Subject: Re: RFC: Change SWI number base?
To: None <email@example.com>
From: David Laight <David.Laight@btinternet.com>
Date: 01/08/2002 16:07:03
> > Ok - make the thumb code:
> > .balign 4
> > .short syscall & 0xffff
> > entry:
> > swi syscall >> 16
> > bcs __cerror
> > ret
> > And do a 32 bit load from pc-4 (as in the current ARM version).
> That is a truly evil hack. Well done.
How we (netbsd) does the system call is something we will have to live
with - and own up to...
The above code has the simplicity of making the thumb code match the
existing arm interface.
It also leaves the glorious hack in one .S file.
I suspect it is also almost the cheapest, the swi code has the user PC
lurking - so the only issue is reading the syscall number into the data
Loading a 24bit constant takes 3 instructions on ARM (or a data read),
Doing it without a data read in thumb must be more - likely to spill
the stub into a second cache line?
I'd vote for the small stub...
> Unfortunately, it won't work on big-endian systems (which we might well
> end up supporting some day).
Has anyone built any hardware that expects to run arm big-endian?
I guess (like 68k) it would require byte ports on D24-31, so you
can't (sensibly) run anything other than what the hardware designer