[gentoo-user] CONFIG_IOMMU_SUPPORT???

[gentoo-user] CONFIG_IOMMU_SUPPORT???

[Mark Knecht]

Hi all,
   Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
It is apparently a kernel config option no required by
virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
suspect there's something else I need to enable before this option
becomes available? Unfortunately I haven't been able to Google what I
need to do to find it.

Thanks,
Mark

* Messages for package app-emulation/virtualbox-modules-4.1.2:

 *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
 * Please check to make sure these options are set correctly.
 * Failure to do so may cause unexpected problems.
<SNIP>

k2 linux # cat .config | grep IOMMU
CONFIG_GART_IOMMU=y
# CONFIG_CALGARY_IOMMU is not set
CONFIG_AMD_IOMMU=y
# CONFIG_AMD_IOMMU_STATS is not set
CONFIG_IOMMU_HELPER=y
CONFIG_IOMMU_API=y
# CONFIG_IOMMU_DEBUG is not set
# CONFIG_IOMMU_STRESS is not set
k2 linux #

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Dale]

Mark Knecht wrote:
> Hi all,
>     Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
> It is apparently a kernel config option no required by
> virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
> suspect there's something else I need to enable before this option
> becomes available? Unfortunately I haven't been able to Google what I
> need to do to find it.
>
> Thanks,
> Mark
>
> * Messages for package app-emulation/virtualbox-modules-4.1.2:
>
>   *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
>   * Please check to make sure these options are set correctly.
>   * Failure to do so may cause unexpected problems.
> <SNIP>
>
> k2 linux # cat .config | grep IOMMU
> CONFIG_GART_IOMMU=y
> # CONFIG_CALGARY_IOMMU is not set
> CONFIG_AMD_IOMMU=y
> # CONFIG_AMD_IOMMU_STATS is not set
> CONFIG_IOMMU_HELPER=y
> CONFIG_IOMMU_API=y
> # CONFIG_IOMMU_DEBUG is not set
> # CONFIG_IOMMU_STRESS is not set
> k2 linux #
>
>
>    

There are a few of them in different places.  I'm not sure which one it 
claims to need as none seems to match the name exactly.  Here is a way 
to search tho.  When you type in make menuconfig and it comes up, hit 
the / key.  That would the the question mark WITHOUT pressing the shift 
key.  Type in IOMMU and hit return.  From there you can scroll though 
with the arrow keys to find out which one you think fits.

This si my results:

root@fireball /usr/src/linux-3.0.3-gentoo # cat .config | grep IOMMU
CONFIG_GART_IOMMU=y
# CONFIG_CALGARY_IOMMU is not set
CONFIG_AMD_IOMMU=y
# CONFIG_AMD_IOMMU_STATS is not set
CONFIG_IOMMU_HELPER=y
CONFIG_IOMMU_API=y
# CONFIG_IOMMU_DEBUG is not set
# CONFIG_IOMMU_STRESS is not set
root@fireball

Yours looks the same as mine so not sure what to suggest other then the 
search above.  Maybe you are Intel based where I am AMD based?

Hope that helps.

Dale

:-)  :-)

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Hilco Wijbenga]

On 23 August 2011 20:38, Mark Knecht <markknecht@gmail.com> wrote:
> Hi all,
>   Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
> It is apparently a kernel config option no required by
> virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
> suspect there's something else I need to enable before this option
> becomes available? Unfortunately I haven't been able to Google what I
> need to do to find it.
>
> Thanks,
> Mark
>
> * Messages for package app-emulation/virtualbox-modules-4.1.2:
>
>  *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
>  * Please check to make sure these options are set correctly.
>  * Failure to do so may cause unexpected problems.
> <SNIP>
>
> k2 linux # cat .config | grep IOMMU
> CONFIG_GART_IOMMU=y
> # CONFIG_CALGARY_IOMMU is not set
> CONFIG_AMD_IOMMU=y
> # CONFIG_AMD_IOMMU_STATS is not set
> CONFIG_IOMMU_HELPER=y
> CONFIG_IOMMU_API=y
> # CONFIG_IOMMU_DEBUG is not set
> # CONFIG_IOMMU_STRESS is not set
> k2 linux #

I noticed the same. There does not appear to be any such option but
luckily its absence does not appear to affect running VB at all.

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Gregory Woodbury]

[-- Attachment #1: Type: text/plain, Size: 890 bytes --]

On Wed, Aug 24, 2011 at 12:18 AM, Hilco Wijbenga
<hilco.wijbenga@gmail.com>wrote:

> On 23 August 2011 20:38, Mark Knecht <markknecht@gmail.com> wrote:
> > Hi all,
> >   Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
> > It is apparently a kernel config option no required by
> > virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
> > suspect there's something else I need to enable before this option
> > becomes available? Unfortunately I haven't been able to Google what I
> > need to do to find it.
>

IOMMU_SUPPORT is a derived value that becomes set when any of the various
IOMMU
devices are enabled.  I enabled the IBM "CALGARY" IOMMU and the AMD IOMMU to
get the
VBox build to stop complaining.

I was also getting complaints about PCI_STUB support from VBox, but the
running VMs don't
need the vboxpcistub.ko driver to work.

-- 
G.Wolfe Woodbury

[-- Attachment #2: Type: text/html, Size: 1295 bytes --]

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Michael Orlitzky]

On 08/23/11 23:38, Mark Knecht wrote:
> 
> * Messages for package app-emulation/virtualbox-modules-4.1.2:
> 
>  *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
>  * Please check to make sure these options are set correctly.
>  * Failure to do so may cause unexpected problems.
> <SNIP>
> 
> k2 linux # cat .config | grep IOMMU

The .config file does not necessarily contain all possible options.

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Mark Knecht]

On Tue, Aug 23, 2011 at 9:18 PM, Hilco Wijbenga
<hilco.wijbenga@gmail.com> wrote:
> On 23 August 2011 20:38, Mark Knecht <markknecht@gmail.com> wrote:
>> Hi all,
>>   Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
>> It is apparently a kernel config option no required by
>> virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
>> suspect there's something else I need to enable before this option
>> becomes available? Unfortunately I haven't been able to Google what I
>> need to do to find it.
>>
>> Thanks,
>> Mark
>>
>> * Messages for package app-emulation/virtualbox-modules-4.1.2:
>>
>>  *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
>>  * Please check to make sure these options are set correctly.
>>  * Failure to do so may cause unexpected problems.
>> <SNIP>
>>
>> k2 linux # cat .config | grep IOMMU
>> CONFIG_GART_IOMMU=y
>> # CONFIG_CALGARY_IOMMU is not set
>> CONFIG_AMD_IOMMU=y
>> # CONFIG_AMD_IOMMU_STATS is not set
>> CONFIG_IOMMU_HELPER=y
>> CONFIG_IOMMU_API=y
>> # CONFIG_IOMMU_DEBUG is not set
>> # CONFIG_IOMMU_STRESS is not set
>> k2 linux #
>
> I noticed the same. There does not appear to be any such option but
> luckily its absence does not appear to affect running VB at all.
>

At the time of writing this I haven't tried Gregory's item about the
CONFIG_CALGARY_IOMMU option. I'll try that later when I get to the
machine but the reading I did on the web sounded like that was for IBM
specific hardware.

Here it seems that on the machine I was working - Intel based, i920 -
VB won't install Win 7 64-bit without it. I get a bunch of messages
that it will install but many cause problems. Are you running any
64-bit VMs?

However I've got another Intel machine where I can install Win 7
64-bit in VB just fine with no complaints at all. Here's the IOMMU
stuff from that box:

mark@c2stable ~ $ uname -a
Linux c2stable 3.0.1-gentoo #1 SMP PREEMPT Mon Aug 8 13:05:19 PDT 2011
x86_64 Intel(R) Core(TM) i7 CPU X 980 @ 3.33GHz GenuineIntel GNU/Linux
mark@c2stable ~ $ zcat /proc/config.gz | grep IOMMU
CONFIG_GART_IOMMU=y
# CONFIG_CALGARY_IOMMU is not set
# CONFIG_AMD_IOMMU is not set
CONFIG_IOMMU_HELPER=y
CONFIG_IOMMU_API=y
# CONFIG_IOMMU_DEBUG is not set
# CONFIG_IOMMU_STRESS is not set
mark@c2stable ~ $

Except for the AMD option they seem identical unless I'm misreading something.

I'm now wondering if there is both a BIOS and chipset component to
making this work? I'm pretty sure I've got all the Vt-d stuff enabled
but I'll have to wait to go into BIOS on both of them to compare.

I'll try to do a diff later today between both machine's .config files
to look for differences.

Cheers,
Mark

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Hilco Wijbenga]

On 24 August 2011 10:55, Mark Knecht <markknecht@gmail.com> wrote:
> On Tue, Aug 23, 2011 at 9:18 PM, Hilco Wijbenga
> <hilco.wijbenga@gmail.com> wrote:
>> On 23 August 2011 20:38, Mark Knecht <markknecht@gmail.com> wrote:
>>> Hi all,
>>>   Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
>>> It is apparently a kernel config option no required by
>>> virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
>>> suspect there's something else I need to enable before this option
>>> becomes available? Unfortunately I haven't been able to Google what I
>>> need to do to find it.
>>>
>>> Thanks,
>>> Mark
>>>
>>> * Messages for package app-emulation/virtualbox-modules-4.1.2:
>>>
>>>  *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
>>>  * Please check to make sure these options are set correctly.
>>>  * Failure to do so may cause unexpected problems.
>>> <SNIP>
>>>
>>> k2 linux # cat .config | grep IOMMU
>>> CONFIG_GART_IOMMU=y
>>> # CONFIG_CALGARY_IOMMU is not set
>>> CONFIG_AMD_IOMMU=y
>>> # CONFIG_AMD_IOMMU_STATS is not set
>>> CONFIG_IOMMU_HELPER=y
>>> CONFIG_IOMMU_API=y
>>> # CONFIG_IOMMU_DEBUG is not set
>>> # CONFIG_IOMMU_STRESS is not set
>>> k2 linux #
>>
>> I noticed the same. There does not appear to be any such option but
>> luckily its absence does not appear to affect running VB at all.
>>
>
> At the time of writing this I haven't tried Gregory's item about the
> CONFIG_CALGARY_IOMMU option. I'll try that later when I get to the
> machine but the reading I did on the web sounded like that was for IBM
> specific hardware.
>
> Here it seems that on the machine I was working - Intel based, i920 -
> VB won't install Win 7 64-bit without it. I get a bunch of messages
> that it will install but many cause problems. Are you running any
> 64-bit VMs?

Sort of. I'm working on installing Gentoo AMD64 in a VB machine. No
problems so far but I've only just started (it's a weekend project).
Certainly no Windows.

> However I've got another Intel machine where I can install Win 7
> 64-bit in VB just fine with no complaints at all. Here's the IOMMU
> stuff from that box:
>
> mark@c2stable ~ $ uname -a
> Linux c2stable 3.0.1-gentoo #1 SMP PREEMPT Mon Aug 8 13:05:19 PDT 2011
> x86_64 Intel(R) Core(TM) i7 CPU X 980 @ 3.33GHz GenuineIntel GNU/Linux
> mark@c2stable ~ $ zcat /proc/config.gz | grep IOMMU
> CONFIG_GART_IOMMU=y
> # CONFIG_CALGARY_IOMMU is not set
> # CONFIG_AMD_IOMMU is not set
> CONFIG_IOMMU_HELPER=y
> CONFIG_IOMMU_API=y
> # CONFIG_IOMMU_DEBUG is not set
> # CONFIG_IOMMU_STRESS is not set
> mark@c2stable ~ $
>
> Except for the AMD option they seem identical unless I'm misreading something.
>
> I'm now wondering if there is both a BIOS and chipset component to
> making this work? I'm pretty sure I've got all the Vt-d stuff enabled
> but I'll have to wait to go into BIOS on both of them to compare.
>
> I'll try to do a diff later today between both machine's .config files
> to look for differences.

Can't you simply install the Calgary IOMMU as well? I know that it
seems IBM based stuff (I noticed it myself when I was going through
the same process) which you probably do not need but I assume it can't
hurt (apart from making the kernel a bit bigger). But if it fixes your
problem... Probably worth a try.

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Mark Knecht]

On Wed, Aug 24, 2011 at 11:58 AM, Hilco Wijbenga
<hilco.wijbenga@gmail.com> wrote:
> On 24 August 2011 10:55, Mark Knecht <markknecht@gmail.com> wrote:
>> On Tue, Aug 23, 2011 at 9:18 PM, Hilco Wijbenga
>> <hilco.wijbenga@gmail.com> wrote:
>>> On 23 August 2011 20:38, Mark Knecht <markknecht@gmail.com> wrote:
>>>> Hi all,
>>>>   Can someone point me toward somehow enabling CONFIG_IOMMU_SUPPORT?
>>>> It is apparently a kernel config option no required by
>>>> virtualbox-drivers but I'm not finding it in the 3.0.3 kernel. I
>>>> suspect there's something else I need to enable before this option
>>>> becomes available? Unfortunately I haven't been able to Google what I
>>>> need to do to find it.
>>>>
>>>> Thanks,
>>>> Mark
>>>>
>>>> * Messages for package app-emulation/virtualbox-modules-4.1.2:
>>>>
>>>>  *   CONFIG_IOMMU_SUPPORT:       is not set when it should be.
>>>>  * Please check to make sure these options are set correctly.
>>>>  * Failure to do so may cause unexpected problems.
>>>> <SNIP>
>>>>
>>>> k2 linux # cat .config | grep IOMMU
>>>> CONFIG_GART_IOMMU=y
>>>> # CONFIG_CALGARY_IOMMU is not set
>>>> CONFIG_AMD_IOMMU=y
>>>> # CONFIG_AMD_IOMMU_STATS is not set
>>>> CONFIG_IOMMU_HELPER=y
>>>> CONFIG_IOMMU_API=y
>>>> # CONFIG_IOMMU_DEBUG is not set
>>>> # CONFIG_IOMMU_STRESS is not set
>>>> k2 linux #
>>>
>>> I noticed the same. There does not appear to be any such option but
>>> luckily its absence does not appear to affect running VB at all.
>>>
>>
>> At the time of writing this I haven't tried Gregory's item about the
>> CONFIG_CALGARY_IOMMU option. I'll try that later when I get to the
>> machine but the reading I did on the web sounded like that was for IBM
>> specific hardware.
>>
>> Here it seems that on the machine I was working - Intel based, i920 -
>> VB won't install Win 7 64-bit without it. I get a bunch of messages
>> that it will install but many cause problems. Are you running any
>> 64-bit VMs?
>
> Sort of. I'm working on installing Gentoo AMD64 in a VB machine. No
> problems so far but I've only just started (it's a weekend project).
> Certainly no Windows.
>
>> However I've got another Intel machine where I can install Win 7
>> 64-bit in VB just fine with no complaints at all. Here's the IOMMU
>> stuff from that box:
>>
>> mark@c2stable ~ $ uname -a
>> Linux c2stable 3.0.1-gentoo #1 SMP PREEMPT Mon Aug 8 13:05:19 PDT 2011
>> x86_64 Intel(R) Core(TM) i7 CPU X 980 @ 3.33GHz GenuineIntel GNU/Linux
>> mark@c2stable ~ $ zcat /proc/config.gz | grep IOMMU
>> CONFIG_GART_IOMMU=y
>> # CONFIG_CALGARY_IOMMU is not set
>> # CONFIG_AMD_IOMMU is not set
>> CONFIG_IOMMU_HELPER=y
>> CONFIG_IOMMU_API=y
>> # CONFIG_IOMMU_DEBUG is not set
>> # CONFIG_IOMMU_STRESS is not set
>> mark@c2stable ~ $
>>
>> Except for the AMD option they seem identical unless I'm misreading something.
>>
>> I'm now wondering if there is both a BIOS and chipset component to
>> making this work? I'm pretty sure I've got all the Vt-d stuff enabled
>> but I'll have to wait to go into BIOS on both of them to compare.
>>
>> I'll try to do a diff later today between both machine's .config files
>> to look for differences.
>
> Can't you simply install the Calgary IOMMU as well? I know that it
> seems IBM based stuff (I noticed it myself when I was going through
> the same process) which you probably do not need but I assume it can't
> hurt (apart from making the kernel a bit bigger). But if it fixes your
> problem... Probably worth a try.
>
>

I Agree and will try it. I was only saying I didn't previously because
all the write ups made it sound like it was for some other purpose.

I think my other question is where does this message about
CONFIG_IOMMU_SUPPORT get generated? Is it in the ebuild? If so then is
the ebuild incorrect because the option doesn't exist? Or does the
option exist but I cannot figure out how to get make menuconfig to
make it available? There are a number of web pages like this one:

http://cateee.net/lkddb/web-lkddb/IOMMU_SUPPORT.html

which seem to indicate that I should see an IOMMU option somewhere in
the drivers section of the kernel config. I do not, however, find that
right now which is why I came here to ask.

Anyway, I will be trying that later today.

Cheers,
Mark

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Paul Hartman]

On Wed, Aug 24, 2011 at 2:13 PM, Mark Knecht <markknecht@gmail.com> wrote:
> http://cateee.net/lkddb/web-lkddb/IOMMU_SUPPORT.html
>
> which seem to indicate that I should see an IOMMU option somewhere in
> the drivers section of the kernel config. I do not, however, find that
> right now which is why I came here to ask.

It looks to me like the specific option CONFIG_IOMMU_SUPPORT was just
added for kernel 3.1, so you won't see it until then. It even mentions
it on the page you linked to:: "found in Linux kernels: 3.1-rc+HEAD"

It appears virtualbox-modules 4.1.0 doesn't check this. Maybe 4.1.2 is
developed against the 3.1 kernel... Maybe you should switch to the
latest git kernel, if you dare, or maybe you could mask the vbox
modules 4.1.2 and use 4.1.0 instead? I'm not a virtualbox user so I
have no idea if that's a reasonable suggestion or not, but I'm just
throwing it out there. :)

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Mark Knecht]

On Wed, Aug 24, 2011 at 12:55 PM, Paul Hartman
<paul.hartman+gentoo@gmail.com> wrote:
> On Wed, Aug 24, 2011 at 2:13 PM, Mark Knecht <markknecht@gmail.com> wrote:
>> http://cateee.net/lkddb/web-lkddb/IOMMU_SUPPORT.html
>>
>> which seem to indicate that I should see an IOMMU option somewhere in
>> the drivers section of the kernel config. I do not, however, find that
>> right now which is why I came here to ask.
>
> It looks to me like the specific option CONFIG_IOMMU_SUPPORT was just
> added for kernel 3.1, so you won't see it until then. It even mentions
> it on the page you linked to:: "found in Linux kernels: 3.1-rc+HEAD"
>
> It appears virtualbox-modules 4.1.0 doesn't check this. Maybe 4.1.2 is
> developed against the 3.1 kernel... Maybe you should switch to the
> latest git kernel, if you dare, or maybe you could mask the vbox
> modules 4.1.2 and use 4.1.0 instead? I'm not a virtualbox user so I
> have no idea if that's a reasonable suggestion or not, but I'm just
> throwing it out there. :)
>
>
Thanks for the insights Paul. I'll roll back to a slightly earlier
version of Virtualbox which will likely solve the problem.

Cheers,
Mark

Re: [gentoo-user] CONFIG_IOMMU_SUPPORT???

[Joost Roeleveld]

On Wednesday, August 24, 2011 01:10:14 PM Mark Knecht wrote:
> On Wed, Aug 24, 2011 at 12:55 PM, Paul Hartman
> 
> <paul.hartman+gentoo@gmail.com> wrote:
> > On Wed, Aug 24, 2011 at 2:13 PM, Mark Knecht <markknecht@gmail.com> wrote:
> >> http://cateee.net/lkddb/web-lkddb/IOMMU_SUPPORT.html
> >> 
> >> which seem to indicate that I should see an IOMMU option somewhere in
> >> the drivers section of the kernel config. I do not, however, find that
> >> right now which is why I came here to ask.
> > 
> > It looks to me like the specific option CONFIG_IOMMU_SUPPORT was just
> > added for kernel 3.1, so you won't see it until then. It even mentions
> > it on the page you linked to:: "found in Linux kernels: 3.1-rc+HEAD"
> > 
> > It appears virtualbox-modules 4.1.0 doesn't check this. Maybe 4.1.2 is
> > developed against the 3.1 kernel... Maybe you should switch to the
> > latest git kernel, if you dare, or maybe you could mask the vbox
> > modules 4.1.2 and use 4.1.0 instead? I'm not a virtualbox user so I
> > have no idea if that's a reasonable suggestion or not, but I'm just
> > throwing it out there. :)
> 
> Thanks for the insights Paul. I'll roll back to a slightly earlier
> version of Virtualbox which will likely solve the problem.
> 
> Cheers,
> Mark

I just checked on my system and with Virtualbox 4.0.4, it works. (ok, it's 
older :)

# " zcat /proc/config.gz | grep IOMMU " gives:
**
# CONFIG_CALGARY_IOMMU is not set
# CONFIG_AMD_IOMMU is not set
CONFIG_IOMMU_HELPER=y
# CONFIG_IOMMU_API is not set
# CONFIG_IOMMU_STRESS is not set
** kernel = gentoo-sources-2.6.34-r6

I can succesfully run 64 and 32bit clients on this machine. CPU is Intel Xeon

On my other machine (with Virtualbox 4.0.12 and AMD Phenom CPU I get the 
following for the same command:
**
CONFIG_GART_IOMMU=y
CONFIG_CALGARY_IOMMU=y
CONFIG_CALGARY_IOMMU_ENABLED_BY_DEFAULT=y
CONFIG_AMD_IOMMU=y
CONFIG_AMD_IOMMU_STATS=y
CONFIG_IOMMU_HELPER=y
CONFIG_IOMMU_API=y
# CONFIG_IOMMU_DEBUG is not set
# CONFIG_IOMMU_STRESS is not set
** kernel = gentoo-sources-2.6.36-r5

I should probably disable some stuff, but not sure why the CALGARY got enabled 
by default.

I haven't gotten round to testing 4.1.2 yet.

--
Joost