Re: [Xen-devel] [PATCH] x86/SVM: correct boot time cpu_data[] handling

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 02.03.17 at 16:19, <boris.ostrovsky@xxxxxxxxxx> wrote:
> On 03/02/2017 10:02 AM, Jan Beulich wrote:
>>>>> On 02.03.17 at 15:58, <boris.ostrovsky@xxxxxxxxxx> wrote:
>>> On 03/02/2017 08:08 AM, Jan Beulich wrote:
>>>> start_svm() already runs after cpu_data[] was set up, so it shouldn't
>>>> modify it anymore (at least not directly). Constify the involved
>>>> pointers.
>>>>
>>>> Furthermore LMSLE feature detection was broken by 566ddbe833 ("x86:
>>>> Fail CPU bringup cleanly if it cannot initialise HVM"), as Andrew
>>>> Cooper has pointed out: c couldn't possibly equal &boot_cpu_data
>>>> anymore.
>>> I can't find Andrew's email --- why is this not possible?
>> Oh, that was in a private irc chat. The issue is that prior to cited
>> commit, &boot_cpu_data would have come into the function. The
>> change done results in the check effectively having become
>>
>>      &cpu_data[0] == &boot_cpu_data
>>
>> which is never true.
> 
> Can't we use 'cpu==0' as BSP test?

While a lot of code does so, it's conceptually wrong, and hence
I'm trying to avoid introducing new instance (also during patch
review): Eventually we will want CPU0 to be offlineable as well,
and with that it could of course also be brought back online.
Linux has gone through this a year or two back, iirc.

Jan


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