Mapping memory regions on s390
A message that is not in the archive (sent off-list or not yet received): 20190221171747.64181ee8@oc2783563651
Quoted 92 lines · On Thu, 21 Feb 2019 17:17:47 +0100 Halil Pasic <[email protected]> wrote:
Quoted 89 lines · On Thu, 21 Feb 2019 15:46:08 +0000 "Dr. David Alan Gilbert"…
Quoted 64 lines · * Cornelia Huck ([email protected]) wrote:
Quoted 52 lines · On Thu, 21 Feb 2019 15:11:50 +0000 "Dr. David Alan Gilbert"…
Quoted 42 lines · * Cornelia Huck ([email protected]) wrote:
Quoted 39 lines · On Thu, 21 Feb 2019 13:39:46 +0000 "Dr. David Alan Gilbert"…
Quoted 27 lines · * Cornelia Huck ([email protected]) wrote:
Quoted 24 lines · On Mon, 18 Feb 2019 15:41:07 +0000 "Dr. David Alan Gilbert"…
Quoted 21 lines · * Cornelia Huck ([email protected]) wrote:
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
We can of course switch the order of mappings
[
0x000000000000000]
... Memory region containing RAM
[ram_size]
... Memory region for memory devices (virtio-pmem, virtio-mem ...) [maxram_size-ram_size]
... Memory region for e.g. special PCI/CCW devices
[ TBD]Quoted 7 lines
Quoted 7 lines
Quoted 2 lines
We can size TBD in a way that we e.g. max out the current page table size before having to switch to more levels.
Yes, that's fine to set some upper limit; you've just got to make sure that the hypervisor knows where it can put stuff and if the guest does PCI that it knows where it's allowed to put stuff and as long as the two don't overlap everyone is happy.
Hm... is that an issue for pci? Do we need to care, as s390 uses special instructions anyway? Or do we want to avoid going through them, so that the guest can use normal read/write?
Quoted 8 lines
Quoted 4 lines
That depends.
The stuff we use for virtio-fs we need the shared region to be accessible by the guest via normal instructions because we're using for DAX. For PCI you might be able to avoid it for most other PCI cases.So,
- virtio-fs regions need to be accessible like normal memory, so they need to show up in the region labeled 'TBD' aboveYes.
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
(it would fine to
communicate the 'where' through pci structures)Hmm, mixing PCI structures into something you're not treating as PCI seems weird to me.
I was thinking about the addresses in the TBD area... they need to go through _some_ pci structure, I assume?
Quoted 8 lines
Well I think it depends how you make it work with CCW;
if the addresses being assigned are assigned by the host then I believe you should use the CCW mechanism you suggested to discover the addresses in the guest.
If the host is going to allocate a block of PCI space and the guest is going to allocate the use within that space and access it with normal instructions then I think it should go via PCI.
Quoted 7 lines
This is getting a bit confusing... let me try to summarize:
- We introduce a special area where shared memory areas are supposed to live.
- If a virtio device accessed via ccw defines shared regions, the driver can discover them via a new ccw that indicates an address in that special area.Right.
Quoted 6 lines
- If a virtio device accessed via pci defines shared regions, the driver will want to discover them via the same mechanism as on other platforms. If I read
https://lists.oasis-open.org/archives/virtio-comment/201901/msg00003.html
correctly, this will mean an offset into a BAR. This will be a normal pci memory region.Right; note that like any other BAR it's upto the guest to write the BAR to select which area of GPA it wants the BAR to map to.
Quoted 3 lines
Now, this sounds to me that we'll have regions in different memory regions, depending on whether they are accessed via ccw or via pci. Not sure if that's a problem.
Quoted 6 lines
Quoted 2 lines
As long as they stay out of each others way it shouldn't be.
I'm not sure. What I'm concerted about is:
* if it is a virtio-ccw device it would need to access the shared memory via normal memory access instructions.
Just for reference, my current draft for the ccw spec looks as follows: