[PATCH 00/20] Create and use hidden aliases in libgmp.so

Torbjorn Granlund tg at gmplib.org
Mon Mar 4 23:07:07 CET 2013

Richard Henderson <rth at twiddle.net> writes:

  The patch set is intended to be fully bisectable.  Each patch in sequence
  should work.  That said, I can't think of any real value in keeping the patches
  separate in the final commit.  The intermediate steps are not useful.

  > I don't see any problem which needs addressing.  I am perfectly happy
  > with how debugging works with the asm code now.
  Matter of taste, I suppose.
I didn't realise the deleted .s file problem.  I always do

  (cd mpn; m4 file.asm)>x.s && gcc -g x.s ...

when I debug, since that effetively zeros turnaround time.

Perhaps we should stop deleting the .s file?

  I generally do that with the assembler itself.  E.g.
  	call	1f
  	.subsection 2
  1:	mov	(%esp), %ebx
  	addl	$_GLOBAL_OFFSET_POINTER_, %ebx
Lucky you to have a powerful assembler all the time...

I actually really like m4.  It is tiny yet really powerful, being Turing
complete.  A friend of mine implemented a bignum library in m4, inspired
by GMP's m4 usage, I think.  Convinced of m4's greatness now?  :-)


More information about the gmp-devel mailing list