This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [RFC][PATCH] ARM: vDSO support
- From: Nathan Lynch <Nathan_Lynch at mentor dot com>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: <libc-alpha at sourceware dot org>
- Date: Tue, 25 Feb 2014 20:56:23 -0600
- Subject: Re: [RFC][PATCH] ARM: vDSO support
- Authentication-results: sourceware.org; auth=none
- References: <1393373216-30253-1-git-send-email-nathan_lynch at mentor dot com> <Pine dot LNX dot 4 dot 64 dot 1402260157500 dot 30025 at digraph dot polyomino dot org dot uk>
On 02/25/2014 08:01 PM, Joseph S. Myers wrote:
> On Tue, 25 Feb 2014, Nathan Lynch wrote:
>
>> Future versions (beginning with 3.15 or 3.16) of Linux on 32-bit ARM
>> are expected to provide fast user-space implementations of the
>> following system calls:
>>
>> - gettimeofday
>> - clock_gettime
>> - clock_getres
>>
>> This patch adds support for the ARM vDSO to glibc. The changes are
>> mostly copied from the aarch64 code, with necessary adjustments to
>> _libc_vdso_platform_setup and INTERNAL_VSYSCALL_NCS (which has
>> __thumb2__ and non-__thumb2__ versions).
>
> It would be best to add this only once the changes are actually in the
> kernel, given that it depends on a kernel/userspace ABI that involves a
> version number.
Agreed. I figured I would need more than one attempt to get this right,
so I posted this as an RFC while things are still in progress on the
kernel side.
>> Tested on OMAP5 (Cortex-A15) with the relevant kernel patch, the latest
>> posted version of which may be found here:
>
> Testing with a kernel without the feature is also important.
Yes, and I will include such results in future submissions.
>> 2014-02-25 Nathan Lynch <nathan_lynch@mentor.com>
>
> Mentor's FSF copyright assignment only applies to people with an
> @codesourcery.com address, so it is advisable to use such an address when
> posting patches to GNU projects.
Okay. Future patches will come from my codesourcery.com address.
>> +int
>> +__gettimeofday (tv, tz)
>> + struct timeval *tv;
>> + struct timezone *tz;
>
> New old-style function definitions should be avoided (I hope at some point
> we can eliminate the existing ones and start using -Wold-style-definition
> by default).
I'll fix this.
Thanks for reviewing.