One reason that we do not build our kernels for
Linux/390 with -gstabs is that we realized, that
gcc produces less efficient code with -gstabs.
But we consider this as a gcc bug and will investigate
this further.
Michael
------------------------------------------------------------------------
Linux/390 Development
Phone: +49-7031-16-2360, Bld 71032-06-109
Email: holzheu@xxxxxxxxxx
"Matt D. Robinson" <yakker@xxxxxxxxxxxxxx> on 07/18/2001 11:21:50 PM
Please respond to "Matt D. Robinson" <yakker@xxxxxxxxxxxxxx>
To: Dave Anderson <anderson@xxxxxxxxxxx>
cc: lkcd@xxxxxxxxxxx
Subject: Re: Module support
Dave Anderson wrote:
>
> Most people won't put up with building their entire kernel -gstabs.
> If that were the case, we would have done that a long time ago. I
> think it would be _GREAT_ to have that, but most people gag on the
> idea of putting all symbols into the kernel, as it creates huge
> kernels.
>
> I've never quite understood this. We build our kernels
> with -g for the whole kernel, and to be sure, it
> creates a much larger vmlinux file -- maybe on the
> order of 4 times as large. But what gets loaded
> into the bzImage file, and subsequently into memory,
> hardly changes at all. In fact, if you also delete
> -fomit_frame_pointer along with adding -g, the kernel
> is actually smaller! (I'm guessing because of less
> aggressive in-lining?).
You omit the frame pointer because it takes an extra register
on x86 systems, which can slow the machine down tremendously
(it has to do more with fewer registers).
> Am I missing something here?
>
> Dave Anderson
No, you're not missing the point. People just don't like to see
large kernels. Trust me, I'd like to see -gstabs as a default
for all systems, but Linus gagged on the idea about two years
ago when I asked. Maybe things are different now ... Heh. :)
Nice to see you again, Dave ... it's been a while. :)
--Matt
|