[PATCH 2/2 v4] efi: ESRT creation tests
AKASHI Takahiro
takahiro.akashi at linaro.org
Mon Mar 1 02:18:25 CET 2021
Heinrich,
On Fri, Feb 26, 2021 at 03:17:42PM +0100, Heinrich Schuchardt wrote:
> On 26.02.21 14:55, Heinrich Schuchardt wrote:
> > On 23.02.21 17:50, Jose Marinho wrote:
> >> This commmit exercises the ESRT creation -- introduced in the previous
> >> commit -- in two tests.
> >>
> >> test 1:
> >> A fake FMP, with TEST_ESRT_NUM_ENTRIES FW images, is installed in the
> >> system leading to the corresponding ESRT entries being populated.
> >> The ESRT entries are checked against the datastructure used to
> >> initialize the FMP.
> >>
> >> test 1 invocation:
> >> make sandbox_capsule_defconfig all
> >> ./u-boot -d arch/sandbox/dts/test.dtb
> >> ut lib
> >>
> >> test 2:
> >> The test is part of test_efi_capsule_fw3.
> >>
> >> In order to run the test the following must be added to
> >> sandbox_defconfig:
> >> +CONFIG_CMD_SF=y
> >> +CONFIG_CMD_MEMORY=y
> >> +CONFIG_CMD_FAT=y
> >> +CONFIG_DFU=y
> >>
> >> The ESRT is printed in the u-boot shell by calling efidebug esrt.
> >> The test ensures that, after the capsule is installed, the ESRT
> >> contains entries with the GUIDs:
> >> - EFI_FIRMWARE_IMAGE_TYPE_UBOOT_FIT_GUID;
> >> - EFI_FIRMWARE_IMAGE_TYPE_UBOOT_RAW_GUID;
> >>
> >> test 2 invocation:
> >> sudo ./test/py/test.py --bd sandbox -k capsule_fw3 -l --build
> >>
> >> Signed-off-by: Jose Marinho <jose.marinho at arm.com>
> >>
> >> CC: Heinrich Schuchardt <xypron.glpk at gmx.de>
> >> CC: Sughosh Ganu <sughosh.ganu at linaro.org>
> >> CC: AKASHI Takahiro <takahiro.akashi at linaro.org>
> >> CC: Ilias Apalodimas <ilias.apalodimas at linaro.org>
> >> CC: Andre Przywara <andre.przywara at arm.com>
> >> CC: Alexander Graf <agraf at csgraf.de>
> >> CC: nd at arm.com
> >>
> >> ---
> >> cmd/efidebug.c | 64 ++++++
> >> test/lib/Makefile | 1 +
> >> test/lib/efi_esrt.c | 191 ++++++++++++++++++
> >> .../test_efi_capsule/test_capsule_firmware.py | 4 +
> >> 4 files changed, 260 insertions(+)
> >> create mode 100644 test/lib/efi_esrt.c
> >>
> >> diff --git a/cmd/efidebug.c b/cmd/efidebug.c
> >> index a7dace2f80..5a9ff2bd9a 100644
> >> --- a/cmd/efidebug.c
> >> +++ b/cmd/efidebug.c
> >> @@ -129,6 +129,61 @@ static int do_efi_capsule_show(struct cmd_tbl *cmdtp, int flag,
> >> return CMD_RET_SUCCESS;
> >> }
> >>
> >> +#ifdef CONFIG_EFI_ESRT
> >> +/**
> >> + * do_efi_capsule_esrt() - manage UEFI capsules
> >> + *
> >> + * @cmdtp: Command table
> >> + * @flag: Command flag
> >> + * @argc: Number of arguments
> >> + * @argv: Argument array
> >> + * Return: CMD_RET_SUCCESS on success,
> >> + * CMD_RET_USAGE or CMD_RET_RET_FAILURE on failure
> >> + *
> >> + * Implement efidebug "capsule esrt" sub-command.
> >> + * The prints the current ESRT table.
> >> + *
> >> + * efidebug capsule esrt
> >> + */
> >> +static int do_efi_capsule_esrt(struct cmd_tbl *cmdtp, int flag,
> >> + int argc, char * const argv[])
> >> +{
> >> + struct efi_system_resource_table *esrt = NULL;
> >> +
> >> + if (argc != 1)
> >> + return CMD_RET_USAGE;
> >> +
> >> + for (int idx = 0; idx < systab.nr_tables; idx++)
> >> + if (!guidcmp(&efi_esrt_guid, &systab.tables[idx].guid))
> >> + esrt = (struct efi_system_resource_table *)systab.tables[idx].table;
> >> +
> >> + if (!esrt)
> >> + return CMD_RET_FAILURE;
> >> +
> >> + printf("========================================\n");
> >> + printf("ESRT: fw_resource_count=%d\n", esrt->fw_resource_count);
> >> + printf("ESRT: fw_resource_count_max=%d\n", esrt->fw_resource_count_max);
> >> + printf("ESRT: fw_resource_version=%lld\n", esrt->fw_resource_version);
> >> +
> >> + for (int idx = 0; idx < esrt->fw_resource_count; idx++) {
> >> + printf("[entry %d]==============================\n", idx);
> >> + printf("ESRT: fw_class=%pUL\n", &esrt->entries[idx].fw_class);
> >> + printf("ESRT: fw_type=%d\n", esrt->entries[idx].fw_type);
> >> + printf("ESRT: fw_version=%d\n", esrt->entries[idx].fw_version);
> >> + printf("ESRT: lowest_supported_fw_version=%d\n",
> >> + esrt->entries[idx].lowest_supported_fw_version);
> >> + printf("ESRT: capsule_flags=%d\n",
> >> + esrt->entries[idx].capsule_flags);
> >> + printf("ESRT: last_attempt_version=%d\n",
> >> + esrt->entries[idx].last_attempt_version);
> >> + printf("ESRT: last_attempt_status=%d\n",
> >> + esrt->entries[idx].last_attempt_status);
> >> + }
> >> + printf("========================================\n");
> >> +
> >> + return CMD_RET_SUCCESS;
> >
> > With which configuration do we see any entry in the list?
> >
> > I only got:
> >
> > => efidebug capsule esrt
> > ========================================
> > ESRT: fw_resource_count=0
> > ESRT: fw_resource_count_max=0
> > ESRT: fw_resource_version=1
> > ========================================
> > =>
> >
> > with
> >
> > CONFIG_EFI_HAVE_CAPSULE_SUPPORT=y
> > CONFIG_EFI_RUNTIME_UPDATE_CAPSULE=y
> > CONFIG_EFI_CAPSULE_ON_DISK=y
> > # CONFIG_EFI_CAPSULE_ON_DISK_EARLY is not set
> > CONFIG_EFI_CAPSULE_FIRMWARE=y
> > CONFIG_EFI_CAPSULE_FIRMWARE_MANAGEMENT=y
> > # CONFIG_EFI_CAPSULE_AUTHENTICATE is not set
> > CONFIG_EFI_CAPSULE_FIRMWARE_FIT=y
> > CONFIG_EFI_CAPSULE_FIRMWARE_RAW=y
> >
> > efidebug dh does not show any FMP protocol installed:
> >
> > => efidebug dh
> > Handle Protocols
> > ================ ====================
> > 000000007eef19c0 Device Path, Device Path To Text, Device Path
> > Utilities, Device-Tree Fixup, Unicode Collation 2, HII String, HII
> > Database, Random Number Generator
> > 000000007eef1bb0 Simple Text Output
> > 000000007eef1c20 Simple Text Input, Simple Text Input Ex
> > 000000007eef1f60 Device Path, Block IO
> > 000000007eef2130 Device Path, Block IO, System Partition, Simple File System
> > 000000007eef2f10 Driver Binding
> > 000000007eef4050 Simple Network, Device Path, PXE Base Code
>
> @Sughosh, Takahiro
>
> If EFI_OS_INDICATIONS_FILE_CAPSULE_DELIVERY_SUPPORTED in OsIndications
> is not set, the FMP protocols are not installed.
First of all, OsIndications is one of variables that should be set properly
by *OS user*, not by the firmware (U-Boot).
In addition, according to "8.5.4 Exchanging information between the OS
and Firmware,"
When submitting capsule via the Mass Storage Device method of Section 8.5.5,
the bit EFI_OS_INDICATIONS_FILE_CAPSULE_DELIVERY_SUPPORTED in OsIndications
variable must be set by submitter to trigger processing of submitted capsule
on next reboot. This bit will be cleared from OsIndications by system
firmware in all cases during processing following reboot.
> Shouldn't the FMP protocols always be installed to process a
> CapsuleUpdate() call? Otherwise efi_capsule_update_firmware() will fail
> with EFI_UNSUPPORTED.
Under the current implementation, capsule update will be expected
to be performed only through a capsule file, ie. UpdateCapsule() API
is *not* supported.
Therefore, FMP (FIT or raw) are not installed if not the case.
-Takahiro Akashi
> Best regards
>
> Heinrich
>
More information about the U-Boot
mailing list