On Tue, May 28, 2024 at 08:59:46AM GMT, Peter Krempa wrote:
On Mon, May 27, 2024 at 19:31:34 +0200, Andrea Bolognani wrote:
Note that TPM supports seems to be currently missing from s390x and loongarch64 QEMU builds. I'm not entirely sure whether there's a good reason for that or it's simply an oversight, but either way we have to skip them for now.
I presume you mean that TPM support was not built into the QEMU builds used to capture the capability dumps, right?
That's just the thing: I don't know :) In both cases the default configuration disables it, but it's unclear to me whether that is something that could be addressed with a simple patch or there are factors at play that make TPM inherently incompatible with these architectures.
+++ b/tests/qemuxmlconfdata/s390x-ccw-default-models.xml @@ -14,6 +14,7 @@ <mac address='52:54:00:09:a4:37'/> </interface> <serial type='pty'/> + <!-- TPM support not currently built on s390x -->
For s390x, Thomas has confirmed that TPM is just not a thing and will never be a thing, so while the comment is technically accurate I agree that stronger language should be used to reflect the situation. Thomas, please correct me if I got this wrong :)
+++ b/tests/qemuxmlconfdata/loongarch64-virt-default-models.xml @@ -14,6 +14,7 @@ <mac address='52:54:00:09:a4:37'/> </interface> <serial type='pty'/> + <!-- TPM support not currently built on loongarch64 -->
For loongarch64, maybe Xianglai Li can provide some insight. Is TPM something that could be flipped on at the QEMU level? -- Andrea Bolognani / Red Hat / Virtualization