On Sat, 5 Jan 2008, Arnaldo Carvalho de Melo wrote:
> Em Sat, Jan 05, 2008 at 03:39:04PM +0200, Ilpo Järvinen escreveu:
> > Hi Dave,
> >
> > After Arnaldo got codiff's inline instrumentation bugs fixed
> > (thanks! :-)), I got my .c-inline-bloat-o-meter to power up
> > reliably after some tweaking and bug fixing on my behalf...
> > It shows some very high readings every now and then in the
> > code under net/.
>
> Thank you for the reports and for showing how these tools can be put to
> good use!
>
> If you have any further suggestions on how to make codiff and the
> dwarves to be of more help or find any other bug, please let me know.
It could use a bit less memory because my header inline checking attempt
on a machine with 2G+2G ends up like this:
$ codiff vmlinux.o.base vmlinux.o
libclasses: out of memory(inline_expansion__new)
$ ls -al vmlinux.o{,.base}
-rw-r----- 1 ijjarvin tkol 633132586 Jan 9 13:11 vmlinux.o
-rw-r----- 1 ijjarvin tkol 633132572 Jan 9 00:58 vmlinux.o.base
$
Considering that's only 0.6G+0.6G I've a problem in understanding why
codiff's number crunching eats up so much memory.
I just hope there isn't any O(n^2) or worse algos in it either once
the memory consumption gets resolved. :-)
--
i.
