Module Name:    src
Committed By:   cherry
Date:           Sun Nov 25 07:48:47 UTC 2012

Modified Files:
        src/sys/arch/xen/xen: evtchn.c

Log Message:
hypervisor_set_ipending() should not set the pending flag across cpus. The 
reason for this is that the pending flag update is not atomic, or protected by 
a lock. Since its current (mis)use in evtchn_do_event() across cpus is to 
notify the remote cpu of an interrupt, we use hypervisor_send_event() instead, 
to trigger an event on the remote cpu, which in turn invokes evtchn_do_event() 
on that cpu and DTRT on it.

On the local cpu, we protect hypervisor_set_ipending() from re-entry via cli(). 
Remove a redundant and unprotected call to hypervisor_set_ipending().

Thanks to jym@ for spotting this discrepancy while reviewing a related patch.

See: http://mail-index.netbsd.org/tech-kern/2012/11/12/msg014374.html


To generate a diff of this commit:
cvs rdiff -u -r1.62 -r1.63 src/sys/arch/xen/xen/evtchn.c

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.

Modified files:

Index: src/sys/arch/xen/xen/evtchn.c
diff -u src/sys/arch/xen/xen/evtchn.c:1.62 src/sys/arch/xen/xen/evtchn.c:1.63
--- src/sys/arch/xen/xen/evtchn.c:1.62	Sun Feb 12 14:24:08 2012
+++ src/sys/arch/xen/xen/evtchn.c	Sun Nov 25 07:48:46 2012
@@ -1,4 +1,4 @@
-/*	$NetBSD: evtchn.c,v 1.62 2012/02/12 14:24:08 jym Exp $	*/
+/*	$NetBSD: evtchn.c,v 1.63 2012/11/25 07:48:46 cherry Exp $	*/
 
 /*
  * Copyright (c) 2006 Manuel Bouyer.
@@ -54,7 +54,7 @@
 
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: evtchn.c,v 1.62 2012/02/12 14:24:08 jym Exp $");
+__KERNEL_RCSID(0, "$NetBSD: evtchn.c,v 1.63 2012/11/25 07:48:46 cherry Exp $");
 
 #include "opt_xen.h"
 #include "isa.h"
@@ -298,15 +298,12 @@ evtchn_do_event(int evtch, struct intrfr
 	ih = evtsource[evtch]->ev_handlers;
 	while (ih != NULL) {
 		if (ih->ih_cpu != ci) {
-			hypervisor_set_ipending(ih->ih_cpu, 1 << ih->ih_level,
-			    evtch >> LONG_SHIFT, evtch & LONG_MASK);
+			hypervisor_send_event(ih->ih_cpu, evtch);
 			iplmask &= ~IUNMASK(ci, ih->ih_level);
 			ih = ih->ih_evt_next;
 			continue;
 		}
 		if (ih->ih_level <= ilevel) {
-			hypervisor_set_ipending(ih->ih_cpu, iplmask,
-			    evtch >> LONG_SHIFT, evtch & LONG_MASK);
 #ifdef IRQ_DEBUG
 		if (evtch == IRQ_DEBUG)
 		    printf("ih->ih_level %d <= ilevel %d\n", ih->ih_level, ilevel);

Reply via email to