From: Phillip Wood <phillip.wood123@gmail.com>
Date: Tue, 16 Jun 2020 15:49:33 GMT
Subject: Re: [PATCH v2 05/12] fast-export: handle overridden main branch names correctly
Message-ID: <8c65ab94-3b38-f8fc-c969-2d5eeff91ef1@gmail.com>
In-Reply-To: <20200616131039.GG666057@coredump.intra.peff.net>

Hi Peff

On 16/06/2020 14:10, Jeff King wrote:
> On Mon, Jun 15, 2020 at 04:05:52PM +0100, Phillip Wood wrote:
> 
>>> @@ -515,14 +515,23 @@ static const char *anonymize_refname(const char *refname)
>>>    	};
>>>    	static struct hashmap refs;
>>>    	static struct strbuf anon = STRBUF_INIT;
>>> +	static char *main_branch;
>> [...]
>>> -	if (!strcmp(refname, "refs/heads/master"))
>>> +	if (!main_branch)
>>> +		main_branch = git_main_branch_name(MAIN_BRANCH_FULL_NAME);
>>> +
>>> +	if (!strcmp(refname, main_branch))
>>>    		return "refs/heads/ref0";
>>
>> This leaks main_branch if it came from git_main_branch_name()
> 
> It's a static that's used over and over, so I think it's intentional to
> essentially memoize it for the life of the program 

Oh you're right, I completely misread the patch

Thanks

Phillip

> (at which point we
> could free it, but don't bother to do so, letting the process exit take
> care of it, and trusting in leak detectors to be aware that it's still
> reachable, as we do for lots of other process-lifetime allocations).
> 
> -Peff
>