linux-kernel - Re: [patch 5/9] x86: Cure per CPU madness on UP

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Tue, 19 Mar 2024 17:21:03 +0100
From: Thomas Gleixner <tglx@...utronix.de>
To: Arnd Bergmann <arnd@...nel.org>, Linus Torvalds
 <torvalds@...uxfoundation.org>
Cc: Guenter Roeck <linux@...ck-us.net>, LKML <linux-kernel@...r.kernel.org>,
 x86@...nel.org, Uros Bizjak <ubizjak@...il.com>,
 linux-sparse@...r.kernel.org, kernel test robot <lkp@...el.com>,
 oe-kbuild-all@...ts.linux.dev
Subject: Re: [patch 5/9] x86: Cure per CPU madness on UP

On Mon, Mar 18 2024 at 20:13, Arnd Bergmann wrote:
> FWIW, I did some experiments a few weeks ago on 32-bit ARM,
> using a fairly minimal kernel in a virtual machine, and
> checking the runtime memory consumption rather than compile-time.
> In a kvm guest with 32MiB RAM, I saw a difference of multiple
> megabytes in memory usage:
>
> Linux testvm 6.8.0-rc4-00410-gc02197fc9076-dirty #1 SMP PREEMPT armv7l
> root@...tvm:~# free
>            	total   used    free  	shared  buff/cache   available
> Mem:       	26932   14956   1732   	    52       12800   	11976
> Swap:      	16360    3632   12728
>
> Linux testvm 6.8.0-rc4-00410-gc02197fc9076-dirty #2 PREEMPT armv7l
> root@...tvm:~# free
>            	total    used  	free  	shared  buff/cache   available
> Mem:       	26932   13744  	5648        32       10092   	13188
> Swap:      	16360    3880  	12480
>
> There is a little difference between runs, but this does seem
> significant enough to keep it. The SMP build was with
> CONFIG_NR_CPUS=2 (the smallest supported compile-time number),
> but running on a single-CPU qemu instance.

With a SMP=y, NR_CPUS=1 build on x86 64bit I get:

               total        used        free      shared  buff/cache   available
Mem:        32882056      498068    32590580        4884      128884    32383988
Swap:         998396           0      998396

Same config just SMP=n:

               total        used        free      shared  buff/cache   available
Mem:        32885804      461704    32635284        4876      119480    32424100
Swap:         998396           0      998396

So the delta for available is ~40 MiB.

But if I look at it with init=/bin/sh on the command line then the delta
is significantly different:

With a SMP=y, NR_CPUS=1 build on x86 64bit I get:

               total        used        free      shared  buff/cache   available
Mem:        32883680      324120    32822728         216       10864    32559560
Swap:              0           0           0

Same config just SMP=n:

               total        used        free      shared  buff/cache   available
Mem:        32885804      326876    32821972         216       11100    32558928
Swap:              0           0           0

Delta available = 632 KiB

I haven't had the time to stare at that in detail, but comparing
/proc/meminfo for the full boot case above does not immediately give me
a hint. It's confusing at best...

Thanks,

        tglx