Project

General

Profile

[logo] 
 
Home
News
Activity
About/Contact
Major Tools
  Dinotrace
  Verilator
  Verilog-mode
  Verilog-Perl
Other Tools
  IPC::Locker
  Parallel::Forker
  Voneline
General Info
  Papers

Activity

From 06/19/2019 to 07/18/2019

07/18/2019

05:13 PM Issue #1479 (Feature): Feature Request: VerilatedVcd callback on rolloverMB
I have the time to implement this feature, and am looking for input.
This is a feature for use in conjunction with...
Marc Jessome
12:02 PM Issue #1478 (Confirmed): -faligned-new does not work under Travis with clang
Wilson Snyder
11:41 AM Issue #1478 (Confirmed): -faligned-new does not work under Travis with clang
It was discovered in bug1468 that the -faligned-new flag causes errors when compiling Verilated code under Travis CI ... Todd Strader
12:01 PM Issue #1468 (Closed): Add Continuous Integration
Awesome work!
Wilson Snyder
11:45 AM Issue #1468: Add Continuous Integration
Squashed, pushed and the tests are green. Also for reference, bug1477 and bug1478 were opened as discussed above. Todd Strader
01:18 AM Issue #1468: Add Continuous Integration
Great, again. "more-travis" looks fine please squash merge it to trunk (or if get it to one commit just normal pull ... Wilson Snyder
12:29 AM Issue #1468: Add Continuous Integration
This is ready to push/discuss next steps:
https://github.com/toddstrader/verilator-dev/tree/more-travis
Here is t...
Todd Strader
10:32 AM Issue #1477 (Assigned): Add macOS coverage in Travis CI
Kevin Kiningham indicated he will look at this in about a month.
Wilson Snyder
09:56 AM Issue #1477 (Assigned): Add macOS coverage in Travis CI
.travis.yml has commented out lines for macOS testing. I was unable to get it to work and don't have a local macOS e... Todd Strader

07/15/2019

01:47 AM Issue #1435 (Resolved): Report column numbers and source text in error messages
Wilson Snyder
01:46 AM Issue #1435: Report column numbers and source text in error messages
Fixed in git towards 4.017. Also added suggestions, which are arguably even more useful.... Wilson Snyder
01:44 AM Issue #1439 (Closed): Add "included from" and macro expansion information in error messages
Macro expansion is very difficult as requires tracking recursive information on every token through the preprocessor.... Wilson Snyder

07/11/2019

04:16 PM Issue #1468: Add Continuous Integration
I was thinking that configure compiles but doesn't link, but it does seem to try a link, so I'm at a loss.
Wilson Snyder
03:55 PM Issue #1468: Add Continuous Integration
I don't know if it's the only clang problem, but there is a systemic problem when using clang under Travis which is c... Todd Strader
12:31 AM Issue #1439: Add "included from" and macro expansion information in error messages
Notes on how GCC tracks defines.
Debug can be seen with "g++ -fdebug-cpp -E -c file.cpp". line-map.c's linemap_du...
Wilson Snyder

07/10/2019

09:58 PM Issue #1305: Error messages do not contain hierarchical information
Reply in bug1315. Wilson Snyder
07:13 PM Issue #1305: Error messages do not contain hierarchical information
So then would we have an AstNode* (or maybe AstConst*) in V3Number and set that pointer inside simulateVisitor::newNu... Todd Strader
09:58 PM Issue #1315: Using an array in a function called from a parameter
The idea is every V3Number would point to the AstConst it's inside (excluding when inside the parser unless we fix th... Wilson Snyder
07:13 PM Issue #1315: Using an array in a function called from a parameter
Wilson,
Some questions regarding #1:
* Just to make sure I understand, are you suggesting that user2/3p would g...
Todd Strader

07/09/2019

11:21 AM Issue #1474: Document parse-order-dependency for config files
Thanks, will give it a go when I have time. Cheers. Shareef Jalloq
02:15 AM Issue #1474: Document parse-order-dependency for config files
An absolute path wasn't really intended but isn't surprising since it was never thought about - would you be willing ... Wilson Snyder

07/08/2019

03:00 PM Issue #1474: Document parse-order-dependency for config files
Creating the testcase shows me that you require an absolute path in the VLT file. Is that by design or a bug? Shareef Jalloq
01:43 PM Issue #1474: Document parse-order-dependency for config files
Yes, the .vlt should be first. If you don't see suppression working please reduce it to a test case and file a new b... Wilson Snyder
01:34 PM Issue #1474: Document parse-order-dependency for config files
I've hit what I think is the same issue today but can't get the waivers to apply. I'm supplying the .wlt configurati... Shareef Jalloq

07/07/2019

12:00 PM Issue #1476 (Duplicate): Please honour CFLAGS and CXXFLAGS provided by user
Thanks, this was earlier reported via Gianfranco Costamagna and is applied in git.
Wilson Snyder
12:49 AM Issue #1476 (Duplicate): Please honour CFLAGS and CXXFLAGS provided by user
It was reported on Debian that verilator does not honor CFLAGS that are provided by user. The attached patch fixes th... أحمد المحمودي

07/06/2019

10:11 PM Issue #1474 (Closed): Document parse-order-dependency for config files
Perfect, thanks, pushed to git towards 4.018.
Wilson Snyder
08:46 PM Issue #1474: Document parse-order-dependency for config files
Hi Wilson,
Sorry for the slow response, looks like the email notification didn't reach me.
An updated patch wit...
Philipp Wagner
08:27 PM Issue #1442 (Resolved): Enum value not made sized when enum sized
Fixed in git towards 4.018.
I tried a few alternatives to preserve the behavior I thought was best for WIDTH warni...
Wilson Snyder
07:27 PM Issue #1468: Add Continuous Integration
Ok, cool. I'm pretty sure it's only happening under 14.04 in Travis (and only --vltmt). I tried to reproduce under ... Todd Strader
05:48 PM Issue #1468: Add Continuous Integration
FYI I disabled t_leak.pl for now when under travis.
Wilson Snyder
07:25 PM Issue #1305: Error messages do not contain hierarchical information
Yes, I think if V3Simulate uses AstConst, then the only standalone number use left is in the parser and that shouldn'... Wilson Snyder
07:15 PM Issue #1305: Error messages do not contain hierarchical information
I currently have it so that V3Numbers point back to their AstNode so at error time they can get location information ... Todd Strader
06:20 PM Issue #1305: Error messages do not contain hierarchical information
I stumbled on the V3Hashed iterator problem earlier too, this is fixed in trunk.
I'm confused, why does AstNode ne...
Wilson Snyder

07/04/2019

03:37 PM Issue #1468: Add Continuous Integration
Feel free to push the badged README.pod. The more_travis branch currently seems to have hackage to debug t_leak so do... Wilson Snyder

07/03/2019

04:26 PM Issue #1468: Add Continuous Integration
Couple more notes:
I'm pretty sure that t_leak is not flaky but just failing under Ubuntu 14.04. It probably look...
Todd Strader
11:53 AM Issue #1468: Add Continuous Integration
Yeah, we only have three workers to play with regardless of OS, compiler, etc. I've set it up so that we can get the... Todd Strader
02:26 PM Issue #1305: Error messages do not contain hierarchical information
OK, here's the latest iteration:
https://github.com/toddstrader/verilator-dev/tree/mod_stack_trace_8
I went ahead...
Todd Strader

07/01/2019

11:19 PM Issue #1315: Using an array in a function called from a parameter
Thanks! I'll try to poke at it and see how far I can get. Patrick Collins
09:52 AM Issue #1468: Add Continuous Integration
> t_leak seems to fail, can you debug?
Sure.
> Have the vltmt & vlt section only do a test_regress, not a test
G...
Todd Strader

06/29/2019

09:31 PM Issue #1468: Add Continuous Integration
1. t_leak seems to fail, can you debug?
2. Have the vltmt & vlt section only do a test_regress, not a test, as thi...
Wilson Snyder
07:25 PM Issue #1468: Add Continuous Integration
FYI got a false failure. I added automatic retry of failures (long overdue). Also turned down the verbosity.
Wilson Snyder
12:26 PM Issue #1468: Add Continuous Integration
FYI Travis caught I botched t_flag_csplit on old makes, fixed (I think).
Wilson Snyder
10:40 AM Issue #1315: Using an array in a function called from a parameter
Note bug1052 also is the same issue, so should be part of the test cases.
Wilson Snyder
09:53 AM Issue #1315: Using an array in a function called from a parameter
Patrick, would love to have you help on this. While certainly it will take a bit of time investment to understand th... Wilson Snyder
12:41 AM Issue #1315: Using an array in a function called from a parameter
I just bumped into this too. Do you think this is a tractable problem for someone new to the Verilator codebase? Patrick Collins
10:39 AM Issue #1052 (Duplicate): Localparam array can't be passed into parameter-called function
bug1315 is the same issue, so tracking there.
Wilson Snyder
10:24 AM Issue #385 (Closed): Dpi exported tasks with array inputs don't compile.
Forgot to close; open array support was added in 3.918 2018-01-02.
Wilson Snyder
10:16 AM Issue #1422 (NoFixNeeded): VPI hierarchy naming
See last msg, think this will make naming compatible, if not please reopen. Wilson Snyder

06/28/2019

05:10 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
I've tried to reduce it along the lines you describe and I've failed. I'm skeptical of this apparoch because simply ... Øyvind Harboe
04:45 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
That helps, but not enough to suggest what to fix so still think we need to get to a testcase. Basically, Verilator ... Wilson Snyder
10:33 AM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
Output w/patch:... Øyvind Harboe
09:59 AM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
Presumably it's --debug that is hiding the issue as the code compiles differently.
Try --debugi 3. Also edit line...
Wilson Snyder
07:36 AM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
I tried with valgrind, the crash was reproducible, but it didn't report anything. Øyvind Harboe
07:36 AM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
I'm afraid that I'm only seeing the crash without "--debug --gdbgt", I checked. Øyvind Harboe
12:06 AM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
The logfile you attached showed no errors. Perhaps you're still in the passing bisect?
Wilson Snyder

06/27/2019

11:34 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
Here's the last lines when running with "--debug --gdbgt":... Øyvind Harboe
11:28 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
Is it possible to split that commit into a sequence of commits, purely for the purpose of bisecting it further or is ... Øyvind Harboe
10:54 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
Bisecting was a good idea, unfortunately that's was a hyper large change, though I wouldn't have expected those edits... Wilson Snyder
10:42 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
I was able to run a bisection to find the first failing version of Verilator:... Øyvind Harboe
10:21 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
This is the only other file in the design in addition to DressRehearsalTestBench.v Øyvind Harboe
10:17 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
The parameterization happens in Chisel, so Chisel spits out unparameterized files, so the only parameterization I hav... Øyvind Harboe
10:13 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
... Øyvind Harboe
10:04 PM Issue #1475 (AskedReporter): V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
There was recently a similar error resulting from parameters, are there parameters in DressRehersalTestBench.v?
Wh...
Wilson Snyder
06:21 PM Issue #1475: V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
This is with 4.016 as well as 4.014:
$ verilator --version
Verilator 4.014 2019-05-08 rev UNKNOWN_REV
Øyvind Harboe
06:19 PM Issue #1475 (Closed): V3Hashed.cpp Called isIdentical on non-hashed nodes, from Gate dedupe()
I'm getting the error below in *BOLD* when I invoke Verilator. DressRehearsalTestBench.v is generated by Chisel: http... Øyvind Harboe
10:00 PM Issue #1474: Document parse-order-dependency for config files
Sorry you had to experiment to figure this out (I didn't realize myself ;) but major thanks for improving the docs.
...
Wilson Snyder
03:03 PM Issue #1474 (Closed): Document parse-order-dependency for config files
When including a *.vlt configuration file with lint waivers, only the
global waivers (file == "*") are applied uncon...
Philipp Wagner
05:06 PM Issue #1468: Add Continuous Integration
OK, the push works now and I see the authorization for Travis. I'm not sure how to tell Travis to start watching the... Todd Strader
04:38 PM Issue #1468: Add Continuous Integration
Fixed a mis-setting, you should be able to push. If other problems might want to use git rather than http as a transp... Wilson Snyder
04:01 PM Issue #1468: Add Continuous Integration
OK, I believe everything is push-able now:
https://github.com/toddstrader/verilator-dev/tree/travis
https://travis-...
Todd Strader

06/26/2019

11:57 PM Issue #1473: How to get no `line pragmas in preprocessor
Wilson Snyder wrote:
> Please see the -P option in the manual.
>
> If you're writing a parser, you might want to ...
Aliaksei Chapyzhenka
10:42 PM Issue #1473 (NoFixNeeded): How to get no `line pragmas in preprocessor
Please see the -P option in the manual.
If you're writing a parser, you might want to consider handling `line for ...
Wilson Snyder
08:47 PM Issue #1473 (NoFixNeeded): How to get no `line pragmas in preprocessor
As I understand, Verilator preprocessor inserts `line pragmas instead of `include and other directives.
It can appea...
Aliaksei Chapyzhenka
11:09 PM Issue #1468: Add Continuous Integration
Sorry, what I meant for t_gantt was just pass --threads 2 always then fix the test. I just pushed that change.
I ...
Wilson Snyder
10:12 PM Issue #1468: Add Continuous Integration
>> On my travis-ci.com dashboard there is a button to request access for GitHub organizations I'm in. Once we land th... Todd Strader
10:50 PM Issue #1472 (AskedReporter): Memory not updating in for loop
Please attach a complete standalone example so I can try it, thanks. Ideally this would be in test_regress format as... Wilson Snyder
12:12 PM Issue #1472 (NotEnoughInfo): Memory not updating in for loop
We are facing an issue with following verilog code
Snippets of the relevant portion of code is shown below :
--...
Somya Dashora

06/25/2019

01:33 PM Issue #1468: Add Continuous Integration
UNOPTTHREADS is a bit finicky, it's based on the parallelism it can find
divided by threads. Can you see if we can ...
Wilson Snyder
12:21 PM Issue #1468: Add Continuous Integration
As advertised, I am now skipping these failing vltmt tests if the core count is < 3.
I'm guessing that the large n...
Todd Strader

06/24/2019

12:17 PM Issue #1468: Add Continuous Integration
OK, so I'm not sure what to do with --vltml. Based on reading the man page and this issue:
https://www.veripool.org...
Todd Strader

06/22/2019

09:04 PM Issue #1422: VPI hierarchy naming
Stumbled into some other code related to handling TOP, note the existing code allows passing "" when constructing a m... Wilson Snyder
09:01 PM Issue #1439: Add "included from" and macro expansion information in error messages
Includes easier than thought as preproc already records the include tree. Also added easy insertion point for adding ... Wilson Snyder
05:01 PM Issue #1439: Add "included from" and macro expansion information in error messages
I haven't done anything yet. Please feel free. Todd Strader
04:52 PM Issue #1439: Add "included from" and macro expansion information in error messages
Have you looked at this?, if not will look at the include part first which seems easier.
Wilson Snyder
04:50 PM Issue #1462: signal redeclaration is not reported
Thanks for checking, was missing many other cases, and also warning on duplicated ports.
Fixed in git towards 4.01...
Wilson Snyder

06/21/2019

10:01 AM Issue #1468: Add Continuous Integration
I guess I just wasn't patient enough. I was looking for the cache to be shared between the test slices (I had one ru... Todd Strader

06/20/2019

10:18 PM Issue #1468: Add Continuous Integration
Woah. I assumed that you couldn't share state between instances because:
https://docs.travis-ci.com/user/build-stag...
Todd Strader
12:39 PM Issue #1468: Add Continuous Integration
Hi,
I just wanted to point out our ongoing activities of LibreCores CI. It is a jenkins-based CI system we are con...
Stefan Wallentowitz
02:07 PM Issue #1470: VPI systemtf
Ah, you are absolutely right. Good, that I didn't start trying to implement anything so far :)
I think we should b...
Stefan Wallentowitz
02:04 PM Issue #1470: VPI systemtf
Adding a way to overload $display would be ok but you'd have a larger problem that display takes arguments and the DP... Wilson Snyder
01:44 PM Issue #1470: VPI systemtf
How would you fancy an implementation that only allows to overwrite systemtf for a limited set of systemtf (in partic... Stefan Wallentowitz
01:41 PM Issue #1470: VPI systemtf
It is actually even worse: It only uses the systemtf to re-direct $display-style system functions to its output.
A...
Stefan Wallentowitz
01:33 PM Issue #1470 (Feature): VPI systemtf
Note Verilator does support a language extension where arbitrary doller-user functions can instead be converted into ... Wilson Snyder
12:58 PM Issue #1470 (Feature): VPI systemtf
Hi,
VPI allows to add and overwrite System Tasks and Functions. I propose to add the infrastructure to allow for t...
Stefan Wallentowitz
01:48 PM Issue #1469: VPI module
Thanks for the background. I agree adding the type to VpiScope seems most preferred.
Wilson Snyder
01:38 PM Issue #1469: VPI module
Oh, I forgot to add: Major reason for this proposal was that nothing outside a VerilatedScope can be accessed anyways... Stefan Wallentowitz
01:34 PM Issue #1469: VPI module
Essentially, cocotb accesses the signals only by their name. It only uses the vpiModule iterator to find toplevel mod... Stefan Wallentowitz
01:26 PM Issue #1469 (Feature): VPI module
My first reading of this was you needed module scopes, but that is already there, so I think you're saying you want s... Wilson Snyder
12:58 PM Issue #1469: VPI module
But we don't want to limit the iteration to just modules, correct? I'm sure I still don't understand enough about ho... Todd Strader
12:51 PM Issue #1469 (Closed): VPI module
Hi,
with the ongoing integration of Verilator with cocotb, one of the first steps is to get a minimal working VPI ...
Stefan Wallentowitz
11:36 AM Issue #1305: Error messages do not contain hierarchical information
Oh, ack. That is part of what VL_LEAK_CHECK/V3Broken::isAllocated is for, right? I clearly hadn't gotten around to ... Todd Strader

06/19/2019

08:58 PM Issue #1305: Error messages do not contain hierarchical information
Clarification on AstNode clean up:
Will AstNodes ever be destroyed without calling deleteNode()? Or is that always ...
Todd Strader
12:45 PM Issue #1468: Add Continuous Integration
I would hope that the ccache is persistent across their farm, or at least it should, otherwise I don't see why they w... Wilson Snyder
 

Also available in: Atom