Hi Vaibhav,
Thanks for taking a look ...
On 11/05/26 6:40 pm, Vaibhav Jain wrote:
GCC 16 tightens diagnostics around const correctness and now correctly
rejects attempts to modify strings referenced through const-qualified
pointers. In kvm_ppc_register_host_cpu_type(), ppc_cpu_aliases[i].model
is defined as const char *, but the code was using strstr() on it and
then modifying the returned pointer in-place to strip
POWERPC_CPU_TYPE_SUFFIX.
This results in a write through a pointer derived from const data,
triggering a build failure with GCC 16:
error: assignment discards 'const' qualifier from pointer target type
[-Werror=discarded-qualifiers]
suffix = strstr(ppc_cpu_aliases[i].model, POWERPC_CPU_TYPE_SUFFIX);
^
Fix this by copying only the needed suffix from the object class-name to
'ppc_cpu_aliases.model' using g_strdup and g_strndup, which maintains the
constness of the allocated buffer assigned to 'ppc_cpu_aliases.model'
member of the 'ppc_cpu_aliases' struct array.
The patch also adds error handling for possible memory allocation failure
while calling g_str{n}dup functions so that the error is properly
propogated back from kvm_arch_init().
Signed-off-by: Amit Machhiwal <[email protected]>
Signed-off-by: Vaibhav Jain (IBM) <[email protected]>
---
Change-log:
v1->v2:
https://lore.kernel.org/all/[email protected]
* Avoid pointer aliasing
* Handle memory allocation failure error
---
target/ppc/kvm.c | 18 ++++++++++--------
1 file changed, 10 insertions(+), 8 deletions(-)
diff --git a/target/ppc/kvm.c b/target/ppc/kvm.c
index 25c28ad089..42cd4f4fff 100644
--- a/target/ppc/kvm.c
+++ b/target/ppc/kvm.c
@@ -170,9 +170,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
cap_rpt_invalidate = kvm_vm_check_extension(s, KVM_CAP_PPC_RPT_INVALIDATE);
cap_ail_mode_3 = kvm_vm_check_extension(s, KVM_CAP_PPC_AIL_MODE_3);
- kvm_ppc_register_host_cpu_type();
- return 0;
+ return kvm_ppc_register_host_cpu_type();
}
int kvm_arch_irqchip_create(KVMState *s)
@@ -2654,14 +2653,17 @@ static int kvm_ppc_register_host_cpu_type(void)
dc = DEVICE_CLASS(ppc_cpu_get_family_class(pvr_pcc));
for (i = 0; ppc_cpu_aliases[i].alias != NULL; i++) {
if (g_ascii_strcasecmp(ppc_cpu_aliases[i].alias, dc->desc) == 0) {
- char *suffix;
+ const gchar *suffix, *cname = object_class_get_name(oc);
+
+ suffix = g_strstr_len(cname, -1, POWERPC_CPU_TYPE_SUFFIX);
+ ppc_cpu_aliases[i].model = unlikely(suffix) ?
+ g_strndup(cname, suffix - cname) : g_strdup(cname);
- ppc_cpu_aliases[i].model = g_strdup(object_class_get_name(oc));
- suffix = strstr(ppc_cpu_aliases[i].model, POWERPC_CPU_TYPE_SUFFIX);
- if (suffix) {
- *suffix = 0;
+ if (!ppc_cpu_aliases[i].model) {
+ return -ENOMEM;
IIUC, g_strndup will abort the program in case of OOM eventually making
it a dead code. We may consider using g_try_strndup at all if needed.
Also, not sure if we really need unlikely() and a typecast may be needed
for (suffix - cname).
However, can we keep OOM related changes in a separate patch and keep
only GCC16 build breakage fix in this patch retaining the original patch
authorship with delta changes attributed alongside respective SoB?
Thanks
Harsh
+ } else {
+ return 0;
}
- break;
}
}
