Re: Koelsch boot broken on 3.18-rc1 ?

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Geert,

On Thursday 23 October 2014 12:59:03 Geert Uytterhoeven wrote:
> On Thu, Oct 23, 2014 at 12:53 PM, Laurent Pinchart wrote:
> > While rebasing my rcar-dmac patches on top of 3.18-rc1 I ran into issues
> > booting the Koelsch board.
> > 
> > Bisecting the problem pointed to
> > 
> > commit 95b0e655f9148881907fdbe5baba6a9f5d094fee
> > Author: Marek Szyprowski <m.szyprowski@samsung.com>
> > Date:   Thu Oct 9 15:26:49 2014 -0700
> > 
> >     ARM: mm: don't limit default CMA region only to low memory
> > 
> > With LPAE enabled this results in the sh-eth driver failing with
> > 
> > [    1.170000] irq 194: nobody cared (try booting with the "irqpoll"
> > option)
> > 
> > and from time to time with an Address Error message as well (which got
> > pushed out of my console log, I'll try to reproduce that again to get the
> > exact message).
> 
> renesas-drivers-2014-10-23-v3.18-rc1 works on Koelsch for me, with LPAE.

I'll test it.

> > This is likely related to a failure earlier in the boot process:
> > 
> > [    0.000000] cma: Reserved 256 MiB at 70000000
> > [    0.000000] cma: Reserved 64 MiB at 6c000000
> > [    0.000000] Memory policy: Data cache writealloc
> > [    0.000000] BUG: mapping for 0xe6e60000 at 0xffe60000 out of vmalloc
> > space
> > 
> > I'm trying to fix this now. Has anyone noticed (and possibly investigated)
> > the issue ? If so please let me know to avoid duplicating efforts.
> 
> The above BUG is harmless, AFAIK. You can get rid of it by using my latest
> DEBUG_LL patch. Or just change the 0xffe60000 to 0xfde60000.

Yes, I've just noticed that, I spoke too fast.

-- 
Regards,

Laurent Pinchart