add transforms to lint module name conflicts and override module names

[albertchen-sifive]

This PR adds the LintConflictingModuleNames lint rule that will check for module name collisions and the RenameDesiredNames transform to override module names.

LintConflictingModuleNames works by looking at DesiredNameAnnotations and checking that all desired names target exactly one module. e.g. if two different modules are annotated with Queue desired name this will result in a lint violation.

RenameModuleNames works by looking for OverrideDesiredNameAnnotations and renaming modules and their associated DesiredNameAnnotations to the name override. If the same module name override targets more than one module then the name override is ignored and the targeted modules are not renamed.

also adds RenameModulesAspect that can be used to emit name overrides and a LintConflictingModuleNamesAspect to collect DesiredNameAnnotations to be checked by the lint pass.

Related issue: chipsalliance/firrtl#1274

Type of change: feature request

Impact: API addition (no impact on existing code)

Development Phase: implementation

Release Notes

add StabilizeModuleNames transform

[mwachs5]

Cool! An immediate reaction... could the hash be prepended with some sort of code that shows which algorithm it is using Maybe something like “hDEADBEEF” for heirqrchical, or “pDEADBEEF” for ports-only, or something? Would be curious to see that in the report at least to understand how names were resolved for each

[albertchen-sifive]

An immediate reaction... could the hash be prepended with some sort of code that shows which algorithm it is using Maybe something like “hDEADBEEF” for heirqrchical, or “pDEADBEEF” for ports-only, or something?

good idea. pushed an update

[mwachs5]

I am not sure I understand the outputs you presented. It says
"===== comparing freechips.rocketchip.system.DefaultSmallConfig.stabilized.v =====
shared names (147):
AXI4Buffer_p5AA54DAA
AXI4Buffer_pD0BF562D
...
"

What are you comparing it to? When it says
names only in original: 129
names only in freechips.rocketchip.system.TinyConfig.stabilized.v: 98

What is original?

[albertchen-sifive]

What is original?

freechips.rocketchip.system.DefaultConfig

[albertchen-sifive]

anyone know how to fix The command "travis_wait 80 make emulator-ndebug -C regression SUITE=Miscellaneous JVM_MEMORY=3G" exited with 137. https://travis-ci.org/github/chipsalliance/rocket-chip/jobs/684537271#L5385? I've already restarted once. googling it looks like its because it takes too many resources?

[mwachs5]

Uh oh maybe my sudo = false change was premature

[albertchen-sifive]

maybe split up the Miscellaneous test suite to to smaller ones?

[mwachs5]

Sorry, didn't realize this review was still sitting in "pending" state

[mwachs5]

Also, can you update the PR Title/description at the top to reflect the new scope of this?

[azidar]

Looking better :) I think we should get rid of the "strategy" concept entirely as its just distracting now.

[azidar]

Great! I think removing the strategy stuff made the PR a lot cleaner. I'm happy to approve except for two minor nits (update recommendedFix and add a comment to the optionalPrereqs). I think Megan has a few unresolved comments about the tests, so once those are addressed happy to merge!

[mwachs5]

@azidar do you need to re-review given your comments? Merging is blocked on your "request changes". I am reviewing again now.

[mwachs5]

LGTM!

[albertchen-sifive]

force-pushed because rerun workflow button doesn't seem to work for me