diff mbox series

migration: fix stringop-truncation warning

Message ID 1576074210-52834-3-git-send-email-pbonzini@redhat.com
State New
Headers show
Series migration: fix stringop-truncation warning | expand

Commit Message

Paolo Bonzini Dec. 11, 2019, 2:23 p.m. UTC 
From: Marc-André Lureau <marcandre.lureau@redhat.com>

../migration/global_state.c: In function ‘global_state_store_running’:
../migration/global_state.c:47:5: error: ‘strncpy’ specified bound 100 equals destination size [-Werror=stringop-truncation]
   47 |     strncpy((char *)global_state.runstate,
      |     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   48 |            state, sizeof(global_state.runstate));
      |            ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The assert() above allows to call with strlen(src)+1.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 migration/global_state.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

Comments

Richard Henderson Dec. 12, 2019, 12:55 a.m. UTC | #1 
On 12/11/19 6:23 AM, Paolo Bonzini wrote:
> @@ -44,8 +44,7 @@ void global_state_store_running(void)
>  {
>      const char *state = RunState_str(RUN_STATE_RUNNING);
>      assert(strlen(state) < sizeof(global_state.runstate));
> -    strncpy((char *)global_state.runstate,
> -           state, sizeof(global_state.runstate));
> +    memcpy(global_state.runstate, state, strlen(state) + 1);

We should assign the strlen result to a local variable rather than compute it
twice.


r~
Paolo Bonzini Dec. 12, 2019, 1:06 a.m. UTC | #2 
On 12/12/19 01:55, Richard Henderson wrote:
> On 12/11/19 6:23 AM, Paolo Bonzini wrote:
>> @@ -44,8 +44,7 @@ void global_state_store_running(void)
>>  {
>>      const char *state = RunState_str(RUN_STATE_RUNNING);
>>      assert(strlen(state) < sizeof(global_state.runstate));
>> -    strncpy((char *)global_state.runstate,
>> -           state, sizeof(global_state.runstate));
>> +    memcpy(global_state.runstate, state, strlen(state) + 1);
> 
> We should assign the strlen result to a local variable rather than compute it
> twice.

We could even strcpy since the assertion ensures it's valid, but perhaps
it's better to do nothing, since the best alternative would be
memset+strcpy, i.e. back to strncpy.

We generally are quite mindful about our uses of strncpy, but maybe we
could fold the assertion and strncpy into a qemu_strncpy function,
and/or qemu_strncpy_nonul for when you're copying into an array that
does _not_ need to be nul-terminated (so the assertion can become <=
rather than <).  I'll add it to BiteSizedTasks.

Paolo
diff mbox series

Patch

diff --git a/migration/global_state.c b/migration/global_state.c
index 2531147..a39a876 100644
--- a/migration/global_state.c
+++ b/migration/global_state.c
@@ -44,8 +44,7 @@  void global_state_store_running(void)
 {
     const char *state = RunState_str(RUN_STATE_RUNNING);
     assert(strlen(state) < sizeof(global_state.runstate));
-    strncpy((char *)global_state.runstate,
-           state, sizeof(global_state.runstate));
+    memcpy(global_state.runstate, state, strlen(state) + 1);
 }
 
 bool global_state_received(void)