GNUBoot
/
coreboot-kgpe-d16
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

spi/tpm: Make sure AP properly syncs up with Cr50 

When Cr50 TPM is being reset, it continues replying to the SPI bus
requests, sends wrong register values in response to read requests.

This patch makes sure that the TPM driver does not proceed unless
proper value is read from the TPM device identification register.

If the read value is still wrong after 10 retries taken with 10 ms
intervals, the driver gives up and declares TPM broken/unavailable.

BRANCH=cr50
BUG=b:68012381
TEST=ran a script resetting the Fizz device as soon as the "index
     0x1007 return code 0" string shows up in the AP console output.
     The script keeps rebooting the Fizz indefinitely, before this
     script Fizz would fail to read TPM properly and fall into
     recovery after no more than four reboots.

Change-Id: I7e67ec62c2bf31077b9ae558e09214d07eccf96b
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: https://review.coreboot.org/22231
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This commit is contained in:
Vadim Bendebury 2017-10-25 15:45:00 -07:00
parent d2c636582d
commit 9a506d5c9a
1 changed files with 36 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
src/drivers/spi/tpm/tpm.c
View File

@ -280,7 +280,8 @@ static int tpm2_write_reg(unsigned reg_number, const void *buffer, size_t bytes)
* To read a register, start transaction, transfer data from the TPM, deassert
* CS when done.
*
* Returns one to indicate success, zero to indicate failure.
* Returns one to indicate success, zero to indicate failure. In case of
* failure zero out the user buffer.
*/
static int tpm2_read_reg(unsigned reg_number, void *buffer, size_t bytes)
{
@ -368,21 +369,52 @@ static int tpm2_claim_locality(void)
return 1;
}
/* Device/vendor ID values of the TPM devices this driver supports. */
static const uint32_t supported_did_vids[] = {
0x00281ae0 /* H1 based Cr50 security chip. */
};
int tpm2_init(struct spi_slave *spi_if)
{
uint32_t did_vid, status;
uint8_t cmd;
int retries;
struct tpm2_info *tpm_info = car_get_var_ptr(&g_tpm_info);
struct spi_slave *spi_slave = car_get_var_ptr(&g_spi_slave);
memcpy(spi_slave, spi_if, sizeof(*spi_if));
/*
* It is enough to check the first register read error status to bail
* out in case of malfunctioning TPM.
* 150 ms should be enough to synchronize with the TPM even under the
* worst nested reset request conditions. In vast majority of cases
* there would be no wait at all.
*/
if (!tpm2_read_reg(TPM_DID_VID_REG, &did_vid, sizeof(did_vid)))
printk(BIOS_INFO, "Probing TPM: ");
for (retries = 15; retries > 0; retries--) {
int i;
/* In case of falure to read div_vid is set to zero. */
tpm2_read_reg(TPM_DID_VID_REG, &did_vid, sizeof(did_vid));
for (i = 0; i < ARRAY_SIZE(supported_did_vids); i++)
if (did_vid == supported_did_vids[i])
break; /* Tpm is up and ready. */
if (i < ARRAY_SIZE(supported_did_vids))
break;
/* TPM might be resetting, let's retry in a bit. */
mdelay(10);
printk(BIOS_INFO, ".");
}
if (!retries) {
printk(BIOS_ERR, "\n%s: Failed to connect to the TPM\n",
__func__);
return -1;
}
printk(BIOS_INFO, " done!\n");
/* Claim locality 0. */
if (!tpm2_claim_locality())