Re: [Xen-devel] [PATCH v12 00/10] x86: multiboot2 protocol support

[Doug Goldstein <cardoe@xxxxxxxxxx>]

On 1/23/17 9:45 AM, Daniel Kiper wrote:
> On Mon, Jan 23, 2017 at 09:35:55AM -0600, Doug Goldstein wrote:
>> On 1/23/17 7:08 AM, Daniel Kiper wrote:
>>> On Fri, Jan 20, 2017 at 10:54:12PM +0100, Daniel Kiper wrote:
>>>> On Fri, Jan 20, 2017 at 02:42:30PM -0500, Doug Goldstein wrote:
>>>>> On 1/19/17 8:34 PM, Daniel Kiper wrote:
>>>>>> Hi,
>>>>>>
>>>>>> I am sending twelfth version of multiboot2 protocol support for
>>>>>> legacy BIOS and EFI platforms. This patch series release contains
>>>>>> fixes for all known/confirmed issues.
>>>>>
>>>>> With my fix to efi_multiboot2() in 5/10 and the entire series applied, I
>>>>> get the following on some of the systems I have access to:
>>>>>
>>>>> (XEN) [    2.000533] HVM: HAP page sizes: 4kB, 2MB, 1GB
>>>>> (XEN) [    7.012109] Stuck ??
>>>>> (XEN) [    7.012129] Failed to bring up CPU 1 (error -5)
>>>>> (XEN) [   12.023606] Stuck ??
>>>>> (XEN) [   12.023622] Failed to bring up CPU 2 (error -5)
>>>>> (XEN) [   17.035099] Stuck ??
>>>>> (XEN) [   17.035115] Failed to bring up CPU 3 (error -5)
>>>>> (XEN) [   17.035116] Brought up 1 CPUs
>>>>>
>>>>> On other machines they reset when setting PAGING into cr0 (actually the
>>>>> jmp following it) on line 124 of trampoline.S
>>>>
>>>> Thanks! I will take a look.
>>>>
>>>>> If I drop the series to just 2-5 against staging (since patch 1 has
>>>>> already gone in) and apply the fix to efi_multiboot2() then all the
>>>>> machines I presently have access to boot.
>>>>
>>>> Great!
>>>>
>>>>> Effectively the fix to efi_multiboot2() gets us back to the same level
>>>>> of hardware support that v11 + my v5 was at for 1-5. So I will extend my:
>>>>>
>>>>> Reviewed-by: Doug Goldstein <cardoe@xxxxxxxxxx>
>>>>> Tested-by: Doug Goldstein <cardoe@xxxxxxxxxx>
>>>>
>>>> Thanks!
>>>>
>>>>> on the condition that the fix is applied to 5/10 prior to commit.
>>>>
>>>> Will do.
>>>>
>>>> By the way, I have asked my team colleagues to do more tests of this 
>>>> series.
>>>> I will come back to you if I have something in hand.
>>>
>>> Once you told me that you applied some patches on top of my patch series to 
>>> get
>>> it working. Is it still true? If you still use some extra patches could you 
>>> send
>>> me them? What about ExitBootServices() call? Did you disabled it? If yes on 
>>> which
>>> machines it have to be disabled?
>>>
>>> Daniel
>>>
>>
>> I previously used the patch that I linked to you authored by Konrad. I
>> have since switched to the patches that XenServer uses to no-op
>> efi_get_time() and to map additional ranges of reserved memory to make
>> EBS() work.
> 
> Could you send me them?
> 
> Daniel
> 

I apply them only for 2 out of the roughly dozen machines. Its not those
changes.

https://github.com/xenserver/xen-4.7.pg/blob/master/master/0002-efi-Ensure-incorrectly-typed-runtime-services-get-ma.patch

https://github.com/xenserver/xen-4.7.pg/blob/master/master/0001-x86-time-Don-t-use-EFI-s-GetTime-call.patch

-- 
Doug Goldstein

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel