[CRIU] [PATCH 1/2] DO NOT MERGE: Move x86 VDSO code to AArch64

[Christopher Covington]

Hi Cyrill,

Thanks for looking at the changes. Adding the mailing list.

On 08/01/2014 11:20 AM, Cyrill Gorcunov wrote:
> On Fri, Aug 01, 2014 at 05:47:44PM +0400, Cyrill Gorcunov wrote:
>> On Fri, Aug 01, 2014 at 09:46:12AM -0400, Christopher Covington wrote:
>>>
>>> I was seeing multiple regions of memory identified as the vdso, but only one
>>> of them matched what was reported elsewhere in the logs. When I added this
>>> test, only the apparently correct region got processed as the vdso. Would it
>>> be helpful to show some logs?
>>
>> yes please

I'm working to regenerate such logs.

> Christopher, you know I've looked into a number of approaches and it seems
> the best way would be not to share the code but rather copy it. Simply
> because it's quite likely the vdso structure may be changed in future so
> that even Elf parsing might be different between arm and x86 (you remember
> that previously I've been using static Elf structure parsing which didn't
> consider dynamic symbols).

What makes you predict the code will become more different between
architectures? The interactions I've seen on the Linux kernel mailing lists
indicate that the developers (Andy, Will) are now working more closely with
each other and their changes are making things more similar across architectures.

>   So, I think
> 
>  - copy the files you need into ARM's arch
>  - tune them up to be workable on ARM
> 
> this is safest from my POV.

I'll submit patches like this in the sort term, based on your vdso/vvar
re-ordering series. Are there any changes needed to them before they can be
merged?

Thanks,
Christopher

-- 
Employee of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by the Linux Foundation.