On 18/03/15 09:09, Poul-Henning Kamp wrote: > I'm not confident enough in the stack_start/stack_end values > to expose it like this, and I'd rather we remove them again if > at all possible. > > This stuff originated in the PCRE recursion problem, and I thought > the consensus was that newer PCRE JIT worked well enough that we > wouldn't need this stack-hack anyway ?
Regarding PCRE, I've tried to attack the design questions in the email "solve the pcre stack overflow #1576 - was: stack vs workspace for pcre and others" as of 2015-01-19. There was no feedback on the mailing list so far and I'd still appreciate any. Regarding the stack, I think the punchline is that as long as we use pcre on the stack (see "solve the pcre..." email for details), checking the stack limits and handle errors before running into an overflow is the best we can do, so at this point I think we should really keep the stack start/end value and rather make more use of them then less. Nils _______________________________________________ varnish-dev mailing list [email protected] https://www.varnish-cache.org/lists/mailman/listinfo/varnish-dev
