[PATCH RFC] Support newer kernels lacking noat syscalls

Mark Salter msalter at redhat.com
Tue Apr 24 16:02:31 UTC 2012


On Tue, 2012-04-24 at 17:48 +0200, Bernhard Reutner-Fischer wrote:
> On 24 April 2012 17:35, Mark Salter <msalter at redhat.com> wrote:
> > On Tue, 2012-04-24 at 17:00 +0200, Bernhard Reutner-Fischer wrote:
> >> On 24 April 2012 16:50, Mark Salter <msalter at redhat.com> wrote:
> >> > Upstream Linux kernel development is requiring new architecture ports to
> >> > use only the default set of generic syscalls. This means familiar syscalls
> >>
> >> http://lists.uclibc.org/pipermail/uclibc/2011-September/045726.html
> >>
> >
> > Heh. I knew there had to be patch out there. I saw some other older ones
> > for specific syscalls, but never saw yours in my searching.
> >
> > Looks pretty close to what I've been using but doesn't seem to have
> > generated much interest. I missed the no_cancel bits, but I figured
> > there would be such things broken in my patch too.
> >
> > So, is there any other concerns besides the cancellation? I saw some
> > concern about bloat, but if the approach is to use the noat syscalls
> > if they exist, it shouldn't bloat ports using older kernels.
> 
> Do you have size(1) measures for using the *at() versus noat?

I have some numbers, but they are not apples to apples (different uClibc
versions). I can generate some more useful numbers and post them.




More information about the uClibc mailing list