On 06.05.20 16:03, Janosch Frank wrote:
> On 5/6/20 2:46 PM, Marc Hartmayer wrote:
>> Add support for Protected Virtual Machine (PVM) tests. For starting a
>> PVM guest we must be able to generate a PVM image by using the
>> `genprotimg` tool from the s390-tools collection. This requires the
>> ability to pass a machine-specific host-key document, so the option
>> `--host-key-document` is added to the configure script.
>>
>> Signed-off-by: Marc Hartmayer <mhartmay@xxxxxxxxxxxxx>
>> ---
>> .gitignore | 1 +
>> configure | 8 ++++++++
>> s390x/Makefile | 16 +++++++++++++---
>> s390x/unittests.cfg | 20 ++++++++++++++++++++
>> scripts/common.bash | 30 +++++++++++++++++++++++++++++-
>> 5 files changed, 71 insertions(+), 4 deletions(-)
>>
>> diff --git a/.gitignore b/.gitignore
>> index 784cb2ddbcb8..1fa5c0c0ea76 100644
>> --- a/.gitignore
>> +++ b/.gitignore
>> @@ -4,6 +4,7 @@
>> *.o
>> *.flat
>> *.elf
>> +*.img
>> .pc
>> patches
>> .stgit-*
>> diff --git a/configure b/configure
>> index 5d2cd90cd180..29191f4b0994 100755
>> --- a/configure
>> +++ b/configure
>> @@ -18,6 +18,7 @@ u32_long=
>> vmm="qemu"
>> errata_force=0
>> erratatxt="errata.txt"
>> +host_key_document=
>>
>> usage() {
>> cat <<-EOF
>> @@ -40,6 +41,8 @@ usage() {
>> no environ is provided by the user (enabled by default)
>> --erratatxt=FILE specify a file to use instead of errata.txt. Use
>> '--erratatxt=' to ensure no file is used.
>> + --host-key-document=HOST_KEY_DOCUMENT
>> + host-key-document to use (s390x only)
>> EOF
>> exit 1
>> }
>> @@ -91,6 +94,9 @@ while [[ "$1" = -* ]]; do
>> --erratatxt)
>> erratatxt="$arg"
>> ;;
>> + --host-key-document)
>> + host_key_document="$arg"
>> + ;;
>> --help)
>> usage
>> ;;
>> @@ -207,6 +213,8 @@ PRETTY_PRINT_STACKS=$pretty_print_stacks
>> ENVIRON_DEFAULT=$environ_default
>> ERRATATXT=$erratatxt
>> U32_LONG_FMT=$u32_long
>> +GENPROTIMG=genprotimg
>> +HOST_KEY_DOCUMENT=$host_key_document
>> EOF
>>
>> cat <<EOF > lib/config.h
>> diff --git a/s390x/Makefile b/s390x/Makefile
>> index ddb4b48ecbf9..a57655dcce10 100644
>> --- a/s390x/Makefile
>> +++ b/s390x/Makefile
>> @@ -17,12 +17,19 @@ tests += $(TEST_DIR)/stsi.elf
>> tests += $(TEST_DIR)/skrf.elf
>> tests += $(TEST_DIR)/smp.elf
>> tests += $(TEST_DIR)/sclp.elf
>> -tests_binary = $(patsubst %.elf,%.bin,$(tests))
>>
>> -all: directories test_cases test_cases_binary
>> +tests_binary = $(patsubst %.elf,%.bin,$(tests))
>> +ifneq ($(HOST_KEY_DOCUMENT),)
>> +tests_pv_img = $(patsubst %.elf,%.pv.img,$(tests))
>> +else
>> +tests_pv_img =
>> +endif
>> +
>> +all: directories test_cases test_cases_binary test_cases_pv
>>
>> test_cases: $(tests)
>> test_cases_binary: $(tests_binary)
>> +test_cases_pv: $(tests_pv_img)
>>
>> CFLAGS += -std=gnu99
>> CFLAGS += -ffreestanding
>> @@ -68,8 +75,11 @@ FLATLIBS = $(libcflat)
>> %.bin: %.elf
>> $(OBJCOPY) -O binary $< $@
>>
>> +%.pv.img: %.bin $(HOST_KEY_DOCUMENT)
>> + $(GENPROTIMG) --host-key-document $(HOST_KEY_DOCUMENT) --no-verify --image $< -o $@
>> +
>> arch_clean: asm_offsets_clean
>> - $(RM) $(TEST_DIR)/*.{o,elf,bin} $(TEST_DIR)/.*.d lib/s390x/.*.d
>> + $(RM) $(TEST_DIR)/*.{o,elf,bin,img} $(TEST_DIR)/.*.d lib/s390x/.*.d
>>
>> generated-files = $(asm-offsets)
>> $(tests:.elf=.o) $(cstart.o) $(cflatobjs): $(generated-files)
>> diff --git a/s390x/unittests.cfg b/s390x/unittests.cfg
>> index b307329354f6..6beaca45fb20 100644
>> --- a/s390x/unittests.cfg
>> +++ b/s390x/unittests.cfg
>> @@ -16,6 +16,8 @@
>> # # a test. The check line can contain multiple files
>> # # to check separated by a space but each check
>> # # parameter needs to be of the form <path>=<value>
>> +# pv_support = 0|1 # Optionally specify whether a test supports the
>> +# # execution as a PV guest.
>> ##############################################################################
>>
>> [selftest-setup]
>> @@ -25,62 +27,80 @@ extra_params = -append 'test 123'
>>
>> [intercept]
>> file = intercept.elf
>> +pv_support = 1
>
> So, let's do this discussion once more:
> Why would we need a opt-in for something which works on all our current
> tests? I'd much rather have a opt-out or just a bail-out when running
> the test like I already implemented for the storage key related tests...
>
> I don't see any benefit for this right now other than forcing me to add
> another line to this file that was not needed before..
>
Exactly my thought. I would assume that most tests that properly test
for feature availability should just work?
--
Thanks,
David / dhildenb
