Source-Changes-HG archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
[src/trunk]: src/sys/arch/alpha Slight tweak to previous changes:
details: https://anonhg.NetBSD.org/src/rev/a75315de5e23
branches: trunk
changeset: 961451:a75315de5e23
user: thorpej <thorpej%NetBSD.org@localhost>
date: Tue Apr 20 00:09:45 2021 +0000
description:
Slight tweak to previous changes:
Rather than simply increment the interrupt depth for the clock interrupt,
we add 0x10. Why? Because while we only use a single Alpha IPL (4) for
IPL_{BIO,NET,TTY,VM}, technically the architecture specification suports
two in the OSF/1 PALcode (3 [low-pri] and 4 [high-pri]), meaning we could
conceiveably have intrdepth > 1 just for device interrupts.
Adding 0x10 here means that cpu_intr_p() can check for "intrdepth != 0" for
"in interrupt context" and CLKF_INTR() can check "(intrdepth & 0xf) != 0" for
"was processing interrupts when the clock interrupt happened".
diffstat:
sys/arch/alpha/alpha/interrupt.c | 22 ++++++++++++++++++----
sys/arch/alpha/include/cpu.h | 4 ++--
2 files changed, 20 insertions(+), 6 deletions(-)
diffs (68 lines):
diff -r 072d021be652 -r a75315de5e23 sys/arch/alpha/alpha/interrupt.c
--- a/sys/arch/alpha/alpha/interrupt.c Mon Apr 19 23:51:42 2021 +0000
+++ b/sys/arch/alpha/alpha/interrupt.c Tue Apr 20 00:09:45 2021 +0000
@@ -1,4 +1,4 @@
-/* $NetBSD: interrupt.c,v 1.93 2021/04/15 00:19:52 rin Exp $ */
+/* $NetBSD: interrupt.c,v 1.94 2021/04/20 00:09:45 thorpej Exp $ */
/*-
* Copyright (c) 2000, 2001 The NetBSD Foundation, Inc.
@@ -65,7 +65,7 @@
#include <sys/cdefs.h> /* RCS ID & Copyright macro defns */
-__KERNEL_RCSID(0, "$NetBSD: interrupt.c,v 1.93 2021/04/15 00:19:52 rin Exp $");
+__KERNEL_RCSID(0, "$NetBSD: interrupt.c,v 1.94 2021/04/20 00:09:45 thorpej Exp $");
#include <sys/param.h>
#include <sys/systm.h>
@@ -212,7 +212,21 @@
break;
case ALPHA_INTR_CLOCK: /* clock interrupt */
- atomic_inc_ulong(&ci->ci_intrdepth);
+ /*
+ * Rather than simply increment the interrupt depth
+ * for the clock interrupt, we add 0x10. Why? Because
+ * while we only call out a single device interrupt
+ * level, technically the architecture specification
+ * suports two, meaning we could have intrdepth > 1
+ * just for device interrupts.
+ *
+ * Adding 0x10 here means that cpu_intr_p() can check
+ * for "intrdepth != 0" for "in interrupt context" and
+ * CLKF_INTR() can check "(intrdepth & 0xf) != 0" for
+ * "was processing interrupts when the clock interrupt
+ * happened".
+ */
+ atomic_add_long(&ci->ci_intrdepth, 0x10);
sc->sc_evcnt_clock.ev_count++;
ci->ci_data.cpu_nintr++;
if (platform.clockintr) {
@@ -237,7 +251,7 @@
schedhz != 0)
schedclock(ci->ci_curlwp);
}
- atomic_dec_ulong(&ci->ci_intrdepth);
+ atomic_add_long(&ci->ci_intrdepth, -0x10);
break;
case ALPHA_INTR_ERROR: /* Machine Check or Correctable Error */
diff -r 072d021be652 -r a75315de5e23 sys/arch/alpha/include/cpu.h
--- a/sys/arch/alpha/include/cpu.h Mon Apr 19 23:51:42 2021 +0000
+++ b/sys/arch/alpha/include/cpu.h Tue Apr 20 00:09:45 2021 +0000
@@ -1,4 +1,4 @@
-/* $NetBSD: cpu.h,v 1.100 2021/04/15 08:23:24 rin Exp $ */
+/* $NetBSD: cpu.h,v 1.101 2021/04/20 00:09:45 thorpej Exp $ */
/*-
* Copyright (c) 1998, 1999, 2000, 2001 The NetBSD Foundation, Inc.
@@ -206,7 +206,7 @@
* like this stastic has to be extremely accurate.
*/
#define CLKF_INTR(framep) \
- (curcpu()->ci_intrdepth > 1) /* one for clock interrupt itself */
+ ((curcpu()->ci_intrdepth & 0xf) != 0) /* see interrupt() */
/*
* This is used during profiling to integrate system time. It can safely
Home |
Main Index |
Thread Index |
Old Index