Re: [Bugme-new] [Bug 11490] New: Old mtrr bug comeback

[Andrew Morton <akpm@linux-foundation.org>]

(switched to email.  Please respond via emailed reply-to-all, not via the
bugzilla web interface).

On Wed,  3 Sep 2008 08:44:27 -0700 (PDT) bugme-daemon@bugzilla.kernel.org wrote:

> http://bugzilla.kernel.org/show_bug.cgi?id=11490
> 
>            Summary: Old mtrr bug comeback
>            Product: Memory Management
>            Version: 2.5
>      KernelVersion: v2.6.27-rc5
>           Platform: All
>         OS/Version: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: normal
>           Priority: P1
>          Component: MTTR
>         AssignedTo: akpm@osdl.org
>         ReportedBy: ak@axet.ru
> 
> 
> Latest working kernel version:v2.6.26.3
> 
> Old mtrr bug comeback, when i boot my notebook with last kerenl they going to
> work really slow as turttle. to boot on normal speed i need use kernel param
> mem=1000M
> 
> when i use 2.6.26.3 everyting going fine. as i remember that bug fixed in
> 2.6.26 stable release and was in 2.6.25 and earlier.
> 
> probably you can find more details in bugzilla.redhat.com or
> bugzilla.mozilla.com i do no remember where is exactly i have been posted that
> old bug.
> 
> here the point (warning on 2.6.26.3 kernel):
> 
> WARNING: BIOS bug: CPU MTRRs don't cover all of memory, losing 7MB of RAM.
> ------------[ cut here ]------------
> WARNING: at arch/x86/kernel/cpu/mtrr/main.c:706
> mtrr_trim_uncached_memory+0x103/0x168()
> Modules linked in:
> Pid: 0, comm: swapper Not tainted 2.6.26.3 #1
>  [<c06260e9>] ? printk+0xf/0x16
>  [<c0427749>] warn_on_slowpath+0x41/0x7b
>  [<c0628171>] ? _spin_unlock_irqrestore+0x10/0x14
>  [<c0628171>] ? _spin_unlock_irqrestore+0x10/0x14
>  [<c0427e4f>] ? release_console_sem+0x181/0x189
>  [<c04282d3>] ? vprintk+0x2e6/0x30b
>  [<c04282d3>] ? vprintk+0x2e6/0x30b
>  [<c040dbfd>] ? generic_get_mtrr+0x53/0x8a
>  [<c0756177>] mtrr_trim_uncached_memory+0x103/0x168
>  [<c075606c>] ? mtrr_bp_init+0x204/0x20c
>  [<c07539a9>] setup_arch+0x27b/0x6ce
>  [<c06260e9>] ? printk+0xf/0x16
>  [<c074d5c0>] start_kernel+0x64/0x2da
>  [<c074d008>] __init_begin+0x8/0xa
>  =======================
> ---[ end trace 4eaa2a86a8e2da22 ]---
> update e820 for mtrr
> modified physical RAM map:
> 

A post-2.6.26 regression.

We have a couple of mtrr fixes pending:

http://userweb.kernel.org/~akpm/mmotm/broken-out/x86-delay-early-cpu-initialization-until-cpuid-is-done.patch

http://userweb.kernel.org/~akpm/mmotm/broken-out/x86-move-mtrr-cpu-cap-setting-early-in-early_init_xxxx.patch

but I don't know if they'll fix this regression?