* Re: [PATCH] arm64: add kc_offset_to_vaddr and kc_vaddr_to_offset macro
@ 2015-09-02 10:24 ` Will Deacon
0 siblings, 0 replies; 10+ messages in thread
From: Will Deacon @ 2015-09-02 10:24 UTC (permalink / raw)
To: yalin wang
Cc: Catalin Marinas, akpm@linux-foundation.org,
kirill.shutemov@linux.intel.com, ard.biesheuvel@linaro.org,
minchan@kernel.org, fkan@apm.com, steve.capper@linaro.org,
lauraa@codeaurora.org, linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
On Sun, Aug 30, 2015 at 07:19:59AM +0100, yalin wang wrote:
> This patch add kc_offset_to_vaddr() and kc_vaddr_to_offset(),
> the default version doesn't work on arm64, because arm64 kernel address
> is below the PAGE_OFFSET, like module address and vmemmap address are
> all below PAGE_OFFSET address.
>
> Signed-off-by: yalin wang <yalin.wang2010@gmail.com>
> ---
> arch/arm64/include/asm/pgtable.h | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
> index d374191..ef1ed5c 100644
> --- a/arch/arm64/include/asm/pgtable.h
> +++ b/arch/arm64/include/asm/pgtable.h
> @@ -660,6 +660,8 @@ static inline void update_mmu_cache(struct vm_area_struct *vma,
>
> #define update_mmu_cache_pmd(vma, address, pmd) do { } while (0)
>
> +#define kc_vaddr_to_offset(v) ((v) & ((1UL << VA_BITS) - 1))
> +#define kc_offset_to_vaddr(o) ((o) | ~((1UL << VA_BITS) - 1))
I think it would be cleaner to define something like VIRTUAL_BASE, and
then define the vmalloc area in terms of that as well as these kcore
macros.
Will
^ permalink raw reply [flat|nested] 10+ messages in thread* [PATCH] arm64: add kc_offset_to_vaddr and kc_vaddr_to_offset macro
2015-09-02 10:24 ` Will Deacon
@ 2015-09-04 16:51 ` Catalin Marinas
-1 siblings, 0 replies; 10+ messages in thread
From: Catalin Marinas @ 2015-09-04 16:51 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Sep 02, 2015 at 11:24:54AM +0100, Will Deacon wrote:
> On Sun, Aug 30, 2015 at 07:19:59AM +0100, yalin wang wrote:
> > This patch add kc_offset_to_vaddr() and kc_vaddr_to_offset(),
> > the default version doesn't work on arm64, because arm64 kernel address
> > is below the PAGE_OFFSET, like module address and vmemmap address are
> > all below PAGE_OFFSET address.
> >
> > Signed-off-by: yalin wang <yalin.wang2010@gmail.com>
> > ---
> > arch/arm64/include/asm/pgtable.h | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
> > index d374191..ef1ed5c 100644
> > --- a/arch/arm64/include/asm/pgtable.h
> > +++ b/arch/arm64/include/asm/pgtable.h
> > @@ -660,6 +660,8 @@ static inline void update_mmu_cache(struct vm_area_struct *vma,
> >
> > #define update_mmu_cache_pmd(vma, address, pmd) do { } while (0)
> >
> > +#define kc_vaddr_to_offset(v) ((v) & ((1UL << VA_BITS) - 1))
> > +#define kc_offset_to_vaddr(o) ((o) | ~((1UL << VA_BITS) - 1))
>
> I think it would be cleaner to define something like VIRTUAL_BASE, and
> then define the vmalloc area in terms of that as well as these kcore
> macros.
The KASan patches were doing this, defining VA_START and defining
VMALLOC_START in terms of it. But they missed the merging window.
--
Catalin
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [PATCH] arm64: add kc_offset_to_vaddr and kc_vaddr_to_offset macro
@ 2015-09-04 16:51 ` Catalin Marinas
0 siblings, 0 replies; 10+ messages in thread
From: Catalin Marinas @ 2015-09-04 16:51 UTC (permalink / raw)
To: Will Deacon
Cc: yalin wang, fkan@apm.com, steve.capper@linaro.org,
ard.biesheuvel@linaro.org, linux-kernel@vger.kernel.org,
minchan@kernel.org, linux-arm-kernel@lists.infradead.org,
lauraa@codeaurora.org, akpm@linux-foundation.org,
kirill.shutemov@linux.intel.com
On Wed, Sep 02, 2015 at 11:24:54AM +0100, Will Deacon wrote:
> On Sun, Aug 30, 2015 at 07:19:59AM +0100, yalin wang wrote:
> > This patch add kc_offset_to_vaddr() and kc_vaddr_to_offset(),
> > the default version doesn't work on arm64, because arm64 kernel address
> > is below the PAGE_OFFSET, like module address and vmemmap address are
> > all below PAGE_OFFSET address.
> >
> > Signed-off-by: yalin wang <yalin.wang2010@gmail.com>
> > ---
> > arch/arm64/include/asm/pgtable.h | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
> > index d374191..ef1ed5c 100644
> > --- a/arch/arm64/include/asm/pgtable.h
> > +++ b/arch/arm64/include/asm/pgtable.h
> > @@ -660,6 +660,8 @@ static inline void update_mmu_cache(struct vm_area_struct *vma,
> >
> > #define update_mmu_cache_pmd(vma, address, pmd) do { } while (0)
> >
> > +#define kc_vaddr_to_offset(v) ((v) & ((1UL << VA_BITS) - 1))
> > +#define kc_offset_to_vaddr(o) ((o) | ~((1UL << VA_BITS) - 1))
>
> I think it would be cleaner to define something like VIRTUAL_BASE, and
> then define the vmalloc area in terms of that as well as these kcore
> macros.
The KASan patches were doing this, defining VA_START and defining
VMALLOC_START in terms of it. But they missed the merging window.
--
Catalin
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH] arm64: add kc_offset_to_vaddr and kc_vaddr_to_offset macro
2015-09-02 10:24 ` Will Deacon
@ 2015-09-06 4:05 ` yalin wang
-1 siblings, 0 replies; 10+ messages in thread
From: yalin wang @ 2015-09-06 4:05 UTC (permalink / raw)
To: linux-arm-kernel
ok
i will send a V2 patch for this change.
> On Sep 2, 2015, at 18:24, Will Deacon <will.deacon@arm.com> wrote:
>
> On Sun, Aug 30, 2015 at 07:19:59AM +0100, yalin wang wrote:
>> This patch add kc_offset_to_vaddr() and kc_vaddr_to_offset(),
>> the default version doesn't work on arm64, because arm64 kernel address
>> is below the PAGE_OFFSET, like module address and vmemmap address are
>> all below PAGE_OFFSET address.
>>
>> Signed-off-by: yalin wang <yalin.wang2010@gmail.com>
>> ---
>> arch/arm64/include/asm/pgtable.h | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
>> index d374191..ef1ed5c 100644
>> --- a/arch/arm64/include/asm/pgtable.h
>> +++ b/arch/arm64/include/asm/pgtable.h
>> @@ -660,6 +660,8 @@ static inline void update_mmu_cache(struct vm_area_struct *vma,
>>
>> #define update_mmu_cache_pmd(vma, address, pmd) do { } while (0)
>>
>> +#define kc_vaddr_to_offset(v) ((v) & ((1UL << VA_BITS) - 1))
>> +#define kc_offset_to_vaddr(o) ((o) | ~((1UL << VA_BITS) - 1))
>
> I think it would be cleaner to define something like VIRTUAL_BASE, and
> then define the vmalloc area in terms of that as well as these kcore
> macros.
>
> Will
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [PATCH] arm64: add kc_offset_to_vaddr and kc_vaddr_to_offset macro
@ 2015-09-06 4:05 ` yalin wang
0 siblings, 0 replies; 10+ messages in thread
From: yalin wang @ 2015-09-06 4:05 UTC (permalink / raw)
To: Will Deacon
Cc: Catalin Marinas, akpm@linux-foundation.org,
kirill.shutemov@linux.intel.com, ard.biesheuvel@linaro.org,
minchan@kernel.org, fkan@apm.com, steve.capper@linaro.org,
lauraa@codeaurora.org, linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
ok
i will send a V2 patch for this change.
> On Sep 2, 2015, at 18:24, Will Deacon <will.deacon@arm.com> wrote:
>
> On Sun, Aug 30, 2015 at 07:19:59AM +0100, yalin wang wrote:
>> This patch add kc_offset_to_vaddr() and kc_vaddr_to_offset(),
>> the default version doesn't work on arm64, because arm64 kernel address
>> is below the PAGE_OFFSET, like module address and vmemmap address are
>> all below PAGE_OFFSET address.
>>
>> Signed-off-by: yalin wang <yalin.wang2010@gmail.com>
>> ---
>> arch/arm64/include/asm/pgtable.h | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
>> index d374191..ef1ed5c 100644
>> --- a/arch/arm64/include/asm/pgtable.h
>> +++ b/arch/arm64/include/asm/pgtable.h
>> @@ -660,6 +660,8 @@ static inline void update_mmu_cache(struct vm_area_struct *vma,
>>
>> #define update_mmu_cache_pmd(vma, address, pmd) do { } while (0)
>>
>> +#define kc_vaddr_to_offset(v) ((v) & ((1UL << VA_BITS) - 1))
>> +#define kc_offset_to_vaddr(o) ((o) | ~((1UL << VA_BITS) - 1))
>
> I think it would be cleaner to define something like VIRTUAL_BASE, and
> then define the vmalloc area in terms of that as well as these kcore
> macros.
>
> Will
^ permalink raw reply [flat|nested] 10+ messages in thread