linux-kernel - Re: [PATCH 2/2] KVM: x86: fix edge EOI and IOAPIC reconfig race

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: <20151006203338.GB4508@potion.brq.redhat.com>
Date:	Tue, 6 Oct 2015 22:33:39 +0200
From:	Radim Krčmář <rkrcmar@...hat.com>
To:	Paolo Bonzini <pbonzini@...hat.com>
Cc:	linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
	Steve Rutherford <srutherford@...gle.com>,
	stable@...r.kernel.org
Subject: Re: [PATCH 2/2] KVM: x86: fix edge EOI and IOAPIC reconfig race

2015-08-15 02:00+0200, Paolo Bonzini:
>On 14/08/2015 10:38, Radim Krčmář wrote:
>>> How do you reproduce the bug?
>> I run rhel4 (2.6.9) kernel on 2 VCPUs and frequently alternate
>> smp_affinity of "timer".  The bug is hit within seconds.
> 
> Nice, I'll try to make a unit test for it on the plane. :)

(Time on planes is best spent by sleeping ;])

What do you think about the series?

I made a prototype kvm-unit-test ...
(Reproduces with APICv or split irqchip builds.)
---8<---
x86: test IO-APIC dest_id modification before

Regression test for kvm commit $TODO.
Run with '-smp 2'.

I had problems with handle_irq so this version uses asm workaround;
will fix that in final version.  Initialization also presumes that
everything will work as it did on my machines :)
---
 x86/ioapic.c | 47 +++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 45 insertions(+), 2 deletions(-)

diff --git a/x86/ioapic.c b/x86/ioapic.c
index 1fe1ccc9eadd..55f8eea03496 100644
--- a/x86/ioapic.c
+++ b/x86/ioapic.c
@@ -4,21 +4,27 @@
 #include "smp.h"
 #include "desc.h"
 #include "isr.h"
+#include "io.h"
 
 #define EDGE_TRIGGERED 0
 #define LEVEL_TRIGGERED 1
 
-static void set_ioapic_redir(unsigned line, unsigned vec, unsigned trig_mode)
+static void __set_ioapic_redir(unsigned line, u8 vec, bool trig_mode, u8 dest_id)
 {
 	ioapic_redir_entry_t e = {
 		.vector = vec,
-		.delivery_mode = 0,
 		.trig_mode = trig_mode,
+		.dest_id = dest_id,
 	};
 
 	ioapic_write_redir(line, e);
 }
 
+static void set_ioapic_redir(unsigned line, unsigned vec, unsigned trig_mode)
+{
+	__set_ioapic_redir(line, vec, trig_mode, 0);
+}
+
 static void set_irq_line(unsigned line, int val)
 {
 	asm volatile("out %0, %1" : : "a"((u8)val), "d"((u16)(0x2000 + line)));
@@ -298,6 +304,41 @@ static void test_ioapic_level_retrigger_mask(void)
 	set_mask(0x0e, false);
 }
 
+static volatile int pit_working = -1;
+
+static void __attribute__((used)) pit_isr_fn(void)
+{
+	if (!pit_working++);
+		__set_ioapic_redir(0x2, 0x84, EDGE_TRIGGERED, 1);
+	eoi();
+}
+
+void pit_isr(void);
+asm (
+     "pit_isr: \n"
+     "   call pit_isr_fn \n"
+#ifndef __x86_64__
+     "   iret"
+#else
+     "   iretq"
+#endif
+     );
+
+static void test_ioapic_eoi_bug(void)
+{
+	if (cpu_count() < 2)
+		return;
+
+	set_idt_entry(0x84, pit_isr, 0);
+	__set_ioapic_redir(0x2, 0x84, EDGE_TRIGGERED, 0);
+
+	outb(0x34, 0x43);
+
+	for (unsigned loops = 100000000; loops && pit_working < 1; loops--)
+		asm volatile ("nop");
+
+	report("dest_id reconfiguration before EOI", pit_working >= 1);
+}
 
 int main(void)
 {
@@ -325,5 +366,7 @@ int main(void)
 	test_ioapic_level_mask();
 	test_ioapic_level_retrigger_mask();
 
+	test_ioapic_eoi_bug();
+
 	return report_summary();
 }
-- 
2.5.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/