linux-kernel - Re: [PATCH] Option to disable AMD C1E (allows dynticks to work)

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:	Sat, 29 Dec 2007 13:19:40 +0200
From:	Eduard-Gabriel Munteanu <eduard.munteanu@...ux360.ro>
To:	Richard Harman <richard@...hardharman.com>
Cc:	LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] Option to disable AMD C1E (allows dynticks to work)

On Sat, 29 Dec 2007 04:09:41 -0500
Richard Harman <richard@...hardharman.com> wrote:

> Right now, with your patch and the 'noirqdebug' option or disabling
> nohz the system appears to be stable.  This laptop otherwise locks up
> trying to configure apic/lapic, or locks up solid later with NO
> oops/bug and nothing will bring it out of it (eg, sysrq has no
> effect).  NMI watchdog was also inoperable, which I believe was due
> to the apic/lapic being disabled.
> 
> With your patch, the lapic is used:
> 
> ACPI: LAPIC (acpi_id[0x00] lapic_id[0x00] enabled)
> ACPI: LAPIC (acpi_id[0x01] lapic_id[0x01] enabled)

The kernel should display these messages even without my patch.

> These kernel dmesg logs are hand-transcribed, and are from 2.6.22.5,
> but this shows where the kernel would lock up good and solid:
> 
> http://www.richardharman.com/pavilion/2.6.22.5/dmesg.txt

I doubt it is my patch that makes your laptop work. C1E detection was
added by the following commit:

$ git-describe fb79d22e1d4b06385796cc0db0084a2e07beccee
v2.6.23-2294-gfb79d22

Try building a 2.6.24-rc6 without my patch and see if it still works
(apply it again in reverse mode, for example). I guess it will work,
because the kernel shouldn't use the LAPIC as a timer at all (except as
a dummy one), with or without dynticks support.

When I started investigating the "nohz is not working problem", I
stumbled across a few threads on LKML where a few people had
similar problems with their laptops and they where instructed to pass
"nolapic" or something similar as a boot parameter to the kernel. I
guess the problem disappeared after the commit I mentioned previously,
since such reports weren't that recent. IIRC, that was on kernels
previous to 2.6.24-*, where x86 and x86-64 weren't unified and there
were no dynticks on the latter.

To be more clear, my patch disables C1E instead of disabling the
LAPIC, as these two don't agree with each other.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/