Re: [PATCH v2 06/11] tpm_crb: move ACPI table building to device interface

[Stefan Berger <stefanb@linux.ibm.com>]

On 7/31/23 23:02, Joelle van Dyne wrote:
> On Mon, Jul 17, 2023 at 6:42 AM Igor Mammedov <imammedo@redhat.com> wrote:
>>
>> On Fri, 14 Jul 2023 13:21:33 -0400
>> Stefan Berger <stefanb@linux.ibm.com> wrote:
>>
>>> On 7/14/23 03:09, Joelle van Dyne wrote:
>>>> This logic is similar to TPM TIS ISA device. Since TPM CRB can only
>>>> support TPM 2.0 backends, we check for this in realize.
>>>>
>>>> Signed-off-by: Joelle van Dyne <j@getutm.app>
>>>
>>> This patch changes the order of in which the ACPI table elements are created but doesn't matter and also doesn't seem to upset ACPI test cases from what I saw:
>>
>> it seems we do have tests for TIS only (which I added when I was refactoring it to TYPE_ACPI_DEV_AML_IF)
>> perhaps add a test for CRB before this patch a follow process described in bios-tables-test.c
>> for updating expected blob
> I read the file and looked at the commits for TIS tests but I'm not
> sure I understand how it works. At what point do I specify that the
> CRB device should be created for the test?

For me it would be a bit of trial an error as well. So here's my best guess:

Did you look at b193e5f9cccb322b0febd5a2aba486? You basically have to find out which files
are going to change due to extending the tests, so doing something like in that patch comes
after you found out which files are changing and iirc the tests are going to complain about
those files. So I would try to first add CRB tests similar to the following to tests/qtest/bios-tables-test.c.
in one patch, then run the test cases and they will tell you which files changed, and then
add a patch similar to b193e5f9cccb322b0febd5a2aba486 before the test-enabling patch.

             if (tpm_model_is_available("-machine q35", "tpm-tis")) {
                 qtest_add_func("acpi/q35/tpm2-tis", test_acpi_q35_tcg_tpm2_tis);
                 qtest_add_func("acpi/q35/tpm12-tis",
                                test_acpi_q35_tcg_tpm12_tis);
             }

I would try to something like the above to aarch64 here:

     } else if (strcmp(arch, "aarch64") == 0) {
         if (has_tcg && qtest_has_device("virtio-blk-pci")) {
             qtest_add_func("acpi/virt", test_acpi_virt_tcg);
             qtest_add_func("acpi/virt/acpihmatvirt",
                             test_acpi_virt_tcg_acpi_hmat);


Then you run the tests again then it should create those files with the ACPI data and you copy them
to their destination (like in ca745d2277496464b54fd832c15c45d0227325bb) and remove the changes from
tests/qtest/bios-tables-test-allowed-diff.h and that becomes your 3rd patch. Once you run the tests
again with the 3rd patch there should be no more complaints about ACPI related changes.

Since CRB ACPI tests are not enabled right now you can add these patches somewhere in the middle of
the series or also at the end.


I hope this helps.

    Stefan