[PATCHv2 6/6] arm64: Add support for CONFIG_DEBUG_VIRTUAL

Mark Rutland mark.rutland at arm.com
Thu Nov 3 08:57:54 PDT 2016


On Wed, Nov 02, 2016 at 06:05:38PM -0600, Laura Abbott wrote:
> On 11/02/2016 05:06 PM, Mark Rutland wrote:
> >On Wed, Nov 02, 2016 at 03:00:54PM -0600, Laura Abbott wrote:
> >>+CFLAGS_physaddr.o		:= -DTEXT_OFFSET=$(TEXT_OFFSET)
> >>+obj-$(CONFIG_DEBUG_VIRTUAL)	+= physaddr.o

> >>+	/*
> >>+	 * This is intentionally different than above to be a tighter check
> >>+	 * for symbols.
> >>+	 */
> >>+	VIRTUAL_BUG_ON(x < kimage_vaddr + TEXT_OFFSET || x > (unsigned long) _end);
> >
> >Can't we use _text instead of kimage_vaddr + TEXT_OFFSET? That way we don't
> >need CFLAGS_physaddr.o.
> >
> >Or KERNEL_START / KERNEL_END from <asm/memory.h>?
> >
> >Otherwise, this looks good to me (though I haven't grokked the need for
> >__pa_symbol() yet).
> 
> I guess it's a question of what's clearer. I like kimage_vaddr +
> TEXT_OFFSET because it clearly states we are checking from the
> start of the kernel image vs. _text only shows the start of the
> text region. Yes, it's technically the same but a little less
> obvious. I suppose that could be solved with some more elaboration
> in the comment.

Sure, it's arguable either way.

I do think that KERNEL_START/KERNEL_END are a better choice, with the comment
you suggest, and/or renamed to KERNEL_IMAGE_*. They already describe the bounds
of the image (though the naming doesn't make that entirely clear).

Thanks,
Mark.



More information about the linux-arm-kernel mailing list