Bug #8485

Consider including an x86 kernel with SMP support

Added by Dr_Whax 2014-12-25 12:59:20 . Updated 2016-09-20 16:50:01 .

Status:
Resolved
Priority:
Normal
Assignee:
Category:
Hardware support
Target version:
Start date:
2014-12-25
Due date:
% Done:

100%

Feature Branch:
feature/10298-linux-4.x-aufs
Type of work:
Research
Blueprint:

Starter:
Affected tool:
Deliverable for:

Description

Currently, we use two kernels, one for x86 and one for x86_64. The x86 one doesn’t seem to have any SMP support, however, there are x86 processors, which are only 32bit, but contain 2 cores, like the one’s that can be found in x60’s. We currently use the following kernel: linux-image-3.16.0-4-586 which doesn’t have SMP support enabled in the kernel.

There are a few solutions possible as we see here:

1) Include linux-image-3.2.0-4-686-pae as a third kernel for SMP and find out a way to detect CPU’s who support this in some way.
2) Replace linux-image-3.16.0-4-586 with linux-image-3.2.0-4-686-pae and still support a large array of devices from say, pentium3 and on.
3) Fill Debian bug to get SMP support in for linux-image-3.16.0-4-586 and drop including linux-image-3.2.0-4-686-pae


Subtasks


Related issues

Blocked by Tails - Feature #10298: Upgrade to Linux 4.x Resolved 2015-08-11

History

#1 Updated by intrigeri 2015-01-01 18:38:20

  • Subject changed from Consider including an x86 Debian kernel with SMP support. to Consider including an x86 kernel with SMP support
  • Category set to Hardware support
  • Status changed from New to Confirmed
  • Assignee set to Dr_Whax
  • Type of work changed from Discuss to Research

#2 Updated by intrigeri 2015-01-01 18:41:08

  • related to Feature #8183: Ship a 64-bit (x86_64) instead of 32-bit userspace added

#3 Updated by intrigeri 2015-01-01 18:44:04

  • How much would it add to the ISO size to add a third kernel? Note that it would also bloat most IUKs quite a bit. Given how few SMP 32-bit systems are still in use these days, and the relatively low benefit of giving them access to their 2nd core, I doubt it’s worth making the upgrade UX worse for everyone.
  • Requiring PAE would kill support for some more recent CPUs too, e.g. the ThinkPad X32 generation: not all CPUs post Pentium III have PAE => not sure the 2nd option is acceptable yet. I think that’s strongly related to Feature #8183.
  • The 3rd option (get SMP into Debian’s “586” kernel) seems interesting. Can you try building such a kernel and see if it works?

#4 Updated by intrigeri 2016-06-08 08:52:58

FWIW, the –586 kernel is gone in Debian’s 4.x, there’s now –686 and –686-pae. If I understand the source right, both have SMP enabled, so we’ll get 32-bit SMP for free when we resolve Feature #10298. I’m going to express that in Redmine-speak.

Except that 1. I didn’t check if the X60 work with that –686 kernel so that might or might not be enough to solve the use case this ticket was created for; and 2. given how Feature #10298 is slowed down by Feature #8183, it might be that in the end we switch to 64-bit userspace at the same time as to Linux 4.x, and then this ticket is moot.

#5 Updated by intrigeri 2016-06-08 08:53:16

  • related to deleted (Feature #8183: Ship a 64-bit (x86_64) instead of 32-bit userspace)

#6 Updated by intrigeri 2016-06-08 08:53:19

#7 Updated by intrigeri 2016-06-10 12:17:00

  • Status changed from Confirmed to In Progress
  • Assignee changed from Dr_Whax to anonym
  • Target version set to Tails_2.6
  • % Done changed from 0 to 50
  • QA Check set to Ready for QA
  • Feature Branch set to feature/10298-linux-4.x-aufs

As explained above, the 686 kernel shipped on feature/10298-linux-4.x-aufs has SMP, so that branch will fix this ticket.

#8 Updated by Dr_Whax 2016-07-21 16:38:51

I tested this out on an Thinkpad x60 with the Intel l2400(http://ark.intel.com/fr/products/27229/Intel-Core-Duo-Processor-L2400-2M-Cache-1_66-GHz-667-MHz-FSB), I see two CPU’s appearing in /proc/cpuinfo.

However, when attempting to shutdown the machine, I got a kernel panic on memory wiping, something that didn’t happen before.

See kernel panic log here: http://jurrevanbergen.nl/x60_smp_kernel_panic.jpg

#9 Updated by intrigeri 2016-07-22 05:05:25

> I tested this out on an Thinkpad x60 with the Intel l2400(http://ark.intel.com/fr/products/27229/Intel-Core-Duo-Processor-L2400-2M-Cache-1_66-GHz-667-MHz-FSB), I see two CPU’s appearing in /proc/cpuinfo.

Thanks, so this problem will be solved by the topic branch. Good!

> However, when attempting to shutdown the machine, I got a kernel panic on memory wiping, something that didn’t happen before. I’ll make a separate ticket for that.

It’s worth documenting on known issues, but other than that, I doubt that opening a ticket about that will be a good use of our time (and todo list), unless this regression occurs on a lot more hardware.

#10 Updated by anonym 2016-08-23 08:28:10

  • Status changed from In Progress to Fix committed
  • Assignee deleted (anonym)
  • % Done changed from 50 to 100
  • QA Check changed from Ready for QA to Pass

#11 Updated by anonym 2016-09-20 16:50:01

  • Status changed from Fix committed to Resolved