Re: [PATCH] Improve DST handling (Bug 23102, Bug 21942, Bug 18018, Bug, 23259, CVE-2011-0536 ).

[Carlos O'Donell <carlos at redhat dot com>]

On 06/07/2018 07:38 AM, Florian Weimer wrote:
> On 06/06/2018 10:18 PM, Carlos O'Donell wrote:
>> +is_dst (const char *input, const char *ref)
> 
> Maybe we should fix the strncmp bug and use that?
> 
> In rtld, the actual definitions are not built with the proper name,
> but aliases for use in IFUNCs, which we don't use in rtld.
> 
> The attached patch fixes that for x86-64 and i686 (multi-arch in both
> cases).  I don't think other architectures are affected (verified for
> ppc64le, build-many-glibcs.py is still running).
> 
> If you want to use strncmp to express more clearly what the code is
> about, I can commit the fix separately (with a proper ChangeLog entry
> etc.).

I would like to exclude this for now.

I am happy to discuss using strncmp *after* we get the first patch in
which we all agree implements the semantics we want.

Note, I just took your new comment bit "(excluding the $ sign but 
including curly braces, if any)" and added it to my v4 patch which
I'll post after going through your other review. Thanks for that.

Cheers,
Carlos.