Date: Fri, 30 Apr 1999 12:54:01 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: John Polstra <jdp@polstra.com> Cc: Poul-Henning Kamp <phk@critter.freebsd.dk>, freebsd-current@FreeBSD.ORG Subject: Re: Any action on PR 10570 ? getting closer to 65K :-( Message-ID: <199904301954.MAA39863@apollo.backplane.com> References: <XFMail.990430113648.jdp@polstra.com>
next in thread | previous in thread | raw e-mail | index | archive | help
:If we're going to split hairs, how about this: To make a reference
:count exceed 2^32, you need to have >2^32 different pointers pointing
:to it. A pointer takes 2^2 bytes on the i386. So that's 2^34 bytes
:of memory you'd need just to store the pointers.
:
:You'd better hope you don't get a panic on that mother! It might
:take quite awhile to write a 16 GByte crash dump. :-)
:
:> you certainly cannot create 2^32 routes without having other
:> significant problems, and while I agree with Rod that the overflow
:> should be checked, I think it should be done with a KASSERT() if not
:> just with a comment.
:
:A check would be worthwhile to detect bugs in the code (increments
:without matching decrements). If you want to check for bona-fide
:overflows, you'd best be prepared to check every counter in the
:system.
:
:John
:---
: John Polstra jdp@polstra.com
We do check some critical reference-count fields for overflow. But it
doesn't make sense to check all of them.
This particular reference count has no related bugs that we know of
apart from legally bumping past 65536, so it would be kinda silly to
check it for an overflow. If one insisted, I suppose we could add a
check if INVARIANTS is enabled but I don't see any advantage to doing
so in this case.
-Matt
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199904301954.MAA39863>