
- tcg_out_opc_imm(s, OPC_XORI, ret, ret, 1);
- }
- break;
-
- default:
- g_assert_not_reached();
- break;
- }
+ int tmpflags = tcg_out_setcond_int(s, cond, ret, arg1, arg2);
+ tcg_out_setcond_end(s, ret, tmpflags);
}
static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
@@ -948,9 +960,7 @@ static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
[TCG_COND_GE] = OPC_BGEZ,
};
- MIPSInsn s_opc = OPC_SLTU;
- MIPSInsn b_opc;
- int cmp_map;
+ MIPSInsn b_opc = 0;
switch (cond) {
case TCG_COND_EQ:
@@ -959,7 +969,6 @@ static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
case TCG_COND_NE:
b_opc = OPC_BNE;
break;
-
case TCG_COND_LT:
case TCG_COND_GT:
case TCG_COND_LE:
@@ -968,133 +977,76 @@ static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
b_opc = b_zero[cond];
arg2 = arg1;
arg1 = 0;
- break;
}
- s_opc = OPC_SLT;
- /* FALLTHRU */
-
- case TCG_COND_LTU:
- case TCG_COND_GTU:
- case TCG_COND_LEU:
- case TCG_COND_GEU:
- cmp_map = mips_cmp_map[cond];
- if (cmp_map & MIPS_CMP_SWAP) {
- TCGReg t = arg1;
- arg1 = arg2;
- arg2 = t;
- }
- tcg_out_opc_reg(s, s_opc, TCG_TMP0, arg1, arg2);
- b_opc = (cmp_map & MIPS_CMP_INV ? OPC_BEQ : OPC_BNE);
- arg1 = TCG_TMP0;
- arg2 = TCG_REG_ZERO;
break;
-
default:
- g_assert_not_reached();
break;
}
+ if (b_opc == 0) {
+ int tmpflags = tcg_out_setcond_int(s, cond, TCG_TMP0, arg1, arg2);
+
+ arg2 = TCG_REG_ZERO;
+ arg1 = tmpflags & ~SETCOND_FLAGS;
+ b_opc = tmpflags & SETCOND_INV ? OPC_BEQ : OPC_BNE;
+ }
+
+ tcg_out_reloc(s, s->code_ptr, R_MIPS_PC16, l, 0);
tcg_out_opc_br(s, b_opc, arg1, arg2);
- tcg_out_reloc(s, s->code_ptr - 1, R_MIPS_PC16, l, 0);
tcg_out_nop(s);
}
-static TCGReg tcg_out_reduce_eq2(TCGContext *s, TCGReg tmp0, TCGReg tmp1,
- TCGReg al, TCGReg ah,
- TCGReg bl, TCGReg bh)
+static int tcg_out_setcond2_int(TCGContext *s, TCGCond cond, TCGReg ret,
+ TCGReg al, TCGReg ah, TCGReg bl, TCGReg bh)
{
- /* Merge highpart comparison into AH. */
- if (bh != 0) {
- if (ah != 0) {
- tcg_out_opc_reg(s, OPC_XOR, tmp0, ah, bh);
- ah = tmp0;
- } else {
- ah = bh;
- }
+ int flags = 0;
+
+ switch (cond) {
+ case TCG_COND_EQ:
+ flags |= SETCOND_INV;
+ /* fall through */
+ case TCG_COND_NE:
+ flags |= SETCOND_NEZ;
+ tcg_out_opc_reg(s, OPC_XOR, TCG_TMP0, al, bl);
+ tcg_out_opc_reg(s, OPC_XOR, TCG_TMP1, ah, bh);
+ tcg_out_opc_reg(s, OPC_OR, ret, TCG_TMP0, TCG_TMP1);
+ break;
+
+ default:
+ tcg_out_setcond(s, TCG_COND_EQ, TCG_TMP0, ah, bh);
+ tcg_out_setcond(s, tcg_unsigned_cond(cond), TCG_TMP1, al, bl);
+ tcg_out_opc_reg(s, OPC_AND, TCG_TMP1, TCG_TMP1, TCG_TMP0);
+ tcg_out_setcond(s, tcg_high_cond(cond), TCG_TMP0, ah, bh);
+ tcg_out_opc_reg(s, OPC_OR, ret, TCG_TMP0, TCG_TMP1);
+ break;
}
- /* Merge lowpart comparison into AL. */
- if (bl != 0) {
- if (al != 0) {
- tcg_out_opc_reg(s, OPC_XOR, tmp1, al, bl);
- al = tmp1;
- } else {
- al = bl;
- }
- }
- /* Merge high and low part comparisons into AL. */
- if (ah != 0) {
- if (al != 0) {
- tcg_out_opc_reg(s, OPC_OR, tmp0, ah, al);
- al = tmp0;
- } else {
- al = ah;
- }
- }
- return al;
+ return ret | flags;
}
static void tcg_out_setcond2(TCGContext *s, TCGCond cond, TCGReg ret,
TCGReg al, TCGReg ah, TCGReg bl, TCGReg bh)
{
- TCGReg tmp0 = TCG_TMP0;
- TCGReg tmp1 = ret;
-
- tcg_debug_assert(ret != TCG_TMP0);
- if (ret == ah || ret == bh) {
- tcg_debug_assert(ret != TCG_TMP1);
- tmp1 = TCG_TMP1;
- }
-
- switch (cond) {
- case TCG_COND_EQ:
- case TCG_COND_NE:
- tmp1 = tcg_out_reduce_eq2(s, tmp0, tmp1, al, ah, bl, bh);
- tcg_out_setcond(s, cond, ret, tmp1, TCG_REG_ZERO);
- break;
-
- default:
- tcg_out_setcond(s, TCG_COND_EQ, tmp0, ah, bh);
- tcg_out_setcond(s, tcg_unsigned_cond(cond), tmp1, al, bl);
- tcg_out_opc_reg(s, OPC_AND, tmp1, tmp1, tmp0);
- tcg_out_setcond(s, tcg_high_cond(cond), tmp0, ah, bh);
- tcg_out_opc_reg(s, OPC_OR, ret, tmp1, tmp0);
- break;
- }
+ int tmpflags = tcg_out_setcond2_int(s, cond, ret, al, ah, bl, bh);
+ tcg_out_setcond_end(s, ret, tmpflags);
}
static void tcg_out_brcond2(TCGContext *s, TCGCond cond, TCGReg al, TCGReg ah,
TCGReg bl, TCGReg bh, TCGLabel *l)
{
- TCGCond b_cond = TCG_COND_NE;
- TCGReg tmp = TCG_TMP1;
+ int tmpflags = tcg_out_setcond2_int(s, cond, TCG_TMP0, al, ah, bl, bh);
+ TCGReg tmp = tmpflags & ~SETCOND_FLAGS;
+ MIPSInsn b_opc = tmpflags & SETCOND_INV ? OPC_BEQ : OPC_BNE;
- /* With branches, we emit between 4 and 9 insns with 2 or 3 branches.
- With setcond, we emit between 3 and 10 insns and only 1 branch,
- which ought to get better branch prediction. */
- switch (cond) {
- case TCG_COND_EQ:
- case TCG_COND_NE:
- b_cond = cond;
- tmp = tcg_out_reduce_eq2(s, TCG_TMP0, TCG_TMP1, al, ah, bl, bh);
- break;
-
- default:
- /* Minimize code size by preferring a compare not requiring INV. */
- if (mips_cmp_map[cond] & MIPS_CMP_INV) {
- cond = tcg_invert_cond(cond);
- b_cond = TCG_COND_EQ;
- }
- tcg_out_setcond2(s, cond, tmp, al, ah, bl, bh);
- break;
- }
-
- tcg_out_brcond(s, b_cond, tmp, TCG_REG_ZERO, l);
+ tcg_out_reloc(s, s->code_ptr, R_MIPS_PC16, l, 0);
+ tcg_out_opc_br(s, b_opc, tmp, TCG_REG_ZERO);
+ tcg_out_nop(s);
}
static void tcg_out_movcond(TCGContext *s, TCGCond cond, TCGReg ret,
TCGReg c1, TCGReg c2, TCGReg v1, TCGReg v2)
{
- bool eqz = false;
+ int tmpflags;
+ bool eqz;
/* If one of the values is zero, put it last to match SEL*Z instructions */
if (use_mips32r6_instructions && v1 == 0) {
@@ -1103,27 +1055,9 @@ static void tcg_out_movcond(TCGContext *s, TCGCond cond, TCGReg ret,
cond = tcg_invert_cond(cond);
}
- switch (cond) {
- case TCG_COND_EQ:
- eqz = true;
- /* FALLTHRU */
- case TCG_COND_NE:
- if (c2 != 0) {
- tcg_out_opc_reg(s, OPC_XOR, TCG_TMP0, c1, c2);
- c1 = TCG_TMP0;
- }
- break;
-
- default:
- /* Minimize code size by preferring a compare not requiring INV. */
- if (mips_cmp_map[cond] & MIPS_CMP_INV) {
- cond = tcg_invert_cond(cond);
- eqz = true;
- }
- tcg_out_setcond(s, cond, TCG_TMP0, c1, c2);
- c1 = TCG_TMP0;
- break;
- }
+ tmpflags = tcg_out_setcond_int(s, cond, TCG_TMP0, c1, c2);
+ c1 = tmpflags & ~SETCOND_FLAGS;
+ eqz = tmpflags & SETCOND_INV;
if (use_mips32r6_instructions) {
MIPSInsn m_opc_t = eqz ? OPC_SELEQZ : OPC_SELNEZ;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 21/35] tcg/mips: Always implement movcond
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (19 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 20/35] tcg/mips: Split out tcg_out_setcond_int Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 22/35] tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64} Richard Henderson
` (15 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Expand as branch over move if not supported in the ISA.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-3-richard.henderson@linaro.org>
---
tcg/mips/tcg-target.h | 4 ++--
tcg/mips/tcg-target.c.inc | 19 ++++++++++++++-----
2 files changed, 16 insertions(+), 7 deletions(-)
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index c0576f66d7..0a4083f0d9 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -154,7 +154,7 @@ extern bool use_mips32r2_instructions;
#endif
/* optional instructions detected at runtime */
-#define TCG_TARGET_HAS_movcond_i32 use_movnz_instructions
+#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_bswap16_i32 use_mips32r2_instructions
#define TCG_TARGET_HAS_deposit_i32 use_mips32r2_instructions
#define TCG_TARGET_HAS_extract_i32 use_mips32r2_instructions
@@ -169,7 +169,7 @@ extern bool use_mips32r2_instructions;
#define TCG_TARGET_HAS_qemu_st8_i32 0
#if TCG_TARGET_REG_BITS == 64
-#define TCG_TARGET_HAS_movcond_i64 use_movnz_instructions
+#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_bswap16_i64 use_mips32r2_instructions
#define TCG_TARGET_HAS_bswap32_i64 use_mips32r2_instructions
#define TCG_TARGET_HAS_bswap64_i64 use_mips32r2_instructions
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 89681f00fe..82b078b9c5 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -1070,13 +1070,22 @@ static void tcg_out_movcond(TCGContext *s, TCGCond cond, TCGReg ret,
if (v2 != 0) {
tcg_out_opc_reg(s, OPC_OR, ret, ret, TCG_TMP1);
}
- } else {
+ return;
+ }
+
+ /* This should be guaranteed via constraints */
+ tcg_debug_assert(v2 == ret);
+
+ if (use_movnz_instructions) {
MIPSInsn m_opc = eqz ? OPC_MOVZ : OPC_MOVN;
-
tcg_out_opc_reg(s, m_opc, ret, v1, c1);
-
- /* This should be guaranteed via constraints */
- tcg_debug_assert(v2 == ret);
+ } else {
+ /* Invert the condition in order to branch over the move. */
+ MIPSInsn b_opc = eqz ? OPC_BNE : OPC_BEQ;
+ tcg_out_opc_imm(s, b_opc, c1, TCG_REG_ZERO, 2);
+ tcg_out_nop(s);
+ /* Open-code tcg_out_mov, without the nop-move check. */
+ tcg_out_opc_reg(s, OPC_OR, ret, v1, TCG_REG_ZERO);
}
}
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 22/35] tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64}
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (20 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 21/35] tcg/mips: Always implement movcond Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 23/35] tcg/mips: Implement neg opcodes Richard Henderson
` (14 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
The movcond opcode is now mandatory for backends to implement.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-4-richard.henderson@linaro.org>
---
include/tcg/tcg-opc.h | 4 +--
include/tcg/tcg.h | 1 -
tcg/aarch64/tcg-target.h | 2 --
tcg/arm/tcg-target.h | 1 -
tcg/i386/tcg-target.h | 2 --
tcg/loongarch64/tcg-target.h | 2 --
tcg/mips/tcg-target.h | 2 --
tcg/ppc/tcg-target.h | 2 --
tcg/riscv/tcg-target.h | 2 --
tcg/s390x/tcg-target.h | 2 --
tcg/sparc64/tcg-target.h | 2 --
tcg/tci/tcg-target.h | 2 --
tcg/tcg-op.c | 50 ++++++++----------------------------
tcg/tcg.c | 6 ++---
14 files changed, 14 insertions(+), 66 deletions(-)
diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h
index 6eff3d9106..ecd08db0de 100644
--- a/include/tcg/tcg-opc.h
+++ b/include/tcg/tcg-opc.h
@@ -47,7 +47,7 @@ DEF(mb, 0, 0, 1, 0)
DEF(mov_i32, 1, 1, 0, TCG_OPF_NOT_PRESENT)
DEF(setcond_i32, 1, 2, 1, 0)
DEF(negsetcond_i32, 1, 2, 1, IMPL(TCG_TARGET_HAS_negsetcond_i32))
-DEF(movcond_i32, 1, 4, 1, IMPL(TCG_TARGET_HAS_movcond_i32))
+DEF(movcond_i32, 1, 4, 1, 0)
/* load/store */
DEF(ld8u_i32, 1, 1, 1, 0)
DEF(ld8s_i32, 1, 1, 1, 0)
@@ -113,7 +113,7 @@ DEF(ctpop_i32, 1, 1, 0, IMPL(TCG_TARGET_HAS_ctpop_i32))
DEF(mov_i64, 1, 1, 0, TCG_OPF_64BIT | TCG_OPF_NOT_PRESENT)
DEF(setcond_i64, 1, 2, 1, IMPL64)
DEF(negsetcond_i64, 1, 2, 1, IMPL64 | IMPL(TCG_TARGET_HAS_negsetcond_i64))
-DEF(movcond_i64, 1, 4, 1, IMPL64 | IMPL(TCG_TARGET_HAS_movcond_i64))
+DEF(movcond_i64, 1, 4, 1, IMPL64)
/* load/store */
DEF(ld8u_i64, 1, 1, 1, IMPL64)
DEF(ld8s_i64, 1, 1, 1, IMPL64)
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index fc218fd381..32a208a02e 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -96,7 +96,6 @@ typedef uint64_t TCGRegSet;
#define TCG_TARGET_HAS_extract_i64 0
#define TCG_TARGET_HAS_sextract_i64 0
#define TCG_TARGET_HAS_extract2_i64 0
-#define TCG_TARGET_HAS_movcond_i64 0
#define TCG_TARGET_HAS_negsetcond_i64 0
#define TCG_TARGET_HAS_add2_i64 0
#define TCG_TARGET_HAS_sub2_i64 0
diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index 98727ea53b..352e19aba8 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -85,7 +85,6 @@ typedef enum {
#define TCG_TARGET_HAS_extract_i32 1
#define TCG_TARGET_HAS_sextract_i32 1
#define TCG_TARGET_HAS_extract2_i32 1
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_add2_i32 1
#define TCG_TARGET_HAS_sub2_i32 1
@@ -122,7 +121,6 @@ typedef enum {
#define TCG_TARGET_HAS_extract_i64 1
#define TCG_TARGET_HAS_sextract_i64 1
#define TCG_TARGET_HAS_extract2_i64 1
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 1
#define TCG_TARGET_HAS_add2_i64 1
#define TCG_TARGET_HAS_sub2_i64 1
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index 311a985209..439898efb3 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -115,7 +115,6 @@ extern bool use_neon_instructions;
#define TCG_TARGET_HAS_extract_i32 use_armv7_instructions
#define TCG_TARGET_HAS_sextract_i32 use_armv7_instructions
#define TCG_TARGET_HAS_extract2_i32 1
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_mulu2_i32 1
#define TCG_TARGET_HAS_muls2_i32 1
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index 8417ea4899..7522ce7575 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -149,7 +149,6 @@ typedef enum {
#define TCG_TARGET_HAS_extract_i32 1
#define TCG_TARGET_HAS_sextract_i32 1
#define TCG_TARGET_HAS_extract2_i32 1
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_add2_i32 1
#define TCG_TARGET_HAS_sub2_i32 1
@@ -186,7 +185,6 @@ typedef enum {
#define TCG_TARGET_HAS_extract_i64 1
#define TCG_TARGET_HAS_sextract_i64 0
#define TCG_TARGET_HAS_extract2_i64 1
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 1
#define TCG_TARGET_HAS_add2_i64 1
#define TCG_TARGET_HAS_sub2_i64 1
diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h
index 1bea15b02e..486898c2d3 100644
--- a/tcg/loongarch64/tcg-target.h
+++ b/tcg/loongarch64/tcg-target.h
@@ -97,7 +97,6 @@ typedef enum {
#define TCG_TARGET_CALL_RET_I128 TCG_CALL_RET_NORMAL
/* optional instructions */
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 0
#define TCG_TARGET_HAS_div_i32 1
#define TCG_TARGET_HAS_rem_i32 1
@@ -134,7 +133,6 @@ typedef enum {
#define TCG_TARGET_HAS_qemu_st8_i32 0
/* 64-bit operations */
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 0
#define TCG_TARGET_HAS_div_i64 1
#define TCG_TARGET_HAS_rem_i64 1
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index 0a4083f0d9..5b3fdcc726 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -154,7 +154,6 @@ extern bool use_mips32r2_instructions;
#endif
/* optional instructions detected at runtime */
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_bswap16_i32 use_mips32r2_instructions
#define TCG_TARGET_HAS_deposit_i32 use_mips32r2_instructions
#define TCG_TARGET_HAS_extract_i32 use_mips32r2_instructions
@@ -169,7 +168,6 @@ extern bool use_mips32r2_instructions;
#define TCG_TARGET_HAS_qemu_st8_i32 0
#if TCG_TARGET_REG_BITS == 64
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_bswap16_i64 use_mips32r2_instructions
#define TCG_TARGET_HAS_bswap32_i64 use_mips32r2_instructions
#define TCG_TARGET_HAS_bswap64_i64 use_mips32r2_instructions
diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index 8bfb14998e..a2856afd4d 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -96,7 +96,6 @@ typedef enum {
#define TCG_TARGET_HAS_extract_i32 1
#define TCG_TARGET_HAS_sextract_i32 0
#define TCG_TARGET_HAS_extract2_i32 0
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_mulu2_i32 0
#define TCG_TARGET_HAS_muls2_i32 0
@@ -134,7 +133,6 @@ typedef enum {
#define TCG_TARGET_HAS_extract_i64 1
#define TCG_TARGET_HAS_sextract_i64 0
#define TCG_TARGET_HAS_extract2_i64 0
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 1
#define TCG_TARGET_HAS_add2_i64 1
#define TCG_TARGET_HAS_sub2_i64 1
diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index c1132d178f..f3644a8bc1 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -87,7 +87,6 @@ extern bool have_zbb;
#endif
/* optional instructions */
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_div_i32 1
#define TCG_TARGET_HAS_rem_i32 1
@@ -123,7 +122,6 @@ extern bool have_zbb;
#define TCG_TARGET_HAS_setcond2 1
#define TCG_TARGET_HAS_qemu_st8_i32 0
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 1
#define TCG_TARGET_HAS_div_i64 1
#define TCG_TARGET_HAS_rem_i64 1
diff --git a/tcg/s390x/tcg-target.h b/tcg/s390x/tcg-target.h
index 50e12ef9d6..2c936c1bcb 100644
--- a/tcg/s390x/tcg-target.h
+++ b/tcg/s390x/tcg-target.h
@@ -95,7 +95,6 @@ extern uint64_t s390_facilities[3];
#define TCG_TARGET_HAS_extract_i32 1
#define TCG_TARGET_HAS_sextract_i32 0
#define TCG_TARGET_HAS_extract2_i32 0
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_add2_i32 1
#define TCG_TARGET_HAS_sub2_i32 1
@@ -131,7 +130,6 @@ extern uint64_t s390_facilities[3];
#define TCG_TARGET_HAS_extract_i64 1
#define TCG_TARGET_HAS_sextract_i64 0
#define TCG_TARGET_HAS_extract2_i64 0
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 1
#define TCG_TARGET_HAS_add2_i64 1
#define TCG_TARGET_HAS_sub2_i64 1
diff --git a/tcg/sparc64/tcg-target.h b/tcg/sparc64/tcg-target.h
index 5cfc4b4679..4c286c6006 100644
--- a/tcg/sparc64/tcg-target.h
+++ b/tcg/sparc64/tcg-target.h
@@ -105,7 +105,6 @@ extern bool use_vis3_instructions;
#define TCG_TARGET_HAS_extract_i32 0
#define TCG_TARGET_HAS_sextract_i32 0
#define TCG_TARGET_HAS_extract2_i32 0
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 1
#define TCG_TARGET_HAS_add2_i32 1
#define TCG_TARGET_HAS_sub2_i32 1
@@ -142,7 +141,6 @@ extern bool use_vis3_instructions;
#define TCG_TARGET_HAS_extract_i64 0
#define TCG_TARGET_HAS_sextract_i64 0
#define TCG_TARGET_HAS_extract2_i64 0
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 1
#define TCG_TARGET_HAS_add2_i64 1
#define TCG_TARGET_HAS_sub2_i64 1
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index 91ca33b616..3503fc4a4c 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -69,7 +69,6 @@
#define TCG_TARGET_HAS_not_i32 1
#define TCG_TARGET_HAS_orc_i32 1
#define TCG_TARGET_HAS_rot_i32 1
-#define TCG_TARGET_HAS_movcond_i32 1
#define TCG_TARGET_HAS_negsetcond_i32 0
#define TCG_TARGET_HAS_muls2_i32 1
#define TCG_TARGET_HAS_muluh_i32 0
@@ -104,7 +103,6 @@
#define TCG_TARGET_HAS_not_i64 1
#define TCG_TARGET_HAS_orc_i64 1
#define TCG_TARGET_HAS_rot_i64 1
-#define TCG_TARGET_HAS_movcond_i64 1
#define TCG_TARGET_HAS_negsetcond_i64 0
#define TCG_TARGET_HAS_muls2_i64 1
#define TCG_TARGET_HAS_add2_i32 1
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 9aba103590..26bcd090c1 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -1142,17 +1142,8 @@ void tcg_gen_movcond_i32(TCGCond cond, TCGv_i32 ret, TCGv_i32 c1,
tcg_gen_mov_i32(ret, v1);
} else if (cond == TCG_COND_NEVER) {
tcg_gen_mov_i32(ret, v2);
- } else if (TCG_TARGET_HAS_movcond_i32) {
- tcg_gen_op6i_i32(INDEX_op_movcond_i32, ret, c1, c2, v1, v2, cond);
} else {
- TCGv_i32 t0 = tcg_temp_ebb_new_i32();
- TCGv_i32 t1 = tcg_temp_ebb_new_i32();
- tcg_gen_negsetcond_i32(cond, t0, c1, c2);
- tcg_gen_and_i32(t1, v1, t0);
- tcg_gen_andc_i32(ret, v2, t0);
- tcg_gen_or_i32(ret, ret, t1);
- tcg_temp_free_i32(t0);
- tcg_temp_free_i32(t1);
+ tcg_gen_op6i_i32(INDEX_op_movcond_i32, ret, c1, c2, v1, v2, cond);
}
}
@@ -3011,43 +3002,22 @@ void tcg_gen_movcond_i64(TCGCond cond, TCGv_i64 ret, TCGv_i64 c1,
tcg_gen_mov_i64(ret, v1);
} else if (cond == TCG_COND_NEVER) {
tcg_gen_mov_i64(ret, v2);
- } else if (TCG_TARGET_REG_BITS == 32) {
+ } else if (TCG_TARGET_REG_BITS == 64) {
+ tcg_gen_op6i_i64(INDEX_op_movcond_i64, ret, c1, c2, v1, v2, cond);
+ } else {
TCGv_i32 t0 = tcg_temp_ebb_new_i32();
- TCGv_i32 t1 = tcg_temp_ebb_new_i32();
+ TCGv_i32 zero = tcg_constant_i32(0);
+
tcg_gen_op6i_i32(INDEX_op_setcond2_i32, t0,
TCGV_LOW(c1), TCGV_HIGH(c1),
TCGV_LOW(c2), TCGV_HIGH(c2), cond);
- if (TCG_TARGET_HAS_movcond_i32) {
- tcg_gen_movi_i32(t1, 0);
- tcg_gen_movcond_i32(TCG_COND_NE, TCGV_LOW(ret), t0, t1,
- TCGV_LOW(v1), TCGV_LOW(v2));
- tcg_gen_movcond_i32(TCG_COND_NE, TCGV_HIGH(ret), t0, t1,
- TCGV_HIGH(v1), TCGV_HIGH(v2));
- } else {
- tcg_gen_neg_i32(t0, t0);
+ tcg_gen_movcond_i32(TCG_COND_NE, TCGV_LOW(ret), t0, zero,
+ TCGV_LOW(v1), TCGV_LOW(v2));
+ tcg_gen_movcond_i32(TCG_COND_NE, TCGV_HIGH(ret), t0, zero,
+ TCGV_HIGH(v1), TCGV_HIGH(v2));
- tcg_gen_and_i32(t1, TCGV_LOW(v1), t0);
- tcg_gen_andc_i32(TCGV_LOW(ret), TCGV_LOW(v2), t0);
- tcg_gen_or_i32(TCGV_LOW(ret), TCGV_LOW(ret), t1);
-
- tcg_gen_and_i32(t1, TCGV_HIGH(v1), t0);
- tcg_gen_andc_i32(TCGV_HIGH(ret), TCGV_HIGH(v2), t0);
- tcg_gen_or_i32(TCGV_HIGH(ret), TCGV_HIGH(ret), t1);
- }
tcg_temp_free_i32(t0);
- tcg_temp_free_i32(t1);
- } else if (TCG_TARGET_HAS_movcond_i64) {
- tcg_gen_op6i_i64(INDEX_op_movcond_i64, ret, c1, c2, v1, v2, cond);
- } else {
- TCGv_i64 t0 = tcg_temp_ebb_new_i64();
- TCGv_i64 t1 = tcg_temp_ebb_new_i64();
- tcg_gen_negsetcond_i64(cond, t0, c1, c2);
- tcg_gen_and_i64(t1, v1, t0);
- tcg_gen_andc_i64(ret, v2, t0);
- tcg_gen_or_i64(ret, ret, t1);
- tcg_temp_free_i64(t0);
- tcg_temp_free_i64(t1);
}
}
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 258bd1c10b..d59ff14f0f 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1977,6 +1977,7 @@ bool tcg_op_supported(TCGOpcode op)
case INDEX_op_mov_i32:
case INDEX_op_setcond_i32:
case INDEX_op_brcond_i32:
+ case INDEX_op_movcond_i32:
case INDEX_op_ld8u_i32:
case INDEX_op_ld8s_i32:
case INDEX_op_ld16u_i32:
@@ -1998,8 +1999,6 @@ bool tcg_op_supported(TCGOpcode op)
case INDEX_op_negsetcond_i32:
return TCG_TARGET_HAS_negsetcond_i32;
- case INDEX_op_movcond_i32:
- return TCG_TARGET_HAS_movcond_i32;
case INDEX_op_div_i32:
case INDEX_op_divu_i32:
return TCG_TARGET_HAS_div_i32;
@@ -2072,6 +2071,7 @@ bool tcg_op_supported(TCGOpcode op)
case INDEX_op_mov_i64:
case INDEX_op_setcond_i64:
case INDEX_op_brcond_i64:
+ case INDEX_op_movcond_i64:
case INDEX_op_ld8u_i64:
case INDEX_op_ld8s_i64:
case INDEX_op_ld16u_i64:
@@ -2098,8 +2098,6 @@ bool tcg_op_supported(TCGOpcode op)
case INDEX_op_negsetcond_i64:
return TCG_TARGET_HAS_negsetcond_i64;
- case INDEX_op_movcond_i64:
- return TCG_TARGET_HAS_movcond_i64;
case INDEX_op_div_i64:
case INDEX_op_divu_i64:
return TCG_TARGET_HAS_div_i64;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 23/35] tcg/mips: Implement neg opcodes
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (21 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 22/35] tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64} Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 24/35] tcg/loongarch64: " Richard Henderson
` (13 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-5-richard.henderson@linaro.org>
---
tcg/mips/tcg-target.h | 4 ++--
tcg/mips/tcg-target.c.inc | 8 ++++++++
2 files changed, 10 insertions(+), 2 deletions(-)
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index 5b3fdcc726..20c14224fb 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -184,12 +184,12 @@ extern bool use_mips32r2_instructions;
#endif
/* optional instructions automatically implemented */
-#define TCG_TARGET_HAS_neg_i32 0 /* sub rd, zero, rt */
+#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_ext8u_i32 0 /* andi rt, rs, 0xff */
#define TCG_TARGET_HAS_ext16u_i32 0 /* andi rt, rs, 0xffff */
#if TCG_TARGET_REG_BITS == 64
-#define TCG_TARGET_HAS_neg_i64 0 /* sub rd, zero, rt */
+#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_ext8u_i64 0 /* andi rt, rs, 0xff */
#define TCG_TARGET_HAS_ext16u_i64 0 /* andi rt, rs, 0xffff */
#endif
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 82b078b9c5..8328dbdecc 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -1920,6 +1920,12 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
tcg_out_opc_reg(s, OPC_MFHI, a1, 0, 0);
break;
+ case INDEX_op_neg_i32:
+ i1 = OPC_SUBU;
+ goto do_unary;
+ case INDEX_op_neg_i64:
+ i1 = OPC_DSUBU;
+ goto do_unary;
case INDEX_op_not_i32:
case INDEX_op_not_i64:
i1 = OPC_NOR;
@@ -2144,6 +2150,7 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op)
case INDEX_op_ld16u_i32:
case INDEX_op_ld16s_i32:
case INDEX_op_ld_i32:
+ case INDEX_op_neg_i32:
case INDEX_op_not_i32:
case INDEX_op_bswap16_i32:
case INDEX_op_bswap32_i32:
@@ -2157,6 +2164,7 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op)
case INDEX_op_ld32s_i64:
case INDEX_op_ld32u_i64:
case INDEX_op_ld_i64:
+ case INDEX_op_neg_i64:
case INDEX_op_not_i64:
case INDEX_op_bswap16_i64:
case INDEX_op_bswap32_i64:
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 24/35] tcg/loongarch64: Implement neg opcodes
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (22 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 23/35] tcg/mips: Implement neg opcodes Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 25/35] tcg: Remove TCG_TARGET_HAS_neg_{i32,i64} Richard Henderson
` (12 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-6-richard.henderson@linaro.org>
---
tcg/loongarch64/tcg-target.h | 4 ++--
tcg/loongarch64/tcg-target.c.inc | 9 +++++++++
2 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h
index 486898c2d3..189997644a 100644
--- a/tcg/loongarch64/tcg-target.h
+++ b/tcg/loongarch64/tcg-target.h
@@ -119,7 +119,7 @@ typedef enum {
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_not_i32 1
-#define TCG_TARGET_HAS_neg_i32 0
+#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_andc_i32 1
#define TCG_TARGET_HAS_orc_i32 1
#define TCG_TARGET_HAS_eqv_i32 0
@@ -153,7 +153,7 @@ typedef enum {
#define TCG_TARGET_HAS_bswap32_i64 1
#define TCG_TARGET_HAS_bswap64_i64 1
#define TCG_TARGET_HAS_not_i64 1
-#define TCG_TARGET_HAS_neg_i64 0
+#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_andc_i64 1
#define TCG_TARGET_HAS_orc_i64 1
#define TCG_TARGET_HAS_eqv_i64 0
diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc
index ae13c1f666..a588fb3085 100644
--- a/tcg/loongarch64/tcg-target.c.inc
+++ b/tcg/loongarch64/tcg-target.c.inc
@@ -1441,6 +1441,13 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
}
break;
+ case INDEX_op_neg_i32:
+ tcg_out_opc_sub_w(s, a0, TCG_REG_ZERO, a1);
+ break;
+ case INDEX_op_neg_i64:
+ tcg_out_opc_sub_d(s, a0, TCG_REG_ZERO, a1);
+ break;
+
case INDEX_op_mul_i32:
tcg_out_opc_mul_w(s, a0, a1, a2);
break;
@@ -2076,6 +2083,8 @@ static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op)
case INDEX_op_extrl_i64_i32:
case INDEX_op_extrh_i64_i32:
case INDEX_op_ext_i32_i64:
+ case INDEX_op_neg_i32:
+ case INDEX_op_neg_i64:
case INDEX_op_not_i32:
case INDEX_op_not_i64:
case INDEX_op_extract_i32:
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 25/35] tcg: Remove TCG_TARGET_HAS_neg_{i32,i64}
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (23 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 24/35] tcg/loongarch64: " Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 26/35] tcg: Don't free vector results Richard Henderson
` (11 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
The movcond opcode is now mandatory for backends to implement.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231026041404.1229328-7-richard.henderson@linaro.org>
---
include/tcg/tcg-opc.h | 4 ++--
include/tcg/tcg.h | 1 -
tcg/aarch64/tcg-target.h | 2 --
tcg/arm/tcg-target.h | 1 -
tcg/i386/tcg-target.h | 2 --
tcg/loongarch64/tcg-target.h | 2 --
tcg/mips/tcg-target.h | 2 --
tcg/ppc/tcg-target.h | 2 --
tcg/riscv/tcg-target.h | 2 --
tcg/s390x/tcg-target.h | 2 --
tcg/sparc64/tcg-target.h | 2 --
tcg/tci/tcg-target.h | 2 --
tcg/optimize.c | 4 ++--
tcg/tcg-op.c | 22 +++++++++-------------
tcg/tcg.c | 6 ++----
tcg/tci.c | 2 --
16 files changed, 15 insertions(+), 43 deletions(-)
diff --git a/include/tcg/tcg-opc.h b/include/tcg/tcg-opc.h
index ecd08db0de..b80227fa1c 100644
--- a/include/tcg/tcg-opc.h
+++ b/include/tcg/tcg-opc.h
@@ -100,7 +100,7 @@ DEF(ext16u_i32, 1, 1, 0, IMPL(TCG_TARGET_HAS_ext16u_i32))
DEF(bswap16_i32, 1, 1, 1, IMPL(TCG_TARGET_HAS_bswap16_i32))
DEF(bswap32_i32, 1, 1, 1, IMPL(TCG_TARGET_HAS_bswap32_i32))
DEF(not_i32, 1, 1, 0, IMPL(TCG_TARGET_HAS_not_i32))
-DEF(neg_i32, 1, 1, 0, IMPL(TCG_TARGET_HAS_neg_i32))
+DEF(neg_i32, 1, 1, 0, 0)
DEF(andc_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_andc_i32))
DEF(orc_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_orc_i32))
DEF(eqv_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_eqv_i32))
@@ -171,7 +171,7 @@ DEF(bswap16_i64, 1, 1, 1, IMPL64 | IMPL(TCG_TARGET_HAS_bswap16_i64))
DEF(bswap32_i64, 1, 1, 1, IMPL64 | IMPL(TCG_TARGET_HAS_bswap32_i64))
DEF(bswap64_i64, 1, 1, 1, IMPL64 | IMPL(TCG_TARGET_HAS_bswap64_i64))
DEF(not_i64, 1, 1, 0, IMPL64 | IMPL(TCG_TARGET_HAS_not_i64))
-DEF(neg_i64, 1, 1, 0, IMPL64 | IMPL(TCG_TARGET_HAS_neg_i64))
+DEF(neg_i64, 1, 1, 0, IMPL64)
DEF(andc_i64, 1, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_andc_i64))
DEF(orc_i64, 1, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_orc_i64))
DEF(eqv_i64, 1, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_eqv_i64))
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 32a208a02e..daf2a5bf9e 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -82,7 +82,6 @@ typedef uint64_t TCGRegSet;
#define TCG_TARGET_HAS_bswap16_i64 0
#define TCG_TARGET_HAS_bswap32_i64 0
#define TCG_TARGET_HAS_bswap64_i64 0
-#define TCG_TARGET_HAS_neg_i64 0
#define TCG_TARGET_HAS_not_i64 0
#define TCG_TARGET_HAS_andc_i64 0
#define TCG_TARGET_HAS_orc_i64 0
diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index 352e19aba8..33f15a564a 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -71,7 +71,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_not_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_rot_i32 1
#define TCG_TARGET_HAS_andc_i32 1
#define TCG_TARGET_HAS_orc_i32 1
@@ -107,7 +106,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap32_i64 1
#define TCG_TARGET_HAS_bswap64_i64 1
#define TCG_TARGET_HAS_not_i64 1
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_rot_i64 1
#define TCG_TARGET_HAS_andc_i64 1
#define TCG_TARGET_HAS_orc_i64 1
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index 439898efb3..a712cc80ad 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -101,7 +101,6 @@ extern bool use_neon_instructions;
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_not_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_rot_i32 1
#define TCG_TARGET_HAS_andc_i32 1
#define TCG_TARGET_HAS_orc_i32 0
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index 7522ce7575..fa34deec47 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -135,7 +135,6 @@ typedef enum {
#define TCG_TARGET_HAS_ext16u_i32 1
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_not_i32 1
#define TCG_TARGET_HAS_andc_i32 have_bmi1
#define TCG_TARGET_HAS_orc_i32 0
@@ -171,7 +170,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap16_i64 1
#define TCG_TARGET_HAS_bswap32_i64 1
#define TCG_TARGET_HAS_bswap64_i64 1
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_not_i64 1
#define TCG_TARGET_HAS_andc_i64 have_bmi1
#define TCG_TARGET_HAS_orc_i64 0
diff --git a/tcg/loongarch64/tcg-target.h b/tcg/loongarch64/tcg-target.h
index 189997644a..9c70ebfefc 100644
--- a/tcg/loongarch64/tcg-target.h
+++ b/tcg/loongarch64/tcg-target.h
@@ -119,7 +119,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_not_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_andc_i32 1
#define TCG_TARGET_HAS_orc_i32 1
#define TCG_TARGET_HAS_eqv_i32 0
@@ -153,7 +152,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap32_i64 1
#define TCG_TARGET_HAS_bswap64_i64 1
#define TCG_TARGET_HAS_not_i64 1
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_andc_i64 1
#define TCG_TARGET_HAS_orc_i64 1
#define TCG_TARGET_HAS_eqv_i64 0
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index 20c14224fb..b98ffae1d0 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -184,12 +184,10 @@ extern bool use_mips32r2_instructions;
#endif
/* optional instructions automatically implemented */
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_ext8u_i32 0 /* andi rt, rs, 0xff */
#define TCG_TARGET_HAS_ext16u_i32 0 /* andi rt, rs, 0xffff */
#if TCG_TARGET_REG_BITS == 64
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_ext8u_i64 0 /* andi rt, rs, 0xff */
#define TCG_TARGET_HAS_ext16u_i64 0 /* andi rt, rs, 0xffff */
#endif
diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index a2856afd4d..5295e4f9ab 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -83,7 +83,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_not_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_andc_i32 1
#define TCG_TARGET_HAS_orc_i32 1
#define TCG_TARGET_HAS_eqv_i32 1
@@ -120,7 +119,6 @@ typedef enum {
#define TCG_TARGET_HAS_bswap32_i64 1
#define TCG_TARGET_HAS_bswap64_i64 1
#define TCG_TARGET_HAS_not_i64 1
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_andc_i64 1
#define TCG_TARGET_HAS_orc_i64 1
#define TCG_TARGET_HAS_eqv_i64 1
diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index f3644a8bc1..a4edc3dc74 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -109,7 +109,6 @@ extern bool have_zbb;
#define TCG_TARGET_HAS_bswap16_i32 have_zbb
#define TCG_TARGET_HAS_bswap32_i32 have_zbb
#define TCG_TARGET_HAS_not_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_andc_i32 have_zbb
#define TCG_TARGET_HAS_orc_i32 have_zbb
#define TCG_TARGET_HAS_eqv_i32 have_zbb
@@ -142,7 +141,6 @@ extern bool have_zbb;
#define TCG_TARGET_HAS_bswap32_i64 have_zbb
#define TCG_TARGET_HAS_bswap64_i64 have_zbb
#define TCG_TARGET_HAS_not_i64 1
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_andc_i64 have_zbb
#define TCG_TARGET_HAS_orc_i64 have_zbb
#define TCG_TARGET_HAS_eqv_i64 have_zbb
diff --git a/tcg/s390x/tcg-target.h b/tcg/s390x/tcg-target.h
index 2c936c1bcb..e69b0d2ddd 100644
--- a/tcg/s390x/tcg-target.h
+++ b/tcg/s390x/tcg-target.h
@@ -82,7 +82,6 @@ extern uint64_t s390_facilities[3];
#define TCG_TARGET_HAS_bswap16_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_not_i32 HAVE_FACILITY(MISC_INSN_EXT3)
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_andc_i32 HAVE_FACILITY(MISC_INSN_EXT3)
#define TCG_TARGET_HAS_orc_i32 HAVE_FACILITY(MISC_INSN_EXT3)
#define TCG_TARGET_HAS_eqv_i32 HAVE_FACILITY(MISC_INSN_EXT3)
@@ -117,7 +116,6 @@ extern uint64_t s390_facilities[3];
#define TCG_TARGET_HAS_bswap32_i64 1
#define TCG_TARGET_HAS_bswap64_i64 1
#define TCG_TARGET_HAS_not_i64 HAVE_FACILITY(MISC_INSN_EXT3)
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_andc_i64 HAVE_FACILITY(MISC_INSN_EXT3)
#define TCG_TARGET_HAS_orc_i64 HAVE_FACILITY(MISC_INSN_EXT3)
#define TCG_TARGET_HAS_eqv_i64 HAVE_FACILITY(MISC_INSN_EXT3)
diff --git a/tcg/sparc64/tcg-target.h b/tcg/sparc64/tcg-target.h
index 4c286c6006..f8cf145266 100644
--- a/tcg/sparc64/tcg-target.h
+++ b/tcg/sparc64/tcg-target.h
@@ -91,7 +91,6 @@ extern bool use_vis3_instructions;
#define TCG_TARGET_HAS_ext16u_i32 0
#define TCG_TARGET_HAS_bswap16_i32 0
#define TCG_TARGET_HAS_bswap32_i32 0
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_not_i32 1
#define TCG_TARGET_HAS_andc_i32 1
#define TCG_TARGET_HAS_orc_i32 1
@@ -127,7 +126,6 @@ extern bool use_vis3_instructions;
#define TCG_TARGET_HAS_bswap16_i64 0
#define TCG_TARGET_HAS_bswap32_i64 0
#define TCG_TARGET_HAS_bswap64_i64 0
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_not_i64 1
#define TCG_TARGET_HAS_andc_i64 1
#define TCG_TARGET_HAS_orc_i64 1
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index 3503fc4a4c..2a13816c8e 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -65,7 +65,6 @@
#define TCG_TARGET_HAS_clz_i32 1
#define TCG_TARGET_HAS_ctz_i32 1
#define TCG_TARGET_HAS_ctpop_i32 1
-#define TCG_TARGET_HAS_neg_i32 1
#define TCG_TARGET_HAS_not_i32 1
#define TCG_TARGET_HAS_orc_i32 1
#define TCG_TARGET_HAS_rot_i32 1
@@ -99,7 +98,6 @@
#define TCG_TARGET_HAS_clz_i64 1
#define TCG_TARGET_HAS_ctz_i64 1
#define TCG_TARGET_HAS_ctpop_i64 1
-#define TCG_TARGET_HAS_neg_i64 1
#define TCG_TARGET_HAS_not_i64 1
#define TCG_TARGET_HAS_orc_i64 1
#define TCG_TARGET_HAS_rot_i64 1
diff --git a/tcg/optimize.c b/tcg/optimize.c
index 2db5177c32..6b072d4cdb 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -2001,11 +2001,11 @@ static bool fold_sub_to_neg(OptContext *ctx, TCGOp *op)
switch (ctx->type) {
case TCG_TYPE_I32:
neg_op = INDEX_op_neg_i32;
- have_neg = TCG_TARGET_HAS_neg_i32;
+ have_neg = true;
break;
case TCG_TYPE_I64:
neg_op = INDEX_op_neg_i64;
- have_neg = TCG_TARGET_HAS_neg_i64;
+ have_neg = true;
break;
case TCG_TYPE_V64:
case TCG_TYPE_V128:
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 26bcd090c1..de096a6f93 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -363,9 +363,8 @@ void tcg_gen_sub_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2)
{
- if (arg1 == 0 && TCG_TARGET_HAS_neg_i32) {
- /* Don't recurse with tcg_gen_neg_i32. */
- tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg2);
+ if (arg1 == 0) {
+ tcg_gen_neg_i32(ret, arg2);
} else {
tcg_gen_sub_i32(ret, tcg_constant_i32(arg1), arg2);
}
@@ -383,11 +382,7 @@ void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
void tcg_gen_neg_i32(TCGv_i32 ret, TCGv_i32 arg)
{
- if (TCG_TARGET_HAS_neg_i32) {
- tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg);
- } else {
- tcg_gen_subfi_i32(ret, 0, arg);
- }
+ tcg_gen_op2_i32(INDEX_op_neg_i32, ret, arg);
}
void tcg_gen_and_i32(TCGv_i32 ret, TCGv_i32 arg1, TCGv_i32 arg2)
@@ -1744,9 +1739,8 @@ void tcg_gen_addi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2)
void tcg_gen_subfi_i64(TCGv_i64 ret, int64_t arg1, TCGv_i64 arg2)
{
- if (arg1 == 0 && TCG_TARGET_HAS_neg_i64) {
- /* Don't recurse with tcg_gen_neg_i64. */
- tcg_gen_op2_i64(INDEX_op_neg_i64, ret, arg2);
+ if (arg1 == 0) {
+ tcg_gen_neg_i64(ret, arg2);
} else if (TCG_TARGET_REG_BITS == 64) {
tcg_gen_sub_i64(ret, tcg_constant_i64(arg1), arg2);
} else {
@@ -1772,10 +1766,12 @@ void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2)
void tcg_gen_neg_i64(TCGv_i64 ret, TCGv_i64 arg)
{
- if (TCG_TARGET_HAS_neg_i64) {
+ if (TCG_TARGET_REG_BITS == 64) {
tcg_gen_op2_i64(INDEX_op_neg_i64, ret, arg);
} else {
- tcg_gen_subfi_i64(ret, 0, arg);
+ TCGv_i32 zero = tcg_constant_i32(0);
+ tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret),
+ zero, zero, TCGV_LOW(arg), TCGV_HIGH(arg));
}
}
diff --git a/tcg/tcg.c b/tcg/tcg.c
index d59ff14f0f..d2ea22b397 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1988,6 +1988,7 @@ bool tcg_op_supported(TCGOpcode op)
case INDEX_op_st_i32:
case INDEX_op_add_i32:
case INDEX_op_sub_i32:
+ case INDEX_op_neg_i32:
case INDEX_op_mul_i32:
case INDEX_op_and_i32:
case INDEX_op_or_i32:
@@ -2045,8 +2046,6 @@ bool tcg_op_supported(TCGOpcode op)
return TCG_TARGET_HAS_bswap32_i32;
case INDEX_op_not_i32:
return TCG_TARGET_HAS_not_i32;
- case INDEX_op_neg_i32:
- return TCG_TARGET_HAS_neg_i32;
case INDEX_op_andc_i32:
return TCG_TARGET_HAS_andc_i32;
case INDEX_op_orc_i32:
@@ -2085,6 +2084,7 @@ bool tcg_op_supported(TCGOpcode op)
case INDEX_op_st_i64:
case INDEX_op_add_i64:
case INDEX_op_sub_i64:
+ case INDEX_op_neg_i64:
case INDEX_op_mul_i64:
case INDEX_op_and_i64:
case INDEX_op_or_i64:
@@ -2141,8 +2141,6 @@ bool tcg_op_supported(TCGOpcode op)
return TCG_TARGET_HAS_bswap64_i64;
case INDEX_op_not_i64:
return TCG_TARGET_HAS_not_i64;
- case INDEX_op_neg_i64:
- return TCG_TARGET_HAS_neg_i64;
case INDEX_op_andc_i64:
return TCG_TARGET_HAS_andc_i64;
case INDEX_op_orc_i64:
diff --git a/tcg/tci.c b/tcg/tci.c
index 4640902c88..3cc851b7bd 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -733,12 +733,10 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
regs[r0] = ~regs[r1];
break;
#endif
-#if TCG_TARGET_HAS_neg_i32 || TCG_TARGET_HAS_neg_i64
CASE_32_64(neg)
tci_args_rr(insn, &r0, &r1);
regs[r0] = -regs[r1];
break;
-#endif
#if TCG_TARGET_REG_BITS == 64
/* Load/store operations (64 bit). */
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 26/35] tcg: Don't free vector results
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (24 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 25/35] tcg: Remove TCG_TARGET_HAS_neg_{i32,i64} Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 27/35] tcg/optimize: Pipe OptContext into reset_ts Richard Henderson
` (10 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Song Gao
Avoid reusing vector temporaries so that we may re-use them
when propagating stores to loads.
Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/tcg-op-gvec.c | 112 ++++++++++++++++------------------------------
1 file changed, 38 insertions(+), 74 deletions(-)
diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c
index feb2d3686b..bb88943f79 100644
--- a/tcg/tcg-op-gvec.c
+++ b/tcg/tcg-op-gvec.c
@@ -561,7 +561,6 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz,
tcg_gen_dupi_vec(vece, t_vec, in_c);
}
do_dup_store(type, dofs, oprsz, maxsz, t_vec);
- tcg_temp_free_vec(t_vec);
return;
}
@@ -1024,11 +1023,10 @@ static void expand_2_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
bool load_dest,
void (*fni)(unsigned, TCGv_vec, TCGv_vec))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
if (load_dest) {
tcg_gen_ld_vec(t1, tcg_env, dofs + i);
@@ -1036,8 +1034,6 @@ static void expand_2_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
fni(vece, t1, t0);
tcg_gen_st_vec(t1, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t0);
- tcg_temp_free_vec(t1);
}
/* Expand OPSZ bytes worth of two-vector operands and an immediate operand
@@ -1047,11 +1043,10 @@ static void expand_2i_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
int64_t c, bool load_dest,
void (*fni)(unsigned, TCGv_vec, TCGv_vec, int64_t))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
if (load_dest) {
tcg_gen_ld_vec(t1, tcg_env, dofs + i);
@@ -1059,8 +1054,6 @@ static void expand_2i_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
fni(vece, t1, t0, c);
tcg_gen_st_vec(t1, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t0);
- tcg_temp_free_vec(t1);
}
static void expand_2s_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
@@ -1068,11 +1061,10 @@ static void expand_2s_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
TCGv_vec c, bool scalar_first,
void (*fni)(unsigned, TCGv_vec, TCGv_vec, TCGv_vec))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
if (scalar_first) {
fni(vece, t1, c, t0);
@@ -1081,8 +1073,6 @@ static void expand_2s_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
}
tcg_gen_st_vec(t1, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t0);
- tcg_temp_free_vec(t1);
}
/* Expand OPSZ bytes worth of three-operand operations using host vectors. */
@@ -1091,12 +1081,11 @@ static void expand_3_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
uint32_t tysz, TCGType type, bool load_dest,
void (*fni)(unsigned, TCGv_vec, TCGv_vec, TCGv_vec))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- TCGv_vec t2 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
+ TCGv_vec t2 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
tcg_gen_ld_vec(t1, tcg_env, bofs + i);
if (load_dest) {
@@ -1105,9 +1094,6 @@ static void expand_3_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
fni(vece, t2, t0, t1);
tcg_gen_st_vec(t2, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t2);
- tcg_temp_free_vec(t1);
- tcg_temp_free_vec(t0);
}
/*
@@ -1120,12 +1106,11 @@ static void expand_3i_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
void (*fni)(unsigned, TCGv_vec, TCGv_vec, TCGv_vec,
int64_t))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- TCGv_vec t2 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
+ TCGv_vec t2 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
tcg_gen_ld_vec(t1, tcg_env, bofs + i);
if (load_dest) {
@@ -1134,9 +1119,6 @@ static void expand_3i_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
fni(vece, t2, t0, t1, c);
tcg_gen_st_vec(t2, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t0);
- tcg_temp_free_vec(t1);
- tcg_temp_free_vec(t2);
}
/* Expand OPSZ bytes worth of four-operand operations using host vectors. */
@@ -1146,13 +1128,12 @@ static void expand_4_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
void (*fni)(unsigned, TCGv_vec, TCGv_vec,
TCGv_vec, TCGv_vec))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- TCGv_vec t2 = tcg_temp_new_vec(type);
- TCGv_vec t3 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
+ TCGv_vec t2 = tcg_temp_new_vec(type);
+ TCGv_vec t3 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t1, tcg_env, aofs + i);
tcg_gen_ld_vec(t2, tcg_env, bofs + i);
tcg_gen_ld_vec(t3, tcg_env, cofs + i);
@@ -1162,10 +1143,6 @@ static void expand_4_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
tcg_gen_st_vec(t1, tcg_env, aofs + i);
}
}
- tcg_temp_free_vec(t3);
- tcg_temp_free_vec(t2);
- tcg_temp_free_vec(t1);
- tcg_temp_free_vec(t0);
}
/*
@@ -1178,23 +1155,18 @@ static void expand_4i_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
void (*fni)(unsigned, TCGv_vec, TCGv_vec,
TCGv_vec, TCGv_vec, int64_t))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- TCGv_vec t2 = tcg_temp_new_vec(type);
- TCGv_vec t3 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
+ TCGv_vec t2 = tcg_temp_new_vec(type);
+ TCGv_vec t3 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t1, tcg_env, aofs + i);
tcg_gen_ld_vec(t2, tcg_env, bofs + i);
tcg_gen_ld_vec(t3, tcg_env, cofs + i);
fni(vece, t0, t1, t2, t3, c);
tcg_gen_st_vec(t0, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t3);
- tcg_temp_free_vec(t2);
- tcg_temp_free_vec(t1);
- tcg_temp_free_vec(t0);
}
/* Expand a vector two-operand operation. */
@@ -1732,7 +1704,6 @@ void tcg_gen_gvec_dup_mem(unsigned vece, uint32_t dofs, uint32_t aofs,
TCGv_vec t_vec = tcg_temp_new_vec(type);
tcg_gen_dup_mem_vec(vece, t_vec, tcg_env, aofs);
do_dup_store(type, dofs, oprsz, maxsz, t_vec);
- tcg_temp_free_vec(t_vec);
} else if (vece <= MO_32) {
TCGv_i32 in = tcg_temp_ebb_new_i32();
switch (vece) {
@@ -1766,7 +1737,6 @@ void tcg_gen_gvec_dup_mem(unsigned vece, uint32_t dofs, uint32_t aofs,
for (i = (aofs == dofs) * 16; i < oprsz; i += 16) {
tcg_gen_st_vec(in, tcg_env, dofs + i);
}
- tcg_temp_free_vec(in);
} else {
TCGv_i64 in0 = tcg_temp_ebb_new_i64();
TCGv_i64 in1 = tcg_temp_ebb_new_i64();
@@ -1796,7 +1766,6 @@ void tcg_gen_gvec_dup_mem(unsigned vece, uint32_t dofs, uint32_t aofs,
for (i = (aofs == dofs) * 32; i < oprsz; i += 32) {
tcg_gen_st_vec(in, tcg_env, dofs + i);
}
- tcg_temp_free_vec(in);
} else if (TCG_TARGET_HAS_v128) {
TCGv_vec in0 = tcg_temp_new_vec(TCG_TYPE_V128);
TCGv_vec in1 = tcg_temp_new_vec(TCG_TYPE_V128);
@@ -1807,8 +1776,6 @@ void tcg_gen_gvec_dup_mem(unsigned vece, uint32_t dofs, uint32_t aofs,
tcg_gen_st_vec(in0, tcg_env, dofs + i);
tcg_gen_st_vec(in1, tcg_env, dofs + i + 16);
}
- tcg_temp_free_vec(in0);
- tcg_temp_free_vec(in1);
} else {
TCGv_i64 in[4];
int j;
@@ -3136,15 +3103,14 @@ static void expand_2sh_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
TCGv_i32 shift,
void (*fni)(unsigned, TCGv_vec, TCGv_vec, TCGv_i32))
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
- fni(vece, t0, t0, shift);
- tcg_gen_st_vec(t0, tcg_env, dofs + i);
+ fni(vece, t1, t0, shift);
+ tcg_gen_st_vec(t1, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t0);
}
static void
@@ -3720,18 +3686,16 @@ static void expand_cmp_vec(unsigned vece, uint32_t dofs, uint32_t aofs,
uint32_t bofs, uint32_t oprsz, uint32_t tysz,
TCGType type, TCGCond cond)
{
- TCGv_vec t0 = tcg_temp_new_vec(type);
- TCGv_vec t1 = tcg_temp_new_vec(type);
- uint32_t i;
+ for (uint32_t i = 0; i < oprsz; i += tysz) {
+ TCGv_vec t0 = tcg_temp_new_vec(type);
+ TCGv_vec t1 = tcg_temp_new_vec(type);
+ TCGv_vec t2 = tcg_temp_new_vec(type);
- for (i = 0; i < oprsz; i += tysz) {
tcg_gen_ld_vec(t0, tcg_env, aofs + i);
tcg_gen_ld_vec(t1, tcg_env, bofs + i);
- tcg_gen_cmp_vec(cond, vece, t0, t0, t1);
- tcg_gen_st_vec(t0, tcg_env, dofs + i);
+ tcg_gen_cmp_vec(cond, vece, t2, t0, t1);
+ tcg_gen_st_vec(t2, tcg_env, dofs + i);
}
- tcg_temp_free_vec(t1);
- tcg_temp_free_vec(t0);
}
void tcg_gen_gvec_cmp(TCGCond cond, unsigned vece, uint32_t dofs,
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 27/35] tcg/optimize: Pipe OptContext into reset_ts
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (25 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 26/35] tcg: Don't free vector results Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 28/35] tcg/optimize: Split out cmp_better_copy Richard Henderson
` (9 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Song Gao, Philippe Mathieu-Daudé
Will be needed in the next patch.
Reviewed-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/optimize.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index 6b072d4cdb..cbb095b241 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -123,7 +123,7 @@ static inline bool ts_is_copy(TCGTemp *ts)
}
/* Reset TEMP's state, possibly removing the temp for the list of copies. */
-static void reset_ts(TCGTemp *ts)
+static void reset_ts(OptContext *ctx, TCGTemp *ts)
{
TempOptInfo *ti = ts_info(ts);
TempOptInfo *pi = ts_info(ti->prev_copy);
@@ -138,9 +138,9 @@ static void reset_ts(TCGTemp *ts)
ti->s_mask = 0;
}
-static void reset_temp(TCGArg arg)
+static void reset_temp(OptContext *ctx, TCGArg arg)
{
- reset_ts(arg_temp(arg));
+ reset_ts(ctx, arg_temp(arg));
}
/* Initialize and activate a temporary. */
@@ -239,7 +239,7 @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src)
return true;
}
- reset_ts(dst_ts);
+ reset_ts(ctx, dst_ts);
di = ts_info(dst_ts);
si = ts_info(src_ts);
@@ -702,7 +702,7 @@ static void finish_folding(OptContext *ctx, TCGOp *op)
nb_oargs = def->nb_oargs;
for (i = 0; i < nb_oargs; i++) {
TCGTemp *ts = arg_temp(op->args[i]);
- reset_ts(ts);
+ reset_ts(ctx, ts);
/*
* Save the corresponding known-zero/sign bits mask for the
* first output argument (only one supported so far).
@@ -1215,14 +1215,14 @@ static bool fold_call(OptContext *ctx, TCGOp *op)
for (i = 0; i < nb_globals; i++) {
if (test_bit(i, ctx->temps_used.l)) {
- reset_ts(&ctx->tcg->temps[i]);
+ reset_ts(ctx, &ctx->tcg->temps[i]);
}
}
}
/* Reset temp data for outputs. */
for (i = 0; i < nb_oargs; i++) {
- reset_temp(op->args[i]);
+ reset_temp(ctx, op->args[i]);
}
/* Stop optimizing MB across calls. */
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 28/35] tcg/optimize: Split out cmp_better_copy
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (26 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 27/35] tcg/optimize: Pipe OptContext into reset_ts Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 29/35] tcg/optimize: Optimize env memory operations Richard Henderson
` (8 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel
Compare two temps for "better", split out from finding
the best from a whole list. Use TCGKind, which already
gives the proper priority.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/optimize.c | 29 +++++++++++------------------
1 file changed, 11 insertions(+), 18 deletions(-)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index cbb095b241..118561f56d 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -122,6 +122,11 @@ static inline bool ts_is_copy(TCGTemp *ts)
return ts_info(ts)->next_copy != ts;
}
+static TCGTemp *cmp_better_copy(TCGTemp *a, TCGTemp *b)
+{
+ return a->kind < b->kind ? b : a;
+}
+
/* Reset TEMP's state, possibly removing the temp for the list of copies. */
static void reset_ts(OptContext *ctx, TCGTemp *ts)
{
@@ -174,30 +179,20 @@ static void init_ts_info(OptContext *ctx, TCGTemp *ts)
}
}
-static TCGTemp *find_better_copy(TCGContext *s, TCGTemp *ts)
+static TCGTemp *find_better_copy(TCGTemp *ts)
{
- TCGTemp *i, *g, *l;
+ TCGTemp *i, *ret;
/* If this is already readonly, we can't do better. */
if (temp_readonly(ts)) {
return ts;
}
- g = l = NULL;
+ ret = ts;
for (i = ts_info(ts)->next_copy; i != ts; i = ts_info(i)->next_copy) {
- if (temp_readonly(i)) {
- return i;
- } else if (i->kind > ts->kind) {
- if (i->kind == TEMP_GLOBAL) {
- g = i;
- } else if (i->kind == TEMP_TB) {
- l = i;
- }
- }
+ ret = cmp_better_copy(ret, i);
}
-
- /* If we didn't find a better representation, return the same temp. */
- return g ? g : l ? l : ts;
+ return ret;
}
static bool ts_are_copies(TCGTemp *ts1, TCGTemp *ts2)
@@ -672,12 +667,10 @@ static void init_arguments(OptContext *ctx, TCGOp *op, int nb_args)
static void copy_propagate(OptContext *ctx, TCGOp *op,
int nb_oargs, int nb_iargs)
{
- TCGContext *s = ctx->tcg;
-
for (int i = nb_oargs; i < nb_oargs + nb_iargs; i++) {
TCGTemp *ts = arg_temp(op->args[i]);
if (ts_is_copy(ts)) {
- op->args[i] = temp_arg(find_better_copy(s, ts));
+ op->args[i] = temp_arg(find_better_copy(ts));
}
}
}
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 29/35] tcg/optimize: Optimize env memory operations
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (27 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 28/35] tcg/optimize: Split out cmp_better_copy Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 30/35] tcg: Eliminate duplicate env store operations Richard Henderson
` (7 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Song Gao
Propagate stores to loads, loads to loads.
Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/optimize.c | 264 +++++++++++++++++++++++++++++++++++++++++++++----
1 file changed, 243 insertions(+), 21 deletions(-)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index 118561f56d..b32ef0be0f 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -25,6 +25,7 @@
#include "qemu/osdep.h"
#include "qemu/int128.h"
+#include "qemu/interval-tree.h"
#include "tcg/tcg-op-common.h"
#include "tcg-internal.h"
@@ -37,10 +38,18 @@
glue(glue(case INDEX_op_, x), _i64): \
glue(glue(case INDEX_op_, x), _vec)
+typedef struct MemCopyInfo {
+ IntervalTreeNode itree;
+ QSIMPLEQ_ENTRY (MemCopyInfo) next;
+ TCGTemp *ts;
+ TCGType type;
+} MemCopyInfo;
+
typedef struct TempOptInfo {
bool is_const;
TCGTemp *prev_copy;
TCGTemp *next_copy;
+ QSIMPLEQ_HEAD(, MemCopyInfo) mem_copy;
uint64_t val;
uint64_t z_mask; /* mask bit is 0 if and only if value bit is 0 */
uint64_t s_mask; /* a left-aligned mask of clrsb(value) bits. */
@@ -51,6 +60,9 @@ typedef struct OptContext {
TCGOp *prev_mb;
TCGTempSet temps_used;
+ IntervalTreeRoot mem_copy;
+ QSIMPLEQ_HEAD(, MemCopyInfo) mem_free;
+
/* In flight values from optimization. */
uint64_t a_mask; /* mask bit is 0 iff value identical to first input */
uint64_t z_mask; /* mask bit is 0 iff value bit is 0 */
@@ -127,27 +139,6 @@ static TCGTemp *cmp_better_copy(TCGTemp *a, TCGTemp *b)
return a->kind < b->kind ? b : a;
}
-/* Reset TEMP's state, possibly removing the temp for the list of copies. */
-static void reset_ts(OptContext *ctx, TCGTemp *ts)
-{
- TempOptInfo *ti = ts_info(ts);
- TempOptInfo *pi = ts_info(ti->prev_copy);
- TempOptInfo *ni = ts_info(ti->next_copy);
-
- ni->prev_copy = ti->prev_copy;
- pi->next_copy = ti->next_copy;
- ti->next_copy = ts;
- ti->prev_copy = ts;
- ti->is_const = false;
- ti->z_mask = -1;
- ti->s_mask = 0;
-}
-
-static void reset_temp(OptContext *ctx, TCGArg arg)
-{
- reset_ts(ctx, arg_temp(arg));
-}
-
/* Initialize and activate a temporary. */
static void init_ts_info(OptContext *ctx, TCGTemp *ts)
{
@@ -167,6 +158,7 @@ static void init_ts_info(OptContext *ctx, TCGTemp *ts)
ti->next_copy = ts;
ti->prev_copy = ts;
+ QSIMPLEQ_INIT(&ti->mem_copy);
if (ts->kind == TEMP_CONST) {
ti->is_const = true;
ti->val = ts->val;
@@ -179,6 +171,45 @@ static void init_ts_info(OptContext *ctx, TCGTemp *ts)
}
}
+static MemCopyInfo *mem_copy_first(OptContext *ctx, intptr_t s, intptr_t l)
+{
+ IntervalTreeNode *r = interval_tree_iter_first(&ctx->mem_copy, s, l);
+ return r ? container_of(r, MemCopyInfo, itree) : NULL;
+}
+
+static MemCopyInfo *mem_copy_next(MemCopyInfo *mem, intptr_t s, intptr_t l)
+{
+ IntervalTreeNode *r = interval_tree_iter_next(&mem->itree, s, l);
+ return r ? container_of(r, MemCopyInfo, itree) : NULL;
+}
+
+static void remove_mem_copy(OptContext *ctx, MemCopyInfo *mc)
+{
+ TCGTemp *ts = mc->ts;
+ TempOptInfo *ti = ts_info(ts);
+
+ interval_tree_remove(&mc->itree, &ctx->mem_copy);
+ QSIMPLEQ_REMOVE(&ti->mem_copy, mc, MemCopyInfo, next);
+ QSIMPLEQ_INSERT_TAIL(&ctx->mem_free, mc, next);
+}
+
+static void remove_mem_copy_in(OptContext *ctx, intptr_t s, intptr_t l)
+{
+ while (true) {
+ MemCopyInfo *mc = mem_copy_first(ctx, s, l);
+ if (!mc) {
+ break;
+ }
+ remove_mem_copy(ctx, mc);
+ }
+}
+
+static void remove_mem_copy_all(OptContext *ctx)
+{
+ remove_mem_copy_in(ctx, 0, -1);
+ tcg_debug_assert(interval_tree_is_empty(&ctx->mem_copy));
+}
+
static TCGTemp *find_better_copy(TCGTemp *ts)
{
TCGTemp *i, *ret;
@@ -195,6 +226,80 @@ static TCGTemp *find_better_copy(TCGTemp *ts)
return ret;
}
+static void move_mem_copies(TCGTemp *dst_ts, TCGTemp *src_ts)
+{
+ TempOptInfo *si = ts_info(src_ts);
+ TempOptInfo *di = ts_info(dst_ts);
+ MemCopyInfo *mc;
+
+ QSIMPLEQ_FOREACH(mc, &si->mem_copy, next) {
+ tcg_debug_assert(mc->ts == src_ts);
+ mc->ts = dst_ts;
+ }
+ QSIMPLEQ_CONCAT(&di->mem_copy, &si->mem_copy);
+}
+
+/* Reset TEMP's state, possibly removing the temp for the list of copies. */
+static void reset_ts(OptContext *ctx, TCGTemp *ts)
+{
+ TempOptInfo *ti = ts_info(ts);
+ TCGTemp *pts = ti->prev_copy;
+ TCGTemp *nts = ti->next_copy;
+ TempOptInfo *pi = ts_info(pts);
+ TempOptInfo *ni = ts_info(nts);
+
+ ni->prev_copy = ti->prev_copy;
+ pi->next_copy = ti->next_copy;
+ ti->next_copy = ts;
+ ti->prev_copy = ts;
+ ti->is_const = false;
+ ti->z_mask = -1;
+ ti->s_mask = 0;
+
+ if (!QSIMPLEQ_EMPTY(&ti->mem_copy)) {
+ if (ts == nts) {
+ /* Last temp copy being removed, the mem copies die. */
+ MemCopyInfo *mc;
+ QSIMPLEQ_FOREACH(mc, &ti->mem_copy, next) {
+ interval_tree_remove(&mc->itree, &ctx->mem_copy);
+ }
+ QSIMPLEQ_CONCAT(&ctx->mem_free, &ti->mem_copy);
+ } else {
+ move_mem_copies(find_better_copy(nts), ts);
+ }
+ }
+}
+
+static void reset_temp(OptContext *ctx, TCGArg arg)
+{
+ reset_ts(ctx, arg_temp(arg));
+}
+
+static void record_mem_copy(OptContext *ctx, TCGType type,
+ TCGTemp *ts, intptr_t start, intptr_t last)
+{
+ MemCopyInfo *mc;
+ TempOptInfo *ti;
+
+ mc = QSIMPLEQ_FIRST(&ctx->mem_free);
+ if (mc) {
+ QSIMPLEQ_REMOVE_HEAD(&ctx->mem_free, next);
+ } else {
+ mc = tcg_malloc(sizeof(*mc));
+ }
+
+ memset(mc, 0, sizeof(*mc));
+ mc->itree.start = start;
+ mc->itree.last = last;
+ mc->type = type;
+ interval_tree_insert(&mc->itree, &ctx->mem_copy);
+
+ ts = find_better_copy(ts);
+ ti = ts_info(ts);
+ mc->ts = ts;
+ QSIMPLEQ_INSERT_TAIL(&ti->mem_copy, mc, next);
+}
+
static bool ts_are_copies(TCGTemp *ts1, TCGTemp *ts2)
{
TCGTemp *i;
@@ -221,6 +326,18 @@ static bool args_are_copies(TCGArg arg1, TCGArg arg2)
return ts_are_copies(arg_temp(arg1), arg_temp(arg2));
}
+static TCGTemp *find_mem_copy_for(OptContext *ctx, TCGType type, intptr_t s)
+{
+ MemCopyInfo *mc;
+
+ for (mc = mem_copy_first(ctx, s, s); mc; mc = mem_copy_next(mc, s, s)) {
+ if (mc->itree.start == s && mc->type == type) {
+ return find_better_copy(mc->ts);
+ }
+ }
+ return NULL;
+}
+
static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src)
{
TCGTemp *dst_ts = arg_temp(dst);
@@ -270,6 +387,11 @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src)
si->next_copy = dst_ts;
di->is_const = si->is_const;
di->val = si->val;
+
+ if (!QSIMPLEQ_EMPTY(&si->mem_copy)
+ && cmp_better_copy(src_ts, dst_ts) == dst_ts) {
+ move_mem_copies(dst_ts, src_ts);
+ }
}
return true;
}
@@ -688,6 +810,7 @@ static void finish_folding(OptContext *ctx, TCGOp *op)
ctx->prev_mb = NULL;
if (!(def->flags & TCG_OPF_COND_BRANCH)) {
memset(&ctx->temps_used, 0, sizeof(ctx->temps_used));
+ remove_mem_copy_all(ctx);
}
return;
}
@@ -1213,6 +1336,11 @@ static bool fold_call(OptContext *ctx, TCGOp *op)
}
}
+ /* If the function has side effects, reset mem data. */
+ if (!(flags & TCG_CALL_NO_SIDE_EFFECTS)) {
+ remove_mem_copy_all(ctx);
+ }
+
/* Reset temp data for outputs. */
for (i = 0; i < nb_oargs; i++) {
reset_temp(ctx, op->args[i]);
@@ -2070,6 +2198,83 @@ static bool fold_tcg_ld(OptContext *ctx, TCGOp *op)
return false;
}
+static bool fold_tcg_ld_memcopy(OptContext *ctx, TCGOp *op)
+{
+ TCGTemp *dst, *src;
+ intptr_t ofs;
+ TCGType type;
+
+ if (op->args[1] != tcgv_ptr_arg(tcg_env)) {
+ return false;
+ }
+
+ type = ctx->type;
+ ofs = op->args[2];
+ dst = arg_temp(op->args[0]);
+ src = find_mem_copy_for(ctx, type, ofs);
+ if (src && src->base_type == type) {
+ return tcg_opt_gen_mov(ctx, op, temp_arg(dst), temp_arg(src));
+ }
+
+ reset_ts(ctx, dst);
+ record_mem_copy(ctx, type, dst, ofs, ofs + tcg_type_size(type) - 1);
+ return true;
+}
+
+static bool fold_tcg_st(OptContext *ctx, TCGOp *op)
+{
+ intptr_t ofs = op->args[2];
+ intptr_t lm1;
+
+ if (op->args[1] != tcgv_ptr_arg(tcg_env)) {
+ remove_mem_copy_all(ctx);
+ return false;
+ }
+
+ switch (op->opc) {
+ CASE_OP_32_64(st8):
+ lm1 = 0;
+ break;
+ CASE_OP_32_64(st16):
+ lm1 = 1;
+ break;
+ case INDEX_op_st32_i64:
+ case INDEX_op_st_i32:
+ lm1 = 3;
+ break;
+ case INDEX_op_st_i64:
+ lm1 = 7;
+ break;
+ case INDEX_op_st_vec:
+ lm1 = tcg_type_size(ctx->type) - 1;
+ break;
+ default:
+ g_assert_not_reached();
+ }
+ remove_mem_copy_in(ctx, ofs, ofs + lm1);
+ return false;
+}
+
+static bool fold_tcg_st_memcopy(OptContext *ctx, TCGOp *op)
+{
+ TCGTemp *src;
+ intptr_t ofs, last;
+ TCGType type;
+
+ if (op->args[1] != tcgv_ptr_arg(tcg_env)) {
+ fold_tcg_st(ctx, op);
+ return false;
+ }
+
+ src = arg_temp(op->args[0]);
+ ofs = op->args[2];
+ type = ctx->type;
+ last = ofs + tcg_type_size(type) - 1;
+ remove_mem_copy_in(ctx, ofs, last);
+ record_mem_copy(ctx, type, src, ofs, last);
+ return false;
+}
+
static bool fold_xor(OptContext *ctx, TCGOp *op)
{
if (fold_const2_commutative(ctx, op) ||
@@ -2093,6 +2298,8 @@ void tcg_optimize(TCGContext *s)
TCGOp *op, *op_next;
OptContext ctx = { .tcg = s };
+ QSIMPLEQ_INIT(&ctx.mem_free);
+
/* Array VALS has an element for each temp.
If this temp holds a constant then its value is kept in VALS' element.
If this temp is a copy of other ones then the other copies are
@@ -2214,6 +2421,21 @@ void tcg_optimize(TCGContext *s)
case INDEX_op_ld32u_i64:
done = fold_tcg_ld(&ctx, op);
break;
+ case INDEX_op_ld_i32:
+ case INDEX_op_ld_i64:
+ case INDEX_op_ld_vec:
+ done = fold_tcg_ld_memcopy(&ctx, op);
+ break;
+ CASE_OP_32_64(st8):
+ CASE_OP_32_64(st16):
+ case INDEX_op_st32_i64:
+ done = fold_tcg_st(&ctx, op);
+ break;
+ case INDEX_op_st_i32:
+ case INDEX_op_st_i64:
+ case INDEX_op_st_vec:
+ done = fold_tcg_st_memcopy(&ctx, op);
+ break;
case INDEX_op_mb:
done = fold_mb(&ctx, op);
break;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 30/35] tcg: Eliminate duplicate env store operations
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (28 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 29/35] tcg/optimize: Optimize env memory operations Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 31/35] tcg/optimize: Split out arg_new_constant Richard Henderson
` (6 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Song Gao
Notice when a constant is stored to the same location twice.
Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/optimize.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index b32ef0be0f..a4fe9ee9bb 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -2269,6 +2269,19 @@ static bool fold_tcg_st_memcopy(OptContext *ctx, TCGOp *op)
src = arg_temp(op->args[0]);
ofs = op->args[2];
type = ctx->type;
+
+ /*
+ * Eliminate duplicate stores of a constant.
+ * This happens frequently when the target ISA zero-extends.
+ */
+ if (ts_is_const(src)) {
+ TCGTemp *prev = find_mem_copy_for(ctx, type, ofs);
+ if (src == prev) {
+ tcg_op_remove(ctx->tcg, op);
+ return true;
+ }
+ }
+
last = ofs + tcg_type_size(type) - 1;
remove_mem_copy_in(ctx, ofs, last);
record_mem_copy(ctx, type, src, ofs, last);
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 31/35] tcg/optimize: Split out arg_new_constant
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (29 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 30/35] tcg: Eliminate duplicate env store operations Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 32/35] tcg: Canonicalize subi to addi during opcode generation Richard Henderson
` (5 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Philippe Mathieu-Daudé
Fixes a bug wherein raw uses of tcg_constant_internal
do not have their TempOptInfo initialized.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/optimize.c | 29 ++++++++++++++++++-----------
1 file changed, 18 insertions(+), 11 deletions(-)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index a4fe9ee9bb..d8e437c826 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -338,6 +338,21 @@ static TCGTemp *find_mem_copy_for(OptContext *ctx, TCGType type, intptr_t s)
return NULL;
}
+static TCGArg arg_new_constant(OptContext *ctx, uint64_t val)
+{
+ TCGType type = ctx->type;
+ TCGTemp *ts;
+
+ if (type == TCG_TYPE_I32) {
+ val = (int32_t)val;
+ }
+
+ ts = tcg_constant_internal(type, val);
+ init_ts_info(ctx, ts);
+
+ return temp_arg(ts);
+}
+
static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src)
{
TCGTemp *dst_ts = arg_temp(dst);
@@ -399,16 +414,8 @@ static bool tcg_opt_gen_mov(OptContext *ctx, TCGOp *op, TCGArg dst, TCGArg src)
static bool tcg_opt_gen_movi(OptContext *ctx, TCGOp *op,
TCGArg dst, uint64_t val)
{
- TCGTemp *tv;
-
- if (ctx->type == TCG_TYPE_I32) {
- val = (int32_t)val;
- }
-
/* Convert movi to mov with constant temp. */
- tv = tcg_constant_internal(ctx->type, val);
- init_ts_info(ctx, tv);
- return tcg_opt_gen_mov(ctx, op, dst, temp_arg(tv));
+ return tcg_opt_gen_mov(ctx, op, dst, arg_new_constant(ctx, val));
}
static uint64_t do_constant_folding_2(TCGOpcode op, uint64_t x, uint64_t y)
@@ -1431,7 +1438,7 @@ static bool fold_deposit(OptContext *ctx, TCGOp *op)
op->opc = and_opc;
op->args[1] = op->args[2];
- op->args[2] = temp_arg(tcg_constant_internal(ctx->type, mask));
+ op->args[2] = arg_new_constant(ctx, mask);
ctx->z_mask = mask & arg_info(op->args[1])->z_mask;
return false;
}
@@ -1442,7 +1449,7 @@ static bool fold_deposit(OptContext *ctx, TCGOp *op)
uint64_t mask = deposit64(-1, op->args[3], op->args[4], 0);
op->opc = and_opc;
- op->args[2] = temp_arg(tcg_constant_internal(ctx->type, mask));
+ op->args[2] = arg_new_constant(ctx, mask);
ctx->z_mask = mask & arg_info(op->args[1])->z_mask;
return false;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 32/35] tcg: Canonicalize subi to addi during opcode generation
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (30 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 31/35] tcg/optimize: Split out arg_new_constant Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 33/35] tcg/optimize: Canonicalize subi to addi during optimization Richard Henderson
` (4 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Paolo Bonzini, Philippe Mathieu-Daudé
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231026013945.1152174-2-richard.henderson@linaro.org>
---
tcg/tcg-op.c | 18 ++----------------
1 file changed, 2 insertions(+), 16 deletions(-)
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index de096a6f93..aa6bc6f57d 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -372,12 +372,7 @@ void tcg_gen_subfi_i32(TCGv_i32 ret, int32_t arg1, TCGv_i32 arg2)
void tcg_gen_subi_i32(TCGv_i32 ret, TCGv_i32 arg1, int32_t arg2)
{
- /* some cases can be optimized here */
- if (arg2 == 0) {
- tcg_gen_mov_i32(ret, arg1);
- } else {
- tcg_gen_sub_i32(ret, arg1, tcg_constant_i32(arg2));
- }
+ tcg_gen_addi_i32(ret, arg1, -arg2);
}
void tcg_gen_neg_i32(TCGv_i32 ret, TCGv_i32 arg)
@@ -1752,16 +1747,7 @@ void tcg_gen_subfi_i64(TCGv_i64 ret, int64_t arg1, TCGv_i64 arg2)
void tcg_gen_subi_i64(TCGv_i64 ret, TCGv_i64 arg1, int64_t arg2)
{
- /* some cases can be optimized here */
- if (arg2 == 0) {
- tcg_gen_mov_i64(ret, arg1);
- } else if (TCG_TARGET_REG_BITS == 64) {
- tcg_gen_sub_i64(ret, arg1, tcg_constant_i64(arg2));
- } else {
- tcg_gen_sub2_i32(TCGV_LOW(ret), TCGV_HIGH(ret),
- TCGV_LOW(arg1), TCGV_HIGH(arg1),
- tcg_constant_i32(arg2), tcg_constant_i32(arg2 >> 32));
- }
+ tcg_gen_addi_i64(ret, arg1, -arg2);
}
void tcg_gen_neg_i64(TCGv_i64 ret, TCGv_i64 arg)
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 33/35] tcg/optimize: Canonicalize subi to addi during optimization
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (31 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 32/35] tcg: Canonicalize subi to addi during opcode generation Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 34/35] tcg/optimize: Canonicalize sub2 with constants to add2 Richard Henderson
` (3 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Philippe Mathieu-Daudé
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231026013945.1152174-3-richard.henderson@linaro.org>
---
tcg/optimize.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index d8e437c826..468f827399 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -2166,7 +2166,19 @@ static bool fold_sub_vec(OptContext *ctx, TCGOp *op)
static bool fold_sub(OptContext *ctx, TCGOp *op)
{
- return fold_const2(ctx, op) || fold_sub_vec(ctx, op);
+ if (fold_const2(ctx, op) || fold_sub_vec(ctx, op)) {
+ return true;
+ }
+
+ /* Fold sub r,x,i to add r,x,-i */
+ if (arg_is_const(op->args[2])) {
+ uint64_t val = arg_info(op->args[2])->val;
+
+ op->opc = (ctx->type == TCG_TYPE_I32
+ ? INDEX_op_add_i32 : INDEX_op_add_i64);
+ op->args[2] = arg_new_constant(ctx, -val);
+ }
+ return false;
}
static bool fold_sub2(OptContext *ctx, TCGOp *op)
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 34/35] tcg/optimize: Canonicalize sub2 with constants to add2
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (32 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 33/35] tcg/optimize: Canonicalize subi to addi during optimization Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:48 ` [PATCH 35/35] tcg/sparc64: Implement tcg_out_extrl_i64_i32 Richard Henderson
` (2 subsequent siblings)
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: Philippe Mathieu-Daudé
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231026013945.1152174-4-richard.henderson@linaro.org>
---
tcg/optimize.c | 21 +++++++++++++++++++--
1 file changed, 19 insertions(+), 2 deletions(-)
diff --git a/tcg/optimize.c b/tcg/optimize.c
index 468f827399..f2d01654c5 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -1044,8 +1044,10 @@ static bool fold_add_vec(OptContext *ctx, TCGOp *op)
static bool fold_addsub2(OptContext *ctx, TCGOp *op, bool add)
{
- if (arg_is_const(op->args[2]) && arg_is_const(op->args[3]) &&
- arg_is_const(op->args[4]) && arg_is_const(op->args[5])) {
+ bool a_const = arg_is_const(op->args[2]) && arg_is_const(op->args[3]);
+ bool b_const = arg_is_const(op->args[4]) && arg_is_const(op->args[5]);
+
+ if (a_const && b_const) {
uint64_t al = arg_info(op->args[2])->val;
uint64_t ah = arg_info(op->args[3])->val;
uint64_t bl = arg_info(op->args[4])->val;
@@ -1089,6 +1091,21 @@ static bool fold_addsub2(OptContext *ctx, TCGOp *op, bool add)
tcg_opt_gen_movi(ctx, op2, rh, ah);
return true;
}
+
+ /* Fold sub2 r,x,i to add2 r,x,-i */
+ if (!add && b_const) {
+ uint64_t bl = arg_info(op->args[4])->val;
+ uint64_t bh = arg_info(op->args[5])->val;
+
+ /* Negate the two parts without assembling and disassembling. */
+ bl = -bl;
+ bh = ~bh + !bl;
+
+ op->opc = (ctx->type == TCG_TYPE_I32
+ ? INDEX_op_add2_i32 : INDEX_op_add2_i64);
+ op->args[4] = arg_new_constant(ctx, bl);
+ op->args[5] = arg_new_constant(ctx, bh);
+ }
return false;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PATCH 35/35] tcg/sparc64: Implement tcg_out_extrl_i64_i32
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (33 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 34/35] tcg/optimize: Canonicalize sub2 with constants to add2 Richard Henderson
@ 2023-11-07 2:48 ` Richard Henderson
2023-11-07 2:55 ` [PULL 00/35] tcg patch queue Richard Henderson
2023-11-07 4:59 ` [PATCH " Stefan Hajnoczi
36 siblings, 0 replies; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:48 UTC (permalink / raw)
To: qemu-devel; +Cc: qemu-stable, Philippe Mathieu-Daudé
Build fix for missing symbol.
Cc: qemu-stable@nongnu.org
Fixes: dad2f2f5af ("tcg/sparc64: Disable TCG_TARGET_HAS_extr_i64_i32")
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/sparc64/tcg-target.c.inc | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/tcg/sparc64/tcg-target.c.inc b/tcg/sparc64/tcg-target.c.inc
index 19d9df4a09..a91defd0ac 100644
--- a/tcg/sparc64/tcg-target.c.inc
+++ b/tcg/sparc64/tcg-target.c.inc
@@ -529,6 +529,11 @@ static void tcg_out_extu_i32_i64(TCGContext *s, TCGReg rd, TCGReg rs)
tcg_out_ext32u(s, rd, rs);
}
+static void tcg_out_extrl_i64_i32(TCGContext *s, TCGReg rd, TCGReg rs)
+{
+ tcg_out_ext32u(s, rd, rs);
+}
+
static bool tcg_out_xchg(TCGContext *s, TCGType type, TCGReg r1, TCGReg r2)
{
return false;
--
2.34.1
^ permalink raw reply related [flat|nested] 50+ messages in thread* [PULL 00/35] tcg patch queue
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (34 preceding siblings ...)
2023-11-07 2:48 ` [PATCH 35/35] tcg/sparc64: Implement tcg_out_extrl_i64_i32 Richard Henderson
@ 2023-11-07 2:55 ` Richard Henderson
2023-11-07 3:06 ` Stefan Hajnoczi
2023-11-07 4:59 ` [PATCH " Stefan Hajnoczi
36 siblings, 1 reply; 50+ messages in thread
From: Richard Henderson @ 2023-11-07 2:55 UTC (permalink / raw)
To: qemu-devel, Stefan Hajnoczi
Dangit, forgot the PULL tag.
r~
On 11/6/23 18:48, Richard Henderson wrote:
> The following changes since commit 3e01f1147a16ca566694b97eafc941d62fa1e8d8:
>
> Merge tag 'pull-sp-20231105' of https://gitlab.com/rth7680/qemu into staging (2023-11-06 09:34:22 +0800)
>
> are available in the Git repository at:
>
> https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20231106
>
> for you to fetch changes up to d36ce28be424385fc9f7273bf5c15ce815b5cf4e:
>
> tcg/sparc64: Implement tcg_out_extrl_i64_i32 (2023-11-06 10:48:46 -0800)
>
> ----------------------------------------------------------------
> util: Add cpuinfo for loongarch64
> tcg/loongarch64: Use cpuinfo.h
> tcg/loongarch64: Improve register allocation for INDEX_op_qemu_ld_a*_i128
> host/include/loongarch64: Add atomic16 load and store
> tcg: Move expanders out of line
> tcg/mips: Always implement movcond
> tcg/mips: Implement neg opcodes
> tcg/loongarch64: Implement neg opcodes
> tcg: Make movcond and neg required opcodes
> tcg: Optimize env memory operations
> tcg: Canonicalize sub of immediate to add
> tcg/sparc64: Implement tcg_out_extrl_i64_i32
>
> ----------------------------------------------------------------
> Richard Henderson (35):
> accel/tcg: Move HMP info jit and info opcount code
> tcg: Add C_N2_I1
> tcg/loongarch64: Use C_N2_I1 for INDEX_op_qemu_ld_a*_i128
> util: Add cpuinfo for loongarch64
> tcg/loongarch64: Use cpuinfo.h
> host/include/loongarch64: Add atomic16 load and store
> accel/tcg: Remove redundant case in store_atom_16
> accel/tcg: Fix condition for store_atom_insert_al16
> tcg: Mark tcg_gen_op* as noinline
> tcg: Move tcg_gen_op* out of line
> tcg: Move generic expanders out of line
> tcg: Move 32-bit expanders out of line
> tcg: Move 64-bit expanders out of line
> tcg: Move vec_gen_* declarations to tcg-internal.h
> tcg: Move tcg_gen_opN declarations to tcg-internal.h
> tcg: Unexport tcg_gen_op*_{i32,i64}
> tcg: Move tcg_constant_* out of line
> tcg: Move tcg_temp_new_*, tcg_global_mem_new_* out of line
> tcg: Move tcg_temp_free_* out of line
> tcg/mips: Split out tcg_out_setcond_int
> tcg/mips: Always implement movcond
> tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64}
> tcg/mips: Implement neg opcodes
> tcg/loongarch64: Implement neg opcodes
> tcg: Remove TCG_TARGET_HAS_neg_{i32,i64}
> tcg: Don't free vector results
> tcg/optimize: Pipe OptContext into reset_ts
> tcg/optimize: Split out cmp_better_copy
> tcg/optimize: Optimize env memory operations
> tcg: Eliminate duplicate env store operations
> tcg/optimize: Split out arg_new_constant
> tcg: Canonicalize subi to addi during opcode generation
> tcg/optimize: Canonicalize subi to addi during optimization
> tcg/optimize: Canonicalize sub2 with constants to add2
> tcg/sparc64: Implement tcg_out_extrl_i64_i32
>
> accel/tcg/internal-common.h | 2 -
> host/include/loongarch64/host/atomic128-ldst.h | 52 ++
> host/include/loongarch64/host/cpuinfo.h | 21 +
> .../loongarch64/host/load-extract-al16-al8.h | 39 ++
> host/include/loongarch64/host/store-insert-al16.h | 12 +
> include/exec/cputlb.h | 1 -
> include/tcg/tcg-op-common.h | 538 ++---------------
> include/tcg/tcg-opc.h | 8 +-
> include/tcg/tcg-temp-internal.h | 56 +-
> include/tcg/tcg.h | 82 ---
> tcg/aarch64/tcg-target.h | 4 -
> tcg/arm/tcg-target.h | 2 -
> tcg/i386/tcg-target.h | 4 -
> tcg/loongarch64/tcg-target-con-set.h | 2 +-
> tcg/loongarch64/tcg-target.h | 12 +-
> tcg/mips/tcg-target.h | 4 -
> tcg/ppc/tcg-target.h | 4 -
> tcg/riscv/tcg-target.h | 4 -
> tcg/s390x/tcg-target.h | 4 -
> tcg/sparc64/tcg-target.h | 4 -
> tcg/tcg-internal.h | 18 +
> tcg/tci/tcg-target.h | 4 -
> accel/tcg/cputlb.c | 17 +-
> accel/tcg/monitor.c | 154 +++++
> accel/tcg/translate-all.c | 127 ----
> tcg/optimize.c | 370 ++++++++++--
> tcg/tcg-op-gvec.c | 112 ++--
> tcg/tcg-op.c | 636 ++++++++++++++++-----
> tcg/tcg.c | 131 ++++-
> tcg/tci.c | 2 -
> util/cpuinfo-loongarch.c | 35 ++
> accel/tcg/ldst_atomicity.c.inc | 14 +-
> tcg/loongarch64/tcg-target.c.inc | 34 +-
> tcg/mips/tcg-target.c.inc | 329 +++++------
> tcg/sparc64/tcg-target.c.inc | 5 +
> util/meson.build | 2 +
> 36 files changed, 1535 insertions(+), 1310 deletions(-)
> create mode 100644 host/include/loongarch64/host/atomic128-ldst.h
> create mode 100644 host/include/loongarch64/host/cpuinfo.h
> create mode 100644 host/include/loongarch64/host/load-extract-al16-al8.h
> create mode 100644 host/include/loongarch64/host/store-insert-al16.h
> create mode 100644 util/cpuinfo-loongarch.c
^ permalink raw reply [flat|nested] 50+ messages in thread* Re: [PULL 00/35] tcg patch queue
2023-11-07 2:55 ` [PULL 00/35] tcg patch queue Richard Henderson
@ 2023-11-07 3:06 ` Stefan Hajnoczi
0 siblings, 0 replies; 50+ messages in thread
From: Stefan Hajnoczi @ 2023-11-07 3:06 UTC (permalink / raw)
To: Richard Henderson; +Cc: qemu-devel, Stefan Hajnoczi
On Tue, 7 Nov 2023 at 10:56, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Dangit, forgot the PULL tag.
Okay, thanks for pointing it out. I've added this to the staging branch.
Stefan
>
> r~
>
> On 11/6/23 18:48, Richard Henderson wrote:
> > The following changes since commit 3e01f1147a16ca566694b97eafc941d62fa1e8d8:
> >
> > Merge tag 'pull-sp-20231105' of https://gitlab.com/rth7680/qemu into staging (2023-11-06 09:34:22 +0800)
> >
> > are available in the Git repository at:
> >
> > https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20231106
> >
> > for you to fetch changes up to d36ce28be424385fc9f7273bf5c15ce815b5cf4e:
> >
> > tcg/sparc64: Implement tcg_out_extrl_i64_i32 (2023-11-06 10:48:46 -0800)
> >
> > ----------------------------------------------------------------
> > util: Add cpuinfo for loongarch64
> > tcg/loongarch64: Use cpuinfo.h
> > tcg/loongarch64: Improve register allocation for INDEX_op_qemu_ld_a*_i128
> > host/include/loongarch64: Add atomic16 load and store
> > tcg: Move expanders out of line
> > tcg/mips: Always implement movcond
> > tcg/mips: Implement neg opcodes
> > tcg/loongarch64: Implement neg opcodes
> > tcg: Make movcond and neg required opcodes
> > tcg: Optimize env memory operations
> > tcg: Canonicalize sub of immediate to add
> > tcg/sparc64: Implement tcg_out_extrl_i64_i32
> >
> > ----------------------------------------------------------------
> > Richard Henderson (35):
> > accel/tcg: Move HMP info jit and info opcount code
> > tcg: Add C_N2_I1
> > tcg/loongarch64: Use C_N2_I1 for INDEX_op_qemu_ld_a*_i128
> > util: Add cpuinfo for loongarch64
> > tcg/loongarch64: Use cpuinfo.h
> > host/include/loongarch64: Add atomic16 load and store
> > accel/tcg: Remove redundant case in store_atom_16
> > accel/tcg: Fix condition for store_atom_insert_al16
> > tcg: Mark tcg_gen_op* as noinline
> > tcg: Move tcg_gen_op* out of line
> > tcg: Move generic expanders out of line
> > tcg: Move 32-bit expanders out of line
> > tcg: Move 64-bit expanders out of line
> > tcg: Move vec_gen_* declarations to tcg-internal.h
> > tcg: Move tcg_gen_opN declarations to tcg-internal.h
> > tcg: Unexport tcg_gen_op*_{i32,i64}
> > tcg: Move tcg_constant_* out of line
> > tcg: Move tcg_temp_new_*, tcg_global_mem_new_* out of line
> > tcg: Move tcg_temp_free_* out of line
> > tcg/mips: Split out tcg_out_setcond_int
> > tcg/mips: Always implement movcond
> > tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64}
> > tcg/mips: Implement neg opcodes
> > tcg/loongarch64: Implement neg opcodes
> > tcg: Remove TCG_TARGET_HAS_neg_{i32,i64}
> > tcg: Don't free vector results
> > tcg/optimize: Pipe OptContext into reset_ts
> > tcg/optimize: Split out cmp_better_copy
> > tcg/optimize: Optimize env memory operations
> > tcg: Eliminate duplicate env store operations
> > tcg/optimize: Split out arg_new_constant
> > tcg: Canonicalize subi to addi during opcode generation
> > tcg/optimize: Canonicalize subi to addi during optimization
> > tcg/optimize: Canonicalize sub2 with constants to add2
> > tcg/sparc64: Implement tcg_out_extrl_i64_i32
> >
> > accel/tcg/internal-common.h | 2 -
> > host/include/loongarch64/host/atomic128-ldst.h | 52 ++
> > host/include/loongarch64/host/cpuinfo.h | 21 +
> > .../loongarch64/host/load-extract-al16-al8.h | 39 ++
> > host/include/loongarch64/host/store-insert-al16.h | 12 +
> > include/exec/cputlb.h | 1 -
> > include/tcg/tcg-op-common.h | 538 ++---------------
> > include/tcg/tcg-opc.h | 8 +-
> > include/tcg/tcg-temp-internal.h | 56 +-
> > include/tcg/tcg.h | 82 ---
> > tcg/aarch64/tcg-target.h | 4 -
> > tcg/arm/tcg-target.h | 2 -
> > tcg/i386/tcg-target.h | 4 -
> > tcg/loongarch64/tcg-target-con-set.h | 2 +-
> > tcg/loongarch64/tcg-target.h | 12 +-
> > tcg/mips/tcg-target.h | 4 -
> > tcg/ppc/tcg-target.h | 4 -
> > tcg/riscv/tcg-target.h | 4 -
> > tcg/s390x/tcg-target.h | 4 -
> > tcg/sparc64/tcg-target.h | 4 -
> > tcg/tcg-internal.h | 18 +
> > tcg/tci/tcg-target.h | 4 -
> > accel/tcg/cputlb.c | 17 +-
> > accel/tcg/monitor.c | 154 +++++
> > accel/tcg/translate-all.c | 127 ----
> > tcg/optimize.c | 370 ++++++++++--
> > tcg/tcg-op-gvec.c | 112 ++--
> > tcg/tcg-op.c | 636 ++++++++++++++++-----
> > tcg/tcg.c | 131 ++++-
> > tcg/tci.c | 2 -
> > util/cpuinfo-loongarch.c | 35 ++
> > accel/tcg/ldst_atomicity.c.inc | 14 +-
> > tcg/loongarch64/tcg-target.c.inc | 34 +-
> > tcg/mips/tcg-target.c.inc | 329 +++++------
> > tcg/sparc64/tcg-target.c.inc | 5 +
> > util/meson.build | 2 +
> > 36 files changed, 1535 insertions(+), 1310 deletions(-)
> > create mode 100644 host/include/loongarch64/host/atomic128-ldst.h
> > create mode 100644 host/include/loongarch64/host/cpuinfo.h
> > create mode 100644 host/include/loongarch64/host/load-extract-al16-al8.h
> > create mode 100644 host/include/loongarch64/host/store-insert-al16.h
> > create mode 100644 util/cpuinfo-loongarch.c
>
>
^ permalink raw reply [flat|nested] 50+ messages in thread
* Re: [PATCH 00/35] tcg patch queue
2023-11-07 2:48 [PATCH 00/35] tcg patch queue Richard Henderson
` (35 preceding siblings ...)
2023-11-07 2:55 ` [PULL 00/35] tcg patch queue Richard Henderson
@ 2023-11-07 4:59 ` Stefan Hajnoczi
36 siblings, 0 replies; 50+ messages in thread
From: Stefan Hajnoczi @ 2023-11-07 4:59 UTC (permalink / raw)
To: Richard Henderson; +Cc: qemu-devel
[-- Attachment #1: Type: text/plain, Size: 115 bytes --]
Applied, thanks.
Please update the changelog at https://wiki.qemu.org/ChangeLog/8.2 for any user-visible changes.
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
^ permalink raw reply [flat|nested] 50+ messages in thread