Re: [musl] [PATCH] getopt: fix null pointer arithmetic ub

[Alexey Izbyshev <izbyshev@ispras.ru>]

On 2023-03-10 19:28, Alexander Monakov wrote:
> Hi,
> 
> On Fri, 10 Mar 2023, Alexey Izbyshev wrote:
> 
>> When an option that requires an argument is the last character of
>> argv[argc-1], getopt computes argv[argc] + optpos. While optpos
>> is always zero in this case, adding it to null pointer is still
>> undefined.
>> ---
>>  src/misc/getopt.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>> 
>> diff --git a/src/misc/getopt.c b/src/misc/getopt.c
>> index c3f66995..af12973a 100644
>> --- a/src/misc/getopt.c
>> +++ b/src/misc/getopt.c
>> @@ -87,7 +87,8 @@ int getopt(int argc, char * const argv[], const char 
>> *optstring)
>>  	if (optstring[i] == ':') {
>>  		optarg = 0;
>>  		if (optstring[i+1] != ':' || optpos) {
>> -			optarg = argv[optind++] + optpos;
>> +			optarg = argv[optind++];
>> +			if (optarg) optarg += optpos;
> 
> Can this be written as 'if (optpos) optarg += optpos;' instead? That 
> will be
> folded back into plain addition by the compiler.
> 
Yes, "if (optpos) ..." is actually what I initially wrote before 
changing it to the submitted variant. I'm fine with changing it back; 
thanks for the codegen check.

> (also (unlike the quoted variant) would allow undefined behavior
> instrumentation to catch attempted NULL pointer arithmetic)
> 
Yes, a good point too.

Alexey