pldmtool: Support for Seamless Firmware Servicing

[spargaon]

Adding support for Seamless Firmware Servicing (SFS) protocol. Implement code to support Get Firmware Version and Update Firmware using mctpRaw option of pldmtool.

[spargaon]

pldmtool_test.txt

Test pldmtool commands with the changes.

[spargaon]

Few of the considerations:

Since MCTP layer extracts its header in a response and only sends the payload to the upper layer (e.g. libpldm), there is no IC bit and Message Type information in the payload. I think the SFS protocol probably needs to retain that information as part of its protocol (e.g. a byte). Since the IC bit information is missing, parsing code for SFS response has to either a) disregard checksum and verification of the checksum or b) explicitly expect a checksum in a response and validate the checksum. Currently b) is coded.

Since the Message Type information is not available, the parsing code needs to decide whether the response is for either
a PLDM request or a SFS request, perhaps by considering presence of four bytes of IANA number.

I think existing pldmtool mctpRaw testcases need to be exercised to make sure there is no regression with this change.

[mahkurap]

couple of comments. please check.

[ojayanth]

File Scope:

• Please follow commit message 50(subject(/72(body) Rule,
• Please make sure rin clang-format tool.
• Upstream-status: Inappropriate - Not required here . Is used for patches only

[spargaon]

Except for clang-format tool, where I ran into errors, completed this.

[ojayanth]

Since we are using c++23, inline for constants is generally not required because:

constexpr variables at namespace scope have internal linkage by default (since C++17), so they don’t cause multiple definition issues across translation units.

[spargaon]

@ojayanth, I have addressed most of the comments except:
Please make sure run clang-format tool.

Also, function pldm_edac_crc32() does not accept Use std::span as input, its definition is different. Had a compilation error...

| ../../../../../../workspace/sources/pldm/pldmtool/pldmtool.cpp:214:93: error: cannot convert 'std::span' to 'const void*' | 214 | uint32_t csField = to_big_endian(static_cast<uint32_t>(pldm_edac_crc32(std::span(fData)))); | | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | | | | | std::span

[spargaon]

root@congo-0057:/tmp# pldmtool mctpraw --help
send an MCTP raw request and print response
Usage: pldmtool mctpRaw [OPTIONS]

Options:
-h,--help Print this help message and exit
-m,--mctp_eid UINT MCTP endpoint ID
-v,--verbose
-n,--retry-count UINT Number of retry when PLDM request message is failed
-e,--network-id UINT REQUIRED
MCTP NetworkId
-d,--data UINT ... REQUIRED raw MCTP data
-i,--file-in TEXT Read in the file to be sent
-o,--file-out TEXT Write out the received file
-c,--checksum Append/Validate checksum
-p,--prealloc-tag use pre-allocated MCTP tag for this request

Example: pldmtool mctpRaw -m 21 -e 1
--data 0x7F 0x00 0x00 0x0E 0x78 0x80 0x02
--file-in req.bin --file-out resp.bin --checksum

[spargaon]

Verified with the new code, when the same command is issued:
root@congo-0057:/tmp# pldmtool mctpraw -m 17 -e 1 --data 0x7F 0x00 0x00 0x0E 0x78 0x80 0x02 -i sfs.pkg -c

metadata (e.g. AMD IANA number and file size in Big Endian format) and checksum (also in Big Endian format)
matches with the same in prior coded version of pldmtool

e.g. Some initial bytes with the current vesion
pldmtool: Tx: ff 00 00 0e 78 80 02 00 00 53 80 0a 29 8a 5f fe 02 6d 21 c6 d2 79 1a b2 2e 14 b1 24
checksum 3a 8d cd 87

and this with the prior version
pldmtool: Tx: ff 00 00 0e 78 80 02 00 00 53 80 0a 29 8a 5f fe 02 6d 21 c6 d2 79 1a
checksum: 3a 8d cd 87