On 5/1/26 12:45, Richard Henderson wrote:
On 1/5/26 21:56, Philippe Mathieu-Daudé wrote:
On 5/1/26 01:37, Richard Henderson wrote:
On 12/25/25 03:26, Philippe Mathieu-Daudé wrote:
Rather than ldtul_p() which uses the underlying 'unsigned
long' size, use the ldn() variant, passing the access size
as argument (evaluating TARGET_LONG_BITS / 8).
No need to use #ifdef'ry to check for TARGET_ABI32, since
it is 64-bit:
$ git grep -E '(ABI32|LONG_BITS)' configs/targets/sparc*
configs/targets/sparc-linux-user.mak:5:TARGET_LONG_BITS=32
configs/targets/sparc-softmmu.mak:4:TARGET_LONG_BITS=32
configs/targets/sparc32plus-linux-user.mak:2:TARGET_ABI32=y
configs/targets/sparc32plus-linux-user.mak:8:TARGET_LONG_BITS=64
configs/targets/sparc64-linux-user.mak:8:TARGET_LONG_BITS=64
configs/targets/sparc64-softmmu.mak:6:TARGET_LONG_BITS=64
Directly expand to the big-endian variant (with the '_be' suffix)
since we only build the SPARC targets as big-endian.
Signed-off-by: Philippe Mathieu-Daudé <[email protected]>
---
target/sparc/gdbstub.c | 12 ++----------
1 file changed, 2 insertions(+), 10 deletions(-)
diff --git a/target/sparc/gdbstub.c b/target/sparc/gdbstub.c
index 134617fb232..d265681f6d2 100644
--- a/target/sparc/gdbstub.c
+++ b/target/sparc/gdbstub.c
@@ -112,15 +112,7 @@ int sparc_cpu_gdb_write_register(CPUState *cs,
uint8_t *mem_buf, int n)
{
SPARCCPU *cpu = SPARC_CPU(cs);
CPUSPARCState *env = &cpu->env;
-#if defined(TARGET_ABI32)
- uint32_t tmp;
-
- tmp = ldl_p(mem_buf);
-#else
- target_ulong tmp;
-
- tmp = ldtul_p(mem_buf);
-#endif
+ uint64_t tmp = ldn_be_p(mem_buf, TARGET_LONG_BITS / 8);
No, this changes the behaviour of sparc32plus.
$ git grep TARGET_ABI32 configs/targets/sparc*
configs/targets/sparc32plus-linux-user.mak:2:TARGET_ABI32=y
$ cat configs/targets/sparc32plus-linux-user.mak
TARGET_ABI32=y
TARGET_BIG_ENDIAN=y
TARGET_LONG_BITS=64
Isn't it the same?
No. ABI32 uses uint32_t, not uint64_t, even with 64-bit registers.
Oh right, I missed that.
Which is probably functionally broken, because, history.
And still we maintain it...
