On 28/10/26 21:41, Pierrick Bouvier wrote:
On 10/28/2026 12:39 PM, Pierrick Bouvier wrote:
On 10/28/2026 11:25 AM, Philippe Mathieu-Daudé wrote:
The Makefile needs a toolchain we do not have and qemu-mipsel,
which no longer provides the R5900 CPU. Move the tests next to
the bare-metal harness, register them in meson.build, and remove
the Makefile and the obsolete README.
Signed-off-by: Philippe Mathieu-Daudé <[email protected]>
---
.../system}/test_r5900_div1.c | 0
.../system}/test_r5900_divu1.c | 0
.../system}/test_r5900_madd.c | 0
.../system}/test_r5900_maddu.c | 0
.../system}/test_r5900_mflohi1.c | 0
.../system}/test_r5900_mtlohi1.c | 0
.../system}/test_r5900_mult.c | 0
.../system}/test_r5900_multu.c | 0
tests/tcg/mips/user/isa/r5900/Makefile | 32 -------------------
tests/tcg/mips/user/isa/r5900/README | 6 ----
tests/tcg/mips64el/system/meson.build | 20 ++++++++++++
11 files changed, 20 insertions(+), 38 deletions(-)
rename tests/tcg/{mips/user/isa/r5900 => mips64el/system}/test_r5900_div1.c
(100%)
rename tests/tcg/{mips/user/isa/r5900 => mips64el/system}/test_r5900_divu1.c
(100%)
rename tests/tcg/{mips/user/isa/r5900 => mips64el/system}/test_r5900_madd.c
(100%)
rename tests/tcg/{mips/user/isa/r5900 => mips64el/system}/test_r5900_maddu.c
(100%)
rename tests/tcg/{mips/user/isa/r5900 =>
mips64el/system}/test_r5900_mflohi1.c (100%)
rename tests/tcg/{mips/user/isa/r5900 =>
mips64el/system}/test_r5900_mtlohi1.c (100%)
rename tests/tcg/{mips/user/isa/r5900 => mips64el/system}/test_r5900_mult.c
(100%)
rename tests/tcg/{mips/user/isa/r5900 => mips64el/system}/test_r5900_multu.c
(100%)
delete mode 100644 tests/tcg/mips/user/isa/r5900/Makefile
delete mode 100644 tests/tcg/mips/user/isa/r5900/README
+# o32: 32-bit pointers, 32-bit GPRs are enough
+foreach t: [
+ 'div1',
+ 'divu1',
+ 'madd',
+ 'maddu',
+ 'mflohi1',
+ 'mtlohi1',
+ 'mult',
+ 'multu',
+]
+ tests += {
+ 'test_r5900_' + t + '.c': {
+ 'cflags': cflags + ['-mabi=32'],
+ 'qemu_args': qemu_args,
+ 'gdb_test': gdb_test,
+ }
+ }
Just a small nit, please keep full file name in the original array
(s/div1/div1.c), All other meson.build for tcg/tests do this also.
(s/div1/test_5900_div1.c)
Squashed:
-- >8 --
diff --git a/tests/tcg/mips64el/system/meson.build
b/tests/tcg/mips64el/system/meson.build
index fd5a759c20fa..c16a21f57aef 100644
--- a/tests/tcg/mips64el/system/meson.build
+++ b/tests/tcg/mips64el/system/meson.build
@@ -21,17 +21,17 @@ gdb_test = ['--gdb-args=-ex "set remoteaddresssize 64"',
# o32: 32-bit pointers, 32-bit GPRs are enough
foreach t: [
- 'div1',
- 'divu1',
- 'madd',
- 'maddu',
- 'mflohi1',
- 'mtlohi1',
- 'mult',
- 'multu',
+ 'test_r5900_div1.c',
+ 'test_r5900_divu1.c',
+ 'test_r5900_madd.c',
+ 'test_r5900_maddu.c',
+ 'test_r5900_mflohi1.c',
+ 'test_r5900_mtlohi1.c',
+ 'test_r5900_mult.c',
+ 'test_r5900_multu.c',
]
tests += {
- 'test_r5900_' + t + '.c': {
+ t: {
'cflags': cflags + ['-mabi=32'],
'qemu_args': qemu_args,
'gdb_test': gdb_test,
---
Reviewed-by: Pierrick Bouvier <[email protected]>
Thanks!
