linux-kernel - Re: [PATCH v2 char-misc 1/1] x86/hyperv: Fix PIT shutdown quirk

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: <alpine.DEB.2.21.1810300859420.5984@nanos.tec.linutronix.de>
Date:   Tue, 30 Oct 2018 09:35:59 +0100 (CET)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Michael Kelley <mikelley@...rosoft.com>
cc:     "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "x86@...nel.org" <x86@...nel.org>,
        "daniel.lezcano@...aro.org" <daniel.lezcano@...aro.org>,
        "olaf@...fle.de" <olaf@...fle.de>,
        "apw@...onical.com" <apw@...onical.com>,
        vkuznets <vkuznets@...hat.com>,
        "jasowang@...hat.com" <jasowang@...hat.com>,
        "marcelo.cerri@...onical.com" <marcelo.cerri@...onical.com>,
        Stephen Hemminger <sthemmin@...rosoft.com>,
        KY Srinivasan <kys@...rosoft.com>
Subject: Re: [PATCH v2 char-misc 1/1] x86/hyperv: Fix PIT shutdown quirk

Michael,

On Mon, 22 Oct 2018, Michael Kelley wrote:

> pit_shutdown() doesn't work on Hyper-V because of a quirk in the
> PIT emulation. After shutdown the emulated PIT continues to interrupt
> @18.2 HZ. This problem exists in all versions of Hyper-V and just
> had not previously been noticed. So replace the normal PIT shutdown
> function with an alternate version that works on Hyper-V.
> 
> The alternate shutdown function could go in the i8253.c module,
> but it seems better to keep the hack with the Hyper-V code rather
> than clutter up i8253.c with #ifdef CONFIG_HYPERV.

You can avoid the ideffery completely. Something like the uncompiled patch
(lacks comments) below keeps everything in i8253 and should just work.

Thanks,

	tglx

8<-------------------

diff --git a/drivers/clocksource/i8253.c b/drivers/clocksource/i8253.c
index 9c38895542f4..9b5d2259c2e5 100644
--- a/drivers/clocksource/i8253.c
+++ b/drivers/clocksource/i8253.c
@@ -10,6 +10,7 @@
 #include <linux/module.h>
 #include <linux/i8253.h>
 #include <linux/smp.h>
+#include <linux/hypervisor.h>
 
 /*
  * Protects access to I/O ports
@@ -109,9 +110,10 @@ static int pit_shutdown(struct clock_event_device *evt)
 	raw_spin_lock(&i8253_lock);
 
 	outb_p(0x30, PIT_MODE);
-	outb_p(0, PIT_CH0);
-	outb_p(0, PIT_CH0);
-
+	if (!hypervisor_is_mshyperv()) {
+		outb_p(0, PIT_CH0);
+		outb_p(0, PIT_CH0);
+	}
 	raw_spin_unlock(&i8253_lock);
 	return 0;
 }
diff --git a/include/linux/hypervisor.h b/include/linux/hypervisor.h
index fc08b433c856..2ab7303675f3 100644
--- a/include/linux/hypervisor.h
+++ b/include/linux/hypervisor.h
@@ -10,6 +10,7 @@
 #ifdef CONFIG_X86
 
 #include <asm/jailhouse_para.h>
+#include <asm/hypervisor.h>
 #include <asm/x86_init.h>
 
 static inline void hypervisor_pin_vcpu(int cpu)
@@ -17,6 +18,11 @@ static inline void hypervisor_pin_vcpu(int cpu)
 	x86_platform.hyper.pin_vcpu(cpu);
 }
 
+static inline bool hypervisor_is_mshyperv(void)
+{
+	return hypervisor_is_type(X86_HYPER_MS_HYPERV);
+}
+
 #else /* !CONFIG_X86 */
 
 #include <linux/of.h>
@@ -30,6 +36,11 @@ static inline bool jailhouse_paravirt(void)
 	return of_find_compatible_node(NULL, NULL, "jailhouse,cell");
 }
 
+static inline bool hypervisor_is_mshyperv(void)
+{
+	return false;
+}
+
 #endif /* !CONFIG_X86 */
 
 #endif /* __LINUX_HYPEVISOR_H */