[PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[Andy Lutomirski]

The new symbols provide the same API as the 64-bit variants, so they
should have the same symbol version name.  This can't break
userspace, since these symbols are new for 32-bit Linux.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
---
 arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S b/arch/x86/vdso/vdso32/vdso32.lds.S
index 28c4607..aadb8b9 100644
--- a/arch/x86/vdso/vdso32/vdso32.lds.S
+++ b/arch/x86/vdso/vdso32/vdso32.lds.S
@@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
  */
 VERSION
 {
+	LINUX_2.6 {
+	global:
+		__vdso_clock_gettime;
+		__vdso_gettimeofday;
+		__vdso_time;
+	};
+
 	LINUX_2.5 {
 	global:
 		__kernel_vsyscall;
 		__kernel_sigreturn;
 		__kernel_rt_sigreturn;
-		__vdso_clock_gettime;
-		__vdso_gettimeofday;
-		__vdso_time;
 	local: *;
 	};
 }
-- 
1.9.0

Re: [PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[Stefani Seibold]

Am Samstag, den 29.03.2014, 13:15 -0700 schrieb Andy Lutomirski:
> The new symbols provide the same API as the 64-bit variants, so they
> should have the same symbol version name.  This can't break
> userspace, since these symbols are new for 32-bit Linux.
> 

This breaks my glibc patch... What is the benefit for doing this in that
way?

> Signed-off-by: Andy Lutomirski <luto@amacapital.net>
> ---
>  arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S b/arch/x86/vdso/vdso32/vdso32.lds.S
> index 28c4607..aadb8b9 100644
> --- a/arch/x86/vdso/vdso32/vdso32.lds.S
> +++ b/arch/x86/vdso/vdso32/vdso32.lds.S
> @@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
>   */
>  VERSION
>  {
> +	LINUX_2.6 {
> +	global:
> +		__vdso_clock_gettime;
> +		__vdso_gettimeofday;
> +		__vdso_time;
> +	};
> +
>  	LINUX_2.5 {
>  	global:
>  		__kernel_vsyscall;
>  		__kernel_sigreturn;
>  		__kernel_rt_sigreturn;
> -		__vdso_clock_gettime;
> -		__vdso_gettimeofday;
> -		__vdso_time;
>  	local: *;
>  	};
>  }

Re: [PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[Andy Lutomirski]

On Sat, Mar 29, 2014 at 11:33 PM, Stefani Seibold <stefani@seibold.net> wrote:
> Am Samstag, den 29.03.2014, 13:15 -0700 schrieb Andy Lutomirski:
>> The new symbols provide the same API as the 64-bit variants, so they
>> should have the same symbol version name.  This can't break
>> userspace, since these symbols are new for 32-bit Linux.
>>
>
> This breaks my glibc patch... What is the benefit for doing this in that
> way?

The functions do the same thing as they do in the 64-bit and x32
cases, so the versions should be the same.

>
>> Signed-off-by: Andy Lutomirski <luto@amacapital.net>
>> ---
>>  arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
>>  1 file changed, 7 insertions(+), 3 deletions(-)
>>
>> diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S b/arch/x86/vdso/vdso32/vdso32.lds.S
>> index 28c4607..aadb8b9 100644
>> --- a/arch/x86/vdso/vdso32/vdso32.lds.S
>> +++ b/arch/x86/vdso/vdso32/vdso32.lds.S
>> @@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
>>   */
>>  VERSION
>>  {
>> +     LINUX_2.6 {
>> +     global:
>> +             __vdso_clock_gettime;
>> +             __vdso_gettimeofday;
>> +             __vdso_time;
>> +     };
>> +
>>       LINUX_2.5 {
>>       global:
>>               __kernel_vsyscall;
>>               __kernel_sigreturn;
>>               __kernel_rt_sigreturn;
>> -             __vdso_clock_gettime;
>> -             __vdso_gettimeofday;
>> -             __vdso_time;
>>       local: *;
>>       };
>>  }
>
>



-- 
Andy Lutomirski
AMA Capital Management, LLC

Re: [PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[H. Peter Anvin]

Has your glibc patch been accepted upstream yet? I'm assuming not, in which case this is a legitimate patch.

On March 29, 2014 11:33:37 PM PDT, Stefani Seibold <stefani@seibold.net> wrote:
>Am Samstag, den 29.03.2014, 13:15 -0700 schrieb Andy Lutomirski:
>> The new symbols provide the same API as the 64-bit variants, so they
>> should have the same symbol version name.  This can't break
>> userspace, since these symbols are new for 32-bit Linux.
>> 
>
>This breaks my glibc patch... What is the benefit for doing this in
>that
>way?
>
>> Signed-off-by: Andy Lutomirski <luto@amacapital.net>
>> ---
>>  arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
>>  1 file changed, 7 insertions(+), 3 deletions(-)
>> 
>> diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S
>b/arch/x86/vdso/vdso32/vdso32.lds.S
>> index 28c4607..aadb8b9 100644
>> --- a/arch/x86/vdso/vdso32/vdso32.lds.S
>> +++ b/arch/x86/vdso/vdso32/vdso32.lds.S
>> @@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
>>   */
>>  VERSION
>>  {
>> +	LINUX_2.6 {
>> +	global:
>> +		__vdso_clock_gettime;
>> +		__vdso_gettimeofday;
>> +		__vdso_time;
>> +	};
>> +
>>  	LINUX_2.5 {
>>  	global:
>>  		__kernel_vsyscall;
>>  		__kernel_sigreturn;
>>  		__kernel_rt_sigreturn;
>> -		__vdso_clock_gettime;
>> -		__vdso_gettimeofday;
>> -		__vdso_time;
>>  	local: *;
>>  	};
>>  }

-- 
Sent from my mobile phone.  Please pardon brevity and lack of formatting.

Re: [PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[Stefani Seibold]

I think it is in review. I still wait for the confirmation of my assign
form.

Am Sonntag, den 30.03.2014, 09:59 -0700 schrieb H. Peter Anvin:
> Has your glibc patch been accepted upstream yet? I'm assuming not, in which case this is a legitimate patch.
> 
> On March 29, 2014 11:33:37 PM PDT, Stefani Seibold <stefani@seibold.net> wrote:
> >Am Samstag, den 29.03.2014, 13:15 -0700 schrieb Andy Lutomirski:
> >> The new symbols provide the same API as the 64-bit variants, so they
> >> should have the same symbol version name.  This can't break
> >> userspace, since these symbols are new for 32-bit Linux.
> >> 
> >
> >This breaks my glibc patch... What is the benefit for doing this in
> >that
> >way?
> >
> >> Signed-off-by: Andy Lutomirski <luto@amacapital.net>
> >> ---
> >>  arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
> >>  1 file changed, 7 insertions(+), 3 deletions(-)
> >> 
> >> diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S
> >b/arch/x86/vdso/vdso32/vdso32.lds.S
> >> index 28c4607..aadb8b9 100644
> >> --- a/arch/x86/vdso/vdso32/vdso32.lds.S
> >> +++ b/arch/x86/vdso/vdso32/vdso32.lds.S
> >> @@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
> >>   */
> >>  VERSION
> >>  {
> >> +	LINUX_2.6 {
> >> +	global:
> >> +		__vdso_clock_gettime;
> >> +		__vdso_gettimeofday;
> >> +		__vdso_time;
> >> +	};
> >> +
> >>  	LINUX_2.5 {
> >>  	global:
> >>  		__kernel_vsyscall;
> >>  		__kernel_sigreturn;
> >>  		__kernel_rt_sigreturn;
> >> -		__vdso_clock_gettime;
> >> -		__vdso_gettimeofday;
> >> -		__vdso_time;
> >>  	local: *;
> >>  	};
> >>  }
> 

Re: [PATCH] x86: Fix the symbol versions on the 32-bit vDSO

[H. Peter Anvin]

Well, I applied the patch.  This is not ABI yet and Andy is right it makes more sense.

On March 30, 2014 12:00:40 PM PDT, Stefani Seibold <stefani@seibold.net> wrote:
>I think it is in review. I still wait for the confirmation of my assign
>form.
>
>Am Sonntag, den 30.03.2014, 09:59 -0700 schrieb H. Peter Anvin:
>> Has your glibc patch been accepted upstream yet? I'm assuming not, in
>which case this is a legitimate patch.
>> 
>> On March 29, 2014 11:33:37 PM PDT, Stefani Seibold
><stefani@seibold.net> wrote:
>> >Am Samstag, den 29.03.2014, 13:15 -0700 schrieb Andy Lutomirski:
>> >> The new symbols provide the same API as the 64-bit variants, so
>they
>> >> should have the same symbol version name.  This can't break
>> >> userspace, since these symbols are new for 32-bit Linux.
>> >> 
>> >
>> >This breaks my glibc patch... What is the benefit for doing this in
>> >that
>> >way?
>> >
>> >> Signed-off-by: Andy Lutomirski <luto@amacapital.net>
>> >> ---
>> >>  arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
>> >>  1 file changed, 7 insertions(+), 3 deletions(-)
>> >> 
>> >> diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S
>> >b/arch/x86/vdso/vdso32/vdso32.lds.S
>> >> index 28c4607..aadb8b9 100644
>> >> --- a/arch/x86/vdso/vdso32/vdso32.lds.S
>> >> +++ b/arch/x86/vdso/vdso32/vdso32.lds.S
>> >> @@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
>> >>   */
>> >>  VERSION
>> >>  {
>> >> +	LINUX_2.6 {
>> >> +	global:
>> >> +		__vdso_clock_gettime;
>> >> +		__vdso_gettimeofday;
>> >> +		__vdso_time;
>> >> +	};
>> >> +
>> >>  	LINUX_2.5 {
>> >>  	global:
>> >>  		__kernel_vsyscall;
>> >>  		__kernel_sigreturn;
>> >>  		__kernel_rt_sigreturn;
>> >> -		__vdso_clock_gettime;
>> >> -		__vdso_gettimeofday;
>> >> -		__vdso_time;
>> >>  	local: *;
>> >>  	};
>> >>  }
>> 

-- 
Sent from my mobile phone.  Please pardon brevity and lack of formatting.

[tip:x86/vdso] x86, vdso: Fix the symbol versions on the 32-bit vDSO

[tip-bot for Andy Lutomirski]

Commit-ID:  37c975545ec63320789962bf307f000f08fabd48
Gitweb:     http://git.kernel.org/tip/37c975545ec63320789962bf307f000f08fabd48
Author:     Andy Lutomirski <luto@amacapital.net>
AuthorDate: Sat, 29 Mar 2014 13:15:35 -0700
Committer:  H. Peter Anvin <hpa@zytor.com>
CommitDate: Sun, 30 Mar 2014 10:08:38 -0700

x86, vdso: Fix the symbol versions on the 32-bit vDSO

The new symbols provide the same API as the 64-bit variants, so they
should have the same symbol version name.  This can't break
userspace, since these symbols are new for 32-bit Linux.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Cc: Stefani Seibold <stefani@seibold.net>
Link: http://lkml.kernel.org/r/0a869bce03d25619565b1eee7d69a4fd15fd203a.1396124118.git.luto@amacapital.net
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
---
 arch/x86/vdso/vdso32/vdso32.lds.S | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/arch/x86/vdso/vdso32/vdso32.lds.S b/arch/x86/vdso/vdso32/vdso32.lds.S
index 28c4607..aadb8b9 100644
--- a/arch/x86/vdso/vdso32/vdso32.lds.S
+++ b/arch/x86/vdso/vdso32/vdso32.lds.S
@@ -23,14 +23,18 @@ ENTRY(__kernel_vsyscall);
  */
 VERSION
 {
+	LINUX_2.6 {
+	global:
+		__vdso_clock_gettime;
+		__vdso_gettimeofday;
+		__vdso_time;
+	};
+
 	LINUX_2.5 {
 	global:
 		__kernel_vsyscall;
 		__kernel_sigreturn;
 		__kernel_rt_sigreturn;
-		__vdso_clock_gettime;
-		__vdso_gettimeofday;
-		__vdso_time;
 	local: *;
 	};
 }