fstat triggered INVARIANTS panic in memrw()

[Poul-Henning Kamp]

In message <200501281914.j0SJETJH052081 at blue.mwk.domain>, "Mark W. Krentel" writes:
>>>>>> "PHK" == Poul-Henning Kamp <phk at phk.freebsd.dk> writes:
>
>PHK> The only real problem here is that fstat(1) uses kvm in the first place.
>PHK> The necessary information should be exported via sysctls.
>
>PHK> No program used as part of regular system operation should use kvm.
>
>Ok.  That still leaves btsockstat, netstat and trpt (and fstat) as
>setgid kmem programs.  Are there plans to rewrite them?
>
>Until then, I think it's not a bad idea to put a bandaid on fstat(1),
>even if it's temporary.
>
>Also, there's still the bug that kvm_read(3), and kmem(4) don't check
>their arguments for values that lead to address wrap.

I didn't mean to imply that these bugs should not be fixed, I just wanted
to repeat the higher goal so we didn't loose sight of it.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.