tst_clocks: Fix unaddressable byte warning

pevik · pevik · commit 63e8c1ebaafe · 2023-07-28T20:37:09.000+02:00
Valgrind complains about unaddressable byte: tst_test.c:1559: TINFO: Timeout per run is 0h 00m 30s ==28379== Syscall param clock_gettime(tp) points to unaddressable byte(s) ==28379== at 0x497D41D: syscall (in /usr/lib64/libc.so.6) ==28379== by 0x412CCF: syscall_supported_by_kernel (tst_clocks.c:21) ==28379== by 0x412CCF: tst_clock_gettime (tst_clocks.c:71) ==28379== by 0x405A72: heartbeat (tst_test.c:1359) ==28379== by 0x4063CB: testrun (tst_test.c:1443) ==28379== by 0x4063CB: fork_testrun (tst_test.c:1593) ==28379== by 0x40840B: tst_run_tcases (tst_test.c:1687) ==28379== by 0x404CDD: main (tst_test.h:398) ==28379== Address 0x0 is not stack'd, malloc'd or (recently) free'd While this is not a real problem, because it is in the library, this warning appears in every test, which can be confusing. Fixes: 4ac3a9c ("syscalls: Don't pass struct timespec to tst_syscall()") Acked-by: Cyril Hrubis <chrubis@suse.cz> Signed-off-by: Petr Vorel <pvorel@suse.cz>
diff --git a/lib/tst_clocks.c b/lib/tst_clocks.c
@@ -17,8 +17,9 @@ typedef int (*mysyscall)(clockid_t clk_id, void *ts);
 int syscall_supported_by_kernel(long sysnr)
 {
 	int ret;
+	struct timespec foo;
 
-	ret = syscall(sysnr, 0, NULL);
+	ret = syscall(sysnr, 0, &foo);
 	if (ret == -1 && errno == ENOSYS)
 		return 0;
 

Original file line number	Diff line number	Diff line change
`@@ -17,8 +17,9 @@ typedef int (mysyscall)(clockid_t clk_id, void ts);`
`17`	`17`	`int syscall_supported_by_kernel(long sysnr)`
`18`	`18`	`{`
`19`	`19`	`int ret;`
	`20`	`+ struct timespec foo;`
`20`	`21`
`21`		`- ret = syscall(sysnr, 0, NULL);`
	`22`	`+ ret = syscall(sysnr, 0, &foo);`
`22`	`23`	`if (ret == -1 && errno == ENOSYS)`
`23`	`24`	`return 0;`
`24`	`25`