This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Avoid two SSP ABI's for AArch64.
- From: Venkataramanan Kumar <venkataramanan dot kumar at linaro dot org>
- To: "Carlos O'Donell" <carlos at redhat dot com>
- Cc: Richard Earnshaw <rearnsha at arm dot com>, Marcus Shawcroft <marcus dot shawcroft at linaro dot org>, Will Newton <will dot newton at linaro dot org>, "Joseph S. Myers" <joseph at codesourcery dot com>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Tue, 10 Dec 2013 11:06:16 +0530
- Subject: Re: Avoid two SSP ABI's for AArch64.
- Authentication-results: sourceware.org; auth=none
- References: <529CC927 dot 3040800 at redhat dot com> <CABXK9nejJW9EA7Y8edbt9ugC-Y5GvzSJpSBzLKpT7tkKfGqWuw at mail dot gmail dot com> <52A1ECBA dot 4090400 at redhat dot com> <52A1EDB5 dot 3040807 at arm dot com> <52A34409 dot 4060902 at redhat dot com> <52A596BB dot 3030803 at arm dot com> <52A5DE46 dot 7080500 at redhat dot com>
Hi Carlos,
On 9 December 2013 20:44, Carlos O'Donell <carlos@redhat.com> wrote:
> On 12/09/2013 05:08 AM, Richard Earnshaw wrote:
>> Thanks,
>>
>> I do not expect to need to change/add the alternative ABI interface, but
>> knowing that it should be technically possible means I can go forward
>> with what I believe today to be the correct decision with much more
>> confidence.
>>
>> So on that basis I think we should stick with the default ABI -- ie
>> using __stack_chk_guard.
>
> Thank you. In that case we have consensus to use the existing ABI which
> loads the stack guard from a symbol for stack smashing protection.
>
> That means that Kumar need to drop his two patches for gcc and glibc.
>
> I consider the glibc patch dropped (since I was the reviewer).
>
> Cheers,
> Carlos.
>
Thanks Carlos, I am fine with dropping the "glibc" patch, given that
we are going support to default ABI.
On the GCC side, many ports emit stack protection set and testing
code as template of instructions in their back end. Also the register
which loads canary value gets cleared though an extra instruction.
This is however a GCC question to Marcus on doing this for Aarch64.
regards,
Venkat.