When you say "float math isn't working", what do you mean? When you say
that the new binary
only works with the 2.6 kernel, what do you mean? Are you getting bad
answers, or is the program
crashing? If it's crashing, and you run it under gdb, where is it
blowing up, and why? Are you able
to re-create the "old" binary? If you do a static link of the program
using the old tools,
does the resulting binary work better under 2.6?
The FPU emulator has bad some minor cleanup over the past few years, but
it's been pretty
stable, and I can't recall having seen any serious functional or
compatibility problems in a long
time. I think it's more likely that you've got a problem with libraries
and/or build procedures
for the application.
Klatt Uwe wrote:
I have a custom hardware (AU1100) with kernel 2.4 and an working binary using
floats (compiled with gcc 3.3.5).
Now I am testing with kernel 2.6.
When I use the old binary, float math isn't working anymore.
I have to recompile the source with new gcc 4.1.2 but then the new binary is
working only on kernel 2.6.
Can somebody give me some hints, how to chage settings for kernel 2.6 creation
or compiler settings to generate an universal binary.