Subject: Re: CVS commit: src/sys/rump
To: Tom Spindler <dogcow@NetBSD.org>
From: Antti Kantee <pooka@netbsd.org>
List: source-changes
Date: 01/07/2008 09:17:29
On Mon Jan 07 2008 at 06:59:48 +0000, Tom Spindler wrote:
> > Modified Files:
> > 	src/sys/rump/fs: Makefile.rumpfs
> > 	src/sys/rump/fs/lib/libp2k: p2k.c
> > 	src/sys/rump/librump/rumpkern: Makefile auth.c emul.c fstrans_stub.c
> > 	    pool.c rump.c specfs.c vfs.c
> > 	src/sys/rump/librump/rumpuser: rumpuser.c rumpuser.h rumpuser_pth.c
> 
> This fails to build on sparc, because sparc has variable pagesizes, which
> are accessed only by a private-to-the-kernel variable; this is relevant
> because sys_generic.c has sys_ioctl.c which uses IOCPARM_MAX which works
> out to be the pagesize variable.

Sounds like something that could be used for a grant application at the
ministry of silly macros.

> As it looks like the only reason this file is included is for selsysinit()
> (which is two lines long) and for seltrue (one line, and only used by
> a fn used elsewhere in the kernel) - perhaps it would be better to have
> private copies of selsysinit and seltrue rather than pulling in a not-
> insubstantial kernel file? (I have no idea how the select stuff is even
> used, admittedly.)

No, its main purpose is dofileread() and dofilewrite() used to run
puffs itself.

Can you try to see if IOCPARM_MAX can be fixed?  If not, disable rump
for sparc and I'll look at in a couple of days.  I *really* don't have
time now.

thanks,
  antti

-- 
Antti Kantee <pooka@iki.fi>                     Of course he runs NetBSD
http://www.iki.fi/pooka/                          http://www.NetBSD.org/
    "la qualité la plus indispensable du cuisinier est l'exactitude"