[U-Boot] [PATCH v2 11/19] tpm: add TPM2_Clear command support
Simon Glass
sjg at chromium.org
Thu May 3 19:01:52 UTC 2018
Hi Miquel,
On 27 April 2018 at 07:39, Miquel Raynal <miquel.raynal at bootlin.com> wrote:
> Hi Simon,
>
> On Thu, 26 Apr 2018 08:40:27 -0600, Simon Glass <sjg at chromium.org>
> wrote:
>
>> Hi Miquel,
>>
>> On 24 April 2018 at 07:17, Miquel Raynal <miquel.raynal at bootlin.com> wrote:
>> > Hi Simon,
>> >
>> > On Fri, 30 Mar 2018 06:42:32 +0800, Simon Glass <sjg at chromium.org>
>> > wrote:
>> >
>> >> Hi Miquel,
>> >>
>> >> On 29 March 2018 at 15:43, Miquel Raynal <miquel.raynal at bootlin.com> wrote:
>> >> > Add support for the TPM2_Clear command.
>> >> >
>> >> > Change the command file and the help accordingly.
>> >> >
>> >> > Signed-off-by: Miquel Raynal <miquel.raynal at bootlin.com>
>> >> > ---
>> >> > cmd/tpm.c | 29 ++++++++++++++++++++++++++---
>> >> > cmd/tpm_test.c | 6 +++---
>> >> > include/tpm.h | 21 +++++++++++++++++----
>> >> > lib/tpm.c | 42 ++++++++++++++++++++++++++++++++++++++----
>> >> > 4 files changed, 84 insertions(+), 14 deletions(-)
>> >> >
>> >> > diff --git a/cmd/tpm.c b/cmd/tpm.c
>> >> > index fc9ef9d4a3..32921e1a70 100644
>> >> > --- a/cmd/tpm.c
>> >> > +++ b/cmd/tpm.c
>> >> > @@ -454,6 +454,29 @@ static int do_tpm_init(cmd_tbl_t *cmdtp, int flag,
>> >> > return report_return_code(tpm_init());
>> >> > }
>> >> >
>> >> > +static int do_tpm_force_clear(cmd_tbl_t *cmdtp, int flag,
>> >> > + int argc, char * const argv[])
>> >> > +{
>> >> > + u32 handle = 0;
>> >> > + const char *pw = (argc < 3) ? NULL : argv[2];
>> >> > + const ssize_t pw_sz = pw ? strlen(pw) : 0;
>> >> > +
>> >> > + if (argc < 2 || argc > 3)
>> >> > + return CMD_RET_USAGE;
>> >> > +
>> >> > + if (pw_sz > TPM2_DIGEST_LENGTH)
>> >> > + return -EINVAL;
>> >> > +
>> >> > + if (!strcasecmp("TPM2_RH_LOCKOUT", argv[1]))
>> >> > + handle = TPM2_RH_LOCKOUT;
>> >> > + else if (!strcasecmp("TPM2_RH_PLATFORM", argv[1]))
>> >> > + handle = TPM2_RH_PLATFORM;
>> >> > + else
>> >> > + return CMD_RET_USAGE;
>> >> > +
>> >> > + return report_return_code(tpm_force_clear(handle, pw, pw_sz));
>> >> > +}
>> >> > +
>> >> > #define TPM_COMMAND_NO_ARG(cmd) \
>> >> > static int do_##cmd(cmd_tbl_t *cmdtp, int flag, \
>> >> > int argc, char * const argv[]) \
>> >> > @@ -465,7 +488,6 @@ static int do_##cmd(cmd_tbl_t *cmdtp, int flag, \
>> >> >
>> >> > TPM_COMMAND_NO_ARG(tpm_self_test_full)
>> >> > TPM_COMMAND_NO_ARG(tpm_continue_self_test)
>> >> > -TPM_COMMAND_NO_ARG(tpm_force_clear)
>> >> > TPM_COMMAND_NO_ARG(tpm_physical_enable)
>> >> > TPM_COMMAND_NO_ARG(tpm_physical_disable)
>> >> >
>> >> > @@ -951,8 +973,9 @@ U_BOOT_CMD(tpm, CONFIG_SYS_MAXARGS, 1, do_tpm,
>> >> > " physical_set_deactivated 0|1\n"
>> >> > " - Set deactivated flag.\n"
>> >> > "Admin Ownership Commands:\n"
>> >> > -" force_clear\n"
>> >> > -" - Issue TPM_ForceClear command.\n"
>> >> > +" force_clear [<type>]\n"
>> >> > +" - Issue TPM_[Force]Clear command, with <type> one of (TPMv2 only):\n"
>> >> > +" * TPM2_RH_LOCKOUT, TPM2_RH_PLATFORM.\n"
>> >> > " tsc_physical_presence flags\n"
>> >> > " - Set TPM device's Physical Presence flags to <flags>.\n"
>> >> > "The Capability Commands:\n"
>> >> > diff --git a/cmd/tpm_test.c b/cmd/tpm_test.c
>> >> > index 37ad2ff33d..da40dbc423 100644
>> >> > --- a/cmd/tpm_test.c
>> >> > +++ b/cmd/tpm_test.c
>> >> > @@ -176,7 +176,7 @@ static int test_fast_enable(void)
>> >> > TPM_CHECK(tpm_get_flags(&disable, &deactivated, NULL));
>> >> > printf("\tdisable is %d, deactivated is %d\n", disable, deactivated);
>> >> > for (i = 0; i < 2; i++) {
>> >> > - TPM_CHECK(tpm_force_clear());
>> >> > + TPM_CHECK(tpm_force_clear(0, NULL, 0));
>> >> > TPM_CHECK(tpm_get_flags(&disable, &deactivated, NULL));
>> >> > printf("\tdisable is %d, deactivated is %d\n", disable,
>> >> > deactivated);
>> >> > @@ -458,7 +458,7 @@ static int test_write_limit(void)
>> >> > TPM_CHECK(TlclStartupIfNeeded());
>> >> > TPM_CHECK(tpm_self_test_full());
>> >> > TPM_CHECK(tpm_tsc_physical_presence(PRESENCE));
>> >> > - TPM_CHECK(tpm_force_clear());
>> >> > + TPM_CHECK(tpm_force_clear(0, NULL, 0));
>> >> > TPM_CHECK(tpm_physical_enable());
>> >> > TPM_CHECK(tpm_physical_set_deactivated(0));
>> >> >
>> >> > @@ -477,7 +477,7 @@ static int test_write_limit(void)
>> >> > }
>> >> >
>> >> > /* Reset write count */
>> >> > - TPM_CHECK(tpm_force_clear());
>> >> > + TPM_CHECK(tpm_force_clear(0, NULL, 0));
>> >> > TPM_CHECK(tpm_physical_enable());
>> >> > TPM_CHECK(tpm_physical_set_deactivated(0));
>> >> >
>> >> > diff --git a/include/tpm.h b/include/tpm.h
>> >> > index 38d7cb899d..2f17166662 100644
>> >> > --- a/include/tpm.h
>> >> > +++ b/include/tpm.h
>> >> > @@ -43,13 +43,23 @@ enum tpm_startup_type {
>> >> > };
>> >> >
>> >> > enum tpm2_startup_types {
>> >> > - TPM2_SU_CLEAR = 0x0000,
>> >> > - TPM2_SU_STATE = 0x0001,
>> >> > + TPM2_SU_CLEAR = 0x0000,
>> >> > + TPM2_SU_STATE = 0x0001,
>> >> > +};
>> >> > +
>> >> > +enum tpm2_handles {
>> >>
>> >> Please add comment to enum
>> >
>> > Fine, I will document all of them. Just for you to know, these are
>> > values extracted from the (publicly available) specification, I did
>> > not change any of them.
>> >
>> >>
>> >> > + TPM2_RH_OWNER = 0x40000001,
>> >> > + TPM2_RS_PW = 0x40000009,
>> >> > + TPM2_RH_LOCKOUT = 0x4000000A,
>> >> > + TPM2_RH_ENDORSEMENT = 0x4000000B,
>> >> > + TPM2_RH_PLATFORM = 0x4000000C,
>> >> > };
>> >> >
>> >> > enum tpm2_command_codes {
>> >> > TPM2_CC_STARTUP = 0x0144,
>> >> > TPM2_CC_SELF_TEST = 0x0143,
>> >> > + TPM2_CC_CLEAR = 0x0126,
>> >> > + TPM2_CC_CLEARCONTROL = 0x0127,
>> >> > TPM2_CC_GET_CAPABILITY = 0x017A,
>> >> > TPM2_CC_PCR_READ = 0x017E,
>> >> > TPM2_CC_PCR_EXTEND = 0x0182,
>> >> > @@ -567,11 +577,14 @@ uint32_t tpm_tsc_physical_presence(uint16_t presence);
>> >> > uint32_t tpm_read_pubek(void *data, size_t count);
>> >> >
>> >> > /**
>> >> > - * Issue a TPM_ForceClear command.
>> >> > + * Issue a TPM_ForceClear or a TPM2_Clear command.
>> >> > *
>> >> > + * @param handle Handle
>> >> > + * @param pw Password
>> >> > + * @param pw_sz Length of the password
>> >> > * @return return code of the operation
>> >> > */
>> >> > -uint32_t tpm_force_clear(void);
>> >> > +int tpm_force_clear(u32 handle, const char *pw, const ssize_t pw_sz);
>> >> >
>> >> > /**
>> >> > * Issue a TPM_PhysicalEnable command.
>> >> > diff --git a/lib/tpm.c b/lib/tpm.c
>> >> > index 3cc2888267..9a46ac09e6 100644
>> >> > --- a/lib/tpm.c
>> >> > +++ b/lib/tpm.c
>> >> > @@ -608,13 +608,47 @@ uint32_t tpm_read_pubek(void *data, size_t count)
>> >> > return 0;
>> >> > }
>> >> >
>> >> > -uint32_t tpm_force_clear(void)
>> >> > +int tpm_force_clear(u32 handle, const char *pw, const ssize_t pw_sz)
>> >> > {
>> >> > - const uint8_t command[10] = {
>> >> > - 0x0, 0xc1, 0x0, 0x0, 0x0, 0xa, 0x0, 0x0, 0x0, 0x5d,
>> >> > + const u8 command_v1[10] = {
>> >> > + U16_TO_ARRAY(0xc1),
>> >> > + U32_TO_ARRAY(10),
>> >> > + U32_TO_ARRAY(0x5d),
>> >> > };
>> >> > + u8 command_v2[COMMAND_BUFFER_SIZE] = {
>> >> > + U16_TO_ARRAY(TPM2_ST_SESSIONS), /* TAG */
>> >> > + U32_TO_ARRAY(27 + pw_sz), /* Length */
>> >> > + U32_TO_ARRAY(TPM2_CC_CLEAR), /* Command code */
>> >>
>> >> Here we have both v1 and v2 commands. Perhaps we should have a Kconfig
>> >> option to enable v2 support? Or do you think it is not a big addition
>> >> in terms of code side?
>> >
>> > It is a big addition in terms of code side.
>> >
>> >>
>> >> One way to do this would be to have an inline function which can tell
>> >> if you are using v2:
>> >>
>> >> static inline bool tpm_v2(void) {
>> >> return IS_ENABLED(CONFIG_TPM_V2) && further things...
>> >> }
>> >>
>> >> static inline bool tpm_v1(void) {
>> >> return IS_ENABLED(CONFIG_TPM_V1) && further things...
>> >> }
>> >
>> > I like this way of choosing one specification or the other, I could
>> > make them mutually exclusive. It would prevent the need for a global
>> > variable (or an additional field in uclass).
>> >
>> > Would it be fine to actually rename the file (with a 'tpm1' suffix) and
>> > create a new one specific for TPMv2 commands ? Only one file would be
>> > compiled/linked depending on the configuration, avoiding the possibility
>> > to call a v1 command from a v2 chip and vice versa.
>> >
>> > At first I thought a lot of code would be shared but I don't think we
>> > would add so much by having one function per revision now.
>>
>> Well you could have two separate files, if there really is no sharing
>> of commands. But if there are common commands, you should have a
>> 'common' file to implement them, rather than duplicating code.
>>
>> With the above static inline functions we can support:
>>
>> - v1 only (no code-size increment)
>> - v2 only (minimal code size for what will be a common case)
>> - v1 + v2 (e.g. for sandbox testing)
>
> I am changing the whole architecture as you suggested.
>
> Now v1 and v2 are chosen with a Kconfig menu, common code is in a
> tpm-common.c file, and commands/library functions for each
> specification is in tpm-v1.c and tpm-v2.c (with all the needed headers.
> This way TPMv1 code is absolutely untouched while it allows TPMv2
> support to be introduced independently.
>
> You'll tell me how you find this alternative, I will soon send a v3.
>
> Otherwise, as suggested in a first answer, I changed U32_TO_array
I think it works OK as you have it now, with the rename.
Regards,
Simon
More information about the U-Boot
mailing list