Re: [PATCH] VT-d: improve RMRR validity checking

[Noboru Iwamatsu <n_iwamatsu@jp.fujitsu.com>]

 > Seems you inserted a discrete gfx card, then the IGD is disabled. right?

Yes, I'm using Q35 M/B with external PCIe graphic card.
IGD of the M/B is disabled automatically, however,
the entry of its DRHD and RMRR are found.

This issue only occurs when using external VGA.
The rough solution is to ignore the DRHD/RMRR including
BDF[0.2.0] or BDF[0.2.1].
I had been doing this way.

 > can you provide the platform info and BIOS version?

My platform is Fujitsu's private one.

Regards,
Noboru.

> Seems you inserted a discrete gfx card, then the IGD is disabled. right?
> can you provide the platform info and BIOS version?
>
> Regards,
> Weidong
>
> Noboru Iwamatsu wrote:
>> Hi Weidong,
>>
>> Here is the log.
>> Xen version is:
>> xen-unstable c/s 20829 + return.patch + rmrr.patch
>>
>> Regards,
>> Noboru
>>
>>> Noboru Iwamatsu wrote:
>>>> Hi,
>>>>
>>>> After registered invalid DRHDs, Xen hangs in boot time.
>>> Can you post the logs?
>>>
>>> Regards,
>>> Weidong
>>>> About RMRR, I understood the logic.
>>>> In my mainboard, unfortunately, RMRR has non-existent device under
>>>> its scope, and to make matters worse, the RMRR range is invalid.
>>>> So, I think RMRR that has no-existent device is valid.
>>>>
>>>> How do you think about these?
>>>>
>>>>> Hi Noboru,
>>>>>
>>>>> You should not ignore DRHD even if devices under its scope are not
>>>>> pci discoverable. For the sake of security, we still enable these
>>>>> DRHDs but don't set any context mappings. In that case, any DMA that
>>>>> comes from these "supposedly disabled" devices will get blocked by
>>>>> VT-d, and hence avoid any security vulnerability with malicious s/w
>>>>> re-enabling these devices.
>>>>>
>>>>> You RMRR validity fixing is wrong. My RMRR patch is no problem. Pls
>>>>> note that the RMRR checking logic is:
>>>>> If all devices under RMRR's scope are not pci discoverable
>>>>> Ignore the RMRR
>>>>> Else if base_address> end_address
>>>>> Return error
>>>>> Else
>>>>> Register RMRR
>>>>>
>>>>> Regards,
>>>>> Weidong
>>>>>
>>>>>
>>>>> -----Original Message-----
>>>>> From: Noboru Iwamatsu [mailto:n_iwamatsu@jp.fujitsu.com]
>>>>> Sent: Thursday, January 21, 2010 4:26 PM
>>>>> To: Han, Weidong
>>>>> Cc: xen-devel@lists.xensource.com; keir.fraser@eu.citrix.com
>>>>> Subject: Re: [Xen-devel] [PATCH] VT-d: improve RMRR validity checking
>>>>>
>>>>> Hi,
>>>>>
>>>>> Some Q35 mainboard that has buggy BIOS, I have one of this, reports
>>>>> invalid DRHD in addition to the invalid RMRR.
>>>>>
>>>>> Attached patch fixes this DRHD issue in the same way as RMRR.
>>>>> And also, I fixed RMRR validity checking loop.
>>>>>
>>>>> Noboru.
>>>>>
>>>>> Signed-off-by: Noboru Iwamatsu<n_iwamatsu@jp.fujitsu.com>
>>>>>
>>>>>
>>>>> -------- Original Message --------
>>>>> Subject: [Xen-devel] [PATCH] VT-d: improve RMRR validity checking
>>>>> From: Han, Weidong<weidong.han@intel.com>
>>>>> To: xen-devel@lists.xensource.com<xen-devel@lists.xensource.com>
>>>>> Date: Thu Jan 21 2010 11:46:12 GMT+0900
>>>>>
>>>>>> Currently, Xen checks RMRR range and disables VT-d if RMRR range is
>>>>>> set incorrectly in BIOS rigorously. But, actually we can ignore the
>>>>>> RMRR if the device under its scope are not pci discoverable, because
>>>>>> the RMRR won't be used by non-existed or disabled devices.
>>>>>>
>>>>>> This patch ignores the RMRR if the device under its scope are not
>>>>>> pci discoverable, and only checks the validity of RMRRs that are
>>>>>> actually used. In order to avoid duplicate pci device detection
>>>>>> code, this patch defines a function pci_device_detect for it.
>>>>>>
>>>>>> Signed-off-by: Weidong Han<weidong.han@intel.com>
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Xen-devel mailing list
>>>>>> Xen-devel@lists.xensource.com
>>>>>> http://lists.xensource.com/xen-devel
>>
>