diff mbox series

[6/8] target/sparc: Fix VIS fpmerge input registers.

Message ID 20230925050545.30912-7-nbowler@draconx.ca
State New
Headers show
Series SPARC VIS fixes | expand

Commit Message

Nick Bowler Sept. 25, 2023, 5:03 a.m. UTC 
On a real UltraSparc II CPU, the fpmerge instruction reads two
single-precision input registers, but the emulator is reading
from double-precision input registers instead.

These are unlikely to contain the correct data so in most instances
the results of the emulation are just garbage in most instances.

Signed-off-by: Nick Bowler <nbowler@draconx.ca>
---
 target/sparc/helper.h     |  2 +-
 target/sparc/translate.c  |  6 +++++-
 target/sparc/vis_helper.c | 26 +++++++++++++-------------
 3 files changed, 19 insertions(+), 15 deletions(-)

Comments

Richard Henderson Sept. 28, 2023, 9:35 p.m. UTC | #1 
On 9/24/23 01:03, Nick Bowler wrote:
>                   case 0x04b: /* VIS I fpmerge */
>                       CHECK_FPU_FEATURE(dc, VIS1);
> -                    gen_ne_fop_DDD(dc, rd, rs1, rs2, gen_helper_fpmerge);
> +                    cpu_src1_32 = gen_load_fpr_F(dc, rs1);
> +                    cpu_src2_32 = gen_load_fpr_F(dc, rs2);
> +                    cpu_dst_64 = gen_dest_fpr_D(dc, rd);
> +                    gen_helper_fpmerge(cpu_dst_64, cpu_src1_32, cpu_src2_32);
> +                    gen_store_fpr_D(dc, rd, cpu_dst_64);
>                       break;

Use gen_ne_fop_DFF.


r~
Nick Bowler Sept. 29, 2023, 12:32 a.m. UTC | #2 
On 2023-09-28, Richard Henderson <richard.henderson@linaro.org> wrote:
> On 9/24/23 01:03, Nick Bowler wrote:
>>                   case 0x04b: /* VIS I fpmerge */
>>                       CHECK_FPU_FEATURE(dc, VIS1);
>> -                    gen_ne_fop_DDD(dc, rd, rs1, rs2,
>> gen_helper_fpmerge);
>> +                    cpu_src1_32 = gen_load_fpr_F(dc, rs1);
>> +                    cpu_src2_32 = gen_load_fpr_F(dc, rs2);
>> +                    cpu_dst_64 = gen_dest_fpr_D(dc, rd);
>> +                    gen_helper_fpmerge(cpu_dst_64, cpu_src1_32,
>> cpu_src2_32);
>> +                    gen_store_fpr_D(dc, rd, cpu_dst_64);
>>                       break;
>
> Use gen_ne_fop_DFF.

Good catch, I clearly missed that this can use the new helper, I will
respin this one.

Thanks,
  Nick
diff mbox series

Patch

diff --git a/target/sparc/helper.h b/target/sparc/helper.h
index 7a588f3068..b71688079f 100644
--- a/target/sparc/helper.h
+++ b/target/sparc/helper.h
@@ -125,7 +125,7 @@  DEF_HELPER_FLAGS_2(fstox, TCG_CALL_NO_RWG, s64, env, f32)
 DEF_HELPER_FLAGS_2(fdtox, TCG_CALL_NO_RWG, s64, env, f64)
 DEF_HELPER_FLAGS_1(fqtox, TCG_CALL_NO_RWG, s64, env)
 
-DEF_HELPER_FLAGS_2(fpmerge, TCG_CALL_NO_RWG_SE, i64, i64, i64)
+DEF_HELPER_FLAGS_2(fpmerge, TCG_CALL_NO_RWG_SE, i64, i32, i32)
 DEF_HELPER_FLAGS_2(fmul8x16, TCG_CALL_NO_RWG_SE, i64, i32, i64)
 DEF_HELPER_FLAGS_2(fmul8x16al, TCG_CALL_NO_RWG_SE, i64, i32, i32)
 DEF_HELPER_FLAGS_2(fmul8x16au, TCG_CALL_NO_RWG_SE, i64, i32, i32)
diff --git a/target/sparc/translate.c b/target/sparc/translate.c
index cfccd95c3a..241ac429ca 100644
--- a/target/sparc/translate.c
+++ b/target/sparc/translate.c
@@ -4825,7 +4825,11 @@  static void disas_sparc_insn(DisasContext * dc, unsigned int insn)
                     break;
                 case 0x04b: /* VIS I fpmerge */
                     CHECK_FPU_FEATURE(dc, VIS1);
-                    gen_ne_fop_DDD(dc, rd, rs1, rs2, gen_helper_fpmerge);
+                    cpu_src1_32 = gen_load_fpr_F(dc, rs1);
+                    cpu_src2_32 = gen_load_fpr_F(dc, rs2);
+                    cpu_dst_64 = gen_dest_fpr_D(dc, rd);
+                    gen_helper_fpmerge(cpu_dst_64, cpu_src1_32, cpu_src2_32);
+                    gen_store_fpr_D(dc, rd, cpu_dst_64);
                     break;
                 case 0x04c: /* VIS II bshuffle */
                     CHECK_FPU_FEATURE(dc, VIS2);
diff --git a/target/sparc/vis_helper.c b/target/sparc/vis_helper.c
index 306383ba60..029aad3923 100644
--- a/target/sparc/vis_helper.c
+++ b/target/sparc/vis_helper.c
@@ -77,22 +77,22 @@  typedef union {
     float32 f;
 } VIS32;
 
-uint64_t helper_fpmerge(uint64_t src1, uint64_t src2)
+uint64_t helper_fpmerge(uint32_t src1, uint32_t src2)
 {
-    VIS64 s, d;
+    VIS32 s1, s2;
+    VIS64 d;
 
-    s.ll = src1;
-    d.ll = src2;
+    s1.l = src1;
+    s2.l = src2;
 
-    /* Reverse calculation order to handle overlap */
-    d.VIS_B64(7) = s.VIS_B64(3);
-    d.VIS_B64(6) = d.VIS_B64(3);
-    d.VIS_B64(5) = s.VIS_B64(2);
-    d.VIS_B64(4) = d.VIS_B64(2);
-    d.VIS_B64(3) = s.VIS_B64(1);
-    d.VIS_B64(2) = d.VIS_B64(1);
-    d.VIS_B64(1) = s.VIS_B64(0);
-    /* d.VIS_B64(0) = d.VIS_B64(0); */
+    d.VIS_B64(0) = s2.VIS_B32(0);
+    d.VIS_B64(1) = s1.VIS_B32(0);
+    d.VIS_B64(2) = s2.VIS_B32(1);
+    d.VIS_B64(3) = s1.VIS_B32(1);
+    d.VIS_B64(4) = s2.VIS_B32(2);
+    d.VIS_B64(5) = s1.VIS_B32(2);
+    d.VIS_B64(6) = s2.VIS_B32(3);
+    d.VIS_B64(7) = s1.VIS_B32(3);
 
     return d.ll;
 }