Author Topic: AMD Firmware Bug?  (Read 8610 times)

Offline kaptain_zero

  • Full Member
  • ****
  • Posts: 170
AMD Firmware Bug?
« on: May 21, 2013, 01:48:11 AM »
I think I have the right forum for this question, if not, could a mod please move it?

In my battle with this AMD APU, I've come across the following, during the initial bootup of a linux kernel, and no, it does not appear to be PClos specific, but... I keep seeing this every time it boots:

Quote
[Firmware Bug]: cpu 0, try to use APIC500 (LVT offset 0) for vector 0x10400, but the register is already in use for vector 0xf9 on another cpu
[Firmware Bug]: cpu 0, IBS interrupt offset 0 not available (MSRC001103A=0x0000000000000100)
Failed to setup IBS, -22

It seems a rather ominous message, but the computer functions fine, all 4 cores appear to work, so I'm assuming that the kernel has worked around the Firmware bug, but I thought I'd ask if anyone here has any ideas what it actually means?

Regards

Christian
Logged
Asus F2A85-M, AMD A8-5600 APU (Radeon HD 7560D graphics built in), 8GB ram

Asus M5A99FX PRO REV2.0, AMD 8350 8core cpu, Asus GTX 650, 8GB ram.

Samsung SCX-4623F all in one monochrome laser

Offline kjpetrie

  • PCLinuxOS Tester
  • Hero Member
  • *******
  • Posts: 9481
  • Cogito ergo qui est?
    • K J Petrie
Re: AMD Firmware Bug?
« Reply #1 on: May 21, 2013, 07:03:18 AM »
Here's an AMD document about IBS:
http://developer.amd.com/wordpress/media/2012/10/AMD_IBS_paper_EN.pdf

Presumably, without IBS you'll get less efficient use of resources, depending on what you use the computer for.
Logged
KJP
-----------------------------------------------------------
PClos64 2013 (KDE5) on AMD FX 8350 on GA-970A-DS3P (Rev 2.0), 24GB RAM, Asus EN210 (Nvidia) GPU, Maxtor STM325031, 3 SSDs, Amilo LSL 3220T monitor. Also Acer 5810TG.

https://www.kjpetrie.co.uk/

Offline kaptain_zero

  • Full Member
  • ****
  • Posts: 170
Re: AMD Firmware Bug?
« Reply #2 on: May 21, 2013, 01:32:59 PM »
Thanks kjpetrie,

That does shed some light on it.... Question now is, is there some way to tweak the kernel to fix this? Or... do I have a legitimate beef with AMD for buggy firmware in the CPU?

Sadly, I suspect AMD just let it slide... and for 95% of the world, it probably doesn't matter to them, as they never see that warning message considering they run some form of MS OS that hides all that stuff during normal boots.

Logged
Asus F2A85-M, AMD A8-5600 APU (Radeon HD 7560D graphics built in), 8GB ram

Asus M5A99FX PRO REV2.0, AMD 8350 8core cpu, Asus GTX 650, 8GB ram.

Samsung SCX-4623F all in one monochrome laser

Offline kjpetrie

  • PCLinuxOS Tester
  • Hero Member
  • *******
  • Posts: 9481
  • Cogito ergo qui est?
    • K J Petrie
Re: AMD Firmware Bug?
« Reply #3 on: May 21, 2013, 03:30:13 PM »
I don't think it's a question of tweaking the kernel. The kernel detects the bug and does without. If you want the IBS to work you probably need a firmware upgrade. I don't know how that would be done or whether there are any available, but that's a matter for AMD. Nor do I know what practical difference it would make. Unless you're noticing performance shortfalls it might not be worth the trouble, as a failed firmware upgrade could brick the machine.
Logged
KJP
-----------------------------------------------------------
PClos64 2013 (KDE5) on AMD FX 8350 on GA-970A-DS3P (Rev 2.0), 24GB RAM, Asus EN210 (Nvidia) GPU, Maxtor STM325031, 3 SSDs, Amilo LSL 3220T monitor. Also Acer 5810TG.

https://www.kjpetrie.co.uk/

Offline daniel

  • Global Moderator
  • Hero Member
  • *********
  • Posts: 4937
  • God knows, i'm not an Angel!
    • Tipps und Tricks
Re: AMD Firmware Bug?
« Reply #4 on: May 21, 2013, 03:50:22 PM »
We are not along...
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1152484

My post about it on a another thread...
http://www.pclinuxos.com/forum/index.php/topic,116320.msg991628.html#msg991628

Will search a little on web, may there is the answer available...


Regards
Daniel
Logged

Offline kaptain_zero

  • Full Member
  • ****
  • Posts: 170
Re: AMD Firmware Bug?
« Reply #5 on: May 21, 2013, 04:15:08 PM »
Ah... the plot thickens.... that first link was very interesting, Daniel.

As mentioned before, I was just scratching my head about all this and wondering if I needed to worry about it. The PC functions fine, Gkrellm reports normal activity on all 4 cores under use and I have no issues with restoring from sleep, so I'm going to stop loosing more sleep over this.  ;D  I'll just start wearing sunglasses in the basement during the boot sequence. <grin>

I did find this post by Robert Richter in that link quite interesting though:

Quote

This occurs after suspend/resume on AMD family 10h systems. The upstream kernel (v3.9-rc3) should be also affected.

The BIOS is known to report wrong (zero'ed) lvt offsets for mce and ibs which are set to the same value of 0. The BIOS wont be fixed. There is a quirk in the kernel that fixes this (force_ibs_eilvt_setup()). The fix reassigns the ibs lvt offet to 1 (instead of 0 that conflicts with mce threshold). The register reassignment must be applied on each cpu.

Current quirk does this during boot time, but after suspend/resume the register contents get lost and is not reinitialized anymore. Thus, after resume the apic setup code detects the offset conflict caused be the reset registers and throws the [hardware error] message.

In a result the ibs lvt offset is not properly initialized and using ibs (used for hardware monitoring by the perf and oprofile tool) may fail. The system should otherwise function correctly.

I will try to submit a patch upstream next weeks that fixes the problem.

-Robert

So thanks everyone, I now go back to sleeping peacefully.....

Regards

Christian
Logged
Asus F2A85-M, AMD A8-5600 APU (Radeon HD 7560D graphics built in), 8GB ram

Asus M5A99FX PRO REV2.0, AMD 8350 8core cpu, Asus GTX 650, 8GB ram.

Samsung SCX-4623F all in one monochrome laser