Skip to content

Pressing either Ctrl-\ of Ctrl-4 core dumps R

5 messages · Henrik Bengtsson, William Dunlap, Luke Tierney +1 more

Original
Henrik Bengtsson
# 
When running R from the terminal on Linux (Ubuntu 16.04), it core
dumps whenever / wherever I press Ctrl-4 or Ctrl-\.  You get thrown
back to the terminal with "Quit (core dump)" being the only message.
Grepping the R source code, it doesn't look like that message is
generated by R itself.  Over on Twitter, it has been confirmed to also
happen on macOS.

$ R -d valgrind --vanilla --quiet
==979== Memcheck, a memory error detector
==979== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==979== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info
==979== Command: /usr/lib/R/bin/exec/R --vanilla --quiet
==979==
[1] 3

# At next prompt I press Ctrl-\. The same happens also when done in
the middle of an entry.
==979== Process terminating with default action of signal 3 (SIGQUIT)
==979==    at 0x576C9C3: __select_nocancel (syscall-template.S:84)
==979==    by 0x502EABE: R_SelectEx (in /usr/lib/R/lib/libR.so)
==979==    by 0x502EDDF: R_checkActivityEx (in /usr/lib/R/lib/libR.so)
==979==    by 0x502F32B: ??? (in /usr/lib/R/lib/libR.so)
==979==    by 0x4F6988B: Rf_ReplIteration (in /usr/lib/R/lib/libR.so)
==979==    by 0x4F69CF0: ??? (in /usr/lib/R/lib/libR.so)
==979==    by 0x4F69DA7: run_Rmainloop (in /usr/lib/R/lib/libR.so)
==979==    by 0x4007CA: main (in /usr/lib/R/bin/exec/R)
==979==
==979== HEAP SUMMARY:
==979==     in use at exit: 28,981,596 bytes in 13,313 blocks
==979==   total heap usage: 27,002 allocs, 13,689 frees, 49,025,684
bytes allocated
==979==
==979== LEAK SUMMARY:
==979==    definitely lost: 0 bytes in 0 blocks
==979==    indirectly lost: 0 bytes in 0 blocks
==979==      possibly lost: 0 bytes in 0 blocks
==979==    still reachable: 28,981,596 bytes in 13,313 blocks
==979==         suppressed: 0 bytes in 0 blocks
==979== Rerun with --leak-check=full to see details of leaked memory
==979==
==979== For counts of detected and suppressed errors, rerun with: -v
==979== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)
Quit (core dumped)

$ R --version
R version 3.3.2 (2016-10-31) -- "Sincere Pumpkin Patch"
Copyright (C) 2016 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

/Henrik
William Dunlap
# 
Control-backslash is the default way to generate SIGQUIT from the
keyboard on Unix and SIGQUIT, by default, aborts the process and
causes it to produce a core dump.  Do you want R to catch SIGQUIT?

% stty --all
speed 38400 baud; rows 24; columns 64; line = 0;
intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^D;
eol = <undef>; eol2 = <undef>; swtch = <undef>; start = ^Q;
stop = ^S; susp = ^Z; rprnt = ^R; werase = ^W; lnext = ^V;
flush = ^O; min = 1; time = 0;
-parenb -parodd -cmspar cs8 -hupcl -cstopb cread -clocal -crtscts
-ignbrk -brkint -ignpar -parmrk -inpck -istrip -inlcr -igncr
icrnl ixon -ixoff -iuclc -ixany -imaxbel -iutf8
opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0
tab0 bs0 vt0 ff0
isig icanon iexten echo echoe echok -echonl -noflsh -xcase
-tostop -echoprt echoctl echoke
Bill Dunlap
TIBCO Software
wdunlap tibco.com


On Fri, Feb 10, 2017 at 10:40 AM, Henrik Bengtsson
<henrik.bengtsson at gmail.com> wrote:
Luke Tierney
# 
So do a number of other interactive programs when working in a
terminal (e.g. python) since it looks like your terminal is configured
for those two actions to send the SIGQUIT signal. Whether R should
ignore that signal, under some circumstances at least, is another
question.

Best,

luke
On Fri, 10 Feb 2017, Henrik Bengtsson wrote:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    
2 days later

    

      
      

        


  

        

      Henrik Bengtsson
    

    

      
      #
    

  


  

      
Thanks for these explanations - it all makes sense, that is, the
default behavior for a process that does not capture SIGQUIT is to
quit and perform a core dump
(https://en.wikipedia.org/wiki/Unix_signal#SIGQUIT).

Then the remaining question, as Luke says, is: should R handle this
signal?  For instance, in interactive mode, SIGQUIT could maybe bring
up:

Possible actions:
1: abort (with core dump, if enabled)
2: normal R exit
3: exit R without saving workspace
4: exit R saving workspace
5: ignore SIGQUIT (continue evaluation)
Selection:

giving the option to ignore a SIGQUIT send by mistake.  Not sure how
big of a problem this is (I'm surprise I never hit Ctrl+\ by mistake
previously).


Also, I'm sharing my notes about stty and SIGQUIT in case someone else
finds them useful:

My terminal (Linux / Ubuntu 16.04) settings are the same as Bill's
(stty --all). Thus, pressing Ctrl+\ causes the terminal to signals
SIGQUIT to the running process (= R).  Since R does not handle /
capture this specifically, this results in the process to quit and
perform a core dump.  Pressing Ctrl+\ is effectively the same a
calling 'kill -s QUIT <pid>'.

One can disable the QUIT signal send by the terminal by:

$ stty quit ''

such that one gets:

$ stty --all
speed 38400 baud; rows 33; columns 80; line = 0;
intr = ^C; quit = <undef>; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;
eol2 = <undef>; swtch = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R;
werase = ^W; lnext = ^V; discard = ^O; min = 1; time = 0;
-parenb -parodd -cmspar cs8 -hupcl -cstopb cread -clocal -crtscts
-ignbrk -brkint -ignpar -parmrk -inpck -istrip -inlcr -igncr icrnl ixon -ixoff
-iuclc -ixany -imaxbel iutf8
opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab0 bs0 vt0 ff0
isig icanon iexten echo echoe echok -echonl -noflsh -xcase -tostop -echoprt
echoctl echoke -flusho -extproc

This will prevent QUIT to be signalled when pressing Ctrl+\ and
thereby R (Python, ...) won't core dump.  One can of course still use
kill -s QUIT <pid>.

To reset the above, either restart the terminal or use either of:

$ stty quit ^\\    ## caret notation (espaced ^\)
$ stty quit 0x1c   ## hexadecimal notation
$ stty quit 034    ## octal notation
$ stty quit 28     ## decimal notation

I still don't understand why the terminal treats keypress Ctrl+4 the
same as Ctrl+\, but at least I'm not alone;
https://catern.com/posts/terminal_quirks.html#fn.3.

Thanks

Henrik

        
On Fri, Feb 10, 2017 at 11:00 AM,  <luke-tierney at uiowa.edu> wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Peter Dalgaard
    

    

      
      #
    

  


  

      
On 12 Feb 2017, at 23:54 , Henrik Bengtsson <henrik.bengtsson at gmail.com> wrote:

        

            

      
      
      
    

        
I would guess that this was just to get certain escape chars within reach on non-US keyboard layouts, e.g. the [\] characters are replaced by some permutation of the three extra letters in Scandinavian languages (as Henrik surely knows all about). So the awkward ones were reassigned/duplicated at Ctrl-2 -- Ctrl-8.

On, say, a current Mac Terminal.app, Ctrl-\ should be Ctrl-Shift-Alt-7, but that key combination actually just generates "7". I also recall terminals where some characters could only be obtained via compose sequences, e.g. compose-/-/ for "\", and there was no obvious way to add a Ctrl modifier to that.

-pd