Subject: Re: kern/35278: veriexec sometimes feeds user va to log(9)
To: None <kern-bug-people@netbsd.org, gnats-admin@netbsd.org,>
From: Elad Efrat <elad@NetBSD.org>
List: netbsd-bugs
Date: 12/23/2006 17:25:01
The following reply was made to PR kern/35278; it has been noted by GNATS.
From: Elad Efrat <elad@NetBSD.org>
To: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
Cc: gnats-bugs@NetBSD.org
Subject: Re: kern/35278: veriexec sometimes feeds user va to log(9)
Date: Sat, 23 Dec 2006 19:22:43 +0200
YAMAMOTO Takashi wrote:
>> ugh! any plans in sight to fix that?
>
> that == VOP interfaces?
> i don't think there's a quick way.
specifically, fix nfs, so it works okay with SAVENAME from namei()
VOP_FOO...
>> or should we fall back to any
>> of the previous ideas?
>
> maybe, unless you want to add more ugliness to namei. :)
>
>> if the latter, unless you object, I'd like just do the PNBUF_GET() +
>> copyinstr() + PNBUF_PUT() dance where we need them for netbsd-4, and
>> on the trunk we can look into changing NDINIT()/namei().
>>
>
> do you mean to leave the trunk broken for now?
I'll commit the kern_exec.c changes because it's okay there like you
say.
the possibilities for sys_unlink/rename_files are:
- change NDINIT()/namei() to work with a pathname_t interface similar
to what you suggested. if we do that, we probably don't want to
pull it up to netbsd-4, and we'll have to pull up a PNBUF_GET() +
copyinstr() + PNBUF_PUT() fix.
- use a pathname_t interface similar to the one I suggested. this
can be pulled up to netbsd-4.
personally, if we can't use SAVENAME (that was quite elegant imho :)
then I don't mind what we choose to do.
-e.