On Tue, 15 Jan 2019 21:18:18 +0100
Laszlo Ersek <ler...@redhat.com> wrote:
> On 01/15/19 16:41, Igor Mammedov wrote:
> > For testcase to use UEFI firmware, one needs to provide and specify
> > firmware and varstore blobs names in test_data { uefi_fl1, uefi_fl2) }
> > fields respectively.
> >
> > Signed-off-by: Igor Mammedov <imamm...@redhat.com>
> > ---
> > tests/bios-tables-test.c | 36 ++++++++++++++++++++++++++----------
> > 1 file changed, 26 insertions(+), 10 deletions(-)
> >
> > diff --git a/tests/bios-tables-test.c b/tests/bios-tables-test.c
> > index 8887319..d290dd2 100644
> > --- a/tests/bios-tables-test.c
> > +++ b/tests/bios-tables-test.c
> > @@ -26,6 +26,8 @@
> > typedef struct {
> > const char *machine;
> > const char *variant;
> > + const char *uefi_fl1;
> > + const char *uefi_fl2;
> > uint64_t rsdp_addr;
> > uint8_t rsdp_table[36 /* ACPI 2.0+ RSDP size */];
> > GArray *tables;
> > @@ -519,21 +521,35 @@ static void test_smbios_structs(test_data *data)
> > static void test_acpi_one(const char *params, test_data *data)
> > {
> > char *args;
> > -
> > - /* Disable kernel irqchip to be able to override apic irq0. */
> > - args = g_strdup_printf("-machine %s,accel=%s,kernel-irqchip=off "
> > - "-net none -display none %s "
> > - "-drive id=hd0,if=none,file=%s,format=raw "
> > - "-device ide-hd,drive=hd0 ",
> > - data->machine, "kvm:tcg",
> > - params ? params : "", disk);
> > + bool use_uefi = data->uefi_fl1 && data->uefi_fl2;
> > +
> > + if (use_uefi) {
> > + args = g_strdup_printf("-machine %s,accel=%s -nodefaults
> > -nographic "
> > + "-drive if=pflash,format=raw,file=%s/%s,readonly "
> > + "-drive if=pflash,format=raw,file=%s/%s,snapshot=on %s",
>
> Today I Learned: about "snapshot=on". Thanks :) The command line looks good.
This way one doesn't have to make images in C (which is messy), keep track of
temporary images and clean up (which isn't reliable when test crashes).
Makefile magic can to all of that and in a much cleaner way.
> > + data->machine, "kvm:tcg", data_dir, data->uefi_fl1, data_dir,
>
> You could open-code "kvm:tcg" in the format string at once (unless you
> turn that into a parameter in a later patch in the series). But, I see
> the pre-patch code passes "kvm:tcg" as an argument too.
That was my line of reasoning as well, so I've kept it for consistence
with original code. If someone would insist I can throw in a separate
patch to open-code it before this one.
> > + data->uefi_fl2, params ? params : "");
> > +
> > + } else {
> > + /* Disable kernel irqchip to be able to override apic irq0. */
> > + args = g_strdup_printf("-machine %s,accel=%s,kernel-irqchip=off "
> > + "-net none -display none %s "
> > + "-drive id=hd0,if=none,file=%s,format=raw "
> > + "-device ide-hd,drive=hd0 ",
> > + data->machine, "kvm:tcg", params ? params : "", disk);
> > + }
> >
> > data->qts = qtest_init(args);
> >
> > - boot_sector_test(data->qts);
> > + if (use_uefi) {
> > + data->rsdp_addr = uefi_find_rsdp_addr(data->qts,
> > + 0x40000000ULL, 128ULL * 1024 * 1024);
>
> I think open-coding the DRAM size is valid; after all, it depends on the
> QEMU command line, and you control the QEMU command line above. However,
> do we really want to open-code the DRAM base here? That's
> board-specific. Should we pass that too through the "test_data" structure?
I've missed this one. In addition to your suggestion, RAM size is also
a good candidate for "test_data" structure.
>
> > + } else {
> > + boot_sector_test(data->qts);
> > + test_acpi_rsdp_address(data);
> > + }
> >
> > data->tables = g_array_new(false, true, sizeof(AcpiSdtTable));
> > - test_acpi_rsdp_address(data);
> > test_acpi_rsdp_table(data);
> > test_acpi_rxsdt_table(data);
> > test_acpi_fadt_table(data);
> >
>
> Thanks,
> Laszlo
