On Wed, 6 Jun 2007, Jesse Barnes wrote:
On some machines, buggy BIOSes don't properly setup WB MTRRs to
cover all available RAM, meaning the last few megs (or even gigs)
of memory will be marked uncached. Since Linux tends to allocate
from high memory addresses first, this causes the machine to be
unusably slow as soon as the kernel starts really using memory
(i.e. right around init time).
This patch works around the problem by scanning the MTRRs at
boot and figuring out whether the current end_pfn value (setup
by early e820 code) goes beyond the highest WB MTRR range, and
if so, trimming it to match. A fairly obnoxious KERN_WARNING
is printed too, letting the user know that not all of their
memory is available due to a likely BIOS bug.
Something similar could be done on i386 if needed, but the boot
ordering would be slightly different, since the MTRR code on i386
depends on the boot_cpu_data structure being setup.
Justin, can you please test and make sure this patch works for
you too? It'll only work around the problem, but it's better
than having to do mem= by hand or waiting for a fix from your
BIOS vendor.
Thanks,
Jesse
Jesse, it worked.
With mem=8832M (without your patch): 2.6.22-rc4:
top - 17:39:02 up 1 day, 8:07, 25 users, load average: 2.33, 0.76, 0.30
Tasks: 325 total, 11 running, 314 sleeping, 0 stopped, 0 zombie
Cpu(s): 80.0%us, 20.0%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 0.0%si,
0.0%st
Mem: 8039620k total, 7936472k used, 103148k free, 708k buffers
Swap: 16787768k total, 128k used, 16787640k free, 6646248k cached
With no mem= in append line (with your patch): 2.6.22-rc4:
top - 17:44:01 up 1 min, 1 user, load average: 0.97, 0.25, 0.08
Tasks: 145 total, 1 running, 144 sleeping, 0 stopped, 0 zombie
Cpu(s): 5.2%us, 3.0%sy, 1.2%ni, 86.8%id, 3.8%wa, 0.0%hi, 0.0%si,
0.0%st
Mem: 8039608k total, 969380k used, 7070228k free, 1232k buffers
Swap: 16787768k total, 0k used, 16787768k free, 109448k cached
Odd, remote netconsole did not capture the dmesg the E820 memory map.
Jun 6 17:43:03 p34 [ 53.598611] ahci 0000:00:1f.2: AHCI 0001.0100 32 slots
6 ports 3 Gbps 0x3f impl SATA mode Jun 6 17:43:03 p34 [ 53.598683] ahci
0000:00:1f.2: flags: 64bit ncq led clo pio slum part Jun 6 17:43:03 p34 [
53.598986] scsi0 : ahci Jun 6 17:43:03 p34 [ 53.599131] scsi1 : ahci Jun
6 17:43:03 p34 [ 53.599239] scsi2 : ahci Jun 6 17:43:03 p34 [ 53.599340]
scsi3 : ahci Jun 6 17:43:03 p34 [ 53.599438] scsi4 : ahci
I will run with this patch for a while but so far, no issues, everything
looks great.
Will it make it into 2.6.22-rc5? :)
Justin.
