Linus Torvalds wrote:
>
> On Sat, 9 Feb 2002, Andrew Morton wrote:
> > >
> >
> > Is better, except the filename gets expanded multipe times into
> > the object file. How about:
> >
> > #define BUG() \
> > asm( "ud2\n" \
> > "\t.word %0\n" \
> > "\t.long %1\n" \
> > : : "i" (__LINE__), "i" (__FILE__))
>
> Even better.
>
> That way you can actually totally remove the "verbose bug" config option,
> because even the verbose BUG's aren't actually using up any noticeable
> amounts of space.
>
> This is all assuming that gcc doesn't create the string for inline
> functions that aren't used, which it probably cannot, so maybe this
> doesn't work out.
>
gcc generally get it wrong - unreferenced strings still appear
in the object code from multiple usage patterns. I think this
was fixed about six months ago.
But yes, the verbose BUG overhead is now six bytes per BUG, plus
a few bytes per file for the filename. And it's my opinion that
the non-verbose BUG option is undesirable - it's making the
developers' job harder. Seems that with this change, the reasons
for CONFIG_DEBUG_BUGVERBOSE are no longer with us, and it can
disappear.
-
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Feb 15 2002 - 21:00:28 EST