GCC 15 reports the below false positive '-Wmaybe-uninitialized' warning
in vphn_unpack_associativity() when building the powerpc selftests.
# make -C tools/testing/selftests TARGETS="powerpc"
[...]
CC test-vphn
In file included from test-vphn.c:3:
In function ‘vphn_unpack_associativity’,
inlined from ‘test_one’ at test-vphn.c:371:2,
inlined from ‘test_vphn’ at test-vphn.c:399:9:
test-vphn.c:10:33: error: ‘be_packed’ may be used uninitialized
[-Werror=maybe-uninitialized]
10 | #define be16_to_cpup(x) bswap_16(*x)
| ^~~~~~~~
vphn.c:42:27: note: in expansion of macro ‘be16_to_cpup’
42 | u16 new = be16_to_cpup(field++);
| ^~~~~~~~~~~~
In file included from test-vphn.c:19:
vphn.c: In function ‘test_vphn’:
vphn.c:27:16: note: ‘be_packed’ declared here
27 | __be64 be_packed[VPHN_REGISTER_COUNT];
| ^~~~~~~~~
cc1: all warnings being treated as errors
When vphn_unpack_associativity() is called from hcall_vphn(), this error
is not seen during compilation because GCC 15 seems to consider 'retbuf'
always populated from the hypervisor which is eventually referred by
'be_packed'. However, GCC 15's dataflow analysis can’t prove the same
before the first dereference when vphn_unpack_associativity() is called
from test_one() with pre-initialized array of 'struct test'. This
results in a false positive warning which is promoted to an error under
'-Werror'. This problem is not seen when the compilation is performed
with GCC 13 and 14.
Suppress the warning locally around the offending statement when
building with GCC 15 using a diagnostic pragma. This keeps the build
working while limiting the scope of the suppression to the specific
statement that triggers the false positive. An issue [1] has also been
created on GCC bugzilla.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124427
Fixes: 58dae82843f5 ("selftests/powerpc: Add test for VPHN")
Reviewed-by: Vaibhav Jain <[email protected]>
Signed-off-by: Amit Machhiwal <[email protected]>
---
arch/powerpc/platforms/pseries/vphn.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/arch/powerpc/platforms/pseries/vphn.c
b/arch/powerpc/platforms/pseries/vphn.c
index 3f85ece3c872..9bc891143fec 100644
--- a/arch/powerpc/platforms/pseries/vphn.c
+++ b/arch/powerpc/platforms/pseries/vphn.c
@@ -39,7 +39,22 @@ static int vphn_unpack_associativity(const long *packed,
__be32 *unpacked)
be_packed[i] = cpu_to_be64(packed[i]);
for (i = 1; i < VPHN_ASSOC_BUFSIZE; i++) {
+/*
+ * When this function is called from hcall_vphn(), GCC 15 seems to consider
+ * 'retbuf' always populated from the hypervisor which is eventually referred
by
+ * 'be_packed'. However, GCC 15's dataflow analysis can’t prove the same before
+ * the first dereference when this function is called from test_one() with
+ * pre-initialized array of 'struct test'. This results in a false positive
+ * '-Wmaybe-uninitialized' warning which is promoted to an error under
+ * '-Werror'. This problem is not seen when the compilation is performed with
+ * older GCC versions.
+ */
+#pragma GCC diagnostic push
+#if defined(__GNUC__) && __GNUC__ >= 15
+#pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
+#endif
u16 new = be16_to_cpup(field++);
+#pragma GCC diagnostic pop
if (is_32bit) {
/*
base-commit: 1f318b96cc84d7c2ab792fcc0bfd42a7ca890681
--
2.50.1 (Apple Git-155)
