On Sat, Mar 25, 2006 at 02:28:41AM -0000, Yan Seiner wrote:
> > Are you using crunch, hardfpa, softvfp, something else?
>
> I don't know.... I am using the standard 2.4 ts-9 kernel floating
> point....
Right.
FPA is the floating point coprocessor that nobody has anymore, but
gcc emits code for by default (if you use hard-float), and the linux
kernel emulates (nwfpe/fastfpe.)
The ep93xx have a hardware floating point coprocessor, which is called
MaverickCrunch or simply crunch.
You can also ask gcc to do all floating point computations 'in software'
(i.e. without using a hardware floating point coprocessor or trapping to
the linux kernel.) This is called soft-float. When in this mode, there
are two possible word orders it can use: FPA word ordering (which is
compatible with the FPA coprocessor, where the byte order is native-endian
(so that's little-endian on LE ep93xx systems) but big-endian word order),
and VFP word ordering (which is fully native byte and word ordering.)
So, since there are so many possibilities, whenever there is a floating
point problem on an ARM box the first problem is to find out which floating
point method is actually being used!
Which gcc version are you using? Do a "gcc -dumpspecs" and then check
what it gives for "multilib_defaults". This is from a big-endian IXP
box running debian:
*multilib_defaults:
marm mbig-endian mhard-float mapcs-32 mno-thumb-interwork
--L
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/ts-7000/
<*> To unsubscribe from this group, send an email to:
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|