Module Name: src
Committed By: martin
Date: Tue Jun 27 18:51:47 UTC 2023
Modified Files:
src/sys/kern [netbsd-10]: kern_timeout.c
Log Message:
Pull up following revision(s) (requested by pho in ticket #219):
sys/kern/kern_timeout.c: revision 1.74
sys/kern/kern_timeout.c: revision 1.75
sys/kern/kern_timeout.c: revision 1.76
callout(9): Fix panic() in callout_destroy() (kern/57226)
The culprit was callout_halt(). "(c->c_flags & CALLOUT_FIRED) != 0" wasn't
the correct way to check if a callout is running. It failed to wait for a
running callout to finish in the following scenario:
1. cpu0 initializes a callout and schedules it.
2. cpu0 invokes callout_softlock() and fires the callout, setting the flag
CALLOUT_FIRED.
3. The callout invokes callout_schedule() to re-schedule itself.
4. callout_schedule_locked() clears the flag CALLOUT_FIRED, and releases
the lock.
5. Before the lock is re-acquired by callout_softlock(), cpu1 decides to
destroy the callout. It first invokes callout_halt() to make sure the
callout finishes running.
6. But since CALLOUT_FIRED has been cleared, callout_halt() thinks it's not
running and therefore returns without invoking callout_wait().
7. cpu1 proceeds to invoke callout_destroy() while it's still running on
cpu0. callout_destroy() detects that and panics.
callout(9): Tidy up the condition for "callout is running on another LWP"
No functional changes.
callout(9): Delete the unused member cc_cancel from struct callout_cpu
I see no reason why it should be there, and believe its a leftover from
some old code.
To generate a diff of this commit:
cvs rdiff -u -r1.73 -r1.73.2.1 src/sys/kern/kern_timeout.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/kern/kern_timeout.c
diff -u src/sys/kern/kern_timeout.c:1.73 src/sys/kern/kern_timeout.c:1.73.2.1
--- src/sys/kern/kern_timeout.c:1.73 Sat Oct 29 00:19:21 2022
+++ src/sys/kern/kern_timeout.c Tue Jun 27 18:51:47 2023
@@ -1,4 +1,4 @@
-/* $NetBSD: kern_timeout.c,v 1.73 2022/10/29 00:19:21 riastradh Exp $ */
+/* $NetBSD: kern_timeout.c,v 1.73.2.1 2023/06/27 18:51:47 martin Exp $ */
/*-
* Copyright (c) 2003, 2006, 2007, 2008, 2009, 2019 The NetBSD Foundation, Inc.
@@ -59,7 +59,7 @@
*/
#include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: kern_timeout.c,v 1.73 2022/10/29 00:19:21 riastradh Exp $");
+__KERNEL_RCSID(0, "$NetBSD: kern_timeout.c,v 1.73.2.1 2023/06/27 18:51:47 martin Exp $");
/*
* Timeouts are kept in a hierarchical timing wheel. The c_time is the
@@ -174,7 +174,6 @@ struct callout_cpu {
u_int cc_ticks;
lwp_t *cc_lwp;
callout_impl_t *cc_active;
- callout_impl_t *cc_cancel;
struct evcnt cc_ev_late;
struct evcnt cc_ev_block;
struct callout_circq cc_todo; /* Worklist */
@@ -263,6 +262,17 @@ callout_lock(callout_impl_t *c)
}
/*
+ * Check if the callout is currently running on an LWP that isn't curlwp.
+ */
+static inline bool
+callout_running_somewhere_else(callout_impl_t *c, struct callout_cpu *cc)
+{
+ KASSERT(c->c_cpu == cc);
+
+ return cc->cc_active == c && cc->cc_lwp != curlwp;
+}
+
+/*
* callout_startup:
*
* Initialize the callout facility, called at system startup time.
@@ -378,7 +388,7 @@ callout_destroy(callout_t *cs)
KASSERTMSG((c->c_flags & CALLOUT_PENDING) == 0,
"pending callout %p: c_func (%p) c_flags (%#x) destroyed from %p",
c, c->c_func, c->c_flags, __builtin_return_address(0));
- KASSERTMSG(c->c_cpu->cc_lwp == curlwp || c->c_cpu->cc_active != c,
+ KASSERTMSG(!callout_running_somewhere_else(c, c->c_cpu),
"running callout %p: c_func (%p) c_flags (%#x) destroyed from %p",
c, c->c_func, c->c_flags, __builtin_return_address(0));
c->c_magic = 0;
@@ -496,7 +506,6 @@ bool
callout_stop(callout_t *cs)
{
callout_impl_t *c = (callout_impl_t *)cs;
- struct callout_cpu *cc;
kmutex_t *lock;
bool expired;
@@ -509,16 +518,6 @@ callout_stop(callout_t *cs)
expired = ((c->c_flags & CALLOUT_FIRED) != 0);
c->c_flags &= ~(CALLOUT_PENDING|CALLOUT_FIRED);
- cc = c->c_cpu;
- if (cc->cc_active == c) {
- /*
- * This is for non-MPSAFE callouts only. To synchronize
- * effectively we must be called with kernel_lock held.
- * It's also taken in callout_softclock.
- */
- cc->cc_cancel = c;
- }
-
SDT_PROBE5(sdt, kernel, callout, stop,
c, c->c_func, c->c_arg, c->c_flags, expired);
@@ -542,7 +541,6 @@ callout_halt(callout_t *cs, void *interl
{
callout_impl_t *c = (callout_impl_t *)cs;
kmutex_t *lock;
- int flags;
KASSERT(c->c_magic == CALLOUT_MAGIC);
KASSERT(!cpu_intr_p());
@@ -552,11 +550,10 @@ callout_halt(callout_t *cs, void *interl
lock = callout_lock(c);
SDT_PROBE4(sdt, kernel, callout, halt,
c, c->c_func, c->c_arg, c->c_flags);
- flags = c->c_flags;
- if ((flags & CALLOUT_PENDING) != 0)
+ if ((c->c_flags & CALLOUT_PENDING) != 0)
CIRCQ_REMOVE(&c->c_list);
- c->c_flags = flags & ~(CALLOUT_PENDING|CALLOUT_FIRED);
- if (__predict_false(flags & CALLOUT_FIRED)) {
+ c->c_flags &= ~(CALLOUT_PENDING|CALLOUT_FIRED);
+ if (__predict_false(callout_running_somewhere_else(c, c->c_cpu))) {
callout_wait(c, interlock, lock);
return true;
}
@@ -592,7 +589,7 @@ callout_wait(callout_impl_t *c, void *in
* - the callout itself has called callout_halt() (nice!)
*/
cc = c->c_cpu;
- if (__predict_true(cc->cc_active != c || cc->cc_lwp == l))
+ if (__predict_true(!callout_running_somewhere_else(c, cc)))
break;
/* It's running - need to wait for it to complete. */
