[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <f39de4a1c484f7ba593cd674da72ad85@smtp.hushmail.com>
Date: Sat, 11 May 2013 19:24:53 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: Unused global symbols in jumbo
I made the MD5crypt helper functions in sse-intrinsics.c static, which made no difference in itself. So I also added MAYBE_INLINE. There was a 1-2% boost from that here.
magnum
On 11 May, 2013, at 19:03 , magnum <john.magnum@...hmail.com> wrote:
> Lots of real "issues" are shown though. Why do we not declare the mmxput stuff in sse-intrinsics.c as static? I hope this doesn't mean they're not inlined if otherwise applicable... And there seem to be lot's of missing "static" for functions in EPI_fmt_plug.c and other files.
>
> magnum
>
>
> On 11 May, 2013, at 18:35 , <jfoug@....net> wrote:
>
>> Many of these have to be global. There were issues in some compiling env's (cygwin??) where it would not align static variables. Thus, we have #defines in each source, which mangles the normal name (crypt) into something unique to the format (MD5Gen_crypt, etc). Yes, they are globals, but steps have been taken to make sure they are linker safe.
>>
>> There is little/no gain worrying about unused globals. The only time it matters at all, is at link time. The linker will tell you if there are problems.
>>
>> Jim.
>>
>> ---- Alexander Cherepanov <cherepan@...me.ru> wrote:
>>> Hi, all!
>>>
>>> There are 331 unused globals in unstable-jumbo -- list attached. It's
>>> generated by compiling unstable with debugging information and running
>>> the following command:
>>>
>>> $ nm -olg *.o | perl unused-globals.pl | sort -k1.40 > unused-globals.txt
>>>
>>> unused-globals.pl is also attached.
>>>
>>> --
>>> Alexander Cherepanov
>>
>>
>
>
>
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
