CAPABILITIES: Add SupportedAlgorithms #2968

ShitalJumbad · 2025-01-25T02:00:17Z

fix #2279

steven-bellock

Still need to figure out how this information is conveyed to the Integrator.

include/industry_standard/spdm.h

include/internal/libspdm_common_lib.h

library/spdm_requester_lib/libspdm_req_get_capabilities.c

library/spdm_responder_lib/libspdm_rsp_capabilities.c

include/internal/libspdm_common_lib.h

library/spdm_responder_lib/libspdm_rsp_capabilities.c

library/spdm_requester_lib/libspdm_req_get_capabilities.c

include/library/spdm_requester_lib.h

library/spdm_responder_lib/libspdm_rsp_capabilities.c

doc/api/requester_api.md

include/library/spdm_requester_lib.h

library/spdm_responder_lib/libspdm_rsp_capabilities.c

library/spdm_requester_lib/libspdm_req_get_capabilities.c

doc/api/requester_api.md

steven-bellock · 2025-03-24T15:22:04Z

include/industry_standard/spdm.h

+} spdm_supported_algorithms_block_t;
+
+/* Specification states that total Extended algorithms count is less than or equal to 20*/
+#define SPDM_ALGORITHMS_MAX_NUM_EXT_ASYM_COUNT 20


Actually can just use SPDM_NEGOTIATE_ALGORITHMS_MAX_NUM_STRUCT_TABLE_ALG instead of SPDM_ALGORITHMS_MAX_NUM_STRUCT_TABLE_ALG.

I am not sure if we need to add SPDM version number here.

The MAX counter value maybe changed in different version.

It is in the older specifications but it is an equation.

ExtAsymCount + ExtHashCount + ExtAlgCount2 + ExtAlgCount3 + ExtAlgCount4 + ExtAlgCount5 <= 20

So for spdm.h there can be

/* Maximum value for the sum of ExtAsymCount + ExtHashCount + ExtAlgCount2 + ExtAlgCount3 + ExtAlgCount4 + ExtAlgCount5 */ #define SPDM_ALGORITHMS_EXT_ALG_MAX_SUM 20

but that can be a separate issue. libspdm currently does not check that, but that does not need to be in this pull request. @ShitalJumbad you can remove SPDM_ALGORITHMS_MAX_NUM_EXT_ASYM_COUNT and SPDM_ALGORITHMS_MAX_NUM_EXT_HASH_COUNT from this pull request.

Filed #3019.

library/spdm_requester_lib/libspdm_req_get_capabilities.c

include/library/spdm_requester_lib.h

include/internal/libspdm_common_lib.h

include/internal/libspdm_requester_lib.h

include/library/spdm_requester_lib.h

include/internal/libspdm_common_lib.h

include/library/spdm_requester_lib.h

jyao1 · 2025-06-17T06:48:47Z

include/industry_standard/spdm.h

@@ -177,6 +210,8 @@ typedef struct {
    /* Below field is added in 1.2.*/
    uint32_t data_transfer_size;
    uint32_t max_spdm_msg_size;
+    /* Below field is added in 1.3.*/
+    spdm_supported_algorithms_block_t supported_algorithms;


I gave feedback before to remove spdm_supported_algorithms_block_t from spdm_capabilities_response_t.

You can comment this line here.

Thank you for your feedback and for referencing the earlier discussion.t - cbda82c#r2151455599

To summarize, I understand the concern about including spdm_supported_algorithms_block_t in spdm_capabilities_response_t, especially since it’s a variable-length structure and may be considered a special case.

However, I added it here because, based on my reading of the SPDM 1.3 specification, it seems that spdm_supported_algorithms_block_t was introduced specifically for version 1.3, and including it in the structure aligns with the spec for that version. That said, I may be wrong, so I’m just trying to understand how we make this decision.

I’m also looping in @steven-bellock to get his perspective.

No, I don't there is any requirement to add anything defined in the spec.
If a structure is variable size, then it is OK to comment it here.

I prefer to not include it for convenience.

library/spdm_requester_lib/libspdm_req_get_capabilities.c

include/internal/libspdm_requester_lib.h

jyao1 · 2025-06-17T06:55:47Z

include/internal/libspdm_requester_lib.h

@@ -129,10 +129,37 @@ libspdm_return_t libspdm_get_version(libspdm_context_t *spdm_context,
                                     uint8_t *version_number_entry_count,
                                     spdm_version_number_t *version_number_entry);

+#define LIBSPDM_MAX_EXT_ALG_COUNT 5


libspdm does not support ext_alg. I think we can ignore them totally.

See #2968 (comment).

No, I don't suggest to keep the complexity.

As we already said, libspdm does not support ext_alg.
What is the value to return ext_alg to the integrator?
The integrator cannot use ext_algo with libspdm anyway.

include/internal/libspdm_requester_lib.h

include/internal/libspdm_common_lib.h

jyao1 · 2025-06-17T07:01:28Z

include/library/spdm_requester_lib.h

+libspdm_return_t libspdm_get_supported_algorithms(void *spdm_context,
+                                                  size_t *responder_supported_algorithms_length,
+                                                  void *responder_supported_algorithms_buffer,
+                                                  uint8_t *spdm_version);


If you include spdm_version to libspdm_responder_supported_algorithms_13_t, then you can remove spdm_version parameter here.

then you can remove spdm_version parameter here.

The Integrator still needs to know what libspdm_responder_supported_algorithms_*_t to apply to responder_supported_algorithms_buffer. Without this parameter they would need to libspdm_get_data LIBSPDM_DATA_SPDM_VERSION, which is not the end of the world, but this makes it more convenient.

Thanks for the suggestion! I think having the spdm_version parameter improves the convenience and usability of the API. Let me know if you have a strong preference here.

If we can guarantee that spdm_version is already the first byte of responder_supported_algorithms_buffer, then the integrator can just check the first byte.

Or we can define following libspdm_responder_supported_algorithms_common_header structure to ensure that every libspdm_responder_supported_algorithms_13_t or libspdm_responder_supported_algorithms_14_t will use the same common header.

typedef struct { uint8_t spdm_version; /* the rest of the field is spdm_version specific */ } libspdm_responder_supported_algorithms_common_header;

I really cannot see the need to have uint8_t *spdm_version

fix DMTF#2279 Signed-off-by: Shital Jumbad <sjumbad@nvidia.com>

ShitalJumbad · 2025-07-03T22:52:31Z

library/spdm_requester_lib/libspdm_req_get_capabilities.c

@@ -382,6 +423,61 @@ static libspdm_return_t libspdm_try_get_capabilities(libspdm_context_t *spdm_con
        spdm_context->connection_info.capability.max_spdm_msg_size = 0;
    }

+    /* Copy algorithms if requested and received */
+    if (get_supported_algorithms && supported_algs != NULL &&
+        spdm_response->header.spdm_version >= SPDM_MESSAGE_VERSION_13 &&


Instead of using >= SPDM_MESSAGE_VERSION_13 here, should I use == SPDM_MESSAGE_VERSION_13 since I am typecasting to libspdm_responder_supported_algorithms_13_t on line 436?

good question!

I think it is OK to use >= now.
If change is needed for new version, we need to update the code anyway.

jyao1 · 2025-07-07T06:24:16Z

library/spdm_requester_lib/libspdm_req_get_capabilities.c

+
+        *supported_algs_length = sizeof(libspdm_responder_supported_algorithms_13_t);
+        dst->spdm_version = spdm_response->header.spdm_version;
+        dst->struct_table_count = src->param1;


I think we need to check if this param1 exceeds the allowed struct_table_count.
Otherwise, it may cause buffer overflow.

ShitalJumbad requested review from jyao1 and steven-bellock as code owners January 25, 2025 02:00

ShitalJumbad force-pushed the fix-2279 branch 3 times, most recently from a061ee9 to c0d1cd5 Compare January 25, 2025 03:08

steven-bellock requested changes Jan 28, 2025

View reviewed changes

ShitalJumbad force-pushed the fix-2279 branch 11 times, most recently from d8be4bf to edf4eaa Compare February 13, 2025 01:06

ShitalJumbad requested a review from steven-bellock February 13, 2025 01:09

steven-bellock requested changes Feb 13, 2025

View reviewed changes

hiption reviewed Feb 17, 2025

View reviewed changes

doc/api/requester_api.md Outdated Show resolved Hide resolved

This was referenced Feb 17, 2025

Update test case 2 (Capabilities) to support SPDM 1.3 DMTF/SPDM-Responder-Validator#146

Merged

Add 1.3 test case for CAPABILITIES SupportedAlgorithms DMTF/SPDM-Responder-Validator#149

Open

ShitalJumbad force-pushed the fix-2279 branch 5 times, most recently from b1dc2d7 to da7747c Compare March 20, 2025 01:16

ShitalJumbad requested a review from steven-bellock March 20, 2025 20:56

steven-bellock requested changes Mar 24, 2025

View reviewed changes

jyao1 reviewed Mar 25, 2025

View reviewed changes

include/library/spdm_requester_lib.h Outdated Show resolved Hide resolved

jyao1 reviewed Apr 9, 2025

View reviewed changes

include/internal/libspdm_common_lib.h Outdated Show resolved Hide resolved

jyao1 reviewed Apr 9, 2025

View reviewed changes

include/internal/libspdm_requester_lib.h Outdated Show resolved Hide resolved

jyao1 reviewed Apr 9, 2025

View reviewed changes

include/library/spdm_requester_lib.h Outdated Show resolved Hide resolved

steven-bellock reviewed Apr 22, 2025

View reviewed changes

include/internal/libspdm_common_lib.h Outdated Show resolved Hide resolved

include/library/spdm_requester_lib.h Outdated Show resolved Hide resolved

ShitalJumbad force-pushed the fix-2279 branch 2 times, most recently from dc9f8ed to 1a52978 Compare June 13, 2025 22:03

jyao1 reviewed Jun 17, 2025

View reviewed changes

library/spdm_requester_lib/libspdm_req_get_capabilities.c Outdated Show resolved Hide resolved

jyao1 reviewed Jun 17, 2025

View reviewed changes

library/spdm_requester_lib/libspdm_req_get_capabilities.c Outdated Show resolved Hide resolved

jyao1 reviewed Jun 17, 2025

View reviewed changes

include/internal/libspdm_requester_lib.h Outdated Show resolved Hide resolved

jyao1 reviewed Jun 17, 2025

View reviewed changes

include/internal/libspdm_requester_lib.h Show resolved Hide resolved

jyao1 reviewed Jun 17, 2025

View reviewed changes

include/internal/libspdm_common_lib.h Outdated Show resolved Hide resolved

jyao1 reviewed Jun 17, 2025

View reviewed changes

ShitalJumbad force-pushed the fix-2279 branch 4 times, most recently from b2ceb49 to 16488b0 Compare July 2, 2025 01:25

ShitalJumbad marked this pull request as draft July 2, 2025 01:29

ShitalJumbad force-pushed the fix-2279 branch 4 times, most recently from d5d8e89 to 9e0ae18 Compare July 3, 2025 20:37

CAPABILITIES: Add SupportedAlgorithms

cbda82c

fix DMTF#2279 Signed-off-by: Shital Jumbad <sjumbad@nvidia.com>

ShitalJumbad force-pushed the fix-2279 branch from 9e0ae18 to cbda82c Compare July 3, 2025 22:02

ShitalJumbad commented Jul 3, 2025

View reviewed changes

ShitalJumbad marked this pull request as ready for review July 3, 2025 23:10

ShitalJumbad requested review from steven-bellock and jyao1 July 3, 2025 23:10

jyao1 reviewed Jul 7, 2025

View reviewed changes

CAPABILITIES: Add SupportedAlgorithms #2968

Are you sure you want to change the base?

CAPABILITIES: Add SupportedAlgorithms #2968

Uh oh!

Conversation

ShitalJumbad commented Jan 25, 2025

Uh oh!

steven-bellock 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!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

steven-bellock Mar 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

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!

Uh oh!

Uh oh!

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!

Uh oh!

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!

Uh oh!

steven-bellock Mar 24, 2025 •

edited

Loading