linux-kernel - [RFC PATCH v2 05/12] x86/ioapic: Refactor the delay logic in timer_irq

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <2058afd49c1ecec0fb300da71727d04793812c01.1492589715.git.douly.fnst@cn.fujitsu.com>
Date:   Wed, 19 Apr 2017 17:05:19 +0800
From:   Dou Liyang <douly.fnst@...fujitsu.com>
To:     <x86@...nel.org>, <linux-kernel@...r.kernel.org>
CC:     <tglx@...utronix.de>, <mingo@...nel.org>, <hpa@...or.com>,
        <ebiederm@...ssion.com>, <bhe@...hat.com>,
        <izumi.taku@...fujitsu.com>, Dou Liyang <douly.fnst@...fujitsu.com>
Subject: [RFC PATCH v2 05/12] x86/ioapic: Refactor the delay logic in timer_irq_works()

Timer_irq_works() calls mdelay(10) to delay ten ticks and checks whether
the timer IRQ works or not. The mdelay() depends on the loops_per_jiffy
which is set up in calibrate_delay().

Timer_irq_works() has been in advance of calibrate_delay(). the mdelay()
doesn't work well in timer_irq_works().

Refactor the delay logic by replace the mdelay() with __delay() and
regard 2**MAX_BAND as the maximum threshold of CPU frequency.

Signed-off-by: Dou Liyang <douly.fnst@...fujitsu.com>
---
 arch/x86/kernel/apic/io_apic.c | 47 ++++++++++++++++++++++++++++++++++++------
 1 file changed, 41 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 347bb9f..d3b0268 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1608,6 +1608,18 @@ static int __init notimercheck(char *s)
 __setup("no_timer_check", notimercheck);
 
 /*
+ * The loops_per_jiffy is not be set, udelay() or mdelay() cannot
+ * be called here.
+ * Set 16GHZ(loops_per_jiffy almost is 2**24) as the maximum threshold
+ * of CPU frequency. Exceed the threshold indicates the timer interrupt
+ * doesn't work.
+ *
+ * FIXME: Explore a new way to check the timer interrupt work.
+ */
+
+#define MAX_BAND 24
+
+/*
  * There is a nasty bug in some older SMP boards, their mptable lies
  * about the timer IRQ. We do the following to work around the situation:
  *
@@ -1617,28 +1629,51 @@ __setup("no_timer_check", notimercheck);
  */
 static int __init timer_irq_works(void)
 {
-	unsigned long t1 = jiffies;
+	unsigned long ticks;
 	unsigned long flags;
+	/* Set up approx 2 Bo*oMips to start */
+	int band = 11;
+	int loop_times;
 
 	if (no_timer_check)
 		return 1;
 
 	local_save_flags(flags);
 	local_irq_enable();
-	/* Let ten ticks pass... */
-	mdelay((10 * 1000) / HZ);
-	local_irq_restore(flags);
+	/*
+	 * Estimate the loops in (1,2] jiffies.
+	 *
+	 * After the loops (2**12+2**13+...+2**MAX_BAND) times,
+	 * if time_before(jiffies, ticks + 2) is also n, That the
+	 * timer interrupt did not work, return 0.
+	 */
+	ticks = jiffies;
+	do {
+		if (band++ == MAX_BAND) {
+			local_irq_restore(flags);
+			return 0;
+		}
+
+		__delay(1 << band);
+	} while (time_before(jiffies, ticks + 2));
 
 	/*
 	 * Expect a few ticks at least, to be sure some possible
 	 * glue logic does not lock up after one or two first
 	 * ticks in a non-ExtINT mode.  Also the local APIC
-	 * might have cached one ExtINT interrupt.  Finally, at
+	 * might have cached one ExtINT interrupt. Finally, at
 	 * least one tick may be lost due to delays.
+	 *
+	 * As 2**12+2**13+...+2**band < 2**(1 + band),
+	 * it certainly delayed for more than 4 ticks.
 	 */
+	for (loop_times = 0; loop_times < 4; loop_times++)
+		__delay(1 << (1 + band));
+
+	local_irq_restore(flags);
 
 	/* jiffies wrap? */
-	if (time_after(jiffies, t1 + 4))
+	if (time_after(jiffies, ticks + 4))
 		return 1;
 	return 0;
 }
-- 
2.5.5