Re: [PATCH] show_trace() module_end = 0?

Hugh Dickins (hugh@veritas.com)
Thu, 12 Jul 2001 18:29:50 +0100 (BST)


On Thu, 12 Jul 2001, Linus Torvalds wrote:
> On Thu, 12 Jul 2001, Hugh Dickins wrote:
> > show_trace() contains an erroneous line, introduced in 2.4.6-pre4,
> > which disables trace on module text: appears to be from temporary
> > testing, since code and comments for tracing module text remain.
>
> It as actually disabled on purpose.
>
> It's there because without it the backtrace is sometimes so full of crud
> that it is almost impossible to read.
>
> I chose to disable the module back-trace, because what we should _really_
> do is to walk the vmalloc space and verify whether it's a valid address or
> not. But as I don't use modules myself, I didn't have much incentive to do
> so, or to test that it worked.

Thanks for owning up to that line! I see your point.

I fear that most values found on the stack and within the vmalloc address
range would actually turn out to be valid addresses. We probably should
not re-enable show_trace() on module text without something (in vm_struct?)
to distinguish module text from data, bss, and from other uses of vmalloc.

Hugh

-
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/