[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Linux kernel and floating point.

"Aneesh Kumar K.V" <aneesh.kumar@digital.com> writes:

> Hi, 
>  Can i  afford to have floating point calculation inside Alpha kernel. I
> found some information saying that floating point is not allowed. Right
> now the situation is if i compile the code with -mno-fp-regs i get lot
> of unresolved symbols where as removing this option from the CFLAGS
> makes every thing work properly. 
> Is the only solution is to change the floating point to fixed
> arithmetic.The code under discussion is  in openmosix/mosix project 
> mos/balance.c file 

The kernel does not use the FPU at all. The benefit is that it doesn't
need to save/restore the fpu registers inside the kernel.

You can probably get away with storing, using and restoring registers
in certain well controled conditions (like inside pal code?) But you
better make damn sure your realy can.

Most likely you would have to disable all interrupt for the duration
of the fpu usage. It might be way saver and equally fast to use
fixpoint arithmetic.


To UNSUBSCRIBE, email to debian-alpha-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: