Re: [PATCH][BZ #13724] Do not segfault in pthread_setname_np (x, NULL)

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

On 10/09/2013 01:57 PM, Jeff Law wrote:
> On 10/09/13 11:53, Carlos O'Donell wrote:
>> On 10/08/2013 03:04 PM, Jeff Law wrote:
>>>> It seems incredibly useful to enable the asserts and trigger
>>>> these violations as early as possible. If you don't care you
>>>> can disable the asserts?
>>> Another approach would be similar to what we're doing with
>>> memstomp. ie, build a set of wrappers which check for these
>>> argument goofs and allow users to dl-preload DSOs with the
>>> wrappers.
>>> 
>>> When I first proposed the idea for these sanity checking
>>> dl-preload libraries for Fedora I envisioned that we could go
>>> beyond just checking for overlapping memory areas in the mem* and
>>> str* functions. There could be a set of pthread wrapper functions
>>> that check for whatever invariants we can in the pthread*
>>> functions without a huge performance hit.
>> 
>> Right, you could wrap everything up and check all inputs against 
>> non-NULL where the headers say __attribute__ ((__nonnull__
>> params)) or __nonnull(params) (short-hand in glibc).
> But using the attributes are compile-time checks only.  Using an
> LD_PRELOAD provides runtime checking

I agree, what I want to say was that to automatically generate the
wrappers and check against null parameters you can parse the public
headers and look for glibc's __nonnull markups?

Cheers,
Carlos.