This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [v3] libstdc++/44413 (for ext/vstring)
- From: Doug Semler <dougsemler at gmail dot com>
- To: Paolo Carlini <paolo dot carlini at oracle dot com>
- Cc: "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>, "libstdc++" <libstdc++ at gcc dot gnu dot org>
- Date: Wed, 9 Jun 2010 09:52:18 -0400
- Subject: Re: [v3] libstdc++/44413 (for ext/vstring)
- References: <4C0F5D67.9010405@oracle.com>
On Wed, Jun 9, 2010 at 5:22 AM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> Hi,
>
> let's do this for ext/vstring first, and let's see how people react. I'm
> still reluctant to do it for basic_string, given that the compare
> functions are exported and legacy apps are at risk of behaving
> differently all of a sudden. Committed to mainline.
>
I don't see it. You're really hurting the 32 bit side with this. The
original is reduced to a subtraction, while the new code adds:
xorl %eax, %eax
cmpl $0, %edx
je .L3
setg %al
movzbl %al, %eax
leal -1(%eax,%eax), %eax
.L3
ret
In addition, it seems that the attempt only reduces the code by one move?