Re: [Intel-gfx] [PATCH 0/6] KVM: Remove uses of struct page from x86 and arm64 MMU
Excerpts from Paolo Bonzini's message of June 25, 2021 1:35 am: > On 24/06/21 14:57, Nicholas Piggin wrote: >> KVM: Fix page ref underflow for regions with valid but non-refcounted pages > > It doesn't really fix the underflow, it disallows mapping them in the > first place. Since in principle things can break, I'd rather be > explicit, so let's go with "KVM: do not allow mapping valid but > non-reference-counted pages". > >> It's possible to create a region which maps valid but non-refcounted >> pages (e.g., tail pages of non-compound higher order allocations). These >> host pages can then be returned by gfn_to_page, gfn_to_pfn, etc., family >> of APIs, which take a reference to the page, which takes it from 0 to 1. >> When the reference is dropped, this will free the page incorrectly. >> >> Fix this by only taking a reference on the page if it was non-zero, > > s/on the page/on valid pages/ (makes clear that invalid pages are fine > without refcounting). That seems okay, you can adjust the title or changelog as you like. > Thank you *so* much, I'm awful at Linux mm. Glad to help. Easy to see why you were taking this approach because the API really does need to be improved and even a pretty intwined with mm subsystem like KVM shouldn't _really_ be doing this kind of trick (and it should go away when old API is removed). Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 0/6] KVM: Remove uses of struct page from x86 and arm64 MMU
Excerpts from Paolo Bonzini's message of June 24, 2021 10:41 pm:
> On 24/06/21 13:42, Nicholas Piggin wrote:
>> Excerpts from Nicholas Piggin's message of June 24, 2021 8:34 pm:
>>> Excerpts from David Stevens's message of June 24, 2021 1:57 pm:
>>>> KVM supports mapping VM_IO and VM_PFNMAP memory into the guest by using
>>>> follow_pte in gfn_to_pfn. However, the resolved pfns may not have
>>>> assoicated struct pages, so they should not be passed to pfn_to_page.
>>>> This series removes such calls from the x86 and arm64 secondary MMU. To
>>>> do this, this series modifies gfn_to_pfn to return a struct page in
>>>> addition to a pfn, if the hva was resolved by gup. This allows the
>>>> caller to call put_page only when necessated by gup.
>>>>
>>>> This series provides a helper function that unwraps the new return type
>>>> of gfn_to_pfn to provide behavior identical to the old behavior. As I
>>>> have no hardware to test powerpc/mips changes, the function is used
>>>> there for minimally invasive changes. Additionally, as gfn_to_page and
>>>> gfn_to_pfn_cache are not integrated with mmu notifier, they cannot be
>>>> easily changed over to only use pfns.
>>>>
>>>> This addresses CVE-2021-22543 on x86 and arm64.
>>>
>>> Does this fix the problem? (untested I don't have a POC setup at hand,
>>> but at least in concept)
>>
>> This one actually compiles at least. Unfortunately I don't have much
>> time in the near future to test, and I only just found out about this
>> CVE a few hours ago.
>
> And it also works (the reproducer gets an infinite stream of userspace
> exits and especially does not crash). We can still go for David's
> solution later since MMU notifiers are able to deal with this pages, but
> it's a very nice patch for stable kernels.
Oh nice, thanks for testing. How's this?
Thanks,
Nick
---
KVM: Fix page ref underflow for regions with valid but non-refcounted pages
It's possible to create a region which maps valid but non-refcounted
pages (e.g., tail pages of non-compound higher order allocations). These
host pages can then be returned by gfn_to_page, gfn_to_pfn, etc., family
of APIs, which take a reference to the page, which takes it from 0 to 1.
When the reference is dropped, this will free the page incorrectly.
Fix this by only taking a reference on the page if it was non-zero,
which indicates it is participating in normal refcounting (and can be
released with put_page).
Signed-off-by: Nicholas Piggin
---
virt/kvm/kvm_main.c | 19 +--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 6a6bc7af0e28..46fb042837d2 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2055,6 +2055,13 @@ static bool vma_is_valid(struct vm_area_struct *vma,
bool write_fault)
return true;
}
+static int kvm_try_get_pfn(kvm_pfn_t pfn)
+{
+ if (kvm_is_reserved_pfn(pfn))
+ return 1;
+ return get_page_unless_zero(pfn_to_page(pfn));
+}
+
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
unsigned long addr, bool *async,
bool write_fault, bool *writable,
@@ -2104,13 +2111,21 @@ static int hva_to_pfn_remapped(struct vm_area_struct
*vma,
* Whoever called remap_pfn_range is also going to call e.g.
* unmap_mapping_range before the underlying pages are freed,
* causing a call to our MMU notifier.
+*
+* Certain IO or PFNMAP mappings can be backed with valid
+* struct pages, but be allocated without refcounting e.g.,
+* tail pages of non-compound higher order allocations, which
+* would then underflow the refcount when the caller does the
+* required put_page. Don't allow those pages here.
*/
- kvm_get_pfn(pfn);
+ if (!kvm_try_get_pfn(pfn))
+ r = -EFAULT;
out:
pte_unmap_unlock(ptep, ptl);
*p_pfn = pfn;
- return 0;
+
+ return r;
}
/*
--
2.23.0
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 0/6] KVM: Remove uses of struct page from x86 and arm64 MMU
Excerpts from Nicholas Piggin's message of June 24, 2021 8:34 pm:
> Excerpts from David Stevens's message of June 24, 2021 1:57 pm:
>> KVM supports mapping VM_IO and VM_PFNMAP memory into the guest by using
>> follow_pte in gfn_to_pfn. However, the resolved pfns may not have
>> assoicated struct pages, so they should not be passed to pfn_to_page.
>> This series removes such calls from the x86 and arm64 secondary MMU. To
>> do this, this series modifies gfn_to_pfn to return a struct page in
>> addition to a pfn, if the hva was resolved by gup. This allows the
>> caller to call put_page only when necessated by gup.
>>
>> This series provides a helper function that unwraps the new return type
>> of gfn_to_pfn to provide behavior identical to the old behavior. As I
>> have no hardware to test powerpc/mips changes, the function is used
>> there for minimally invasive changes. Additionally, as gfn_to_page and
>> gfn_to_pfn_cache are not integrated with mmu notifier, they cannot be
>> easily changed over to only use pfns.
>>
>> This addresses CVE-2021-22543 on x86 and arm64.
>
> Does this fix the problem? (untested I don't have a POC setup at hand,
> but at least in concept)
This one actually compiles at least. Unfortunately I don't have much
time in the near future to test, and I only just found out about this
CVE a few hours ago.
---
It's possible to create a region which maps valid but non-refcounted
pages (e.g., tail pages of non-compound higher order allocations). These
host pages can then be returned by gfn_to_page, gfn_to_pfn, etc., family
of APIs, which take a reference to the page, which takes it from 0 to 1.
When the reference is dropped, this will free the page incorrectly.
Fix this by only taking a reference on the page if it was non-zero,
which indicates it is participating in normal refcounting (and can be
released with put_page).
---
virt/kvm/kvm_main.c | 19 +--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 6a6bc7af0e28..46fb042837d2 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2055,6 +2055,13 @@ static bool vma_is_valid(struct vm_area_struct *vma,
bool write_fault)
return true;
}
+static int kvm_try_get_pfn(kvm_pfn_t pfn)
+{
+ if (kvm_is_reserved_pfn(pfn))
+ return 1;
+ return get_page_unless_zero(pfn_to_page(pfn));
+}
+
static int hva_to_pfn_remapped(struct vm_area_struct *vma,
unsigned long addr, bool *async,
bool write_fault, bool *writable,
@@ -2104,13 +2111,21 @@ static int hva_to_pfn_remapped(struct vm_area_struct
*vma,
* Whoever called remap_pfn_range is also going to call e.g.
* unmap_mapping_range before the underlying pages are freed,
* causing a call to our MMU notifier.
+*
+* Certain IO or PFNMAP mappings can be backed with valid
+* struct pages, but be allocated without refcounting e.g.,
+* tail pages of non-compound higher order allocations, which
+* would then underflow the refcount when the caller does the
+* required put_page. Don't allow those pages here.
*/
- kvm_get_pfn(pfn);
+ if (!kvm_try_get_pfn(pfn))
+ r = -EFAULT;
out:
pte_unmap_unlock(ptep, ptl);
*p_pfn = pfn;
- return 0;
+
+ return r;
}
/*
--
2.23.0
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 3/6] KVM: x86/mmu: avoid struct page in MMU
Excerpts from Marc Zyngier's message of June 24, 2021 8:06 pm: > On Thu, 24 Jun 2021 09:58:00 +0100, > Nicholas Piggin wrote: >> >> Excerpts from David Stevens's message of June 24, 2021 1:57 pm: >> > From: David Stevens >> > out_unlock: >> >if (is_tdp_mmu_root(vcpu->kvm, vcpu->arch.mmu->root_hpa)) >> >read_unlock(>kvm->mmu_lock); >> >else >> >write_unlock(>kvm->mmu_lock); >> > - kvm_release_pfn_clean(pfn); >> > + if (pfnpg.page) >> > + put_page(pfnpg.page); >> >return r; >> > } >> >> How about >> >> kvm_release_pfn_page_clean(pfnpg); > > I'm not sure. I always found kvm_release_pfn_clean() ugly, because it > doesn't mark the page 'clean'. I find put_page() more correct. > > Something like 'kvm_put_pfn_page()' would make more sense, but I'm so > bad at naming things that I could just as well call it 'bob()'. That seems like a fine name to me. A little better than bob. Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 2/6] KVM: mmu: also return page from gfn_to_pfn
Excerpts from Paolo Bonzini's message of June 24, 2021 8:21 pm: > On 24/06/21 12:17, Nicholas Piggin wrote: >>> If all callers were updated that is one thing, but from the changelog >>> it sounds like that would not happen and there would be some gfn_to_pfn >>> users left over. >>> >>> But yes in the end you would either need to make gfn_to_pfn never return >>> a page found via follow_pte, or change all callers to the new way. If >>> the plan is for the latter then I guess that's fine. >> >> Actually in that case anyway I don't see the need -- the existence of >> gfn_to_pfn is enough to know it might be buggy. It can just as easily >> be grepped for as kvm_pfn_page_unwrap. > > Sure, but that would leave us with longer function names > (gfn_to_pfn_page* instead of gfn_to_pfn*). So the "safe" use is the one > that looks worse and the unsafe use is the one that looks safe. The churn isn't justified because of function name length. Choose g2pp() if you want a non-descriptive but short name. The existing name isn't good anyway because it not only looks up a pfn but also a page, and more importantly it gets a ref on the page. The name should be changed if you introduce a new API. >> And are gfn_to_page cases also >> vulernable to the same issue? > > No, they're just broken for the VM_IO|VM_PFNMAP case. No they aren't vulnerable, or they are vunlerable but also broken in other cases? Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 0/6] KVM: Remove uses of struct page from x86 and arm64 MMU
Excerpts from David Stevens's message of June 24, 2021 1:57 pm:
> KVM supports mapping VM_IO and VM_PFNMAP memory into the guest by using
> follow_pte in gfn_to_pfn. However, the resolved pfns may not have
> assoicated struct pages, so they should not be passed to pfn_to_page.
> This series removes such calls from the x86 and arm64 secondary MMU. To
> do this, this series modifies gfn_to_pfn to return a struct page in
> addition to a pfn, if the hva was resolved by gup. This allows the
> caller to call put_page only when necessated by gup.
>
> This series provides a helper function that unwraps the new return type
> of gfn_to_pfn to provide behavior identical to the old behavior. As I
> have no hardware to test powerpc/mips changes, the function is used
> there for minimally invasive changes. Additionally, as gfn_to_page and
> gfn_to_pfn_cache are not integrated with mmu notifier, they cannot be
> easily changed over to only use pfns.
>
> This addresses CVE-2021-22543 on x86 and arm64.
Does this fix the problem? (untested I don't have a POC setup at hand,
but at least in concept)
I have no problem with improving the API and probably in the direction
of your series is good. But there seems to be a lot of unfixed arch code
and broken APIs remaining left to do after your series too. This might
be most suitable to backport and as a base for your series that can take
more time to convert to new APIs.
Thanks,
Nick
---
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 6a6bc7af0e28..e208c279d903 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2104,13 +2104,21 @@ static int hva_to_pfn_remapped(struct vm_area_struct
*vma,
* Whoever called remap_pfn_range is also going to call e.g.
* unmap_mapping_range before the underlying pages are freed,
* causing a call to our MMU notifier.
+*
+* Certain IO or PFNMAP mappings can be backed with valid
+* struct pages, but be allocated without refcounting e.g.,
+* tail pages of non-compound higher order allocations, which
+* would then underflow the refcount when the caller does the
+* required put_page. Don't allow those pages here.
*/
- kvm_get_pfn(pfn);
+ if (!kvm_try_get_pfn(pfn))
+ r = -EFAULT;
out:
pte_unmap_unlock(ptep, ptl);
*p_pfn = pfn;
- return 0;
+
+ return r;
}
/*
@@ -2487,6 +2495,13 @@ void kvm_set_pfn_accessed(kvm_pfn_t pfn)
}
EXPORT_SYMBOL_GPL(kvm_set_pfn_accessed);
+static int kvm_try_get_pfn(kvm_pfn_t pfn)
+{
+ if (kvm_is_reserved_pfn(pfn))
+ return 1;
+ return get_page_unless_zero(pfn_to_page(pfn));
+}
+
void kvm_get_pfn(kvm_pfn_t pfn)
{
if (!kvm_is_reserved_pfn(pfn))
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 2/6] KVM: mmu: also return page from gfn_to_pfn
Excerpts from Nicholas Piggin's message of June 24, 2021 7:57 pm: > Excerpts from Paolo Bonzini's message of June 24, 2021 7:42 pm: >> On 24/06/21 10:52, Nicholas Piggin wrote: >>>> For now, wrap all calls to gfn_to_pfn functions in the new helper >>>> function. Callers which don't need the page struct will be updated in >>>> follow-up patches. >>> Hmm. You mean callers that do need the page will be updated? Normally >>> if there will be leftover users that don't need the struct page then >>> you would go the other way and keep the old call the same, and add a new >>> one (gfn_to_pfn_page) just for those that need it. >> >> Needing kvm_pfn_page_unwrap is a sign that something might be buggy, so >> it's a good idea to move the short name to the common case and the ugly >> kvm_pfn_page_unwrap(gfn_to_pfn(...)) for the weird one. In fact I'm not >> sure there should be any kvm_pfn_page_unwrap in the end. > > If all callers were updated that is one thing, but from the changelog > it sounds like that would not happen and there would be some gfn_to_pfn > users left over. > > But yes in the end you would either need to make gfn_to_pfn never return > a page found via follow_pte, or change all callers to the new way. If > the plan is for the latter then I guess that's fine. Actually in that case anyway I don't see the need -- the existence of gfn_to_pfn is enough to know it might be buggy. It can just as easily be grepped for as kvm_pfn_page_unwrap. And are gfn_to_page cases also vulernable to the same issue? So I think it could be marked deprecated or something if not everything will be converted in the one series, and don't need to touch all that arch code with this patch. Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 2/6] KVM: mmu: also return page from gfn_to_pfn
Excerpts from Paolo Bonzini's message of June 24, 2021 7:42 pm: > On 24/06/21 10:52, Nicholas Piggin wrote: >>> For now, wrap all calls to gfn_to_pfn functions in the new helper >>> function. Callers which don't need the page struct will be updated in >>> follow-up patches. >> Hmm. You mean callers that do need the page will be updated? Normally >> if there will be leftover users that don't need the struct page then >> you would go the other way and keep the old call the same, and add a new >> one (gfn_to_pfn_page) just for those that need it. > > Needing kvm_pfn_page_unwrap is a sign that something might be buggy, so > it's a good idea to move the short name to the common case and the ugly > kvm_pfn_page_unwrap(gfn_to_pfn(...)) for the weird one. In fact I'm not > sure there should be any kvm_pfn_page_unwrap in the end. If all callers were updated that is one thing, but from the changelog it sounds like that would not happen and there would be some gfn_to_pfn users left over. But yes in the end you would either need to make gfn_to_pfn never return a page found via follow_pte, or change all callers to the new way. If the plan is for the latter then I guess that's fine. Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 3/6] KVM: x86/mmu: avoid struct page in MMU
Excerpts from David Stevens's message of June 24, 2021 1:57 pm: > From: David Stevens > out_unlock: > if (is_tdp_mmu_root(vcpu->kvm, vcpu->arch.mmu->root_hpa)) > read_unlock(>kvm->mmu_lock); > else > write_unlock(>kvm->mmu_lock); > - kvm_release_pfn_clean(pfn); > + if (pfnpg.page) > + put_page(pfnpg.page); > return r; > } How about kvm_release_pfn_page_clean(pfnpg); Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 2/6] KVM: mmu: also return page from gfn_to_pfn
Excerpts from David Stevens's message of June 24, 2021 1:57 pm: > From: David Stevens > > Return a struct kvm_pfn_page containing both a pfn and an optional > struct page from the gfn_to_pfn family of functions. This differentiates > the gup and follow_fault_pfn cases, which allows callers that only need > a pfn to avoid touching the page struct in the latter case. For callers > that need a struct page, introduce a helper function that unwraps a > struct kvm_pfn_page into a struct page. This helper makes the call to > kvm_get_pfn which had previously been in hva_to_pfn_remapped. > > For now, wrap all calls to gfn_to_pfn functions in the new helper > function. Callers which don't need the page struct will be updated in > follow-up patches. Hmm. You mean callers that do need the page will be updated? Normally if there will be leftover users that don't need the struct page then you would go the other way and keep the old call the same, and add a new one (gfn_to_pfn_page) just for those that need it. Most kernel code I look at passes back multiple values by updating pointers to struct or variables rather than returning a struct, I suppose that's not really a big deal and a matter of taste. Thanks, Nick ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 1/6] KVM: x86/mmu: release audited pfns
Excerpts from David Stevens's message of June 24, 2021 1:57 pm: > From: David Stevens Changelog? This looks like a bug, should it have a Fixes: tag? Thanks, Nick > > Signed-off-by: David Stevens > --- > arch/x86/kvm/mmu/mmu_audit.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/arch/x86/kvm/mmu/mmu_audit.c b/arch/x86/kvm/mmu/mmu_audit.c > index cedc17b2f60e..97ff184084b4 100644 > --- a/arch/x86/kvm/mmu/mmu_audit.c > +++ b/arch/x86/kvm/mmu/mmu_audit.c > @@ -121,6 +121,8 @@ static void audit_mappings(struct kvm_vcpu *vcpu, u64 > *sptep, int level) > audit_printk(vcpu->kvm, "levels %d pfn %llx hpa %llx " >"ent %llxn", vcpu->arch.mmu->root_level, pfn, >hpa, *sptep); > + > + kvm_release_pfn_clean(pfn); > } > > static void inspect_spte_has_rmap(struct kvm *kvm, u64 *sptep) > -- > 2.32.0.93.g670b81a890-goog > > ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
[Intel-gfx] [PATCH] drm/i915: select CRC32
kbuild test robot found a build failure when building with thin archives: http://marc.info/?l=linux-kbuild=149802285009737=2 Signed-off-by: Nicholas Piggin <npig...@gmail.com> --- drivers/gpu/drm/i915/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/i915/Kconfig b/drivers/gpu/drm/i915/Kconfig index a5cd5dacf055..e9e64e8e9765 100644 --- a/drivers/gpu/drm/i915/Kconfig +++ b/drivers/gpu/drm/i915/Kconfig @@ -21,6 +21,7 @@ config DRM_I915 select ACPI_BUTTON if ACPI select SYNC_FILE select IOSF_MBI + select CRC32 help Choose this option if you have a system that has "Intel Graphics Media Accelerator" or "HD Graphics" integrated graphics, -- 2.11.0 ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH] drm/i915: select CRC32
On Wed, 21 Jun 2017 10:15:56 +0100
Chris Wilson <ch...@chris-wilson.co.uk> wrote:
> Quoting Daniel Vetter (2017-06-21 10:13:41)
> > On Wed, Jun 21, 2017 at 04:34:20PM +1000, Nicholas Piggin wrote:
> > > kbuild test robot found a build failure when building with thin
> > > archives:
> > >
> > > http://marc.info/?l=linux-kbuild=149802285009737=2
> > >
> > > Signed-off-by: Nicholas Piggin <npig...@gmail.com>
> >
> > Pushed to drm-intel for 4.14, thanks.
>
> Probably deserves to be in fixes if we identify what introduced the CRC
> dep. mst-dp? pipe crc?
Looks like
579cea5f3 ("drm/i915/gvt: golden virtual HW state management")
Thanks,
Nick
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
