Re: Intel Plumas 533 (E7501) and PAE support

Hmm,

I used memmap=8GB!4GB as per https://docs.pmem.io/getting-started-guide/creating-development-environments/linux-environments/linux-memmap

# cat /proc/cmdline

BOOT_IMAGE=/vmlinuz-4.9.0-8-686-pae root=UUID=xxx ro quiet memmap=8G!4G

But while the pae kernel logs look changed

[ 0.000000] e820: BIOS-provided physical RAM map:

[ 0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009fbff] usable

[ 0.000000] BIOS-e820: [mem 0x000000000009fc00-0x000000000009ffff] reserved

[ 0.000000] BIOS-e820: [mem 0x00000000000f0000-0x00000000000fffff] reserved

[ 0.000000] BIOS-e820: [mem 0x0000000000100000-0x00000000f7feffff] usable

[ 0.000000] BIOS-e820: [mem 0x00000000f7ff0000-0x00000000f7ff7fff] ACPI data

[ 0.000000] BIOS-e820: [mem 0x00000000f7ff8000-0x00000000f7ffffff] ACPI NVS

[ 0.000000] BIOS-e820: [mem 0x00000000fec00000-0x00000000fecfffff] reserved

[ 0.000000] BIOS-e820: [mem 0x00000000fee00000-0x00000000fee00fff] reserved

[ 0.000000] BIOS-e820: [mem 0x00000000ffb00000-0x00000000ffbfffff] reserved

[ 0.000000] BIOS-e820: [mem 0x00000000fff80000-0x00000000ffffffff] reserved

[ 0.000000] BIOS-e820: [mem 0x0000000100000000-0x04000000fbffffff] usable

[ 0.000000] Notice: NX (Execute Disable) protection missing in CPU!

[ 0.000000] e820: user-defined physical RAM map:

[ 0.000000] user: [mem 0x0000000000000000-0x000000000009fbff] usable

[ 0.000000] user: [mem 0x000000000009fc00-0x000000000009ffff] reserved

[ 0.000000] user: [mem 0x00000000000f0000-0x00000000000fffff] reserved

[ 0.000000] user: [mem 0x0000000000100000-0x00000000f7feffff] usable

[ 0.000000] user: [mem 0x00000000f7ff0000-0x00000000f7ff7fff] ACPI data

[ 0.000000] user: [mem 0x00000000f7ff8000-0x00000000f7ffffff] ACPI NVS

[ 0.000000] user: [mem 0x00000000fec00000-0x00000000fecfffff] reserved

[ 0.000000] user: [mem 0x00000000fee00000-0x00000000fee00fff] reserved

[ 0.000000] user: [mem 0x00000000ffb00000-0x00000000ffbfffff] reserved

[ 0.000000] user: [mem 0x00000000fff80000-0x00000000ffffffff] reserved

[ 0.000000] user: [mem 0x0000000100000000-0x00000002ffffffff] persistent (type 12)

[ 0.000000] user: [mem 0x0000000300000000-0x04000000fbffffff] usable

[ 0.000000] Legacy DMI 2.3 present.

I don't have any pmem* device

# ls -la /dev/*me*

crw-r----- 1 root kmem 1, 1 Feb 17 06:07 /dev/mem

crw------- 1 root root 10, 59 Feb 17 06:07 /dev/memory_bandwidth

Even after loading the module

# lsmod | grep pm

nd_pmem 16384 0

nd_btt 24576 1 nd_pmem

libnvdimm 94208 3 nd_btt,nd_pmem,nd_e820

Does the *pmem module require some extra parameters? How can I enable logging for it? (to see what it does/tries to do?)

Thanks again!

Ciprian

On Sun, Feb 17, 2019 at 12:27 PM Ciprian Manea <ciprian.manea@gmail.com> wrote:

Hi,

Thank you for your quick response Ben!

I'm attaching here my current kernel ouput (first pages) if any of you can check it out and tell me if anything seems wrong.

I couldn't see in this old BIOS any settings related to memory 🤔

I vaguely remember more than 5 years ago when I tried last time to use more than the 4GB, that I had to disable ACPI if I remember correctly, force some MTRR parameters and of course use a PAE kernel. Actually, the only thing which remotely worked was an enterprise kernel from CentOS/RedHat 5.x, but with those forced parameters even if the free command was showing 8 or 12 GB RAM, then kernel was somehow busy looping with "something" 30-40% of the time, under no user load.

Thanks!
Ciprian

On Sat, Feb 16, 2019 at 10:39 PM Ben Hutchings <ben@decadent.org.uk> wrote:
On Sat, 2019-02-16 at 11:31 +0200, Ciprian Manea wrote:
> Hi there,
>
> I have this old server with 4x Intel Plumas (32 bit) running a
> recent 4.9.0-8-686-pae stock kernel
>
> And while the BIOS sees and tests 12GB of RAM, the Debian PAE enabled
> kernel sees only 4GB
>
> What are the PAE tricks to get to see/use more than 4GB RAM in the latest
> Debian 9.7?

You shouldn't need to do anything special. At a guess, the BIOS has a
compatibility setting that limits the RAM size it reports to the OS,
and you'll need to disable that.

If you know what you're doing, you can override the detected memory
size and physical addresses with the mem= and memmap= kernel
parameters.

However, beware that less than 1 GiB of the 12 GiB RAM will be "lowmem"
(directly accessible by the kernel) and the Linux virtual memory
manager no longer works well with such low ratios of lowmem to total
RAM.

Ben.

--
Ben Hutchings
When in doubt, use brute force. - Ken Thompson