[ot] hw/opentitan: ot_spi_device: Implement the TPM mode #249

engdoreis · 2025-10-22T10:35:57Z

ThIs PR does not cover all the TPM implementation, the HW registers will be covered by a follow up PR.

To test use the Opentitan PR lowRISC/opentitan#28542 and run:

bazel test --test_output=streamed --build_tests_only \
    $(bazel query //sw/device/tests:all | grep qemu | grep spi_device) \
    --test_tag_filters="qemu" --override_repository="+qemu+qemu_opentitan=$qemu_repo"

rivos-eblot · 2025-10-22T16:00:30Z

I'll review it soon; I've noticed there are a couple of type mismatches (e.g. bool x; ... x = 1;) and missing unsigned constant suffixes (u)

rivos-eblot · 2025-10-22T16:02:02Z

see also the PR title, as they now follow the

`component`: ...

formatting, i.e. ot_spi_device: Implement the TPM mode

AlexJones0

I got part way through reviewing but then @engdoreis told me this is still being actively worked on as a draft. I'll leave the comments I had for now though to aid development, but I haven't looked at any of the TPM logic itself yet.

hw/opentitan/ot_spi_device.c

Based on the table in the spi_device/doc/programmers_guide Signed-off-by: Douglas Reis <[email protected]>

Signed-off-by: Douglas Reis <[email protected]>

rivos-eblot

Note: I would recommend to add new trace_ot_spi_device_...
Traces are very useful for debugging the emulation itself, but they are also very useful when developping guest SW (RISC-V code) and remote host peer SW. Thanks.

rivos-eblot · 2025-10-23T15:38:50Z

hw/opentitan/ot_spi_device.c

-#define SPI_SRAM_MBX_SIZE     0x400u
-#define SPI_SRAM_SFDP_OFFSET  (SPI_SRAM_MBX_OFFSET + SPI_SRAM_MBX_SIZE)
-#define SPI_SRAM_SFDP_SIZE    0x100u
+


try to remove extra empty lines.

rivos-eblot · 2025-10-23T15:39:25Z

hw/opentitan/ot_spi_device.c

-#define SPI_SRAM_PAYLOAD_OFFSET (SPI_SRAM_SFDP_OFFSET + SPI_SRAM_SFDP_SIZE)
+#define SPI_SRAM_INGRESS_OFFSET 0xE00u
+
+static_assert(SPI_SRAM_INGRESS_OFFSET >=


try to move the assertions after the constant definitions

rivos-eblot · 2025-10-23T15:40:14Z

hw/opentitan/ot_spi_device.c

 } OtSpiFlashState;

+typedef enum {
+    SPI_TPM_IDLE, // Wait for the fist byte of the header.


QEMU comments only use C style (/* */)

rivos-eblot · 2025-10-23T15:40:51Z

hw/opentitan/ot_spi_device.c

+    OtSpiTpmState state;
+    uint32_t can_receive;
+    uint32_t transfer_size;
+    bool read;


try to move it near the other bool

rivos-eblot · 2025-10-23T15:43:18Z

hw/opentitan/ot_spi_device.c

+ */
+static bool ot_spi_device_is_tpm_mode_crb(const OtSPIDeviceState *s)
+{
+    return (


very weird formatting :-)
there is an explci cast to an enumerated balue but you're returning a bool?
Looks like the cast is copy/pasted.

rivos-eblot · 2025-10-23T15:57:43Z

hw/opentitan/ot_spi_device.c

+    SpiDeviceTpm *tpm = &s->tpm;
+
+    if (size != tpm->can_receive) {
+        error_setg(&error_fatal, "TPM expected %u bytes, got %u\n",


Suspicious fatal error (just checking): Is this an internal error or something that can be triggered from the peer?

rivos-eblot · 2025-10-23T15:58:15Z

hw/opentitan/ot_spi_device.c

+    if (size != tpm->can_receive) {
+        error_setg(&error_fatal, "TPM expected %u bytes, got %u\n",
+                   tpm->can_receive, size);
+        tpm->state = SPI_TPM_END;


won't be reached :-)

rivos-eblot · 2025-10-23T15:58:34Z

hw/opentitan/ot_spi_device.c

+        return;
+    }
+
+    tpm->reg = buf[1] << 8 | buf[2];


missing us

rivos-eblot · 2025-10-23T15:58:52Z

hw/opentitan/ot_spi_device.c

+    tpm->locality = tpm->reg >> 12u;
+    s->tpm_regs[R_TPM_CMD_ADDR] =
+        (tpm->opcode << R_TPM_CMD_ADDR_CMD_SHIFT) | tpm->reg;
+    if (tpm->read) { // When read, we immidiatelly release the software to fill


C++ style comment + typo

rivos-eblot · 2025-10-23T16:00:19Z

hw/opentitan/ot_spi_device.c

+    if (tpm->read) {
+        tpm->state = tpm->should_sw_handle ? SPI_TPM_READ : SPI_TPM_READ_HW_REG;
+    }
+    tx_buf[size - 1] = 0x01u;


engdoreis force-pushed the ot_spi_tpm branch 2 times, most recently from 2ed5386 to aa2670c Compare October 22, 2025 14:01

engdoreis mentioned this pull request Oct 22, 2025

[qemu] Earlgrey 1.0.0 spi_device TPM support lowRISC/opentitan#27930

Open

engdoreis force-pushed the ot_spi_tpm branch 4 times, most recently from 8ba3a5c to 8cc7204 Compare October 22, 2025 15:26

AlexJones0 reviewed Oct 22, 2025

View reviewed changes

engdoreis force-pushed the ot_spi_tpm branch from 8cc7204 to 352f19d Compare October 22, 2025 16:11

engdoreis changed the title ~~[DRAFT][OT spi_device] Implement the TPM mode~~ [ot] hw/opentitan: ot_spi_device: Implement the TPM mode Oct 22, 2025

engdoreis marked this pull request as ready for review October 22, 2025 16:28

engdoreis force-pushed the ot_spi_tpm branch 3 times, most recently from c004a06 to 8675744 Compare October 23, 2025 11:05

engdoreis added 2 commits October 23, 2025 16:02

[ot] hw/opentitan: ot_spi_device: Update the SRAM definition

9ef37a1

Based on the table in the spi_device/doc/programmers_guide Signed-off-by: Douglas Reis <[email protected]>

[ot] hw/opentitan: ot_spi_device: Implement the tpm mode

aa54714

Signed-off-by: Douglas Reis <[email protected]>

engdoreis force-pushed the ot_spi_tpm branch from 8675744 to aa54714 Compare October 23, 2025 15:03

rivos-eblot requested changes Oct 23, 2025

View reviewed changes

[ot] hw/opentitan: ot_spi_device: Implement the TPM mode #249

Are you sure you want to change the base?

[ot] hw/opentitan: ot_spi_device: Implement the TPM mode #249

Conversation

engdoreis commented Oct 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rivos-eblot commented Oct 22, 2025

Uh oh!

rivos-eblot commented Oct 22, 2025

Uh oh!

AlexJones0 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rivos-eblot left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

engdoreis commented Oct 22, 2025 •

edited

Loading