----- [email protected] wrote:
> On Wed, Mar 13, 2013 at 09:25:44AM -0400, Boris Ostrovsky wrote:
> > On 03/01/2013 07:14 AM, Josh Boyer wrote:
> > >On Thu, Feb 28, 2013 at 04:52:20PM -0800, H. Peter Anvin wrote:
> > >>On 02/28/2013 04:42 PM, Josh Boyer wrote:
> > >>>On Fri, Mar 01, 2013 at 01:36:29AM +0100, Borislav Petkov wrote:
> > >>>>On Thu, Feb 28, 2013 at 04:15:45PM -0800, H. Peter Anvin wrote:
> > >>>>>>I'll try to get someone to test this tomorrow.
> > >>>>Btw, you'd need to apply that other patch too
> > >>>>
> > >>>>http://marc.info/?l=xen-devel&m=136206183814547&w=2
> > >>>>
> > >>>>so that arch_flush_lazy_mmu_mode() has at least one caller on
> x86_64.
> > >>>Yeah, we already have that applied. It stops crashes in xen
> > >>>environments so we pulled it in as a bugfix. Thanks though!
> > >>>
> > >>Who are "we"?
> > >Sorry, Fedora. That patch has a link to a bug in it. We applied
> the
> > >patch for that bug. I'll apply Boris' patch on top and get the
> same
> > >people to test it.
> >
> > Josh, have you had a chance to test this?
>
> I've tested it on bare metal for a while now. No problems noticed at
> all. I've not heard back from Krishna who was testing it in the Xen
> environment. Krishna?
Any updates?
Thanks.
-boris
On Wed, Mar 20, 2013 at 06:53:55AM -0700, Boris Ostrovsky wrote:
>
> ----- [email protected] wrote:
>
> > On Wed, Mar 13, 2013 at 09:25:44AM -0400, Boris Ostrovsky wrote:
> > > On 03/01/2013 07:14 AM, Josh Boyer wrote:
> > > >On Thu, Feb 28, 2013 at 04:52:20PM -0800, H. Peter Anvin wrote:
> > > >>On 02/28/2013 04:42 PM, Josh Boyer wrote:
> > > >>>On Fri, Mar 01, 2013 at 01:36:29AM +0100, Borislav Petkov wrote:
> > > >>>>On Thu, Feb 28, 2013 at 04:15:45PM -0800, H. Peter Anvin wrote:
> > > >>>>>>I'll try to get someone to test this tomorrow.
> > > >>>>Btw, you'd need to apply that other patch too
> > > >>>>
> > > >>>>http://marc.info/?l=xen-devel&m=136206183814547&w=2
> > > >>>>
> > > >>>>so that arch_flush_lazy_mmu_mode() has at least one caller on
> > x86_64.
> > > >>>Yeah, we already have that applied. It stops crashes in xen
> > > >>>environments so we pulled it in as a bugfix. Thanks though!
> > > >>>
> > > >>Who are "we"?
> > > >Sorry, Fedora. That patch has a link to a bug in it. We applied
> > the
> > > >patch for that bug. I'll apply Boris' patch on top and get the
> > same
> > > >people to test it.
> > >
> > > Josh, have you had a chance to test this?
> >
> > I've tested it on bare metal for a while now. No problems noticed at
> > all. I've not heard back from Krishna who was testing it in the Xen
> > environment. Krishna?
>
>
> Any updates?
No. I've still not heard from Krishna.
At this point I've tested it on bare metal quite a bit, and Konrad has
tested it on both bare metal and Xen. That should already cover the
case Krishna was going to test anyway. I suggest we move forward and
take the patch.
josh
On Wed, Mar 20, 2013 at 08:08:45PM -0400, Josh Boyer wrote:
> On Wed, Mar 20, 2013 at 06:53:55AM -0700, Boris Ostrovsky wrote:
> >
> > ----- [email protected] wrote:
> >
> > > On Wed, Mar 13, 2013 at 09:25:44AM -0400, Boris Ostrovsky wrote:
> > > > On 03/01/2013 07:14 AM, Josh Boyer wrote:
> > > > >On Thu, Feb 28, 2013 at 04:52:20PM -0800, H. Peter Anvin wrote:
> > > > >>On 02/28/2013 04:42 PM, Josh Boyer wrote:
> > > > >>>On Fri, Mar 01, 2013 at 01:36:29AM +0100, Borislav Petkov wrote:
> > > > >>>>On Thu, Feb 28, 2013 at 04:15:45PM -0800, H. Peter Anvin wrote:
> > > > >>>>>>I'll try to get someone to test this tomorrow.
> > > > >>>>Btw, you'd need to apply that other patch too
> > > > >>>>
> > > > >>>>http://marc.info/?l=xen-devel&m=136206183814547&w=2
> > > > >>>>
> > > > >>>>so that arch_flush_lazy_mmu_mode() has at least one caller on
> > > x86_64.
> > > > >>>Yeah, we already have that applied. It stops crashes in xen
> > > > >>>environments so we pulled it in as a bugfix. Thanks though!
> > > > >>>
> > > > >>Who are "we"?
> > > > >Sorry, Fedora. That patch has a link to a bug in it. We applied
> > > the
> > > > >patch for that bug. I'll apply Boris' patch on top and get the
> > > same
> > > > >people to test it.
> > > >
> > > > Josh, have you had a chance to test this?
> > >
> > > I've tested it on bare metal for a while now. No problems noticed at
> > > all. I've not heard back from Krishna who was testing it in the Xen
> > > environment. Krishna?
> >
> >
> > Any updates?
>
> No. I've still not heard from Krishna.
>
> At this point I've tested it on bare metal quite a bit, and Konrad has
> tested it on both bare metal and Xen. That should already cover the
> case Krishna was going to test anyway. I suggest we move forward and
> take the patch.
Peter?
Would you like me or Boris to clean up the two patches with the
appropiate Acks and send them to you?
>
> josh
Sure.
Konrad Rzeszutek Wilk <[email protected]> wrote:
>On Wed, Mar 20, 2013 at 08:08:45PM -0400, Josh Boyer wrote:
>> On Wed, Mar 20, 2013 at 06:53:55AM -0700, Boris Ostrovsky wrote:
>> >
>> > ----- [email protected] wrote:
>> >
>> > > On Wed, Mar 13, 2013 at 09:25:44AM -0400, Boris Ostrovsky wrote:
>> > > > On 03/01/2013 07:14 AM, Josh Boyer wrote:
>> > > > >On Thu, Feb 28, 2013 at 04:52:20PM -0800, H. Peter Anvin
>wrote:
>> > > > >>On 02/28/2013 04:42 PM, Josh Boyer wrote:
>> > > > >>>On Fri, Mar 01, 2013 at 01:36:29AM +0100, Borislav Petkov
>wrote:
>> > > > >>>>On Thu, Feb 28, 2013 at 04:15:45PM -0800, H. Peter Anvin
>wrote:
>> > > > >>>>>>I'll try to get someone to test this tomorrow.
>> > > > >>>>Btw, you'd need to apply that other patch too
>> > > > >>>>
>> > > > >>>>http://marc.info/?l=xen-devel&m=136206183814547&w=2
>> > > > >>>>
>> > > > >>>>so that arch_flush_lazy_mmu_mode() has at least one caller
>on
>> > > x86_64.
>> > > > >>>Yeah, we already have that applied. It stops crashes in xen
>> > > > >>>environments so we pulled it in as a bugfix. Thanks though!
>> > > > >>>
>> > > > >>Who are "we"?
>> > > > >Sorry, Fedora. That patch has a link to a bug in it. We
>applied
>> > > the
>> > > > >patch for that bug. I'll apply Boris' patch on top and get
>the
>> > > same
>> > > > >people to test it.
>> > > >
>> > > > Josh, have you had a chance to test this?
>> > >
>> > > I've tested it on bare metal for a while now. No problems
>noticed at
>> > > all. I've not heard back from Krishna who was testing it in the
>Xen
>> > > environment. Krishna?
>> >
>> >
>> > Any updates?
>>
>> No. I've still not heard from Krishna.
>>
>> At this point I've tested it on bare metal quite a bit, and Konrad
>has
>> tested it on both bare metal and Xen. That should already cover the
>> case Krishna was going to test anyway. I suggest we move forward and
>> take the patch.
>
>Peter?
>
>Would you like me or Boris to clean up the two patches with the
>appropiate Acks and send them to you?
>>
>> josh
--
Sent from my mobile phone. Please excuse brevity and lack of formatting.
From: Boris Ostrovsky <[email protected]>
Invoking arch_flush_lazy_mmu_mode() results in calls to
preempt_enable()/disable() which may have performance impact.
Since lazy MMU is not used on bare metal we can patch away
arch_flush_lazy_mmu_mode() so that it is never called in such
environment.
Signed-off-by: Boris Ostrovsky <[email protected]>
Tested-by: Josh Boyer <[email protected]>
Tested-by: Konrad Rzeszutek Wilk <[email protected]>
Acked-by: Borislav Petkov <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
---
arch/x86/include/asm/paravirt.h | 5 ++++-
arch/x86/include/asm/paravirt_types.h | 2 ++
arch/x86/kernel/paravirt.c | 25 +++++++++++++------------
arch/x86/lguest/boot.c | 1 +
arch/x86/xen/mmu.c | 1 +
5 files changed, 21 insertions(+), 13 deletions(-)
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 5edd174..7361e47 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -703,7 +703,10 @@ static inline void arch_leave_lazy_mmu_mode(void)
PVOP_VCALL0(pv_mmu_ops.lazy_mode.leave);
}
-void arch_flush_lazy_mmu_mode(void);
+static inline void arch_flush_lazy_mmu_mode(void)
+{
+ PVOP_VCALL0(pv_mmu_ops.lazy_mode.flush);
+}
static inline void __set_fixmap(unsigned /* enum fixed_addresses */ idx,
phys_addr_t phys, pgprot_t flags)
diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
index 142236e..b3b0ec1 100644
--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -91,6 +91,7 @@ struct pv_lazy_ops {
/* Set deferred update mode, used for batching operations. */
void (*enter)(void);
void (*leave)(void);
+ void (*flush)(void);
};
struct pv_time_ops {
@@ -679,6 +680,7 @@ void paravirt_end_context_switch(struct task_struct *next);
void paravirt_enter_lazy_mmu(void);
void paravirt_leave_lazy_mmu(void);
+void paravirt_flush_lazy_mmu(void);
void _paravirt_nop(void);
u32 _paravirt_ident_32(u32);
diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c
index 17fff18..8bfb335 100644
--- a/arch/x86/kernel/paravirt.c
+++ b/arch/x86/kernel/paravirt.c
@@ -263,6 +263,18 @@ void paravirt_leave_lazy_mmu(void)
leave_lazy(PARAVIRT_LAZY_MMU);
}
+void paravirt_flush_lazy_mmu(void)
+{
+ preempt_disable();
+
+ if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
+ arch_leave_lazy_mmu_mode();
+ arch_enter_lazy_mmu_mode();
+ }
+
+ preempt_enable();
+}
+
void paravirt_start_context_switch(struct task_struct *prev)
{
BUG_ON(preemptible());
@@ -292,18 +304,6 @@ enum paravirt_lazy_mode paravirt_get_lazy_mode(void)
return this_cpu_read(paravirt_lazy_mode);
}
-void arch_flush_lazy_mmu_mode(void)
-{
- preempt_disable();
-
- if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
- arch_leave_lazy_mmu_mode();
- arch_enter_lazy_mmu_mode();
- }
-
- preempt_enable();
-}
-
struct pv_info pv_info = {
.name = "bare hardware",
.paravirt_enabled = 0,
@@ -475,6 +475,7 @@ struct pv_mmu_ops pv_mmu_ops = {
.lazy_mode = {
.enter = paravirt_nop,
.leave = paravirt_nop,
+ .flush = paravirt_nop,
},
.set_fixmap = native_set_fixmap,
diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
index 1cbd89c..7114c63 100644
--- a/arch/x86/lguest/boot.c
+++ b/arch/x86/lguest/boot.c
@@ -1334,6 +1334,7 @@ __init void lguest_init(void)
pv_mmu_ops.read_cr3 = lguest_read_cr3;
pv_mmu_ops.lazy_mode.enter = paravirt_enter_lazy_mmu;
pv_mmu_ops.lazy_mode.leave = lguest_leave_lazy_mmu_mode;
+ pv_mmu_ops.lazy_mode.flush = paravirt_flush_lazy_mmu;
pv_mmu_ops.pte_update = lguest_pte_update;
pv_mmu_ops.pte_update_defer = lguest_pte_update;
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index e8e3493..f4f4105 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -2197,6 +2197,7 @@ static const struct pv_mmu_ops xen_mmu_ops __initconst = {
.lazy_mode = {
.enter = paravirt_enter_lazy_mmu,
.leave = xen_leave_lazy_mmu,
+ .flush = paravirt_flush_lazy_mmu,
},
.set_fixmap = xen_set_fixmap,
--
1.8.0.2
From: Samu Kallio <[email protected]>
In paravirtualized x86_64 kernels, vmalloc_fault may cause an oops
when lazy MMU updates are enabled, because set_pgd effects are being
deferred.
One instance of this problem is during process mm cleanup with memory
cgroups enabled. The chain of events is as follows:
- zap_pte_range enables lazy MMU updates
- zap_pte_range eventually calls mem_cgroup_charge_statistics,
which accesses the vmalloc'd mem_cgroup per-cpu stat area
- vmalloc_fault is triggered which tries to sync the corresponding
PGD entry with set_pgd, but the update is deferred
- vmalloc_fault oopses due to a mismatch in the PUD entries
The OOPs usually looks as so:
------------[ cut here ]------------
kernel BUG at arch/x86/mm/fault.c:396!
invalid opcode: 0000 [#1] SMP
.. snip ..
CPU 1
Pid: 10866, comm: httpd Not tainted 3.6.10-4.fc18.x86_64 #1
RIP: e030:[<ffffffff816271bf>] [<ffffffff816271bf>] vmalloc_fault+0x11f/0x208
.. snip ..
Call Trace:
[<ffffffff81627759>] do_page_fault+0x399/0x4b0
[<ffffffff81004f4c>] ? xen_mc_extend_args+0xec/0x110
[<ffffffff81624065>] page_fault+0x25/0x30
[<ffffffff81184d03>] ? mem_cgroup_charge_statistics.isra.13+0x13/0x50
[<ffffffff81186f78>] __mem_cgroup_uncharge_common+0xd8/0x350
[<ffffffff8118aac7>] mem_cgroup_uncharge_page+0x57/0x60
[<ffffffff8115fbc0>] page_remove_rmap+0xe0/0x150
[<ffffffff8115311a>] ? vm_normal_page+0x1a/0x80
[<ffffffff81153e61>] unmap_single_vma+0x531/0x870
[<ffffffff81154962>] unmap_vmas+0x52/0xa0
[<ffffffff81007442>] ? pte_mfn_to_pfn+0x72/0x100
[<ffffffff8115c8f8>] exit_mmap+0x98/0x170
[<ffffffff810050d9>] ? __raw_callee_save_xen_pmd_val+0x11/0x1e
[<ffffffff81059ce3>] mmput+0x83/0xf0
[<ffffffff810624c4>] exit_mm+0x104/0x130
[<ffffffff8106264a>] do_exit+0x15a/0x8c0
[<ffffffff810630ff>] do_group_exit+0x3f/0xa0
[<ffffffff81063177>] sys_exit_group+0x17/0x20
[<ffffffff8162bae9>] system_call_fastpath+0x16/0x1b
Calling arch_flush_lazy_mmu_mode immediately after set_pgd makes the
changes visible to the consistency checks.
CC: [email protected]
RedHat-Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=914737
Tested-by: Josh Boyer <[email protected]>
Reported-and-Tested-by: Krishna Raman <[email protected]>
Signed-off-by: Samu Kallio <[email protected]>
Tested-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
---
arch/x86/mm/fault.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 2b97525..0e88336 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -378,10 +378,12 @@ static noinline __kprobes int vmalloc_fault(unsigned long address)
if (pgd_none(*pgd_ref))
return -1;
- if (pgd_none(*pgd))
+ if (pgd_none(*pgd)) {
set_pgd(pgd, *pgd_ref);
- else
+ arch_flush_lazy_mmu_mode();
+ } else {
BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref));
+ }
/*
* Below here mismatches are bugs because these lower tables
--
1.8.0.2
On 03/23/2013 09:36 AM, Konrad Rzeszutek Wilk wrote:
> From: Boris Ostrovsky <[email protected]>
>
> Invoking arch_flush_lazy_mmu_mode() results in calls to
> preempt_enable()/disable() which may have performance impact.
>
> Since lazy MMU is not used on bare metal we can patch away
> arch_flush_lazy_mmu_mode() so that it is never called in such
> environment.
>
> Signed-off-by: Boris Ostrovsky <[email protected]>
> Tested-by: Josh Boyer <[email protected]>
> Tested-by: Konrad Rzeszutek Wilk <[email protected]>
> Acked-by: Borislav Petkov <[email protected]>
> Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Peter, what's the status of these two patches? They are not going into
3.9, right?
Thanks.
-boris
> ---
> arch/x86/include/asm/paravirt.h | 5 ++++-
> arch/x86/include/asm/paravirt_types.h | 2 ++
> arch/x86/kernel/paravirt.c | 25 +++++++++++++------------
> arch/x86/lguest/boot.c | 1 +
> arch/x86/xen/mmu.c | 1 +
> 5 files changed, 21 insertions(+), 13 deletions(-)
>
> diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
> index 5edd174..7361e47 100644
> --- a/arch/x86/include/asm/paravirt.h
> +++ b/arch/x86/include/asm/paravirt.h
> @@ -703,7 +703,10 @@ static inline void arch_leave_lazy_mmu_mode(void)
> PVOP_VCALL0(pv_mmu_ops.lazy_mode.leave);
> }
>
> -void arch_flush_lazy_mmu_mode(void);
> +static inline void arch_flush_lazy_mmu_mode(void)
> +{
> + PVOP_VCALL0(pv_mmu_ops.lazy_mode.flush);
> +}
>
> static inline void __set_fixmap(unsigned /* enum fixed_addresses */ idx,
> phys_addr_t phys, pgprot_t flags)
> diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
> index 142236e..b3b0ec1 100644
> --- a/arch/x86/include/asm/paravirt_types.h
> +++ b/arch/x86/include/asm/paravirt_types.h
> @@ -91,6 +91,7 @@ struct pv_lazy_ops {
> /* Set deferred update mode, used for batching operations. */
> void (*enter)(void);
> void (*leave)(void);
> + void (*flush)(void);
> };
>
> struct pv_time_ops {
> @@ -679,6 +680,7 @@ void paravirt_end_context_switch(struct task_struct *next);
>
> void paravirt_enter_lazy_mmu(void);
> void paravirt_leave_lazy_mmu(void);
> +void paravirt_flush_lazy_mmu(void);
>
> void _paravirt_nop(void);
> u32 _paravirt_ident_32(u32);
> diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c
> index 17fff18..8bfb335 100644
> --- a/arch/x86/kernel/paravirt.c
> +++ b/arch/x86/kernel/paravirt.c
> @@ -263,6 +263,18 @@ void paravirt_leave_lazy_mmu(void)
> leave_lazy(PARAVIRT_LAZY_MMU);
> }
>
> +void paravirt_flush_lazy_mmu(void)
> +{
> + preempt_disable();
> +
> + if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
> + arch_leave_lazy_mmu_mode();
> + arch_enter_lazy_mmu_mode();
> + }
> +
> + preempt_enable();
> +}
> +
> void paravirt_start_context_switch(struct task_struct *prev)
> {
> BUG_ON(preemptible());
> @@ -292,18 +304,6 @@ enum paravirt_lazy_mode paravirt_get_lazy_mode(void)
> return this_cpu_read(paravirt_lazy_mode);
> }
>
> -void arch_flush_lazy_mmu_mode(void)
> -{
> - preempt_disable();
> -
> - if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
> - arch_leave_lazy_mmu_mode();
> - arch_enter_lazy_mmu_mode();
> - }
> -
> - preempt_enable();
> -}
> -
> struct pv_info pv_info = {
> .name = "bare hardware",
> .paravirt_enabled = 0,
> @@ -475,6 +475,7 @@ struct pv_mmu_ops pv_mmu_ops = {
> .lazy_mode = {
> .enter = paravirt_nop,
> .leave = paravirt_nop,
> + .flush = paravirt_nop,
> },
>
> .set_fixmap = native_set_fixmap,
> diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
> index 1cbd89c..7114c63 100644
> --- a/arch/x86/lguest/boot.c
> +++ b/arch/x86/lguest/boot.c
> @@ -1334,6 +1334,7 @@ __init void lguest_init(void)
> pv_mmu_ops.read_cr3 = lguest_read_cr3;
> pv_mmu_ops.lazy_mode.enter = paravirt_enter_lazy_mmu;
> pv_mmu_ops.lazy_mode.leave = lguest_leave_lazy_mmu_mode;
> + pv_mmu_ops.lazy_mode.flush = paravirt_flush_lazy_mmu;
> pv_mmu_ops.pte_update = lguest_pte_update;
> pv_mmu_ops.pte_update_defer = lguest_pte_update;
>
> diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
> index e8e3493..f4f4105 100644
> --- a/arch/x86/xen/mmu.c
> +++ b/arch/x86/xen/mmu.c
> @@ -2197,6 +2197,7 @@ static const struct pv_mmu_ops xen_mmu_ops __initconst = {
> .lazy_mode = {
> .enter = paravirt_enter_lazy_mmu,
> .leave = xen_leave_lazy_mmu,
> + .flush = paravirt_flush_lazy_mmu,
> },
>
> .set_fixmap = xen_set_fixmap,
Commit-ID: 1160c2779b826c6f5c08e5cc542de58fd1f667d5
Gitweb: http://git.kernel.org/tip/1160c2779b826c6f5c08e5cc542de58fd1f667d5
Author: Samu Kallio <[email protected]>
AuthorDate: Sat, 23 Mar 2013 09:36:35 -0400
Committer: H. Peter Anvin <[email protected]>
CommitDate: Wed, 10 Apr 2013 11:25:07 -0700
x86, mm, paravirt: Fix vmalloc_fault oops during lazy MMU updates
In paravirtualized x86_64 kernels, vmalloc_fault may cause an oops
when lazy MMU updates are enabled, because set_pgd effects are being
deferred.
One instance of this problem is during process mm cleanup with memory
cgroups enabled. The chain of events is as follows:
- zap_pte_range enables lazy MMU updates
- zap_pte_range eventually calls mem_cgroup_charge_statistics,
which accesses the vmalloc'd mem_cgroup per-cpu stat area
- vmalloc_fault is triggered which tries to sync the corresponding
PGD entry with set_pgd, but the update is deferred
- vmalloc_fault oopses due to a mismatch in the PUD entries
The OOPs usually looks as so:
------------[ cut here ]------------
kernel BUG at arch/x86/mm/fault.c:396!
invalid opcode: 0000 [#1] SMP
.. snip ..
CPU 1
Pid: 10866, comm: httpd Not tainted 3.6.10-4.fc18.x86_64 #1
RIP: e030:[<ffffffff816271bf>] [<ffffffff816271bf>] vmalloc_fault+0x11f/0x208
.. snip ..
Call Trace:
[<ffffffff81627759>] do_page_fault+0x399/0x4b0
[<ffffffff81004f4c>] ? xen_mc_extend_args+0xec/0x110
[<ffffffff81624065>] page_fault+0x25/0x30
[<ffffffff81184d03>] ? mem_cgroup_charge_statistics.isra.13+0x13/0x50
[<ffffffff81186f78>] __mem_cgroup_uncharge_common+0xd8/0x350
[<ffffffff8118aac7>] mem_cgroup_uncharge_page+0x57/0x60
[<ffffffff8115fbc0>] page_remove_rmap+0xe0/0x150
[<ffffffff8115311a>] ? vm_normal_page+0x1a/0x80
[<ffffffff81153e61>] unmap_single_vma+0x531/0x870
[<ffffffff81154962>] unmap_vmas+0x52/0xa0
[<ffffffff81007442>] ? pte_mfn_to_pfn+0x72/0x100
[<ffffffff8115c8f8>] exit_mmap+0x98/0x170
[<ffffffff810050d9>] ? __raw_callee_save_xen_pmd_val+0x11/0x1e
[<ffffffff81059ce3>] mmput+0x83/0xf0
[<ffffffff810624c4>] exit_mm+0x104/0x130
[<ffffffff8106264a>] do_exit+0x15a/0x8c0
[<ffffffff810630ff>] do_group_exit+0x3f/0xa0
[<ffffffff81063177>] sys_exit_group+0x17/0x20
[<ffffffff8162bae9>] system_call_fastpath+0x16/0x1b
Calling arch_flush_lazy_mmu_mode immediately after set_pgd makes the
changes visible to the consistency checks.
Cc: <[email protected]>
RedHat-Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=914737
Tested-by: Josh Boyer <[email protected]>
Reported-and-Tested-by: Krishna Raman <[email protected]>
Signed-off-by: Samu Kallio <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Tested-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: H. Peter Anvin <[email protected]>
---
arch/x86/mm/fault.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 2b97525..0e88336 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -378,10 +378,12 @@ static noinline __kprobes int vmalloc_fault(unsigned long address)
if (pgd_none(*pgd_ref))
return -1;
- if (pgd_none(*pgd))
+ if (pgd_none(*pgd)) {
set_pgd(pgd, *pgd_ref);
- else
+ arch_flush_lazy_mmu_mode();
+ } else {
BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref));
+ }
/*
* Below here mismatches are bugs because these lower tables
Commit-ID: 511ba86e1d386f671084b5d0e6f110bb30b8eeb2
Gitweb: http://git.kernel.org/tip/511ba86e1d386f671084b5d0e6f110bb30b8eeb2
Author: Boris Ostrovsky <[email protected]>
AuthorDate: Sat, 23 Mar 2013 09:36:36 -0400
Committer: H. Peter Anvin <[email protected]>
CommitDate: Wed, 10 Apr 2013 11:25:10 -0700
x86, mm: Patch out arch_flush_lazy_mmu_mode() when running on bare metal
Invoking arch_flush_lazy_mmu_mode() results in calls to
preempt_enable()/disable() which may have performance impact.
Since lazy MMU is not used on bare metal we can patch away
arch_flush_lazy_mmu_mode() so that it is never called in such
environment.
[ hpa: the previous patch "Fix vmalloc_fault oops during lazy MMU
updates" may cause a minor performance regression on
bare metal. This patch resolves that performance regression. It is
somewhat unclear to me if this is a good -stable candidate. ]
Signed-off-by: Boris Ostrovsky <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Tested-by: Josh Boyer <[email protected]>
Tested-by: Konrad Rzeszutek Wilk <[email protected]>
Acked-by: Borislav Petkov <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: H. Peter Anvin <[email protected]>
Cc: <[email protected]> SEE NOTE ABOVE
---
arch/x86/include/asm/paravirt.h | 5 ++++-
arch/x86/include/asm/paravirt_types.h | 2 ++
arch/x86/kernel/paravirt.c | 25 +++++++++++++------------
arch/x86/lguest/boot.c | 1 +
arch/x86/xen/mmu.c | 1 +
5 files changed, 21 insertions(+), 13 deletions(-)
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 5edd174..7361e47 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -703,7 +703,10 @@ static inline void arch_leave_lazy_mmu_mode(void)
PVOP_VCALL0(pv_mmu_ops.lazy_mode.leave);
}
-void arch_flush_lazy_mmu_mode(void);
+static inline void arch_flush_lazy_mmu_mode(void)
+{
+ PVOP_VCALL0(pv_mmu_ops.lazy_mode.flush);
+}
static inline void __set_fixmap(unsigned /* enum fixed_addresses */ idx,
phys_addr_t phys, pgprot_t flags)
diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
index 142236e..b3b0ec1 100644
--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -91,6 +91,7 @@ struct pv_lazy_ops {
/* Set deferred update mode, used for batching operations. */
void (*enter)(void);
void (*leave)(void);
+ void (*flush)(void);
};
struct pv_time_ops {
@@ -679,6 +680,7 @@ void paravirt_end_context_switch(struct task_struct *next);
void paravirt_enter_lazy_mmu(void);
void paravirt_leave_lazy_mmu(void);
+void paravirt_flush_lazy_mmu(void);
void _paravirt_nop(void);
u32 _paravirt_ident_32(u32);
diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c
index 17fff18..8bfb335 100644
--- a/arch/x86/kernel/paravirt.c
+++ b/arch/x86/kernel/paravirt.c
@@ -263,6 +263,18 @@ void paravirt_leave_lazy_mmu(void)
leave_lazy(PARAVIRT_LAZY_MMU);
}
+void paravirt_flush_lazy_mmu(void)
+{
+ preempt_disable();
+
+ if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
+ arch_leave_lazy_mmu_mode();
+ arch_enter_lazy_mmu_mode();
+ }
+
+ preempt_enable();
+}
+
void paravirt_start_context_switch(struct task_struct *prev)
{
BUG_ON(preemptible());
@@ -292,18 +304,6 @@ enum paravirt_lazy_mode paravirt_get_lazy_mode(void)
return this_cpu_read(paravirt_lazy_mode);
}
-void arch_flush_lazy_mmu_mode(void)
-{
- preempt_disable();
-
- if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
- arch_leave_lazy_mmu_mode();
- arch_enter_lazy_mmu_mode();
- }
-
- preempt_enable();
-}
-
struct pv_info pv_info = {
.name = "bare hardware",
.paravirt_enabled = 0,
@@ -475,6 +475,7 @@ struct pv_mmu_ops pv_mmu_ops = {
.lazy_mode = {
.enter = paravirt_nop,
.leave = paravirt_nop,
+ .flush = paravirt_nop,
},
.set_fixmap = native_set_fixmap,
diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
index 1cbd89c..7114c63 100644
--- a/arch/x86/lguest/boot.c
+++ b/arch/x86/lguest/boot.c
@@ -1334,6 +1334,7 @@ __init void lguest_init(void)
pv_mmu_ops.read_cr3 = lguest_read_cr3;
pv_mmu_ops.lazy_mode.enter = paravirt_enter_lazy_mmu;
pv_mmu_ops.lazy_mode.leave = lguest_leave_lazy_mmu_mode;
+ pv_mmu_ops.lazy_mode.flush = paravirt_flush_lazy_mmu;
pv_mmu_ops.pte_update = lguest_pte_update;
pv_mmu_ops.pte_update_defer = lguest_pte_update;
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 6afbb2c..2f5d687 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -2196,6 +2196,7 @@ static const struct pv_mmu_ops xen_mmu_ops __initconst = {
.lazy_mode = {
.enter = paravirt_enter_lazy_mmu,
.leave = xen_leave_lazy_mmu,
+ .flush = paravirt_flush_lazy_mmu,
},
.set_fixmap = xen_set_fixmap,
On 04/10/2013 08:30 PM, tip-bot for Boris Ostrovsky wrote:
> Commit-ID: 511ba86e1d386f671084b5d0e6f110bb30b8eeb2
> Gitweb: http://git.kernel.org/tip/511ba86e1d386f671084b5d0e6f110bb30b8eeb2
> Author: Boris Ostrovsky <[email protected]>
> AuthorDate: Sat, 23 Mar 2013 09:36:36 -0400
> Committer: H. Peter Anvin <[email protected]>
> CommitDate: Wed, 10 Apr 2013 11:25:10 -0700
>
> x86, mm: Patch out arch_flush_lazy_mmu_mode() when running on bare metal
>
> Invoking arch_flush_lazy_mmu_mode() results in calls to
> preempt_enable()/disable() which may have performance impact.
>
> Since lazy MMU is not used on bare metal we can patch away
> arch_flush_lazy_mmu_mode() so that it is never called in such
> environment.
>
> [ hpa: the previous patch "Fix vmalloc_fault oops during lazy MMU
> updates" may cause a minor performance regression on
> bare metal. This patch resolves that performance regression. It is
> somewhat unclear to me if this is a good -stable candidate. ]
I think this
https://lkml.org/lkml/2013/2/26/420
was also part of lazy mmu set of patches but is missing in the latest
batch of
commits.
-boris
I am sorry for top posting. Haven't figured out how to reply inline in the Android email client.
Boris, could you repost it please? If I recall correctly the only question on that patch was whether to put [email protected] on it and it was decided (by Greg) that was OK.
It is not an urgent patch though as it is only exposed with debug kernels.
Boris Ostrovsky <[email protected]> wrote:
>On 04/10/2013 08:30 PM, tip-bot for Boris Ostrovsky wrote:
>> Commit-ID: 511ba86e1d386f671084b5d0e6f110bb30b8eeb2
>> Gitweb: http://git.kernel.org/tip/511ba86e1d386f671084b5d0e6f110bb30b8eeb2
>> Author: Boris Ostrovsky <[email protected]>
>> AuthorDate: Sat, 23 Mar 2013 09:36:36 -0400
>> Committer: H. Peter Anvin <[email protected]>
>> CommitDate: Wed, 10 Apr 2013 11:25:10 -0700
>>
>> x86, mm: Patch out arch_flush_lazy_mmu_mode() when running on bare metal
>>
>> Invoking arch_flush_lazy_mmu_mode() results in calls to
>> preempt_enable()/disable() which may have performance impact.
>>
>> Since lazy MMU is not used on bare metal we can patch away
>> arch_flush_lazy_mmu_mode() so that it is never called in such
>> environment.
>>
>> [ hpa: the previous patch "Fix vmalloc_fault oops during lazy MMU
>> updates" may cause a minor performance regression on
>> bare metal. This patch resolves that performance regression. It is
>> somewhat unclear to me if this is a good -stable candidate. ]
>
>I think this
>
> https://lkml.org/lkml/2013/2/26/420
>
>was also part of lazy mmu set of patches but is missing in the latest
>batch of
>commits.
>
>-boris
>
????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay? ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m???? ????????I?