All of lore.kernel.org
 help / color / mirror / Atom feed
From: Martin Wilck <martin.wilck@fujitsu-siemens.com>
To: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"H. Peter Anvin" <hpa@zytor.com>,
	"Wichert, Gerhard" <Gerhard.Wichert@fujitsu-siemens.com>,
	"Maciej W. Rozycki" <macro@linux-mips.org>
Subject: Re: [PATCH] x86 (64): make calibrate_APIC_clock() SMI-safe
Date: Thu, 24 Jul 2008 15:55:02 +0200	[thread overview]
Message-ID: <488889B6.9070707@fujitsu-siemens.com> (raw)
In-Reply-To: <20080724120512.GA21804@lenovo>

[-- Attachment #1: Type: text/plain, Size: 1203 bytes --]

Cyrill Gorcunov wrote:

> yes, it will issue some effects but it's better then stuck there.
> More over in 'case of SMI flood with current patch you don't get
> error message printed i think so you better add max iteration
> counter so user will see on console (or whatever) that he is got
> problems.
>                 - Cyrill -

I disagree. If you have a system that generates SMIs in this extreme 
frequency, you're better off stuck than running on such an unstable 
system. The user _will_ see messages on the console if this happens. 
Note that apparently there are few people who have trouble with this. We 
did see problems, but never had more than 1 SMI disturbing the 
calibration procedure.

Anyway, here is another patch that defines max iteration counts. I 
haven't added a "Signed-off:" line, because I prefer the original version.

Martin

-- 
Martin Wilck
PRIMERGY System Software Engineer
FSC IP ESP DEV 6

Fujitsu Siemens Computers GmbH
Heinz-Nixdorf-Ring 1
33106 Paderborn
Germany

Tel:			++49 5251 8 15113
Fax:			++49 5251 8 20209
Email:			mailto:martin.wilck@fujitsu-siemens.com
Internet:		http://www.fujitsu-siemens.com
Company Details:	http://www.fujitsu-siemens.com/imprint.html

[-- Attachment #2: calibrate_APIC_clock-2.diff --]
[-- Type: text/x-patch, Size: 2340 bytes --]

--- arch/x86/kernel/apic_64.c	2008-07-24 15:25:30.000000000 +0200
+++ arch/x86/kernel/apic_64.c.new	2008-07-24 15:26:13.000000000 +0200
@@ -314,11 +314,35 @@ static void setup_APIC_timer(void)
 
 #define TICK_COUNT 100000000
 
+#define MAX_DIFFERENCE 1000UL
+#define MAX_ITER 10
+#define MAX_CALIBRATIONS 10
+static inline int
+__read_tsc_and_apic(unsigned long *tsc, unsigned *apic)
+{
+	unsigned long tsc0, tsc1, diff;
+	int i;
+	for (i = 0; i < MAX_ITER; i++) {
+		rdtscll(tsc0);
+		*apic = apic_read(APIC_TMCCT);
+		rdtscll(tsc1);
+		diff = tsc1 - tsc0;
+		if (diff <= MAX_DIFFERENCE)
+			break;
+	}
+	*tsc = tsc0 + (diff >> 1);
+	if (i == MAX_ITER) {
+		printk(KERN_ERR "unable to read TSC abd APIC simultaneously\n");
+		return -EIO;
+	}
+	return 0;
+}
+
 static void __init calibrate_APIC_clock(void)
 {
 	unsigned apic, apic_start;
 	unsigned long tsc, tsc_start;
-	int result;
+	int result, cnt = 0, err_start, err;
 
 	local_irq_disable();
 
@@ -329,25 +353,41 @@ static void __init calibrate_APIC_clock(
 	 *
 	 * No interrupt enable !
 	 */
+smi_occured:
 	__setup_APIC_LVTT(250000000, 0, 0);
 
-	apic_start = apic_read(APIC_TMCCT);
 #ifdef CONFIG_X86_PM_TIMER
 	if (apic_calibrate_pmtmr && pmtmr_ioport) {
+		apic_start = apic_read(APIC_TMCCT);
 		pmtimer_wait(5000);  /* 5ms wait */
 		apic = apic_read(APIC_TMCCT);
 		result = (apic_start - apic) * 1000L / 5;
 	} else
 #endif
 	{
-		rdtscll(tsc_start);
-
+		err_start = __read_tsc_and_apic(&tsc_start, &apic_start);
 		do {
-			apic = apic_read(APIC_TMCCT);
-			rdtscll(tsc);
+			err = __read_tsc_and_apic(&tsc, &apic);
 		} while ((tsc - tsc_start) < TICK_COUNT &&
 				(apic_start - apic) < TICK_COUNT);
 
+		/*
+		 * If this takes significantly longer than TICK_COUNT,
+		 * some interruption must have occured - retry.
+		 */
+		if (err_start || err || 
+			(tsc - tsc_start) > (TICK_COUNT + TICK_COUNT/1000) ||
+		    	(apic_start - apic) > (TICK_COUNT + TICK_COUNT/1000)) {
+			printk(KERN_ERR
+			       "calibrate_APIC_clock: SMI occured? %lx %x\n",
+			       tsc - tsc_start, apic_start - apic);
+			if (++cnt < MAX_CALIBRATIONS)
+				goto smi_occured;
+			else
+				printk(KERN_CRIT
+				"calibrate_APIC_clock: SMI flood - "
+				"the APIC timer calibration may be wrong!\n");
+		}
 		result = (apic_start - apic) * 1000L * tsc_khz /
 					(tsc - tsc_start);
 	}

  reply	other threads:[~2008-07-24 13:55 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-07-24 10:47 [PATCH] x86 (64): make calibrate_APIC_clock() smp-safe Martin Wilck
2008-07-24 11:16 ` Cyrill Gorcunov
2008-07-24 11:58   ` Martin Wilck
2008-07-24 12:05     ` Cyrill Gorcunov
2008-07-24 13:55       ` Martin Wilck [this message]
2008-07-24 14:31         ` [PATCH] x86 (64): make calibrate_APIC_clock() SMI-safe Cyrill Gorcunov
2008-07-24 15:01           ` Cyrill Gorcunov
2008-07-24 15:13             ` Martin Wilck
2008-07-25  9:02               ` [PATCH] x86 (64): make calibrate_APIC_clock() SMI-safe (take 2) Martin Wilck
2008-07-25 10:08                 ` Cyrill Gorcunov
2008-07-25 12:29                   ` Martin Wilck
2008-07-25 12:59                     ` Cyrill Gorcunov
2008-07-25 13:38                       ` Martin Wilck
2008-07-25 13:48                         ` Cyrill Gorcunov
2008-07-25 14:01                           ` [PATCH] x86 (64): make calibrate_APIC_clock() SMI-safe (take 3) Martin Wilck
2008-07-25 14:15                             ` Cyrill Gorcunov
2008-07-25 15:01                             ` Cyrill Gorcunov
2008-07-25 15:13                               ` Martin Wilck
2008-07-25 15:39                                 ` Cyrill Gorcunov
2008-07-26 15:40                                 ` Ingo Molnar
2009-03-12  9:41                                 ` Jean Delvare
2009-03-12 13:38                                   ` Martin Wilck
2008-07-25 16:51                 ` [PATCH] x86 (64): make calibrate_APIC_clock() SMI-safe (take 2) Olaf Dabrunz
2008-07-24 13:31 ` [PATCH] x86 (64): make calibrate_APIC_clock() smp-safe H. Peter Anvin
2008-07-24 13:42   ` [PATCH] x86 (64): make calibrate_APIC_clock() SMI-safe Martin Wilck

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=488889B6.9070707@fujitsu-siemens.com \
    --to=martin.wilck@fujitsu-siemens.com \
    --cc=Gerhard.Wichert@fujitsu-siemens.com \
    --cc=gorcunov@gmail.com \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=macro@linux-mips.org \
    --cc=tglx@linutronix.de \
    /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.