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

Groups > linux.kernel > #1460992 > unrolled thread

[PATCH v3 0/4] arm64: put objects into proper sections

Back to article view | Back to linux.kernel

Contents

  [PATCH v3 0/4] arm64: put objects into proper sections Jisheng Zhang <jszhang@marvell.com> - 2016-08-12 10:10 +0200
    [PATCH v3 1/4] arm64: vdso: add __init section marker to alloc_vectors_page Jisheng Zhang <jszhang@marvell.com> - 2016-08-12 10:10 +0200
    [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly Jisheng Zhang <jszhang@marvell.com> - 2016-08-12 10:10 +0200
      Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of  __initdata and __read_mostly Ard Biesheuvel <ard.biesheuvel@linaro.org> - 2016-08-12 14:10 +0200
        Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of  __initdata and __read_mostly Jisheng Zhang <jszhang@marvell.com> - 2016-08-15 07:00 +0200
          Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of  __initdata and __read_mostly Jisheng Zhang <jszhang@marvell.com> - 2016-08-15 08:10 +0200
            Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of  __initdata and __read_mostly Ard Biesheuvel <ard.biesheuvel@linaro.org> - 2016-08-15 08:50 +0200

#1460992 — [PATCH v3 0/4] arm64: put objects into proper sections

This series tries to put some objects into proper sections.

patch1 puts alloc_vectors_page() into .init section. This is a clean up
patch.
patch2 constify vm_special_mapping used for aarch32 vectors page.
patch3 fix the incorrect placement of __initdata and __read_mostly
patch4 apply __ro_after_init to some objects

Previously I only want to mark vdso_pages, vdso_spec, vectors_page and
cpu_ops as __read_mostly for performance reason. Then inspired by
Kees's patch[1], I think __ro_after_init is better.

[1] http://www.spinics.net/lists/arm-kernel/msg523188.html

Changes since v2:
  - include <linux/cache.h> for __ro_after_init, thank Mark's suggestions
  - make vdso_spec[0].pages and vdso_spec[1].pages assignment style
    consistent, thank Mark for pointing it out.

Changes since v1:
  - use __ro_after_init instead of __read_mostly
  - find more objects to be suitable for __ro_after_init
  - add one more patch to fix the incorrect placement of __initdata and
    __read_mostly


Jisheng Zhang (4):
  arm64: vdso: add __init section marker to alloc_vectors_page
  arm64: vdso: constify vm_special_mapping used for aarch32 vectors page
  arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly
  arm64: apply __ro_after_init to some objects

 arch/arm64/kernel/cpu_ops.c |  2 +-
 arch/arm64/kernel/kaslr.c   |  4 ++--
 arch/arm64/kernel/vdso.c    | 32 ++++++++++++++++----------------
 arch/arm64/mm/dma-mapping.c |  2 +-
 arch/arm64/mm/init.c        |  4 ++--
 arch/arm64/mm/mmu.c         |  2 +-
 6 files changed, 23 insertions(+), 23 deletions(-)

-- 
2.8.1

[toc] | [next] | [standalone]

#1460993 — [PATCH v3 1/4] arm64: vdso: add __init section marker to alloc_vectors_page

It is not needed after booting, this patch moves the alloc_vectors_page
function to the __init section.

Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
---
 arch/arm64/kernel/vdso.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/vdso.c b/arch/arm64/kernel/vdso.c
index 076312b..e320e8f 100644
--- a/arch/arm64/kernel/vdso.c
+++ b/arch/arm64/kernel/vdso.c
@@ -55,7 +55,7 @@ struct vdso_data *vdso_data = &vdso_data_store.data;
  */
 static struct page *vectors_page[1];
 
-static int alloc_vectors_page(void)
+static int __init alloc_vectors_page(void)
 {
 	extern char __kuser_helper_start[], __kuser_helper_end[];
 	extern char __aarch32_sigret_code_start[], __aarch32_sigret_code_end[];
-- 
2.8.1

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

#1460994 — [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly

__initdata and __read_mostly should be placed after the variable name
for the variable to be placed in the intended section.

Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
---
 arch/arm64/kernel/kaslr.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/kaslr.c b/arch/arm64/kernel/kaslr.c
index b054691..8ebabc4 100644
--- a/arch/arm64/kernel/kaslr.c
+++ b/arch/arm64/kernel/kaslr.c
@@ -20,8 +20,8 @@
 #include <asm/pgtable.h>
 #include <asm/sections.h>
 
-u64 __read_mostly module_alloc_base;
-u16 __initdata memstart_offset_seed;
+u64 module_alloc_base __read_mostly;
+u16 memstart_offset_seed __initdata;
 
 static __init u64 get_kaslr_seed(void *fdt)
 {
-- 
2.8.1

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

#1461138 — Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly

Hi Jisheng,

On 12 August 2016 at 10:01, Jisheng Zhang <jszhang@marvell.com> wrote:
> __initdata and __read_mostly should be placed after the variable name
> for the variable to be placed in the intended section.
>

Why?

> Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
> ---
>  arch/arm64/kernel/kaslr.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/arm64/kernel/kaslr.c b/arch/arm64/kernel/kaslr.c
> index b054691..8ebabc4 100644
> --- a/arch/arm64/kernel/kaslr.c
> +++ b/arch/arm64/kernel/kaslr.c
> @@ -20,8 +20,8 @@
>  #include <asm/pgtable.h>
>  #include <asm/sections.h>
>
> -u64 __read_mostly module_alloc_base;
> -u16 __initdata memstart_offset_seed;
> +u64 module_alloc_base __read_mostly;
> +u16 memstart_offset_seed __initdata;
>
>  static __init u64 get_kaslr_seed(void *fdt)
>  {
> --
> 2.8.1
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

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

#1462547 — Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly

Hi Ard,

On Fri, 12 Aug 2016 14:02:40 +0200 Ard Biesheuvel wrote:

> Hi Jisheng,
> 
> On 12 August 2016 at 10:01, Jisheng Zhang <jszhang@marvell.com> wrote:
> > __initdata and __read_mostly should be placed after the variable name
> > for the variable to be placed in the intended section.
> >  
> 
> Why?

include/linux/init.h says something as:

 * For initialized data:
 * You should insert __initdata or __initconst between the variable name
 * and equal sign followed by value, e.g.:
 *
 * static int init_variable __initdata = 0;
 * static const char linux_logo[] __initconst = { 0x32, 0x36, ... };

and examples in gcc manual also put __attribute__ (...) after variable name.

https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#Common-Variable-Attributes

Then I grep the source, found most lines (especially arch/arm64/*) put the
__initdata and __read_mostly after the variable name.

However, I built the code with three different gcc, the result looks identical
no matter where these markers put. So the commit msg looks wrong, what about
changes it as

"put __initdata and __read_mostly after the variable name
to keep the style consistent"?

Thanks,
Jisheng

> 
> > Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
> > ---
> >  arch/arm64/kernel/kaslr.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/arm64/kernel/kaslr.c b/arch/arm64/kernel/kaslr.c
> > index b054691..8ebabc4 100644
> > --- a/arch/arm64/kernel/kaslr.c
> > +++ b/arch/arm64/kernel/kaslr.c
> > @@ -20,8 +20,8 @@
> >  #include <asm/pgtable.h>
> >  #include <asm/sections.h>
> >
> > -u64 __read_mostly module_alloc_base;
> > -u16 __initdata memstart_offset_seed;
> > +u64 module_alloc_base __read_mostly;
> > +u16 memstart_offset_seed __initdata;
> >
> >  static __init u64 get_kaslr_seed(void *fdt)
> >  {
> > --
> > 2.8.1
> >
> >
> > _______________________________________________
> > linux-arm-kernel mailing list
> > linux-arm-kernel@lists.infradead.org
> > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel  

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

#1462563 — Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly

Hi Ard,

On Mon, 15 Aug 2016 12:52:14 +0800 Jisheng Zhang wrote:

> Hi Ard,
> 
> On Fri, 12 Aug 2016 14:02:40 +0200 Ard Biesheuvel wrote:
> 
> > Hi Jisheng,
> > 
> > On 12 August 2016 at 10:01, Jisheng Zhang <jszhang@marvell.com> wrote:  
> > > __initdata and __read_mostly should be placed after the variable name
> > > for the variable to be placed in the intended section.
> > >    
> > 
> > Why?  
> 
> include/linux/init.h says something as:
> 
>  * For initialized data:
>  * You should insert __initdata or __initconst between the variable name
>  * and equal sign followed by value, e.g.:
>  *
>  * static int init_variable __initdata = 0;
>  * static const char linux_logo[] __initconst = { 0x32, 0x36, ... };
> 
> and examples in gcc manual also put __attribute__ (...) after variable name.
> 
> https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#Common-Variable-Attributes
> 
> Then I grep the source, found most lines (especially arch/arm64/*) put the
> __initdata and __read_mostly after the variable name.
> 
> However, I built the code with three different gcc, the result looks identical
> no matter where these markers put. So the commit msg looks wrong, what about
> changes it as
> 
> "put __initdata and __read_mostly after the variable name
> to keep the style consistent"?

After some consideration, I want to drop this patch in newer version since
it's not a bug, just "style"

Thanks for your reviewing,
Jisheng

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

#1462568 — Re: [PATCH v3 3/4] arm64: kaslr: Fix incorrect placement of __initdata and __read_mostly

On 15 August 2016 at 07:57, Jisheng Zhang <jszhang@marvell.com> wrote:
> Hi Ard,
>
> On Mon, 15 Aug 2016 12:52:14 +0800 Jisheng Zhang wrote:
>
>> Hi Ard,
>>
>> On Fri, 12 Aug 2016 14:02:40 +0200 Ard Biesheuvel wrote:
>>
>> > Hi Jisheng,
>> >
>> > On 12 August 2016 at 10:01, Jisheng Zhang <jszhang@marvell.com> wrote:
>> > > __initdata and __read_mostly should be placed after the variable name
>> > > for the variable to be placed in the intended section.
>> > >
>> >
>> > Why?
>>
>> include/linux/init.h says something as:
>>
>>  * For initialized data:
>>  * You should insert __initdata or __initconst between the variable name
>>  * and equal sign followed by value, e.g.:
>>  *
>>  * static int init_variable __initdata = 0;
>>  * static const char linux_logo[] __initconst = { 0x32, 0x36, ... };
>>
>> and examples in gcc manual also put __attribute__ (...) after variable name.
>>
>> https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#Common-Variable-Attributes
>>
>> Then I grep the source, found most lines (especially arch/arm64/*) put the
>> __initdata and __read_mostly after the variable name.
>>
>> However, I built the code with three different gcc, the result looks identical
>> no matter where these markers put. So the commit msg looks wrong, what about
>> changes it as
>>
>> "put __initdata and __read_mostly after the variable name
>> to keep the style consistent"?
>
> After some consideration, I want to drop this patch in newer version since
> it's not a bug, just "style"
>

OK, thanks for clearing that up. I don't object to the patch, but I
wanted to get confirmation that the current code is also correct.

Thanks,
Ard.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web