[hmac, rtl] Do not skip padding after a hash stop command

[andrea-caforio]

This commit removes some conditional FSM states transition that were deemed undesirable in prim_sha2_pad, see #23936. It made it possible to cancel an ongoing padding operation, which is disallowed by the HMAC module. As a side-effect, the removal plugs some existing FSM transition coverage holes in prim_sha2_pad whose manual exclusions can be removed as well.

This change makes it impossible to transition into the StFifoReceive state from either StPad80, StPad00, StLenHi or StLenLo by asserting the hash_go signal.

[rswarbrick]

Splatting the exclusion list might be right (I haven't reviewed it carefully, but it seems reasonable). But doesn't the RTL change also imply that the UNR list might change?

[andrea-caforio]

Splatting the exclusion list might be right (I haven't reviewed it carefully, but it seems reasonable). But doesn't the RTL change also imply that the UNR list might change?

Thank you for the review @rswarbrick. You're right, the UNR file should be updated too. Unfortunately, we cannot generate these exclusion files with our VCS license on the Zurich workstation.

[vogelpi]

This looks good to me from an RTL perspective. Thanks for the PR @andrea-caforio !

Did you run a full regression for this and inspect the coverage yet?

[andrea-caforio]

This looks good to me from an RTL perspective. Thanks for the PR @andrea-caforio !

Did you run a full regression for this and inspect the coverage yet?

Thank you @vogelpi for the review. I ran a full regression and all the transitions are now properly covered (see screenshot). I further added a second commit that updates the exclusion file with the list in #24692. @martin-velay you might want to double-check this.

[martin-velay]

Hi @andrea-caforio, I would prefer to have the exclusions in a separate PR as there is no direct link between this RTL fix and the exclusions. Thanks in advance !

[andrea-caforio]

Hi @andrea-caforio, I would prefer to have the exclusions in a separate PR as there is no direct link between this RTL fix and the exclusions. Thanks in advance !

Here you go #25696.