[PATCH/RFC] syscall: unify common syscall defines

Mike Frysinger vapier at gentoo.org
Mon Jul 13 10:04:42 UTC 2009


On Friday 10 July 2009 13:43:24 Lennart Sorensen wrote:
> On Thu, Jul 09, 2009 at 06:24:54PM -0400, Mike Frysinger wrote:
> > On Monday 06 July 2009 10:40:35 Lennart Sorensen wrote:
> > > On Sun, Jul 05, 2009 at 06:28:32PM -0400, Mike Frysinger wrote:
> > > > On Friday 03 July 2009 16:50:22 Mike Frysinger wrote:
> > > > > Unify all the common syscall defines in syscalls-common.h and scrub
> > > > > all the duplicated code from relevant ports.  This should also make
> > > > > converting existing ports to INLINE_SYSCALL() much easier as they
> > > > > don't have to get lost in all the unrelated noise, as well as
> > > > > creating new ports.
> > > > > ---
> > > > > can people give this an eye over to make sure i didnt screw up your
> > > > > port ? i'm doing build tests for most arches, but a quick look over
> > > > > by other peeps would be good.
> > > >
> > > > sadly (and ironically), this causes build breakage for the arches
> > > > that do not define INLINE_SYSCALL() due to the weird hacks we have in
> > > > a few files that try to work around ports that have not yet been
> > > > converted. specifically, the fallback code in _exit, ftruncate64,
> > > > truncate64, sync, and llseek.
> > > >
> > > > Lennart: do you need help with this ?  or is this patch sufficient to
> > > > get you started ?
> > >
> > > I can do a compile test to see if it compiles or not.  If it compiles
> > > I can even run it.
> >
> > sorry, what i meant was to give you a basis to easily convert m68k.  this
> > patch isnt going to fix your building problem.  in fact, it introduces
> > more for you :/.
> >
> > the attached patch is pretty much a copy & paste from glibc.  see if it
> > works for you.
>
> I don't understand the macro hell that is involved in syscalls.  I only
> complained that someone added the daemon call for nommu systems without
> making sure that it didn't break builds (which it did on m68k and anything
> else that doesn't have these syscall macros).

there is only so long that unmaintained ports will hold back progress.  in 
this case, m68k has gone past that point.

> Hopefully someone that does understand syscalls and libc stuff can write
> the macros however they are supposed to be done.

i get the syscall stuff, but i dont get m68k assembly.

> > index d1ade80..0dc6b19 100644
> > --- a/libc/sysdeps/linux/m68k/bits/syscalls.h
> > +++ b/libc/sysdeps/linux/m68k/bits/syscalls.h
>
> I can give that a try (although probably not until after the linux
> symposium, unless I get bored).

np
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.busybox.net/pipermail/uclibc/attachments/20090713/95b541b6/attachment.pgp>


More information about the uClibc mailing list