* [PATCH 0/2] A couple of EL2 refcounts fixes
@ 2021-10-04 9:03 Quentin Perret
2021-10-04 9:03 ` [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount Quentin Perret
` (2 more replies)
0 siblings, 3 replies; 7+ messages in thread
From: Quentin Perret @ 2021-10-04 9:03 UTC (permalink / raw)
To: Marc Zyngier, James Morse, Alexandru Elisei, Suzuki K Poulose,
Catalin Marinas, Will Deacon, Quentin Perret, Fuad Tabba,
David Brazdil, linux-arm-kernel, kvmarm, linux-kernel
Cc: kernel-team
Hi all,
This addresses a couple of issues Will has found with the refcounting of
page-tables at EL2. Patch 01 fixes a nasty bug, and probably wants to go
in -stable. Patch 02 fixes a small inconsistency which made it harder to
find refcount-related bugs at EL2.
Feedback welcome !
Thanks,
Quentin
Quentin Perret (2):
KVM: arm64: Fix host stage-2 PGD refcount
KVM: arm64: Report corrupted refcount at EL2
arch/arm64/kvm/hyp/include/nvhe/gfp.h | 1 +
arch/arm64/kvm/hyp/nvhe/mem_protect.c | 6 +++++-
arch/arm64/kvm/hyp/nvhe/page_alloc.c | 15 +++++++++++++++
3 files changed, 21 insertions(+), 1 deletion(-)
--
2.33.0.800.g4c38ced690-goog
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount
2021-10-04 9:03 [PATCH 0/2] A couple of EL2 refcounts fixes Quentin Perret
@ 2021-10-04 9:03 ` Quentin Perret
2021-10-04 9:55 ` Marc Zyngier
2021-10-04 9:03 ` [PATCH 2/2] KVM: arm64: Report corrupted refcount at EL2 Quentin Perret
2021-10-04 9:39 ` [PATCH 0/2] A couple of EL2 refcounts fixes Will Deacon
2 siblings, 1 reply; 7+ messages in thread
From: Quentin Perret @ 2021-10-04 9:03 UTC (permalink / raw)
To: Marc Zyngier, James Morse, Alexandru Elisei, Suzuki K Poulose,
Catalin Marinas, Will Deacon, Quentin Perret, Fuad Tabba,
David Brazdil, linux-arm-kernel, kvmarm, linux-kernel
Cc: kernel-team
The KVM page-table library refcounts the pages of concatenated stage-2
PGDs individually. However, the host's stage-2 PGD is currently managed
by EL2 as a single high-order compound page, which can cause the
refcount of the tail pages to reach 0 when they really shouldn't, hence
corrupting the page-table.
Fix this by introducing a new hyp_split_page() helper in the EL2 page
allocator (matching EL1's split_page() function), and make use of it
from host_s2_zalloc_page().
Fixes: 1025c8c0c6ac ("KVM: arm64: Wrap the host with a stage 2")
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Quentin Perret <qperret@google.com>
---
arch/arm64/kvm/hyp/include/nvhe/gfp.h | 1 +
arch/arm64/kvm/hyp/nvhe/mem_protect.c | 6 +++++-
arch/arm64/kvm/hyp/nvhe/page_alloc.c | 14 ++++++++++++++
3 files changed, 20 insertions(+), 1 deletion(-)
diff --git a/arch/arm64/kvm/hyp/include/nvhe/gfp.h b/arch/arm64/kvm/hyp/include/nvhe/gfp.h
index fb0f523d1492..0a048dc06a7d 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/gfp.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/gfp.h
@@ -24,6 +24,7 @@ struct hyp_pool {
/* Allocation */
void *hyp_alloc_pages(struct hyp_pool *pool, unsigned short order);
+void hyp_split_page(struct hyp_page *page);
void hyp_get_page(struct hyp_pool *pool, void *addr);
void hyp_put_page(struct hyp_pool *pool, void *addr);
diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
index bacd493a4eac..93a79736c283 100644
--- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c
+++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
@@ -35,7 +35,11 @@ const u8 pkvm_hyp_id = 1;
static void *host_s2_zalloc_pages_exact(size_t size)
{
- return hyp_alloc_pages(&host_s2_pool, get_order(size));
+ void *addr = hyp_alloc_pages(&host_s2_pool, get_order(size));
+
+ hyp_split_page(hyp_virt_to_page(addr));
+
+ return addr;
}
static void *host_s2_zalloc_page(void *pool)
diff --git a/arch/arm64/kvm/hyp/nvhe/page_alloc.c b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
index 41fc25bdfb34..a6e874e61a40 100644
--- a/arch/arm64/kvm/hyp/nvhe/page_alloc.c
+++ b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
@@ -193,6 +193,20 @@ void hyp_get_page(struct hyp_pool *pool, void *addr)
hyp_spin_unlock(&pool->lock);
}
+void hyp_split_page(struct hyp_page *p)
+{
+ unsigned short order = p->order;
+ unsigned int i;
+
+ p->order = 0;
+ for (i = 1; i < (1 << order); i++) {
+ struct hyp_page *tail = p + i;
+
+ tail->order = 0;
+ hyp_set_page_refcounted(tail);
+ }
+}
+
void *hyp_alloc_pages(struct hyp_pool *pool, unsigned short order)
{
unsigned short i = order;
--
2.33.0.800.g4c38ced690-goog
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply related [flat|nested] 7+ messages in thread
* [PATCH 2/2] KVM: arm64: Report corrupted refcount at EL2
2021-10-04 9:03 [PATCH 0/2] A couple of EL2 refcounts fixes Quentin Perret
2021-10-04 9:03 ` [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount Quentin Perret
@ 2021-10-04 9:03 ` Quentin Perret
2021-10-04 9:39 ` [PATCH 0/2] A couple of EL2 refcounts fixes Will Deacon
2 siblings, 0 replies; 7+ messages in thread
From: Quentin Perret @ 2021-10-04 9:03 UTC (permalink / raw)
To: Marc Zyngier, James Morse, Alexandru Elisei, Suzuki K Poulose,
Catalin Marinas, Will Deacon, Quentin Perret, Fuad Tabba,
David Brazdil, linux-arm-kernel, kvmarm, linux-kernel
Cc: kernel-team
Some of the refcount manipulation helpers used at EL2 are instrumented
to catch a corrupt state, but not all of them are treated equally. Let's
make things more consistent by instrumenting hyp_page_ref_dec_and_test()
as well.
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Quentin Perret <qperret@google.com>
---
arch/arm64/kvm/hyp/nvhe/page_alloc.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/arm64/kvm/hyp/nvhe/page_alloc.c b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
index a6e874e61a40..0bd7701ad1df 100644
--- a/arch/arm64/kvm/hyp/nvhe/page_alloc.c
+++ b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
@@ -152,6 +152,7 @@ static inline void hyp_page_ref_inc(struct hyp_page *p)
static inline int hyp_page_ref_dec_and_test(struct hyp_page *p)
{
+ BUG_ON(!p->refcount);
p->refcount--;
return (p->refcount == 0);
}
--
2.33.0.800.g4c38ced690-goog
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH 0/2] A couple of EL2 refcounts fixes
2021-10-04 9:03 [PATCH 0/2] A couple of EL2 refcounts fixes Quentin Perret
2021-10-04 9:03 ` [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount Quentin Perret
2021-10-04 9:03 ` [PATCH 2/2] KVM: arm64: Report corrupted refcount at EL2 Quentin Perret
@ 2021-10-04 9:39 ` Will Deacon
2 siblings, 0 replies; 7+ messages in thread
From: Will Deacon @ 2021-10-04 9:39 UTC (permalink / raw)
To: Quentin Perret
Cc: kernel-team, Marc Zyngier, linux-kernel, linux-arm-kernel,
Catalin Marinas, kvmarm
On Mon, Oct 04, 2021 at 10:03:12AM +0100, Quentin Perret wrote:
> This addresses a couple of issues Will has found with the refcounting of
> page-tables at EL2. Patch 01 fixes a nasty bug, and probably wants to go
> in -stable. Patch 02 fixes a small inconsistency which made it harder to
> find refcount-related bugs at EL2.
>
> Feedback welcome !
For both patches:
Acked-by: Will Deacon <will@kernel.org>
Thanks!
Will
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount
2021-10-04 9:03 ` [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount Quentin Perret
@ 2021-10-04 9:55 ` Marc Zyngier
2021-10-04 10:05 ` Quentin Perret
0 siblings, 1 reply; 7+ messages in thread
From: Marc Zyngier @ 2021-10-04 9:55 UTC (permalink / raw)
To: Quentin Perret
Cc: kernel-team, Will Deacon, Catalin Marinas, linux-kernel,
linux-arm-kernel, kvmarm
Hi Quentin,
On Mon, 04 Oct 2021 10:03:13 +0100,
Quentin Perret <qperret@google.com> wrote:
>
> The KVM page-table library refcounts the pages of concatenated stage-2
> PGDs individually. However, the host's stage-2 PGD is currently managed
> by EL2 as a single high-order compound page, which can cause the
> refcount of the tail pages to reach 0 when they really shouldn't, hence
> corrupting the page-table.
nit: this comment only applies to the protected mode, right? As far as
I can tell, 'classic' KVM is just fine.
>
> Fix this by introducing a new hyp_split_page() helper in the EL2 page
> allocator (matching EL1's split_page() function), and make use of it
uber nit: split_page() is not an EL1 function. more of a standard
kernel function.
> from host_s2_zalloc_page().
>
> Fixes: 1025c8c0c6ac ("KVM: arm64: Wrap the host with a stage 2")
> Suggested-by: Will Deacon <will@kernel.org>
> Signed-off-by: Quentin Perret <qperret@google.com>
> ---
> arch/arm64/kvm/hyp/include/nvhe/gfp.h | 1 +
> arch/arm64/kvm/hyp/nvhe/mem_protect.c | 6 +++++-
> arch/arm64/kvm/hyp/nvhe/page_alloc.c | 14 ++++++++++++++
> 3 files changed, 20 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm64/kvm/hyp/include/nvhe/gfp.h b/arch/arm64/kvm/hyp/include/nvhe/gfp.h
> index fb0f523d1492..0a048dc06a7d 100644
> --- a/arch/arm64/kvm/hyp/include/nvhe/gfp.h
> +++ b/arch/arm64/kvm/hyp/include/nvhe/gfp.h
> @@ -24,6 +24,7 @@ struct hyp_pool {
>
> /* Allocation */
> void *hyp_alloc_pages(struct hyp_pool *pool, unsigned short order);
> +void hyp_split_page(struct hyp_page *page);
> void hyp_get_page(struct hyp_pool *pool, void *addr);
> void hyp_put_page(struct hyp_pool *pool, void *addr);
>
> diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> index bacd493a4eac..93a79736c283 100644
> --- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> +++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> @@ -35,7 +35,11 @@ const u8 pkvm_hyp_id = 1;
>
> static void *host_s2_zalloc_pages_exact(size_t size)
> {
> - return hyp_alloc_pages(&host_s2_pool, get_order(size));
> + void *addr = hyp_alloc_pages(&host_s2_pool, get_order(size));
> +
> + hyp_split_page(hyp_virt_to_page(addr));
The only reason this doesn't lead to a subsequent memory leak is that
concatenated page tables are always a power of two, right?
If so, that deserves a comment, because I don't think this works in
the general case unless you actively free the pages that are between
size and (1 << order).
> +
> + return addr;
> }
>
> static void *host_s2_zalloc_page(void *pool)
> diff --git a/arch/arm64/kvm/hyp/nvhe/page_alloc.c b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
> index 41fc25bdfb34..a6e874e61a40 100644
> --- a/arch/arm64/kvm/hyp/nvhe/page_alloc.c
> +++ b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
> @@ -193,6 +193,20 @@ void hyp_get_page(struct hyp_pool *pool, void *addr)
> hyp_spin_unlock(&pool->lock);
> }
>
> +void hyp_split_page(struct hyp_page *p)
> +{
> + unsigned short order = p->order;
> + unsigned int i;
> +
> + p->order = 0;
> + for (i = 1; i < (1 << order); i++) {
> + struct hyp_page *tail = p + i;
> +
> + tail->order = 0;
> + hyp_set_page_refcounted(tail);
> + }
> +}
> +
> void *hyp_alloc_pages(struct hyp_pool *pool, unsigned short order)
> {
> unsigned short i = order;
Thanks,
M.
--
Without deviation from the norm, progress is not possible.
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount
2021-10-04 9:55 ` Marc Zyngier
@ 2021-10-04 10:05 ` Quentin Perret
0 siblings, 0 replies; 7+ messages in thread
From: Quentin Perret @ 2021-10-04 10:05 UTC (permalink / raw)
To: Marc Zyngier
Cc: kernel-team, Will Deacon, Catalin Marinas, linux-kernel,
linux-arm-kernel, kvmarm
Hey Marc,
On Monday 04 Oct 2021 at 10:55:13 (+0100), Marc Zyngier wrote:
> Hi Quentin,
>
> On Mon, 04 Oct 2021 10:03:13 +0100,
> Quentin Perret <qperret@google.com> wrote:
> >
> > The KVM page-table library refcounts the pages of concatenated stage-2
> > PGDs individually. However, the host's stage-2 PGD is currently managed
> > by EL2 as a single high-order compound page, which can cause the
> > refcount of the tail pages to reach 0 when they really shouldn't, hence
> > corrupting the page-table.
>
> nit: this comment only applies to the protected mode, right? As far as
> I can tell, 'classic' KVM is just fine.
Correct, this really only applies to the host stage-2, which implies
we're in protected mode. I'll make that a bit more explicit.
> > Fix this by introducing a new hyp_split_page() helper in the EL2 page
> > allocator (matching EL1's split_page() function), and make use of it
>
> uber nit: split_page() is not an EL1 function. more of a standard
> kernel function.
Fair enough :)
> > from host_s2_zalloc_page().
> >
> > Fixes: 1025c8c0c6ac ("KVM: arm64: Wrap the host with a stage 2")
> > Suggested-by: Will Deacon <will@kernel.org>
> > Signed-off-by: Quentin Perret <qperret@google.com>
> > ---
> > arch/arm64/kvm/hyp/include/nvhe/gfp.h | 1 +
> > arch/arm64/kvm/hyp/nvhe/mem_protect.c | 6 +++++-
> > arch/arm64/kvm/hyp/nvhe/page_alloc.c | 14 ++++++++++++++
> > 3 files changed, 20 insertions(+), 1 deletion(-)
> >
> > diff --git a/arch/arm64/kvm/hyp/include/nvhe/gfp.h b/arch/arm64/kvm/hyp/include/nvhe/gfp.h
> > index fb0f523d1492..0a048dc06a7d 100644
> > --- a/arch/arm64/kvm/hyp/include/nvhe/gfp.h
> > +++ b/arch/arm64/kvm/hyp/include/nvhe/gfp.h
> > @@ -24,6 +24,7 @@ struct hyp_pool {
> >
> > /* Allocation */
> > void *hyp_alloc_pages(struct hyp_pool *pool, unsigned short order);
> > +void hyp_split_page(struct hyp_page *page);
> > void hyp_get_page(struct hyp_pool *pool, void *addr);
> > void hyp_put_page(struct hyp_pool *pool, void *addr);
> >
> > diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> > index bacd493a4eac..93a79736c283 100644
> > --- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> > +++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> > @@ -35,7 +35,11 @@ const u8 pkvm_hyp_id = 1;
> >
> > static void *host_s2_zalloc_pages_exact(size_t size)
> > {
> > - return hyp_alloc_pages(&host_s2_pool, get_order(size));
> > + void *addr = hyp_alloc_pages(&host_s2_pool, get_order(size));
> > +
> > + hyp_split_page(hyp_virt_to_page(addr));
>
> The only reason this doesn't lead to a subsequent memory leak is that
> concatenated page tables are always a power of two, right?
Indeed, and also because the host stage-2 is _never_ freed, so that's
not memory we're going to reclaim anyway -- we don't have an
implementation of ->free_pages_exact() in the host stage-2 mm_ops.
> If so, that deserves a comment, because I don't think this works in
> the general case unless you actively free the pages that are between
> size and (1 << order).
Ack, that'll probably confuse me too in a few weeks, so a comment won't
hurt. I'll re-spin shortly.
Thanks,
Quentin
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH 2/2] KVM: arm64: Report corrupted refcount at EL2
2021-10-05 9:01 Quentin Perret
@ 2021-10-05 9:01 ` Quentin Perret
0 siblings, 0 replies; 7+ messages in thread
From: Quentin Perret @ 2021-10-05 9:01 UTC (permalink / raw)
To: Marc Zyngier, James Morse, Alexandru Elisei, Suzuki K Poulose,
Catalin Marinas, Will Deacon, Quentin Perret, Fuad Tabba,
David Brazdil, linux-arm-kernel, kvmarm, linux-kernel
Cc: kernel-team
Some of the refcount manipulation helpers used at EL2 are instrumented
to catch a corrupt state, but not all of them are treated equally. Let's
make things more consistent by instrumenting hyp_page_ref_dec_and_test()
as well.
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Quentin Perret <qperret@google.com>
---
arch/arm64/kvm/hyp/nvhe/page_alloc.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/arm64/kvm/hyp/nvhe/page_alloc.c b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
index a6e874e61a40..0bd7701ad1df 100644
--- a/arch/arm64/kvm/hyp/nvhe/page_alloc.c
+++ b/arch/arm64/kvm/hyp/nvhe/page_alloc.c
@@ -152,6 +152,7 @@ static inline void hyp_page_ref_inc(struct hyp_page *p)
static inline int hyp_page_ref_dec_and_test(struct hyp_page *p)
{
+ BUG_ON(!p->refcount);
p->refcount--;
return (p->refcount == 0);
}
--
2.33.0.800.g4c38ced690-goog
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
^ permalink raw reply related [flat|nested] 7+ messages in thread
end of thread, other threads:[~2021-10-05 9:02 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-10-04 9:03 [PATCH 0/2] A couple of EL2 refcounts fixes Quentin Perret
2021-10-04 9:03 ` [PATCH 1/2] KVM: arm64: Fix host stage-2 PGD refcount Quentin Perret
2021-10-04 9:55 ` Marc Zyngier
2021-10-04 10:05 ` Quentin Perret
2021-10-04 9:03 ` [PATCH 2/2] KVM: arm64: Report corrupted refcount at EL2 Quentin Perret
2021-10-04 9:39 ` [PATCH 0/2] A couple of EL2 refcounts fixes Will Deacon
2021-10-05 9:01 Quentin Perret
2021-10-05 9:01 ` [PATCH 2/2] KVM: arm64: Report corrupted refcount at EL2 Quentin Perret
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).