[nSLUG] Another debian issue resolved

Peter Cordes peter at cordes.ca
Tue Aug 2 15:17:11 ADT 2005

 You probably have this stuff figured out by now, but I felt like catching
up on my long-ignored NSLUG inbox :)

On Wed, Jun 15, 2005 at 07:53:03AM -0300, Donald Teed wrote:
> Now a question:
> Does anyone know why I get this sort of error in dmesg output:
> mtrr: no MTRR for f2000000,800000 found
> mtrr: no MTRR for f2800000,400000 found
> mtrr: no MTRR for f2c00000,200000 found
> mtrr: no MTRR for f2e00000,100000 found
> mtrr: no MTRR for f2f00000,80000 found
> mtrr: no MTRR for f2f80000,40000 found
> mtrr: no MTRR for f2fc0000,20000 found
> mtrr: no MTRR for f2fe0000,10000 found
> mtrr: no MTRR for f2ff0000,8000 found
> mtrr: no MTRR for f2ff8000,4000 found
> mtrr: no MTRR for f2ffc000,2000 found
> mtrr: no MTRR for f2ffe000,1000 found
> mtrr: base(0xf2000000) is not aligned on a size(0xfff000) boundary

 mtrr = memory type range register, or something like that.  They are
hardware registers in the CPU that control which regions of memory space are
cacheable, etc.  The X server programs them (through the kernel's mtrr
driver) to do write-combining on the region containing video memory.
on a properly-working machine with 16MB of video RAM and 512MB of normal RAM:
llama]~$ cat /proc/mtrr
reg00: base=0x00000000 (   0MB), size= 512MB: write-back, count=1
reg01: base=0xe4000000 (3648MB), size=  64MB: write-combining, count=1
reg05: base=0xe2000000 (3616MB), size=  32MB: write-combining, count=4

> Or alternately, simply how to fix whatever it is.  I believe I had
> mtrr enabled in my Gentoo kernel config, and still have it set up
> in Debian.  The mtrr error doesn't seem to effect the running
> of the system.

 video card stuff might be going somewhat slower than it could be.  Like
video playback and 3D accel.

> I briefly tried a stock Debian kernel image for 2.6.8 and ran into
> 2 issues: everything is a module, it seems, and 2.6 might not
> like an existing mdadm raid set up from a 2.4 kernel.

 2.6 kernels can use md devices configured with 2.4 kernels no problem.
IIRC, it autodetected my RAID arrays without problems.  Oh, but that might
only have been when MD was compiled into the kernel.  modularized md doesn't
autodetect, or something stupid like that.

> The second part might be resolved by giving it a config
> in mdadm.conf (but 2.4 doesn't need it).

 Yeah, that might help.

> In my opinion, md and raid* have little to no practical use
> as modules. They should always be compiled into the kernel.
> I can easily work around that by building my own kernel, but
> for those who don't, this would be much more useful.

 Yeah, I don't like messing with initrd.

#define X(x,y) x##y
Peter Cordes ;  e-mail: X(peter at cor , des.ca)

"The gods confound the man who first found out how to distinguish the hours!
 Confound him, too, who in this place set up a sundial, to cut and hack
 my day so wretchedly into small pieces!" -- Plautus, 200 BC


More information about the nSLUG mailing list