[PATCH] Allocation buffers for NSS result construction

[DJ Delorie]

Florian Weimer <fweimer@redhat.com> writes:
> Oh.  Well.  I don't think I want to paper over *that*.  A crash in 
> strlen seems to be just fine, rather than ignoring the issue and perhaps 
> returning grossly misleading data.

I think you misunderstand - I *intentionally* put bad data in the test
data to make sure the code handles it without crashing, and passes it
along accurately, and that the testsuite can detect and validate it.

A crash in strlen() means I can't use your code, or have to wrap it in
my own checks because I can't trust it.