On Thu, May 14, 2020 at 01:40:23PM -0400, Wietse Venema wrote:
> > A cursory glance at the PCRE2 docs suggests that we can ask libpcre
> > to enforce more conservative limits before it runs out of stack, and
> > it would presumably then unwind and return a recoverable error.
>
> That looks like a guessing game to me, because I doubt that libpcre
> really knows how much stack space remains available, as that depends
> on stack canaries and so on.
>
> Who know what happens when it runs out. Then, it will access an
> invalid page and receive SIGSEGV.
No worse than before, but if we make a conservative guess we can
avoid the segfaults.
Alternatively, we could siglongjmp() out of a segfault handler, enabled
around PCRE lookups, leaking whatever heap space libpcre may have
allocated along the way, and log a more informative message, and thereby
perhaps even avoid occasional service throttling in master that may happen
if the service is killed by a signal. Just a thought...
--
Viktor.
