On 23/08/2021 13:39, Bertrand Marquis wrote:
Hi Julien,
Hi Bertrand,
On 23 Aug 2021, at 12:47, Julien Grall <[email protected]> wrote:
Hi Bertrand,
On 23/08/2021 11:32, Bertrand Marquis wrote:
Replace the code in p2m trying to find a sane value for the VMID size
supported and the PAR to use. We are now using the boot cpuinfo as the
values there are sanitized during boot and the value for those
parameters is now the safest possible value on the system.
On arm32, the system will panic if there are different types of core so
those checks were not needed anyway.
So the assumption is that if you have the same MIDR, then you must have the
same features. I understand this is what Xen assumes today but I never viewed
that check as the truth. It is more to limit the damage on most platform.
This was the assumption before, I did not change anything but just explained in
the commit message why this was not possible to come to this code.
Yes. This is not a new, however I thought I would mention it because I
want to avoid continuing to use wrong assumptions.
However, this code is arm64 only as it is #ifdef right? (Sorry I should
have looked at the code the first time) So ...
So can you confirm whether this is something that Arm guarantees?
For a specific MIDR from Arm (ie a Cortex) the PAR is fixed and VMID size to.
But for an other Arm architecture processor I cannot say.
That said, I am not against removing the code. But I would like the comment to
be amended if this is not a correct assumption.
Would the following be acceptable:
On arm32, Xen is not booting on systems having different MIDR amongst cores and
it is assumed that cores with the same MIDR will have the same PAR and VMID
size.
... I would just drop any mention of arm32 here.
Cheers,
--
Julien Grall
