Coral TPU PCI Passthrough

With regards of Marek patches, I'm actually think it can worth a try (at least the patch seems to treat the problem where MSI-x PBA page is shared with other regs of the device), but there's some cons too:

the patches are quite new (doesn't seems to be integrated yet).
the patches can be applied to more recent Xen (not XCP-ng Xen), and even we could probably backport them, it potentially will require some significant work
we are not 100% sure it's the issue (or the only issue)

So If this is a must have, we can go and do some digging to make it work (but still in the scope of "exeperimental" platform, not the production platform)

olivierlambert

We could probably try on a non-XCP-ng platform with a very recent "vanilla" Xen (+Marek patches) and see if it's fixed. If it is, then we could think about a potential backport when 8.3 will include a more recent Xen version

andSmv

@logical-systems I will check which Xen version the patches are easily applied and If you want I could give you a hand (if needed) to build and install your builded XEN, so you can test if this resolve your issue.

Unfortunatly we don't have the related HW (Coral TPU) to test it by ourselves.

UPDATE: the both patches apply to xen 4.17 (tag RELEASE-4.17.0)

Nornode

@andSmv // @logical-systems

Hi,

I'm researching XCP-NG as an alternative to my homelab VMware hypervisor.
A goal for me is to get proper USB passthrough of the Google Coral TPU.

Did these patches make it work so passthrough to a VM is confirmed to be working?

redakula

@andSmv said in Coral TPU PCI Passthrough:

@logical-systems I will check which Xen version the patches are easily applied and If you want I could give you a hand (if needed) to build and install your builded XEN, so you can test if this resolve your issue.

Unfortunatly we don't have the related HW (Coral TPU) to test it by ourselves.

UPDATE: the both patches apply to xen 4.17 (tag RELEASE-4.17.0)

So the above mentioned patches are included in the 4.17 that is currently available as a test version?

Or did you mean the patches worked on that version?

olivierlambert

Now we have Xen 4.17 in XCP-ng 8.3, that might work (ping @andSmv )

andSmv

@redakula Hello, unfortunately these patches are not in 4.17 Xen (and was never integrated in more recent Xen). So, to test it, you have to manually apply patches (normally should apply as is to 4.17) and rebuild your Xen.

redakula

@andSmv
Damn - i was quick and have a coral m2 A+E coming in a few days

It's just for fun/learning so as long as it doesn't break my homelab too much i will be willing to test so we might get it included
Already using the 4.17 test version without a hitch since it came out.

redakula

@andSmv
As expected the VM with the coral m2 crashes on boot.

Where would i start with building a custom Xen? The Koji docs seem directed at authorized package maintainers so would i need to build the sources directly from Xen?
Feeling old admitting it was in the 2.6 days i last regularly built custom kernels

andSmv

@redakula I'm on it. I keep you posted.

andSmv

@andSmv
Hello,
I integrated Marek's patch and builded a rpm, so you can install (may be need to force rpm install or extract the xen.gz from rpm and install it manually if you prefer)

Obviously there's no guarantee, it'll work in your case. Moreover, I didn't test the patch, so please backup all your data. It should be harmless, but....

Here's the link you can download the rpm (should be operational until the end of the month) https://nextcloud.vates.fr/index.php/s/gd7kMwxHtNEP329

Don't hesitate to ping me if you experience any issue to download/install/... the patched xen.

Hope it helps!

P.S. Be sure you're running 8.3 XCP-ng, as I only uploaded xen hypervisor rpm (and not libs/tools which come within)

redakula

@andSmv Thanks!

I tried to be as uninvasive as possible and changed the symbolic link xen.gz to point to the xen.gz from the RPM you created.

Unfortunately still the same error (It does seem to boot the xen from the RPM as this has version 4.17.3-3 vs. the one currently in the repos which has version 4.17.3-4).

[2024-05-24 17:06:33] (XEN) [  674.051176] Domain 14 (vcpu#2) crashed on cpu#22:
[2024-05-24 17:06:33] (XEN) [  674.051178] ----[ Xen-4.17.3-3  x86_64  debug=n  Not tainted ]----
[2024-05-24 17:06:33] (XEN) [  674.051179] CPU:    22
[2024-05-24 17:06:33] (XEN) [  674.051180] RIP:    0010:[<ffffffffa8581584>]
[2024-05-24 17:06:33] (XEN) [  674.051180] RFLAGS: 0000000000000286   CONTEXT: hvm guest (d14v2)
[2024-05-24 17:06:33] (XEN) [  674.051182] rax: ffffbd9c00149800   rbx: ffff9e9247cc9000   rcx: 0000000000000000
[2024-05-24 17:06:33] (XEN) [  674.051182] rdx: 00000000fee77000   rsi: 0000000000000000   rdi: 0000000000000000
[2024-05-24 17:06:33] (XEN) [  674.051183] rbp: ffffbd9c00327690   rsp: ffffbd9c00327658   r8:  0000000000000000
[2024-05-24 17:06:33] (XEN) [  674.051183] r9:  0000000000000000   r10: 0000000000000000   r11: 0000000000000000
[2024-05-24 17:06:33] (XEN) [  674.051184] r12: ffffbd9c003276ac   r13: 0000000000000011   r14: ffff9e92413390c0
[2024-05-24 17:06:33] (XEN) [  674.051185] r15: 0000000000000077   cr0: 0000000080050033   cr4: 0000000000750ef0
[2024-05-24 17:06:33] (XEN) [  674.051185] cr3: 0000000103806000   cr2: 0000000000000000
[2024-05-24 17:06:33] (XEN) [  674.051186] fsb: 00007b6e7a42a8c0   gsb: ffff9e925b500000   gss: 0000000000000000
[2024-05-24 17:06:33] (XEN) [  674.051186] ds: 0000   es: 0000   fs: 0000   gs: 0000   ss: 0018   cs: 0010

It does appear that there is some movement upstream on this (if i interpret the xen mailing list correctly).
This patch series references the same title as the patch in this thread from 2022 and a bunch of other related work:
https://lore.kernel.org/xen-devel/cover.33fb4385b7dd6c53bda4acf0a9e91748b3d7b1f7.1715313192.git-series.marmarek@invisiblethingslab.com/

andSmv

@redakula
Well, this was unfortunately one of the potential outcome. Unfortunately we don't have the hardware to make more "in deep" debug. I will talk to Marek next week (on Xen Summit) about this patch series and if we could expect it eventually fix the issue with Coral TPU.
Will keep you posted.