Intel Archiecture Software Developer's Manual section 7.1.3 specifies that a
core serializing instruction such as "cpuid" should be executed on _each_ core
before the new instruction is made visible.
Failure to do so can lead to unspecified behavior (Intel XMC erratas include
General Protection Fault in the list), so we should avoid this at all cost.
This problem can affect modified code executed by interrupt handlers after
interrupt are re-enabled at the end of stop_machine, because no core serializing
instruction is executed between the code modification and the moment interrupts
are reenabled.
Because stop_machine_text_poke performs the text modification from the first CPU
decrementing stop_machine_first, modified code executed in thread context is
also affected by this problem. To explain why, we have to split the CPUs in two
categories: the CPU that initiates the text modification (calls text_poke_smp)
and all the others. The scheduler, executed on all other CPUs after
stop_machine, issues an "iret" core serializing instruction, and therefore
handles core serialization for all these CPUs. However, the text modification
initiator can continue its execution on the same thread and access the modified
text without any scheduler call. Given that the CPU that initiates the code
modification is not guaranteed to be the one actually performing the code
modification, it falls into the XMC errata.
Signed-off-by: Mathieu Desnoyers <[email protected]>
CC: Peter Zijlstra <[email protected]>
CC: Arjan van de Ven <[email protected]>
CC: "H. Peter Anvin" <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Steven Rostedt <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: Andrew Morton <[email protected]>
CC: Andi Kleen <[email protected]>
CC: Frederic Weisbecker <[email protected]>
CC: Masami Hiramatsu <[email protected]>
---
arch/x86/kernel/alternative.c | 6 ++++++
1 file changed, 6 insertions(+)
Index: linux-2.6-lttng/arch/x86/kernel/alternative.c
===================================================================
--- linux-2.6-lttng.orig/arch/x86/kernel/alternative.c
+++ linux-2.6-lttng/arch/x86/kernel/alternative.c
@@ -612,6 +612,12 @@ static int __kprobes stop_machine_text_p
flush_icache_range((unsigned long)tpp->addr,
(unsigned long)tpp->addr + tpp->len);
+ /*
+ * Intel Archiecture Software Developer's Manual section 7.1.3 specifies
+ * that a core serializing instruction such as "cpuid" should be
+ * executed on _each_ core before the new instruction is made visible.
+ */
+ sync_core();
return 0;
}
--
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com
(2011/03/01 0:24), Mathieu Desnoyers wrote:
> Intel Archiecture Software Developer's Manual section 7.1.3 specifies that a
> core serializing instruction such as "cpuid" should be executed on _each_ core
> before the new instruction is made visible.
>
> Failure to do so can lead to unspecified behavior (Intel XMC erratas include
> General Protection Fault in the list), so we should avoid this at all cost.
>
> This problem can affect modified code executed by interrupt handlers after
> interrupt are re-enabled at the end of stop_machine, because no core serializing
> instruction is executed between the code modification and the moment interrupts
> are reenabled.
>
> Because stop_machine_text_poke performs the text modification from the first CPU
> decrementing stop_machine_first, modified code executed in thread context is
> also affected by this problem. To explain why, we have to split the CPUs in two
> categories: the CPU that initiates the text modification (calls text_poke_smp)
> and all the others. The scheduler, executed on all other CPUs after
> stop_machine, issues an "iret" core serializing instruction, and therefore
> handles core serialization for all these CPUs. However, the text modification
> initiator can continue its execution on the same thread and access the modified
> text without any scheduler call. Given that the CPU that initiates the code
> modification is not guaranteed to be the one actually performing the code
> modification, it falls into the XMC errata.
Thanks Mathieu!
It seems reasonable change. At least I'm OK :)
Reviewed-by: Masami Hiramatsu <[email protected]>
>
> Signed-off-by: Mathieu Desnoyers <[email protected]>
> CC: Peter Zijlstra <[email protected]>
> CC: Arjan van de Ven <[email protected]>
> CC: "H. Peter Anvin" <[email protected]>
> CC: Thomas Gleixner <[email protected]>
> CC: Steven Rostedt <[email protected]>
> CC: Ingo Molnar <[email protected]>
> CC: Andrew Morton <[email protected]>
> CC: Andi Kleen <[email protected]>
> CC: Frederic Weisbecker <[email protected]>
> CC: Masami Hiramatsu <[email protected]>
> ---
> arch/x86/kernel/alternative.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> Index: linux-2.6-lttng/arch/x86/kernel/alternative.c
> ===================================================================
> --- linux-2.6-lttng.orig/arch/x86/kernel/alternative.c
> +++ linux-2.6-lttng/arch/x86/kernel/alternative.c
> @@ -612,6 +612,12 @@ static int __kprobes stop_machine_text_p
>
> flush_icache_range((unsigned long)tpp->addr,
> (unsigned long)tpp->addr + tpp->len);
> + /*
> + * Intel Archiecture Software Developer's Manual section 7.1.3 specifies
> + * that a core serializing instruction such as "cpuid" should be
> + * executed on _each_ core before the new instruction is made visible.
> + */
> + sync_core();
> return 0;
> }
>
--
Masami HIRAMATSU
2nd Dept. Linux Technology Center
Hitachi, Ltd., Systems Development Laboratory
E-mail: [email protected]
On 02/28/2011 07:24 AM, Mathieu Desnoyers wrote:
>
> Index: linux-2.6-lttng/arch/x86/kernel/alternative.c
> ===================================================================
> --- linux-2.6-lttng.orig/arch/x86/kernel/alternative.c
> +++ linux-2.6-lttng/arch/x86/kernel/alternative.c
> @@ -612,6 +612,12 @@ static int __kprobes stop_machine_text_p
>
> flush_icache_range((unsigned long)tpp->addr,
> (unsigned long)tpp->addr + tpp->len);
> + /*
> + * Intel Archiecture Software Developer's Manual section 7.1.3 specifies
> + * that a core serializing instruction such as "cpuid" should be
> + * executed on _each_ core before the new instruction is made visible.
> + */
> + sync_core();
> return 0;
> }
>
Isn't this executed from an IPI handler, which will return with IRET (a
serializing instruction) anyway?
-hpa
(2011/03/03 15:10), H. Peter Anvin wrote:
> On 02/28/2011 07:24 AM, Mathieu Desnoyers wrote:
>>
>> Index: linux-2.6-lttng/arch/x86/kernel/alternative.c
>> ===================================================================
>> --- linux-2.6-lttng.orig/arch/x86/kernel/alternative.c
>> +++ linux-2.6-lttng/arch/x86/kernel/alternative.c
>> @@ -612,6 +612,12 @@ static int __kprobes stop_machine_text_p
>>
>> flush_icache_range((unsigned long)tpp->addr,
>> (unsigned long)tpp->addr + tpp->len);
>> + /*
>> + * Intel Archiecture Software Developer's Manual section 7.1.3 specifies
>> + * that a core serializing instruction such as "cpuid" should be
>> + * executed on _each_ core before the new instruction is made visible.
>> + */
>> + sync_core();
>> return 0;
>> }
>>
>
> Isn't this executed from an IPI handler, which will return with IRET (a
> serializing instruction) anyway?
No, now stop_machine uses per-cpu workqueue, so that handler will be
executed from worker threads. There is no iret anymore.
BTW, Mathieu, since the latest kernel has batch-text_poke_smp, this patch
needs to be updated.
Thank you,
--
Masami HIRAMATSU
2nd Dept. Linux Technology Center
Hitachi, Ltd., Systems Development Laboratory
E-mail: [email protected]
* Masami Hiramatsu ([email protected]) wrote:
> (2011/03/03 15:10), H. Peter Anvin wrote:
> > On 02/28/2011 07:24 AM, Mathieu Desnoyers wrote:
> >>
> >> Index: linux-2.6-lttng/arch/x86/kernel/alternative.c
> >> ===================================================================
> >> --- linux-2.6-lttng.orig/arch/x86/kernel/alternative.c
> >> +++ linux-2.6-lttng/arch/x86/kernel/alternative.c
> >> @@ -612,6 +612,12 @@ static int __kprobes stop_machine_text_p
> >>
> >> flush_icache_range((unsigned long)tpp->addr,
> >> (unsigned long)tpp->addr + tpp->len);
> >> + /*
> >> + * Intel Archiecture Software Developer's Manual section 7.1.3 specifies
> >> + * that a core serializing instruction such as "cpuid" should be
> >> + * executed on _each_ core before the new instruction is made visible.
> >> + */
> >> + sync_core();
> >> return 0;
> >> }
> >>
> >
> > Isn't this executed from an IPI handler, which will return with IRET (a
> > serializing instruction) anyway?
>
> No, now stop_machine uses per-cpu workqueue, so that handler will be
> executed from worker threads. There is no iret anymore.
>
> BTW, Mathieu, since the latest kernel has batch-text_poke_smp, this patch
> needs to be updated.
OK, I'll update the patch and respin it.
Thanks,
Mathieu
>
> Thank you,
>
> --
> Masami HIRAMATSU
> 2nd Dept. Linux Technology Center
> Hitachi, Ltd., Systems Development Laboratory
> E-mail: [email protected]
--
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com