[Rcpp-devel] memory not mapped with Rcpp on Ubuntu

[John Clarke]

Hi Serguei,

Thank you! This didn't directly solve my problem, but it got me looking in
the right direction. I ended up connecting to a docker container through VS
Code on my host machine (Windows). I didn't wire up the debugger UI (which
is really convenient in VS Code) but was able to identify the issue through
the VS Code interface which is a lot easier to use than a bare terminal.

For the record the issue was a call to *fclose(pOutStream)* where
pOutStream was Null. On Windows and MacOS this didn't seem to cause a
problem, but in Ubuntu it failed. Sadly the traceback information was not
precise enough to identify the offending line of code, so I had to find it
iteratively.

-John

On Thu, Feb 13, 2025 at 4:23 PM Serguei Sokol <serguei.sokol at gmail.com>
wrote:

> Hi John,
>
> Le 13/02/2025 à 15:17, John Clarke a écrit :
> > Hey folks,
> >
> > I've been developing an Rcpp package on my Mac and testing on Mac and
> > Windows. I also have been testing on linux-latest with the standard
> > Github action which runs the CMD check. These passed without any problem
> > until I implemented a test suite (testthat) and increased the test
> > coverage to include additional calls. It turns out that the two main
> > functions (and maybe others) that I wrote cause a segfault on Ubuntu
> > builds. Several versions on Linux are tested and all fail.
> >
> > It appears to be related to Rcpp so I'm wondering if you've encountered
> > this before and how I might go about finding a solution. I can reproduce
> > the error on a local docker image consistently, but... I don't have the
> > debugging tools I"m used to on MacOS via VS Code to help me step through
> > when the error occurs. And I suspect it has to do with a mismatched
> > library version because the tests pass fine on Windows and MacOS. So not
> > sure if debugging would help much.
> >
> > Is there a known configuration/compiling issue/fix that I could try?
> You can see some techniques e.g. here
> https://github.com/wch/r-debug/blob/master/debugging-r.md
> In short, if you already have a docker image, you can compile you code
> there with "-g -O0" options and run it in:
>
> R -d gdb
> (gdb) run
> <run your R/Rcpp (test)code. An error will bring you back to gdb where
> you can check relevant frames, parameters, cals and so on.>
>
> Hoping it helps,
> Best,
> Serguei.
>
>
> > Apart from installing a full fledged instance of Ubuntu with VS Code and
> > debugger, is there a way I can get closer to finding out what the issue
> is?
> >
> > Clip from the failed R error messages when I call
> > shg$runSimFromFixedValues(N, 0, 0, 1940):
> >
> >   *** caught segfault ***
> > address (nil), cause 'memory not mapped'
> >
> > Traceback:
> >   1: .External(list(name = "CppMethod__invoke_notvoid", address =
> > <pointer: 0x55efab6ef710>,     dll = list(name = "Rcpp", path = "/usr/
> > local/lib/R/site-library/Rcpp/libs/Rcpp.so",         dynamicLookup =
> > TRUE, handle = <pointer: 0x55efa9b64b30>,         info = <pointer:
> > 0x55efaa89aaa0>, forceSymbols = FALSE),     numParameters = -1L),
> > <pointer: 0x55efae9ae1f0>, <pointer: 0x55efaf2399d0>,     .pointer, ...)
> >   2: shg$runSimFromFixedValues(N, 0, 0, 1940)
> >
> > Thanks,
> >
> > -John
> >
> > John Clarke | Senior Technical Advisor |
> > Cornerstone Systems Northwest | john.clarke at cornerstonenw.com
> > <mailto:john.clarke at cornerstonenw.com>
> >
> > _______________________________________________
> > Rcpp-devel mailing list
> > Rcpp-devel at lists.r-forge.r-project.org
> > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
>
> _______________________________________________
> Rcpp-devel mailing list
> Rcpp-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20250214/fca50534/attachment.htm>