All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jun Sun <jsun@mvista.com>
To: ralf@oss.sgi.com
Cc: linux-mips@oss.sgi.com
Subject: [PATCH] gettimeoffset for swarm
Date: Mon, 13 May 2002 19:17:52 -0700	[thread overview]
Message-ID: <3CE073D0.8030402@mvista.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 553 bytes --]


By default, swarm will use calibrate_div64_gettimeoffset().  That does
not work in SMP mode because the two cores have different counter register
value.  This patch gives swarm its own and accurate gettimeoffset().

The symptom without this patch is that cpu core 1 does not have micro-second
resolution of gettimeofday().

It applies to 2.4 branch with 1 or 2 line offset warnings.  It does not apply 
to 2.5 branch. 2.5 branch appears to be outdated.

http://linux.junsun.net/patches/oss.sgi.com/submitted/020515.swarm-own-gettimeoffset.patch

Jun



[-- Attachment #2: 020515.swarm-own-gettimeoffset.patch --]
[-- Type: text/plain, Size: 2963 bytes --]


By default, swarm will use calibrate_div64_gettimeoffset().  That does
not work in SMP mode because the two cores have different counter register
value.  This patch gives swarm its own and accurate gettimeoffset().

The symptom without this patch is that cpu core 1 does not have micro-second
resolution of gettimeofday().

Jun

diff -Nru linux/arch/mips/sibyte/sb1250/time.c.orig linux/arch/mips/sibyte/sb1250/time.c
--- linux/arch/mips/sibyte/sb1250/time.c.orig	Tue Apr 30 17:29:25 2002
+++ linux/arch/mips/sibyte/sb1250/time.c	Fri May 10 14:53:39 2002
@@ -34,6 +34,7 @@
 #include <asm/ptrace.h>
 #include <asm/addrspace.h>
 #include <asm/time.h>
+#include <asm/debug.h>
 
 #include <asm/sibyte/sb1250.h>
 #include <asm/sibyte/sb1250_regs.h>
@@ -107,3 +108,19 @@
 	 */
 	ll_local_timer_interrupt(0, regs);
 }
+
+/*
+ * We use our own do_gettimeoffset() instead of the generic one,
+ * because the generic one does not work for SMP case.
+ * In addition, since we use general timer 0 for system time,
+ * we can get accurate intra-jiffy offset without calibration.
+ */
+unsigned long sb1250_gettimeoffset(void)
+{
+	unsigned long count =
+		in64(KSEG1 + A_SCD_TIMER_REGISTER(0, R_SCD_TIMER_CNT));
+
+	db_assert(count <= 1000000 / HZ);
+	return 1000000/HZ - count;
+}
+
diff -Nru linux/arch/mips/sibyte/swarm/setup.c.orig linux/arch/mips/sibyte/swarm/setup.c
--- linux/arch/mips/sibyte/swarm/setup.c.orig	Tue Apr 30 17:38:09 2002
+++ linux/arch/mips/sibyte/swarm/setup.c	Fri May 10 14:54:36 2002
@@ -237,12 +237,15 @@
          * interrupts through low-level (direct) meachanism.
          */
 
+	/* Use our own gettimeoffset() routine */
+	do_gettimeoffset = sb1250_gettimeoffset;
+
         /* We only need to setup the generic timer */
         sb1250_time_init();
 }
 
 extern int xicor_set_time(unsigned long);
-extern unsigned int xicor_get_time(void);
+extern unsigned long xicor_get_time(void);
 
 void __init swarm_setup(void)
 {
diff -Nru linux/include/asm-mips/sibyte/sb1250.h.orig linux/include/asm-mips/sibyte/sb1250.h
--- linux/include/asm-mips/sibyte/sb1250.h.orig	Tue Jan 15 13:25:45 2002
+++ linux/include/asm-mips/sibyte/sb1250.h	Fri May 10 14:51:50 2002
@@ -20,6 +20,7 @@
 #define _ASM_SIBYTE_SB1250_H
 
 extern void sb1250_time_init(void);
+extern unsigned long sb1250_gettimeoffset(void);
 extern void sb1250_mask_irq(int cpu, int irq);
 extern void sb1250_unmask_irq(int cpu, int irq);
 extern void sb1250_smp_finish(void);
diff -Nru linux/include/asm-mips64/sibyte/sb1250.h.orig linux/include/asm-mips64/sibyte/sb1250.h
--- linux/include/asm-mips64/sibyte/sb1250.h.orig	Tue Apr 30 17:30:44 2002
+++ linux/include/asm-mips64/sibyte/sb1250.h	Fri May 10 14:52:16 2002
@@ -20,6 +20,7 @@
 #define _ASM_SIBYTE_SB1250_H
 
 extern void sb1250_time_init(void);
+extern unsigned long sb1250_gettimeoffset(void);
 extern void sb1250_mask_irq(int cpu, int irq);
 extern void sb1250_unmask_irq(int cpu, int irq);
 extern void sb1250_smp_finish(void);

             reply	other threads:[~2002-05-14  2:35 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-05-14  2:17 Jun Sun [this message]
2002-05-14 13:18 ` [PATCH] gettimeoffset for swarm Maciej W. Rozycki

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3CE073D0.8030402@mvista.com \
    --to=jsun@mvista.com \
    --cc=linux-mips@oss.sgi.com \
    --cc=ralf@oss.sgi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.