This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Consensus around kernel syscall wrappers?
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: Joseph Myers <joseph at codesourcery dot com>, "H. Peter Anvin" <hpa at zytor dot com>, Roland McGrath <roland at hack dot frob dot com>
- Cc: Mike Frysinger <vapier at gentoo dot org>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Wed, 2 Sep 2015 17:05:05 -0400
- Subject: Re: Consensus around kernel syscall wrappers?
- Authentication-results: sourceware.org; auth=none
- References: <55CCF8C6 dot 3060007 at redhat dot com> <20150824213059 dot A44862C39E4 at topped-with-meat dot com> <55E75996 dot 7080502 at zytor dot com> <alpine dot DEB dot 2 dot 10 dot 1509022057310 dot 30448 at digraph dot polyomino dot org dot uk>
On 09/02/2015 05:01 PM, Joseph Myers wrote:
> On Wed, 2 Sep 2015, H. Peter Anvin wrote:
>
>> On 08/24/2015 02:30 PM, Roland McGrath wrote:
>>> There is no consensus and I am behind on continuing the previous discussion.
>>
>> OK, so let me ask this: is there a reason these ought to be in glibc as
>> opposed to co-maintained with the Linux kernel?
>
> Cancellation (libc-dependent). errno (libc-dependent). Userspace types
> and error handling choices to fit in properly with other related
> interfaces in libc. Building objects for various ABIs that the kernel is
> never built for or whose ABI distinctions are only relevant at userspace
> level and not for the kernel. All these things naturally fit in as part
> of the glibc build.
Agreed. Cancellation was the first one that comes to mind, and then userspace
types. It's all a natural fit for glibc because we're already doing all of
this.
I have also stated clearly what I think the next step should be:
https://sourceware.org/ml/libc-alpha/2015-08/msg00576.html
Roland, What are we missing?
Cheers,
Carlos.