[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <BANLkTikoQ1E9zJAUGeQTHs-ffLK=R7Wdsw@mail.gmail.com>
Date: Thu, 2 Jun 2011 22:26:22 -0700
From: Yinghai Lu <yinghai@...nel.org>
To: Michael Reinelt <michael@...nelt.co.at>
Cc: lkml <linux-kernel@...r.kernel.org>
Subject: Re: mtrr_cleanup: can not find optimal value
On Thu, Jun 2, 2011 at 10:03 PM, Michael Reinelt <michael@...nelt.co.at> wrote:
> Hi there,
>
> I'm suffering bad graphic performance on my notebook, and when sorting
> things out, I found the following in dmesg:
>
> mtrr_cleanup: can not find optimal value
> please specify mtrr_gran_size/mtrr_chunk_size
>
> and later:
>
> mtrr: type mismatch for e0000000,10000000 old: write-back new:
> write-combining
> [drm] MTRR allocation failed. Graphics performance may suffer.
>
> which is probably due to the fact that MTRR's can't be cleaned (at least I
> suspect so)
>
> The notebook is a Fujitsu Liefbook E8410, Intel Core2 Duo CPU T7700 @
> 2.40GHz, and 8 GB RAM, graphics is Intel Mobile GM965/GL960, kernel is a
> vanilla 2.6.39 X86_64
>
> When I remove 1 RAM module (using 4GB only) the messages above disappear.
> Unfortunately, graphic performance is still bad, but that's another story. I
> want to fix the MTRR issue anyway.
>
> I think the MTRR cleaner is unable to find a really optimal combination, and
> wants me to specify one of the sub-optimal configs. But I have no idea how
> to do that, and found nothing in the net. I found some references to a
> "mtrr-uncover" program, but I cannot compile it (probably because its too
> old). So I'm asking for some advice how to manually tune the MTRR's.
...
> gran_size: 8M chunk_size: 32M num_reg: 7 lose cover RAM: 7M
...
you can append "mtrr_gran_size=8M mtrr_chunk_size=32M" in boot command line.
Thanks
Yinghai
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists