Date: Wed, 2 Jan 2002 04:00:45 +1100 (EST) From: Bruce Evans <bde@zeta.org.au> To: John Baldwin <jhb@FreeBSD.org> Cc: Poul-Henning Kamp <phk@FreeBSD.org>, <cvs-all@FreeBSD.org>, <cvs-committers@FreeBSD.org> Subject: RE: cvs commit: src/sys/i386/i386 trap.c Message-ID: <20020102035156.M10589-100000@gamplex.bde.org> In-Reply-To: <XFMail.011231123911.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 31 Dec 2001, John Baldwin wrote:
> On 30-Dec-01 Poul-Henning Kamp wrote:
> > phk 2001/12/30 11:43:59 PST
> >
> > Modified files:
> > sys/i386/i386 trap.c
> > Log:
> > GC an alternate trap_pfault() which has rotted away behind an "#ifdef
> > notyet"
> > since 21-Mar-95 .
>
> I thought we wanted to switch to using it as it simplifies some of the kernel
> fault handling but just had a few unsafe places in the kernel to fix?
I added the check for disallowing pagefaults for user addresses in kernel
mode except ones for copying functions. It is just the
(!usermode && PCPU_GET(curpcb)->pcb_onfault == NULL)
part of the patch (the other changes are mostly to optimize some copying
functions). I haven't merged any of the simplifications from the old
version.
%%%
Index: trap.c
===================================================================
RCS file: /home/ncvs/src/sys/i386/i386/trap.c,v
retrieving revision 1.210
diff -u -2 -r1.210 trap.c
--- trap.c 30 Dec 2001 19:43:59 -0000 1.210
+++ trap.c 1 Jan 2002 16:39:50 -0000
@@ -688,10 +719,21 @@
/*
* This is a fault on non-kernel virtual memory.
- * vm is initialized above to NULL. If curproc is NULL
- * or curproc->p_vmspace is NULL the fault is fatal.
+ * Do not allow it in kernel mode unless it is for a
+ * a recognized copying function.
*/
- if (p != NULL)
- vm = p->p_vmspace;
+ if (!usermode &&
+ frame->tf_eip != (int)fubyte_access &&
+ frame->tf_eip != (int)fusword_access &&
+ frame->tf_eip != (int)fuword_access &&
+ frame->tf_eip != (int)subyte_access &&
+ frame->tf_eip != (int)susword_access &&
+ frame->tf_eip != (int)suword_access &&
+ PCPU_GET(curpcb)->pcb_onfault == NULL)
+ goto nogo;
+ /*
+ * If curproc->p_vmspace is NULL the fault is fatal.
+ */
+ vm = p->p_vmspace;
if (vm == NULL)
goto nogo;
%%%
This is completely untested (except for removing the p != NULL check).
Bruce
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020102035156.M10589-100000>