[PATCH] handshake: don't print NULL pmksa pointer

[PATCH] handshake: don't print NULL pmksa pointer

[James Prestwood]

This is undefined behavior so if no pmksa is found don't print.
---
 src/handshake.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/handshake.c b/src/handshake.c
index f73f91d1..bee76b26 100644
--- a/src/handshake.c
+++ b/src/handshake.c
@@ -1272,11 +1272,11 @@ void handshake_state_cache_pmksa(struct handshake_state *s)
 {
 	struct pmksa *pmksa = handshake_state_steal_pmksa(s);
 
-	l_debug("%p", pmksa);
-
 	if (!pmksa)
 		return;
 
+	l_debug("%p", pmksa);
+
 	if (L_WARN_ON(pmksa_cache_put(pmksa) < 0))
 		l_free(pmksa);
 }
-- 
2.34.1

Re: [PATCH] handshake: don't print NULL pmksa pointer

[Denis Kenzior]

Hi James,

On 1/6/25 7:46 AM, James Prestwood wrote:
> This is undefined behavior so if no pmksa is found don't print.
> ---
>   src/handshake.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/src/handshake.c b/src/handshake.c
> index f73f91d1..bee76b26 100644
> --- a/src/handshake.c
> +++ b/src/handshake.c
> @@ -1272,11 +1272,11 @@ void handshake_state_cache_pmksa(struct handshake_state *s)
>   {
>   	struct pmksa *pmksa = handshake_state_steal_pmksa(s);
>   
> -	l_debug("%p", pmksa);
> -

Why is this behavior undefined? Would it not just print 0x0000...?

>   	if (!pmksa)
>   		return;
>   
> +	l_debug("%p", pmksa);
> +
>   	if (L_WARN_ON(pmksa_cache_put(pmksa) < 0))
>   		l_free(pmksa);
>   }

Regards,
-Denis

Re: [PATCH] handshake: don't print NULL pmksa pointer

[James Prestwood]

Hi Denis,

On 1/6/25 7:14 AM, Denis Kenzior wrote:
> Hi James,
>
> On 1/6/25 7:46 AM, James Prestwood wrote:
>> This is undefined behavior so if no pmksa is found don't print.
>> ---
>>   src/handshake.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/src/handshake.c b/src/handshake.c
>> index f73f91d1..bee76b26 100644
>> --- a/src/handshake.c
>> +++ b/src/handshake.c
>> @@ -1272,11 +1272,11 @@ void handshake_state_cache_pmksa(struct 
>> handshake_state *s)
>>   {
>>       struct pmksa *pmksa = handshake_state_steal_pmksa(s);
>>   -    l_debug("%p", pmksa);
>> -
>
> Why is this behavior undefined? Would it not just print 0x0000...?

I had a brain fart, I was thinking for some reason it was using %s not 
%p. So yes, this is a valid input to the %p formatter.

I can alter the commit description, but I do think it still makes sense 
to move the print, or at least modify it to make it clear no PMKSA was 
cached, rather than just printing "nul" or "nil" or "0x0000".

>
>>       if (!pmksa)
>>           return;
>>   +    l_debug("%p", pmksa);
>> +
>>       if (L_WARN_ON(pmksa_cache_put(pmksa) < 0))
>>           l_free(pmksa);
>>   }
>
> Regards,
> -Denis

Re: [PATCH] handshake: don't print NULL pmksa pointer

[Denis Kenzior]

Hi James,

 >>
>> Why is this behavior undefined? Would it not just print 0x0000...?
> 
> I had a brain fart, I was thinking for some reason it was using %s not %p. So 
> yes, this is a valid input to the %p formatter.
> 
> I can alter the commit description, but I do think it still makes sense to move 
> the print, or at least modify it to make it clear no PMKSA was cached, rather 
> than just printing "nul" or "nil" or "0x0000".

Sure, that makes sense.

Regards,
-Denis