On Mon, 2025-07-21 at 09:50 -0500, Tom Lendacky wrote: > On 7/21/25 08:08, Tom Lendacky wrote: > > On 7/17/25 16:46, Kai Huang wrote: > > > This series is the latest attempt to support kexec on TDX host following > > > Dave's suggestion to use a percpu boolean to control WBINVD during > > > kexec. > > > > > > Hi Boris/Tom, > > > > > > As requested, I added the first patch to cleanup the last two 'unsigned > > > int' parameters of the relocate_kernel() into one 'unsigned int' and pass > > > flags instead. The patch 2 (patch 1 in v3) also gets updated based on > > > that. Would you help to review? Thanks. > > > > > > I tested that both normal kexec and preserve_context kexec works (using > > > the tools/testing/selftests/kexec/test_kexec_jump.sh). But I don't have > > > SME capable machine to test. > > > > > > Hi Tom, I added your Reviewed-by and Tested-by in the patch 2 anyway > > > since I believe the change is trivial and straightforward). But due to > > > the cleanup patch, I appreciate if you can help to test the first two > > > patches again. Thanks a lot! > > > > Everything is working, Thanks! > > See my comments in patch #1. I didn't test with context preservation, so > that bit was never set. If it was, I think things would have failed. I actually tested the test_kexec_jump.sh in kselftest as mentioned above in a VM, as mentioned above. I got "# kexec_jump succeeded [PASS]" so I think it worked :-) But unfortunately I don't know how to test preserve_context kexec in any other way.
