New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Additional issues with interfaces inside of generate blocks #1001
Comments
Original Redmine Comment Started working on this: https://github.com/toddstrader/verilator-dev/tree/intf_named_gen This fixes t_interface_gen5, but breaks t_genfor_hier. The problem is that I can just slap the new scope name on VarXRefs in the BeginVisitor, but if those VarXRefs were pointing outside of the generate block, now they are broken. I can't play the same game that BeginRelinkVisitor does for VarRefs with VarXRefs, because a VarXref's dotted() is just a string, and I can't check to see if that's been changed or not. Still hacking . . . |
Original Redmine Comment Try adding to V3Begin a AstVarXRef visitor that sets inlinedDots like as in V3Inline. Also with vars fixed, there's probably a similar problem and fix with dotted function/task calls. |
Original Redmine Comment Excellent, that's what I was missing. This works now. I'll take a look at functions too, clean this up and submit the patch. |
Original Redmine Comment VarXRefs work, function references work for the latest version of t_interface_gen5 now, but are broken for a bunch of function tests: https://github.com/toddstrader/verilator-dev I'm currently trying to figure out how to disambiguate between funcrefs that need to have their scope patched in V3Begin and those that don't. |
Original Redmine Comment I found some other issues that weren't previously addressed and have a patch ready here now: https://github.com/toddstrader/verilator-dev/tree/issue1001.2 This doesn't address the function references, as that is not a straight forward copy of the VarXRef fix and I'm still trying to get my head around that one. However, I'd like to propose that this be pulled as-is. If that's acceptable, I'll open another issue to track the function issue. The rationale is that this patch currently gets us further into our codebases with Verilator and I'm trying to minimize the number of out-of-band patches we're juggling. |
Original Redmine Comment Wilson, please hold off on this pull request. There's some issue with arrayed interfaces when using this patch and Johan's issue 879 patch together. We'll sort it out and report back here. |
Original Redmine Comment This branch was provided which also fixes #�. https://github.com/toddstrader/verilator-dev/commits/issue1001.3 This is fixed in git towards 3.880. See some other notes in #�. |
Original Redmine Comment In 3.880. |
Author Name: Todd Strader (@toddstrader)
Original Redmine Issue: 1001 from https://www.veripool.org
Original Date: 2015-11-14
Original Assignee: Todd Strader (@toddstrader)
Additional issue, very similar to Issue #998:
https://github.com/toddstrader/verilator-dev/tree/intf_named_gen
I modified t_interface_gen5 to show the issue. I also added the () to the interface instantiation to make ModelSim happy for cross-checking.
Haven't taken a look at fixing this yet, but any advice is appreciated.
The text was updated successfully, but these errors were encountered: