Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1707829 > unrolled thread

[PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

Started byTycho Andersen <tycho@docker.com>
First post2017-08-09 22:10 +0200
Last post2017-08-14 22:30 +0200
Articles 8 — 3 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled Tycho Andersen <tycho@docker.com> - 2017-08-09 22:10 +0200
    Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Laura Abbott <labbott@redhat.com> - 2017-08-11 19:30 +0200
      Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Tycho Andersen <tycho@docker.com> - 2017-08-11 23:20 +0200
        Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Tycho Andersen <tycho@docker.com> - 2017-08-12 00:00 +0200
        Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Mark Rutland <mark.rutland@arm.com> - 2017-08-12 13:20 +0200
          Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Tycho Andersen <tycho@docker.com> - 2017-08-14 18:30 +0200
            Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Laura Abbott <labbott@redhat.com> - 2017-08-14 20:50 +0200
              Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section  mappings if XPFO is enabled Tycho Andersen <tycho@docker.com> - 2017-08-14 22:30 +0200

#1707829 — [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromTycho Andersen <tycho@docker.com>
Date2017-08-09 22:10 +0200
Subject[PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<ucFsJ-3Hw-11@gated-at.bofh.it>
From: Juerg Haefliger <juerg.haefliger@hpe.com>

XPFO (eXclusive Page Frame Ownership) doesn't support section mappings
yet, so disable it if XPFO is turned on.

Signed-off-by: Juerg Haefliger <juerg.haefliger@canonical.com>
Tested-by: Tycho Andersen <tycho@docker.com>
---
 arch/arm64/mm/mmu.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index f1eb15e0e864..38026b3ccb46 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -176,6 +176,18 @@ static void alloc_init_cont_pte(pmd_t *pmd, unsigned long addr,
 	} while (addr = next, addr != end);
 }
 
+static inline bool use_section_mapping(unsigned long addr, unsigned long next,
+				unsigned long phys)
+{
+	if (IS_ENABLED(CONFIG_XPFO))
+		return false;
+
+	if (((addr | next | phys) & ~SECTION_MASK) != 0)
+		return false;
+
+	return true;
+}
+
 static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
 		     phys_addr_t phys, pgprot_t prot,
 		     phys_addr_t (*pgtable_alloc)(void), int flags)
@@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
 		next = pmd_addr_end(addr, end);
 
 		/* try section mapping first */
-		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
+		if (use_section_mapping(addr, next, phys) &&
 		    (flags & NO_BLOCK_MAPPINGS) == 0) {
 			pmd_set_huge(pmd, phys, prot);
 
-- 
2.11.0

[toc] | [next] | [standalone]

#1709878 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromLaura Abbott <labbott@redhat.com>
Date2017-08-11 19:30 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<udlUZ-6mk-19@gated-at.bofh.it>
In reply to#1707829
On 08/09/2017 01:07 PM, Tycho Andersen wrote:
> From: Juerg Haefliger <juerg.haefliger@hpe.com>
> 
> XPFO (eXclusive Page Frame Ownership) doesn't support section mappings
> yet, so disable it if XPFO is turned on.
> 
> Signed-off-by: Juerg Haefliger <juerg.haefliger@canonical.com>
> Tested-by: Tycho Andersen <tycho@docker.com>
> ---
>  arch/arm64/mm/mmu.c | 14 +++++++++++++-
>  1 file changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
> index f1eb15e0e864..38026b3ccb46 100644
> --- a/arch/arm64/mm/mmu.c
> +++ b/arch/arm64/mm/mmu.c
> @@ -176,6 +176,18 @@ static void alloc_init_cont_pte(pmd_t *pmd, unsigned long addr,
>  	} while (addr = next, addr != end);
>  }
>  
> +static inline bool use_section_mapping(unsigned long addr, unsigned long next,
> +				unsigned long phys)
> +{
> +	if (IS_ENABLED(CONFIG_XPFO))
> +		return false;
> +
> +	if (((addr | next | phys) & ~SECTION_MASK) != 0)
> +		return false;
> +
> +	return true;
> +}
> +
>  static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
>  		     phys_addr_t phys, pgprot_t prot,
>  		     phys_addr_t (*pgtable_alloc)(void), int flags)
> @@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
>  		next = pmd_addr_end(addr, end);
>  
>  		/* try section mapping first */
> -		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
> +		if (use_section_mapping(addr, next, phys) &&
>  		    (flags & NO_BLOCK_MAPPINGS) == 0) {
>  			pmd_set_huge(pmd, phys, prot);
>  
> 

There is already similar logic to disable section mappings for
debug_pagealloc at the start of map_mem, can you take advantage
of that?

Thanks,
Laura

[toc] | [prev] | [next] | [standalone]

#1709970 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromTycho Andersen <tycho@docker.com>
Date2017-08-11 23:20 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<udpvA-6M-21@gated-at.bofh.it>
In reply to#1709878
Hi Laura,

On Fri, Aug 11, 2017 at 10:25:14AM -0700, Laura Abbott wrote:
> On 08/09/2017 01:07 PM, Tycho Andersen wrote:
> > From: Juerg Haefliger <juerg.haefliger@hpe.com>
> > 
> > XPFO (eXclusive Page Frame Ownership) doesn't support section mappings
> > yet, so disable it if XPFO is turned on.
> > 
> > Signed-off-by: Juerg Haefliger <juerg.haefliger@canonical.com>
> > Tested-by: Tycho Andersen <tycho@docker.com>
> > ---
> >  arch/arm64/mm/mmu.c | 14 +++++++++++++-
> >  1 file changed, 13 insertions(+), 1 deletion(-)
> > 
> > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
> > index f1eb15e0e864..38026b3ccb46 100644
> > --- a/arch/arm64/mm/mmu.c
> > +++ b/arch/arm64/mm/mmu.c
> > @@ -176,6 +176,18 @@ static void alloc_init_cont_pte(pmd_t *pmd, unsigned long addr,
> >  	} while (addr = next, addr != end);
> >  }
> >  
> > +static inline bool use_section_mapping(unsigned long addr, unsigned long next,
> > +				unsigned long phys)
> > +{
> > +	if (IS_ENABLED(CONFIG_XPFO))
> > +		return false;
> > +
> > +	if (((addr | next | phys) & ~SECTION_MASK) != 0)
> > +		return false;
> > +
> > +	return true;
> > +}
> > +
> >  static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
> >  		     phys_addr_t phys, pgprot_t prot,
> >  		     phys_addr_t (*pgtable_alloc)(void), int flags)
> > @@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
> >  		next = pmd_addr_end(addr, end);
> >  
> >  		/* try section mapping first */
> > -		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
> > +		if (use_section_mapping(addr, next, phys) &&
> >  		    (flags & NO_BLOCK_MAPPINGS) == 0) {
> >  			pmd_set_huge(pmd, phys, prot);
> >  
> > 
> 
> There is already similar logic to disable section mappings for
> debug_pagealloc at the start of map_mem, can you take advantage
> of that?

You're suggesting something like this instead? Seems to work fine.

diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index 38026b3ccb46..3b2c17bbbf12 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -434,6 +434,8 @@ static void __init map_mem(pgd_t *pgd)
 
 	if (debug_pagealloc_enabled())
 		flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS;
+	if (IS_ENABLED(CONFIG_XPFO))
+		flags |= NO_BLOCK_MAPPINGS;
 
 	/*
 	 * Take care not to create a writable alias for the

Cheers,

Tycho

[toc] | [prev] | [next] | [standalone]

#1709993 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromTycho Andersen <tycho@docker.com>
Date2017-08-12 00:00 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<udq8i-la-15@gated-at.bofh.it>
In reply to#1709970
On Fri, Aug 11, 2017 at 03:13:02PM -0600, Tycho Andersen wrote:
> You're suggesting something like this instead? Seems to work fine.

And in fact, using this patch instead means that booting on 4k pages
works too... I guess because NO_BLOCK_MAPPINGS is looked at in a few
other places that matter too? Anyway, I'll use this patch instead,
thanks for the suggestion!

Tycho

[toc] | [prev] | [next] | [standalone]

#1710227 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromMark Rutland <mark.rutland@arm.com>
Date2017-08-12 13:20 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<udCCu-8pQ-1@gated-at.bofh.it>
In reply to#1709970
Hi,

On Fri, Aug 11, 2017 at 03:13:02PM -0600, Tycho Andersen wrote:
> On Fri, Aug 11, 2017 at 10:25:14AM -0700, Laura Abbott wrote:
> > On 08/09/2017 01:07 PM, Tycho Andersen wrote:
> > > @@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
> > >  		next = pmd_addr_end(addr, end);
> > >  
> > >  		/* try section mapping first */
> > > -		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
> > > +		if (use_section_mapping(addr, next, phys) &&
> > >  		    (flags & NO_BLOCK_MAPPINGS) == 0) {
> > >  			pmd_set_huge(pmd, phys, prot);
> > >  
> > > 
> > 
> > There is already similar logic to disable section mappings for
> > debug_pagealloc at the start of map_mem, can you take advantage
> > of that?
> 
> You're suggesting something like this instead? Seems to work fine.
> 
> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
> index 38026b3ccb46..3b2c17bbbf12 100644
> --- a/arch/arm64/mm/mmu.c
> +++ b/arch/arm64/mm/mmu.c
> @@ -434,6 +434,8 @@ static void __init map_mem(pgd_t *pgd)
>  
>  	if (debug_pagealloc_enabled())
>  		flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS;
> +	if (IS_ENABLED(CONFIG_XPFO))
> +		flags |= NO_BLOCK_MAPPINGS;
>  

IIUC, XPFO carves out individual pages just like DEBUG_PAGEALLOC, so you'll
also need NO_CONT_MAPPINGS.

Thanks
Mark.

[toc] | [prev] | [next] | [standalone]

#1711160 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromTycho Andersen <tycho@docker.com>
Date2017-08-14 18:30 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<ueqpA-6qC-37@gated-at.bofh.it>
In reply to#1710227
On Sat, Aug 12, 2017 at 12:17:34PM +0100, Mark Rutland wrote:
> Hi,
> 
> On Fri, Aug 11, 2017 at 03:13:02PM -0600, Tycho Andersen wrote:
> > On Fri, Aug 11, 2017 at 10:25:14AM -0700, Laura Abbott wrote:
> > > On 08/09/2017 01:07 PM, Tycho Andersen wrote:
> > > > @@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
> > > >  		next = pmd_addr_end(addr, end);
> > > >  
> > > >  		/* try section mapping first */
> > > > -		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
> > > > +		if (use_section_mapping(addr, next, phys) &&
> > > >  		    (flags & NO_BLOCK_MAPPINGS) == 0) {
> > > >  			pmd_set_huge(pmd, phys, prot);
> > > >  
> > > > 
> > > 
> > > There is already similar logic to disable section mappings for
> > > debug_pagealloc at the start of map_mem, can you take advantage
> > > of that?
> > 
> > You're suggesting something like this instead? Seems to work fine.
> > 
> > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
> > index 38026b3ccb46..3b2c17bbbf12 100644
> > --- a/arch/arm64/mm/mmu.c
> > +++ b/arch/arm64/mm/mmu.c
> > @@ -434,6 +434,8 @@ static void __init map_mem(pgd_t *pgd)
> >  
> >  	if (debug_pagealloc_enabled())
> >  		flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS;
> > +	if (IS_ENABLED(CONFIG_XPFO))
> > +		flags |= NO_BLOCK_MAPPINGS;
> >  
> 
> IIUC, XPFO carves out individual pages just like DEBUG_PAGEALLOC, so you'll
> also need NO_CONT_MAPPINGS.

Yes, thanks!

Tycho

[toc] | [prev] | [next] | [standalone]

#1711347 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromLaura Abbott <labbott@redhat.com>
Date2017-08-14 20:50 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<uesB5-7Hs-31@gated-at.bofh.it>
In reply to#1711160
On 08/14/2017 09:22 AM, Tycho Andersen wrote:
> On Sat, Aug 12, 2017 at 12:17:34PM +0100, Mark Rutland wrote:
>> Hi,
>>
>> On Fri, Aug 11, 2017 at 03:13:02PM -0600, Tycho Andersen wrote:
>>> On Fri, Aug 11, 2017 at 10:25:14AM -0700, Laura Abbott wrote:
>>>> On 08/09/2017 01:07 PM, Tycho Andersen wrote:
>>>>> @@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
>>>>>  		next = pmd_addr_end(addr, end);
>>>>>  
>>>>>  		/* try section mapping first */
>>>>> -		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
>>>>> +		if (use_section_mapping(addr, next, phys) &&
>>>>>  		    (flags & NO_BLOCK_MAPPINGS) == 0) {
>>>>>  			pmd_set_huge(pmd, phys, prot);
>>>>>  
>>>>>
>>>>
>>>> There is already similar logic to disable section mappings for
>>>> debug_pagealloc at the start of map_mem, can you take advantage
>>>> of that?
>>>
>>> You're suggesting something like this instead? Seems to work fine.
>>>
>>> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
>>> index 38026b3ccb46..3b2c17bbbf12 100644
>>> --- a/arch/arm64/mm/mmu.c
>>> +++ b/arch/arm64/mm/mmu.c
>>> @@ -434,6 +434,8 @@ static void __init map_mem(pgd_t *pgd)
>>>  
>>>  	if (debug_pagealloc_enabled())
>>>  		flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS;
>>> +	if (IS_ENABLED(CONFIG_XPFO))
>>> +		flags |= NO_BLOCK_MAPPINGS;
>>>  
>>
>> IIUC, XPFO carves out individual pages just like DEBUG_PAGEALLOC, so you'll
>> also need NO_CONT_MAPPINGS.
> 
> Yes, thanks!
> 
> Tycho
> 

Setting NO_CONT_MAPPINGS fixes the TLB conflict aborts I was seeing
on my machine.

Thanks,
Laura

[toc] | [prev] | [next] | [standalone]

#1711422 — Re: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled

FromTycho Andersen <tycho@docker.com>
Date2017-08-14 22:30 +0200
SubjectRe: [kernel-hardening] [PATCH v5 06/10] arm64/mm: Disable section mappings if XPFO is enabled
Message-ID<ueu9Q-gF-27@gated-at.bofh.it>
In reply to#1711347
On Mon, Aug 14, 2017 at 11:42:45AM -0700, Laura Abbott wrote:
> On 08/14/2017 09:22 AM, Tycho Andersen wrote:
> > On Sat, Aug 12, 2017 at 12:17:34PM +0100, Mark Rutland wrote:
> >> Hi,
> >>
> >> On Fri, Aug 11, 2017 at 03:13:02PM -0600, Tycho Andersen wrote:
> >>> On Fri, Aug 11, 2017 at 10:25:14AM -0700, Laura Abbott wrote:
> >>>> On 08/09/2017 01:07 PM, Tycho Andersen wrote:
> >>>>> @@ -190,7 +202,7 @@ static void init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
> >>>>>  		next = pmd_addr_end(addr, end);
> >>>>>  
> >>>>>  		/* try section mapping first */
> >>>>> -		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
> >>>>> +		if (use_section_mapping(addr, next, phys) &&
> >>>>>  		    (flags & NO_BLOCK_MAPPINGS) == 0) {
> >>>>>  			pmd_set_huge(pmd, phys, prot);
> >>>>>  
> >>>>>
> >>>>
> >>>> There is already similar logic to disable section mappings for
> >>>> debug_pagealloc at the start of map_mem, can you take advantage
> >>>> of that?
> >>>
> >>> You're suggesting something like this instead? Seems to work fine.
> >>>
> >>> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
> >>> index 38026b3ccb46..3b2c17bbbf12 100644
> >>> --- a/arch/arm64/mm/mmu.c
> >>> +++ b/arch/arm64/mm/mmu.c
> >>> @@ -434,6 +434,8 @@ static void __init map_mem(pgd_t *pgd)
> >>>  
> >>>  	if (debug_pagealloc_enabled())
> >>>  		flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS;
> >>> +	if (IS_ENABLED(CONFIG_XPFO))
> >>> +		flags |= NO_BLOCK_MAPPINGS;
> >>>  
> >>
> >> IIUC, XPFO carves out individual pages just like DEBUG_PAGEALLOC, so you'll
> >> also need NO_CONT_MAPPINGS.
> > 
> > Yes, thanks!
> > 
> > Tycho
> > 
> 
> Setting NO_CONT_MAPPINGS fixes the TLB conflict aborts I was seeing
> on my machine.

Great, thanks for testing! I've also fixed the lookup_page_ext bug you
noted in the other thread.

Tycho

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web