VTA merge - cmdline
Alexandre Oliva
aoliva@redhat.com
Tue Jun 9 05:39:00 GMT 2009
Hi, Mark,
On Jun 7, 2009, Mark Mitchell <mark@codesourcery.com> wrote:
> --param should be used for tuning parameters for the compiler, i.e.,
> where we would otherwise be tempted to use "magic constants".
Thanks for providing a definition of what “--param” is supposed to stand
for. It was quite convincing to me, so I started changing
-fmin-insn-uid to --param. When the time came to document it under
--param, I found yet another definition, that actually matched my
understanding of the purpose of --param.
I wonder whether there's a misunderstanding as to the meaning of the new
proposed flag, or whether the manual no longer reflects the expected use
of --params.
--param is documented under “Optimization options” as “constants to
control the amount of optimization that is done”.
It doesn't make sense for min-insn-uid=N to be documented as such.
min-insn-uid=N is supposed to bump up the range of INSN_UIDs used by
non-debug insns, so that they start at N, while debug insns take up the
range 1..N-1.
This makes it easier to compare compiler dumps in VTA and non-VTA
compiles. It is not supposed to have any effect on optimization.
Whatever effects it had before were already fixed.
It is not an optimization parameter. It is a dump option, very much
along the lines of -fdump-noaddr, -fdump-unnumbered,
-fdump-unnumbered-links. It just so happens that it takes a number as
an argument, like the other dump option -fsched-verbose.
If --param is to be used for things other than constants to control the
amount of optimization, its documentation must adjusted to reflect this,
and moved where it belongs.
What now?
--
Alexandre Oliva, freedom fighter http://FSFLA.org/~lxoliva/
You must be the change you wish to see in the world. -- Gandhi
Be Free! -- http://FSFLA.org/ FSF Latin America board member
Free Software Evangelist Red Hat Brazil Compiler Engineer
More information about the Gcc-patches
mailing list