Currently exec-target.c file is linked as static and this post a requirement to install libc dev package to build. Without it, build-breaks when compiling selftest/powerpc/benchmark.
CC exec_target /usr/bin/ld: cannot find -lc: No such file or directory collect2: error: ld returned 1 exit status
exec_target.c is using "syscall" library function which could be replaced with a inline assembly and the same is proposed as a fix here.
Suggested-by: Michael Ellerman mpe@ellerman.id.au Signed-off-by: Madhavan Srinivasan maddy@linux.ibm.com --- tools/testing/selftests/powerpc/benchmarks/Makefile | 2 +- .../testing/selftests/powerpc/benchmarks/exec_target.c | 10 ++++++++-- 2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/tools/testing/selftests/powerpc/benchmarks/Makefile b/tools/testing/selftests/powerpc/benchmarks/Makefile index 1321922038d0..ca4483c238b9 100644 --- a/tools/testing/selftests/powerpc/benchmarks/Makefile +++ b/tools/testing/selftests/powerpc/benchmarks/Makefile @@ -18,4 +18,4 @@ $(OUTPUT)/context_switch: LDLIBS += -lpthread
$(OUTPUT)/fork: LDLIBS += -lpthread
-$(OUTPUT)/exec_target: CFLAGS += -static -nostartfiles +$(OUTPUT)/exec_target: CFLAGS += -nostartfiles diff --git a/tools/testing/selftests/powerpc/benchmarks/exec_target.c b/tools/testing/selftests/powerpc/benchmarks/exec_target.c index c14b0fc1edde..20027a23b594 100644 --- a/tools/testing/selftests/powerpc/benchmarks/exec_target.c +++ b/tools/testing/selftests/powerpc/benchmarks/exec_target.c @@ -7,10 +7,16 @@ */
#define _GNU_SOURCE -#include <unistd.h> #include <sys/syscall.h>
void _start(void) { - syscall(SYS_exit, 0); + asm volatile ( + "li %%r0, %[sys_exit];" + "li %%r3, 0;" + "sc;" + : + : [sys_exit] "i" (SYS_exit) + : "r0", "r3" + ); }
Le 05/08/2024 à 10:30, Madhavan Srinivasan a écrit :
That looks ok because SYS_exit() is not supposed to return, but in the general case you should take a lot more precautions regarding which registers get clobbered when using sc.
Maybe it is worth a comment.
Christophe
}
Le 09/08/2024 à 06:25, Madhavan Srinivasan a écrit :
Not really.
sc will clobber r0 and r3-r12, also SO bit in CR.
Here the reason why you have no problem with that is that SYS_exit never returns. At the end, even your "r0" and "r3" clobber are unnecessary because of that.
Christophe
Christophe Leroy christophe.leroy@csgroup.eu writes:
That's my fault for just blurting out that diff on slack without thinking about it too hard.
We should probably just add the proper clobbers, in case anyone copies it in future. Which should be:
asm volatile ( "li %%r0, %[sys_exit];" "li %%r3, 0;" "sc;" : : [sys_exit] "i" (SYS_exit) : "r0", "r3", "r4", "r5", "r6", "r7", "r8", "r9", "r10", "r11", "r12", "r13", "cr0", "ctr", "xer", "memory" );
cheers
linux-kselftest-mirror@lists.linaro.org
-
Christophe Leroy -
Madhavan Srinivasan -
Michael Ellerman