On Fri, Feb 21, 2003 at 12:34:29PM +0100, Pavel Machek wrote:
> Hi!
>
>
> > > Currently, 32-bit emulation in kernel has *5* copies, and its >1000
> > > lines each.
> >
> > Yes :/ Consolidating all these copies into a single layer has been a
> > "project to be" for quite some time.
> >
> > I do not know if it is too late in 2.5.x to begin this work, however.
> > We _are_ in a feature freeze... I suppose it is up to the consensus of
> > arch maintainers, because it [obviously] does not affect ia32.
>
> Actually Andi asked me to do the work. Dave, is it okay with you? What
> about other maintainers?
One issue you need to be careful about is that long long has different
alignment between 32bit and 64bit on ia64 and x86-64. On sparc64/mips64/ppc64
etc. that isn't the case. The x86-64 handlers convert sometimes more than
the later ones because of that.
Also some ioctl handlers are endian dependent, at least in x86-64 (I think
I commented them all). Not sure if Dave did the same.
First step probably is to just get register_ioctl32_translation into
a common header and implementation file.
-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Feb 23 2003 - 22:00:33 EST