GA-890FXA-UD5 IOMMU BIOS Switch?

[joro80]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

A question to the other owners of 890FX motherboards in this thread:

I need to verify a few more boards if they have the IOMMU enabled. So if one of you have a 890FX board and can verify this please send me a note. I can assist with the verification steps.

Thanks.

[joro80]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

I was not able to get iommu to work on my MSI 890FXA-GD70. Before I purchased, an MSI rep told me that the motherboard is supposed to support it, but it looks like they did not implement the necessary bios support either. I am waiting to hear back from them as to when/if they plan to add support, so I would not recommend swapping for this board until I hear more.

Can you also send me your full dmesg and lscpi output so I can see what the BIOS enables for the OS?

Thanks.

[kuzew]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

No, the BIOS vendor just don't support it. You can try to complain to them so that they may support it in a future BIOS version.

Ack, yes -- that's what I intended to say.

I am looking for an 890FX board which has IOMMU support in the BIOS myself. I will try to ask the motherboard vendors too if they want to build in support into their BIOSes.
I also want to have an IOMMU system at home :-)
Cheers.

Oh, cool. Alright, well like I said before I have a ticket in with Gigabyte about this... I /should/ hear from them soon, otherwise I'll need to send it in again. It's a rather nice board, but I don't understand why they wouldn't turn on IOMMU support in the first place, that's one of the key differences between then 890GX and 890FX chipsets are the IOMMU. If I didn't want IOMMU, I could of saved myself some money and went with the 890GX on board.

Cheers.

[leo1981]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

Hi,
I'm looking for a motherboard with AMD 890fx chipset and working IOMMU too.
I found that the ASUS Crosshair IV Formaul has IOMMU support in the BIOS:
Ubiquitous Talk

But ASUS has a bad reputation with Intel VT-d, they don't support Linux.
KVM and Xen cannot use IOMMU in the Intel based motherboards.

Please let us know if you find a motherboard with working IOMMU.

Leonardo

[kuzew]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

Hi,
I'm looking for a motherboard with AMD 890fx chipset and working IOMMU too.
I found that the ASUS Crosshair IV Formaul has IOMMU support in the BIOS:
Ubiquitous Talk

Oh very cool... I still need to play with ZFS on FreeBSD or Solaris...

But ASUS has a bad reputation with Intel VT-d, they don't support Linux.
KVM and Xen cannot use IOMMU in the Intel based motherboards.

VTdHowTo - Xen Wiki -- Might be worth looking at this... The Xen guys have compiled a list of Intel motherboards with VT-d enabled/compatible. It does list 890FX, but I still have yet to see someone have it 1) property supported in the BIOS and 2) working with KVM or Xen. Might want to check out the Intel board list before jumping over to the AMD side of things, until this is sorted.

I was looking at porting coreboot [1] over to my motherboard, but that would be quite a chore and I don't have the time ATM. Reverse engineering/decompiling Gigabyte's AwardBIOS would also be pretty painful, I'm sure they wrote it in C and not in ASM, then there is the matter of the checksum problem... bleh. I'd paid for AwardBIOS to enable this feature if that's what it comes to.

Best of luck Leonardo.
Cheers!

[1] - coreboot

[jdma]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

Can you also send me your full dmesg and lscpi output so I can see what the BIOS enables for the OS?

Thanks.

#xm dmesg
 __  __            _  _    ___   ___  
 \ \/ /___ _ __   | || |  / _ \ / _ \ 
  \  // _ \ '_ \  | || |_| | | | | | |
  /  \  __/ | | | |__   _| |_| | |_| |
 /_/\_\___|_| |_|    |_|(_)___(_)___/ 
                                      
(XEN) Xen version 4.0.0 (@(none)) (gcc version 4.3.4 (Gentoo 4.3.4 p1.1, pie-10.1.5) ) Mon May 17 21:51:45 CDT 2010
(XEN) Latest ChangeSet: unavailable
(XEN) Command line: iommu=verbose
(XEN) Video information:
(XEN)  VGA is text mode 80x25, font 8x16
(XEN)  VBE/DDC methods: V2; EDID transfer time: 1 seconds
(XEN) Disc information:
(XEN)  Found 2 MBR signatures
(XEN)  Found 2 EDD information structures
(XEN) Xen-e820 RAM map:
(XEN)  0000000000000000 - 000000000009f000 (usable)
(XEN)  000000000009f000 - 00000000000a0000 (reserved)
(XEN)  00000000000e4000 - 0000000000100000 (reserved)
(XEN)  0000000000100000 - 00000000c7f90000 (usable)
(XEN)  00000000c7f90000 - 00000000c7f9e000 (ACPI data)
(XEN)  00000000c7f9e000 - 00000000c7fe0000 (ACPI NVS)
(XEN)  00000000c7fe0000 - 00000000c8000000 (reserved)
(XEN)  00000000ffe00000 - 0000000100000000 (reserved)
(XEN)  0000000100000000 - 0000000238000000 (usable)
(XEN) ACPI: RSDP 000FB6A0, 0014 (r0 ACPIAM)
(XEN) ACPI: RSDT C7F90000, 0044 (r1 MSI    OEMSLIC  20100518 MSFT       97)
(XEN) ACPI: FACP C7F90200, 0084 (r1 7640MS A7640100 20100518 MSFT       97)
(XEN) ACPI: DSDT C7F905E0, 9F32 (r1  A7640 A7640100      100 INTL 20051117)
(XEN) ACPI: FACS C7F9E000, 0040
(XEN) ACPI: APIC C7F90390, 0088 (r1 7640MS A7640100 20100518 MSFT       97)
(XEN) ACPI: MCFG C7F90420, 003C (r1 7640MS OEMMCFG  20100518 MSFT       97)
(XEN) ACPI: SLIC C7F90460, 0176 (r1 MSI    OEMSLIC  20100518 MSFT       97)
(XEN) ACPI: OEMB C7F9E040, 0072 (r1 7640MS A7640100 20100518 MSFT       97)
(XEN) ACPI: SRAT C7F9A5E0, 0108 (r3 AMD    FAM_F_10        2 AMD         1)
(XEN) ACPI: HPET C7F9A6F0, 0038 (r1 7640MS OEMHPET  20100518 MSFT       97)
(XEN) ACPI: SSDT C7F9A730, 0DA4 (r1 A M I  POWERNOW        1 AMD         1)
(XEN) System RAM: 8175MB (8372060kB)
(XEN) Domain heap initialised
(XEN) Processor #0 0:10 APIC version 16
(XEN) Processor #1 0:10 APIC version 16
(XEN) Processor #2 0:10 APIC version 16
(XEN) Processor #3 0:10 APIC version 16
(XEN) Processor #4 0:10 APIC version 16
(XEN) Processor #5 0:10 APIC version 16
(XEN) IOAPIC[0]: apic_id 6, version 33, address 0xfec00000, GSI 0-23
(XEN) IOAPIC[1]: apic_id 7, version 33, address 0xfec20000, GSI 24-55
(XEN) Enabling APIC mode:  Flat.  Using 2 I/O APICs
(XEN) Using scheduler: SMP Credit Scheduler (credit)
(XEN) Detected 3200.170 MHz processor.
(XEN) Initing memory sharing.
(XEN) HVM: ASIDs enabled.
(XEN) HVM: SVM enabled
(XEN) HVM: Hardware Assisted Paging detected.
(XEN) AMD-Vi: IOMMU not found!
(XEN) I/O virtualisation disabled
(XEN) Total of 6 processors activated.
(XEN) ENABLING IO-APIC IRQs
(XEN)  -> Using new ACK method
(XEN) TSC is reliable, synchronization unnecessary
(XEN) Platform timer is 14.318MHz HPET
(XEN) Brought up 6 CPUs
(XEN) *** LOADING DOMAIN 0 ***
(XEN)  Xen  kernel: 64-bit, lsb, compat32
(XEN)  Dom0 kernel: 64-bit, lsb, paddr 0x1000000 -> 0x16be000
(XEN) PHYSICAL MEMORY ARRANGEMENT:
(XEN)  Dom0 alloc.:   0000000230000000->0000000232000000 (2031119 pages to be allocated)
(XEN) VIRTUAL MEMORY ARRANGEMENT:
(XEN)  Loaded kernel: ffffffff81000000->ffffffff816be000
(XEN)  Init. ramdisk: ffffffff816be000->ffffffff816be000
(XEN)  Phys-Mach map: ffffea0000000000->ffffea0000f8f078
(XEN)  Start info:    ffffffff816be000->ffffffff816be4b4
(XEN)  Page tables:   ffffffff816bf000->ffffffff816ce000
(XEN)  Boot stack:    ffffffff816ce000->ffffffff816cf000
(XEN)  TOTAL:         ffffffff80000000->ffffffff81800000
(XEN)  ENTRY ADDRESS: ffffffff81000000
(XEN) Dom0 has maximum 6 VCPUs
(XEN) Scrubbing Free RAM: .done.
(XEN) Xen trace buffers: disabled
(XEN) Std. Loglevel: Errors and warnings
(XEN) Guest Loglevel: Nothing (Rate-limited: Errors and warnings)
(XEN) Xen is relinquishing VGA console.
(XEN) *** Serial input -> DOM0 (type 'CTRL-a' three times to switch input to Xen)
(XEN) Freed 164kB init memory.

[joro80]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

(XEN) ACPI: RSDP 000FB6A0, 0014 (r0 ACPIAM)
(XEN) ACPI: RSDT C7F90000, 0044 (r1 MSI    OEMSLIC  20100518 MSFT       97)
(XEN) ACPI: FACP C7F90200, 0084 (r1 7640MS A7640100 20100518 MSFT       97)
(XEN) ACPI: DSDT C7F905E0, 9F32 (r1  A7640 A7640100      100 INTL 20051117)
(XEN) ACPI: FACS C7F9E000, 0040
(XEN) ACPI: APIC C7F90390, 0088 (r1 7640MS A7640100 20100518 MSFT       97)
(XEN) ACPI: MCFG C7F90420, 003C (r1 7640MS OEMMCFG  20100518 MSFT       97)
(XEN) ACPI: SLIC C7F90460, 0176 (r1 MSI    OEMSLIC  20100518 MSFT       97)
(XEN) ACPI: OEMB C7F9E040, 0072 (r1 7640MS A7640100 20100518 MSFT       97)
(XEN) ACPI: SRAT C7F9A5E0, 0108 (r3 AMD    FAM_F_10        2 AMD         1)
(XEN) ACPI: HPET C7F9A6F0, 0038 (r1 7640MS OEMHPET  20100518 MSFT       97)
(XEN) ACPI: SSDT C7F9A730, 0DA4 (r1 A M I  POWERNOW        1 AMD         1)

[ATTACH]2580[/ATTACH]

Ok, no IVRS table and also no IOMMU device in lspci showing up. IOMMU is definitly disabled on this board. Thanks, for the info.

[Lsdmeasap]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

Have you guys sent in a report on this directly to Taiwan?

http://www.gigabyte.com.tw/Support/ServiceCenter.aspx

Just wondering as they may not be aware of the issue, sorry I haven't looked back through the thread to see if you already have or not.

Once Computex is over I will try to get someone there to look into this for you guys if no one there isn't already.

[kuzew]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

...
Just wondering as they may not be aware of the issue, sorry I haven't looked back through the thread to see if you already have or not.
...

I submitted a ticket on Friday, 28th of May... so far the status is "the mail will be processed". If they don't open it/reply by Friday of this week, I will resubmit the ticket again.

I still can't find "official" word that the 890FX really supports IOMMU. No official word from AMD, but I see lots of block diagrams from AMD that have IOMMU written on it. One finds if you look up AMD 890FX Chipset [1] on AMD's website, they don't spec IOMMU. But yet lots of reviewers say that it should have it... so which is it? Lots of mailing lists, reviews of motherboards and other fourms do say this chipset should have IOMMU... so I'm a bit confused.

[1] - AMD 890FX Chipset

[Lsdmeasap]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

It may just be something they have not fully worked out yet so it is not in the current BIOSes is all.

I will let you know what I can find out, hopefully by next week I should know something

[kuzew]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

...
I will let you know what I can find out, hopefully by next week I should know something

Thank you very much for trying to find out what's going on with this, it certainly feels like a mystery.

[leo1981]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

It may just be something they have not fully worked out yet so it is not in the current BIOSes is all.

I will let you know what I can find out, hopefully by next week I should know something

The GA-X58A-UD7 (from your signature) has an Intel X58 chipset.
That chipset is capable of IOMMU (or VT-d), does your BIOS implement IOMMU?

If not, I think Gigabyte doesn't want to implement it.
If yes, probably they have some problems with AMD implementation.

[kuzew]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

Well, Gigabyte technical support finally replied back...

"Dear Customer

Unfortunately the feature is not supported on this board"

I guess that answers that... ):
Not much detail on why or even recommendations for what board they have that has IOMMU enabled. I'm really not too happy with that answer.
But what does this really mean? That the board isn't wired for IOMMU, or they are just not going to support IOMMU even if the board is fully capable of supporting it? Does anyone know how I can tell if this chipset has the ability for IOMMU and they just turned it off/locked it?

[jdma]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

I still can't find "official" word that the 890FX really supports IOMMU. No official word from AMD, but I see lots of block diagrams from AMD that have IOMMU written on it. One finds if you look up AMD 890FX Chipset [1] on AMD's website, they don't spec IOMMU. But yet lots of reviewers say that it should have it... so which is it? Lots of mailing lists, reviews of motherboards and other fourms do say this chipset should have IOMMU... so I'm a bit confused.

I emailed AMD a few days ago to ask because I have read a lot of conflicting information on this subject:

Thank you for contacting AMD. I would like to verify for you that yes, the 890FX chipset does offer the IOMMU feature, however you will want to keep in mind that manufacturers can decide to enable or disable this feature through the BIOS and therefore you will want to verify with manufacturer of your specific board. But in short, yes it does offer the ability to do so.

[kuzew]

Re: GA-890FXA-UD5 IOMMU BIOS Switch?

Oh that is most excellent news! I wonder why they don't have it listed in the features/specs of the chipset...

Well, because porting coreboot to this motherboard would be too much... and I've attempted to disassemble the existing BIOS and failed (I got mostly garbage... didn't look into it further, might be some compression thing they do). I'm now looking into returning my Gigabyte motherboard to Newegg and start to look for a motherboard that has IOMMU enabled in the BIOS. It looks like Gigabyte won't be enabling that feature anytime soon, unless Lsdmeasap finds some information otherwise. I suppose I'll wait around until then.

Thanks jdma for sharing that information. Cheers!