On Wed, Sep 23, 2026 at 02:28:34PM +0000, Surendran Kanagaraj wrote:
> TPM commands need to complete within the command duration defined in
> the TPM2 spec or keep answering TPM2_RC_RETRY for at most
> TPM2_DURATION_LONG (2s).
>
> In rare scenarios NitroTPM can enter a mode where it either sends
> TPM2_RC_RETRY for more than 2s or delays command completion. When
> this happens during auth session start, the driver disables the chip
> and every subsequent request fails:
>
> tpm tpm0: in retry loop
> tpm tpm0: tpm2_load_context: failed with a TPM error 0x0922
>
> Add support for vendor quirks to the CRB driver. The table is keyed by
> the vendor ID read from the interface ID register. For NitroTPM, set
> busy_timeout_ms to 30s, which covers the longest unavailability time
> along with some headroom.
>
> Tested on an EC2 instance with NitroTPM by inducing multiple
> unavailability windows with no TPM errors, and in QEMU with swtpm by
> holding the TPM in TPM2_RC_RETRY and stalling command completion for
> longer than 2s with the quirk applied.
>
> Assisted-by: LLM
> Signed-off-by: Surendran Kanagaraj <[email protected]>
> ---
> drivers/char/tpm/tpm_crb.c | 54 ++++++++++++++++++++++++++++++++++++++
> 1 file changed, 54 insertions(+)
>
> diff --git a/drivers/char/tpm/tpm_crb.c b/drivers/char/tpm/tpm_crb.c
> index ceb4100ba400..4107203beec9 100644
> --- a/drivers/char/tpm/tpm_crb.c
> +++ b/drivers/char/tpm/tpm_crb.c
> @@ -13,6 +13,7 @@
>
> #include <linux/acpi.h>
> #include <linux/highmem.h>
> +#include <linux/io-64-nonatomic-lo-hi.h>
> #include <linux/rculist.h>
> #include <linux/module.h>
> #include <linux/platform_device.h>
> @@ -26,6 +27,15 @@
> #define ACPI_SIG_TPM2 "TPM2"
> #define TPM_CRB_MAX_RESOURCES 3
>
> +/* TPM_CRB_INTF_ID_x vendor ID field, bits 47:32 of the 64-bit register */
> +#define CRB_INTF_ID_VID(intf_id) (((intf_id) >> 32) & 0xffff)
> +
> +/* Amazon NitroTPM */
> +#define CRB_INTF_VID_AMZN 0x0ec2
> +
> +/* Longest unavailability seen from NitroTPM */
> +#define CRB_AMZN_BUSY_TIMEOUT_MS 30000
Use TPM_CRB prefix.
> +
> static const guid_t crb_acpi_start_guid =
> GUID_INIT(0x6BBF6CAB, 0x5463, 0x4714,
> 0xB7, 0xCD, 0xF0, 0x20, 0x3C, 0x03, 0x68, 0xD4);
> @@ -783,6 +793,48 @@ static int crb_map_pluton(struct device *dev, struct
> crb_priv *priv,
> return 0;
> }
>
> +/*
> + * Read the vendor ID from the interface ID register. Returns -ENODEV when
> + * the head registers are not mapped for this start method.
> + */
> +static int crb_vendor_id(struct crb_priv *priv)
> +{
> + u64 intf_id;
> +
> + if (!priv->regs_h)
> + return -ENODEV;
> +
> + intf_id = lo_hi_readq(&priv->regs_h->intf_id);
> +
> + return CRB_INTF_ID_VID(intf_id);
> +}
> +
> +static void crb_amzn_quirk(struct tpm_chip *chip)
> +{
> + /* NitroTPM requires larger timeouts */
> + chip->busy_timeout_ms = CRB_AMZN_BUSY_TIMEOUT_MS;
> +}
> +
> +static const struct crb_vendor_quirk {
> + u16 vendor_id;
> + void (*apply)(struct tpm_chip *chip);
> +} crb_vendor_quirks[] = {
> + { CRB_INTF_VID_AMZN, crb_amzn_quirk },
> +};
> +
> +static void crb_apply_vendor_quirks(struct crb_priv *priv, struct tpm_chip
> *chip)
> +{
> + int vendor_id = crb_vendor_id(priv);
> + unsigned int i;
> +
> + if (vendor_id < 0)
> + return;
> +
> + for (i = 0; i < ARRAY_SIZE(crb_vendor_quirks); i++)
> + if (crb_vendor_quirks[i].vendor_id == vendor_id)
> + crb_vendor_quirks[i].apply(chip);
> +}
> +
> static int crb_acpi_probe(struct platform_device *pdev)
> {
> struct device *dev = &pdev->dev;
> @@ -887,6 +939,8 @@ static int crb_acpi_probe(struct platform_device *pdev)
> chip->acpi_dev_handle = device->handle;
> chip->flags = TPM_CHIP_FLAG_TPM2;
>
> + crb_apply_vendor_quirks(priv, chip);
I.e., all this complexity equals to:
if (crb_vendor_id(priv) == CRB_INTF_VID_AMZN)
chip->busy_timeout_ms = CRB_AMZN_BUSY_TIMEOUT_MS;
?
> +
> rc = tpm_chip_bootstrap(chip);
> if (rc)
> goto out;
> --
> 2.47.3
>
Br, Jarkko
