Skip to content

*** caught bus error *** address 0x114382544, cause 'non-existent physical address'

5 messages · Richard R. Liu, Simon Urbanek

Original
Richard R. Liu
# 
I am experiencing a curious error running R64 (2009-11-18 r50482) on a  
MacBook Pro 17" Unibody early 2009 with 8GB RAM.  In the console  
excerpt below sentences.1000.clustered is a dataframe consisting of  
two vectors, sent and cluster.  sent is a vector of 1000 sentences  
chosen at random from over 3 million, and cluster is a vector of  
integers denoting to which of 5 clusters the sentence belongs.  For  
each cluster I want to display the sentences that belong to it as  
follows:


R version 2.10.0 Patched (2009-11-18 r50482)
Copyright (C) 2009 The R Foundation for Statistical Computing
ISBN 3-900051-07-0

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

   Natural language support but running in an English locale

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

[R.app GUI 1.30 (5521) x86_64-apple-darwin9.8.0]

[Workspace restored from /Users/richard/.RData]

 > library(filehash)
filehash: Simple key-value database (2.0-1 2008-12-19)
 > db <- dbInit("/Volumes/Work on RDR Test Documents/R Databases/ 
DB_TXT", type = "RDS")
 > dbLoad(db)
 >
 > sent.clust <- sentences.1000.clustered
 > i.clust.2 <- which(sent.clust$cluster == 2)
 > attributes(sent.clust)
$names
[1] "sent"    "cluster"

$row.names
    [1]    1    2    3    4    5    6    7    8    9   10   11   12    
13   14   15   16   17   18   19   20   21   22   23   24   25   26    
27   28   29
...
  [987]  987  988  989  990  991  992  993  994  995  996  997  998   
999 1000

$class
[1] "data.frame"

 > i.clust.2 <- which(sent.clust$cluster == 2)
 > nchar(sent.clust$sent[i.clust.2])
   [1]   232   419   787  1010  1926   357  2865   657   560   213   
2175  2363  1300   359   331   965   581   732   398   245  1650  
25893  1289   510
  [25]  1714   524   747  2378   465   651  2253   215   460   253   
2475   554  1130   529   261  1413   418  2168  1569  1538   301   
2506  1163  3036
  [49]  1445   199  1231   370  1307   371   379   232  1297   500    
409   481  2187   314   730  1634   971   860   362   506  1606    
394   145   480
  [73]  1582  1471  1269   308  1113   444   990   298  1644   445    
357  1204   322   638  2182   805  1545   368  1461  1347   649   399   
1094  1624
  [97] 14605  1637   269   503   502  2145  1398   481   122  2248
 > sent.clust$sent[i.clust.2]
   [1] "(patient continuing ...)  
______________________________________________________________________________________________________________________________________ 
  Investigator text for Adverse Events encoded using MedDRA version  
11.0. "
   [2]
  *** caught bus error ***
address 0x114382544, cause 'non-existent physical address'

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

The point of

 > i.clust.2 <- which(sent.clust$cluster == 2)
 > nchar(sent.clust$sent[i.clust.2])

is to try to isolate the problem.  Indeed, the more compact sent.clust 
$sent[which(sent.clust$cluster == 2)] work when I start R in 32-bit  
mode.  In 64-bit mode, attempting to display the sentences in cluster  
5 also fails, whereas for clusters 1, 3, 4 it succeeds.  In 32-bit  
mode, I have experienced no such problem.

Does anybody have any ideas about what is wrong?

Regards,
Richard

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2429 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20091118/840d6930/attachment.bin>
Simon Urbanek
# 
Richard,

in order to track this down, please start R64, type

Sys.getpid()

and note down the number. Then in a separate window open Terminal and  
type

gdb

Then on the gdb console (it will just say "(gdb)") type

attach n

where n is the number from above. It will do something for a while and  
when it comes back in the console, type

c

Then do your work in R until you get a crash. Once you do, you will  
see something happening in the gdb console. Then type

bt

and copy/paste everything that gdb printed since you typed "c" and  
send it to me.

Thanks,
Simon

-----

Practical example -- in R:

 > Sys.getpid()
[1] 19381

in Terminal:

ginaz:~$ gdb
GNU gdb 6.3.50-20050815 (Apple version gdb-962) (Sat Jul 26 08:14:40  
UTC 2008)
[...]
This GDB was configured as "i386-apple-darwin".
(gdb) attach 19381
Attaching to process 19381.
Reading symbols for shared libraries . done
[...]
0x9428a6fa in select$DARWIN_EXTSN ()
(gdb) c
Continuing.

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0x00000003
0x00412d09 in SET_SYMVALUE (x=0x1007110, v=0x0) at ../../../../R-2.10- 
branch/src/main/memory.c:2809
2809	void (SET_SYMVALUE)(SEXP x, SEXP v) { CHECK_OLD_TO_NEW(x, v);  
SYMVALUE(x) = v; }
(gdb) bt
#0  0x00412d09 in SET_SYMVALUE (x=0x1007110, v=0x0) at ../../../../ 
R-2.10-branch/src/main/memory.c:2809
#1  0x0041068e in Rf_ReplIteration (rho=0x1023c38, savestack=0,  
browselevel=0, state=0xbfffe53c) at ../../../../R-2.10-branch/src/main/ 
main.c:263
#2  0x00410a40 in R_ReplConsole (rho=0x1023c38, savestack=0,  
browselevel=0) at ../../../../R-2.10-branch/src/main/main.c:311
#3  0x0041129b in run_Rmainloop () at ../../../../R-2.10-branch/src/ 
main/main.c:964
#4  0x00001ff0 in main (ac=1, av=0xbffff5c8) at ../../../../R-2.10- 
branch/src/main/Rmain.c:33
(gdb) q
The program is running.  Quit anyway (and detach it)? (y or n) y
Detaching from process 19381 thread 0x613.
On Nov 18, 2009, at 14:39 , Richard R. Liu wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    
11 days later

    

      
      

        


  

        

      Richard R. Liu
    

    

      
      #
    

  


  

      
This problem disappeared in patched releases soon after I replied to  
Simon that I was unable to obtain the information that he required  
from the debugger.  After checking a few releases I quite monitoring  
the problem.  I assumed that it had been fixed with comment.  However,  
today, the patched release dated 2009-11-29 has the problem once  
again.  I have tried going back to the patched releases of 2009-11-27  
and 2009-11-22, the only ones for which I have kept installers here at  
work.  Both have the problem.  In neither can I obtain the debugging  
information that Simon requested when the problem initially arose.   
Can somebody tell me the current status of the problem, and when it  
might be fixed (again)?  It occurs at an extremely inconvenient time  
in my project.

Thanks,

Richard R. Liu
Dittingerstr. 33
CH-4053 Basel
Switzerland

Tel.:  +41 61 331 10 47
Mobil: +41 79 708 67 66
Email:  richard.liu at pueo-owl.ch

        
On Nov 18, 2009, at 21:07 , Simon Urbanek wrote:

        

            

      
      
      
    

        

      
      
    

        
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2429 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20091130/f259611b/attachment.bin>

  
  


  


      

    

    

      
      

        


  

        

      Simon Urbanek
    

    

      
      #
    

  


  

      
Richard,

I didn't touch anything, but then you didn't supply a reproducible  
example so we can't do much about it (the trace is unfortunately  
bogus). My next step woul dbe to try it in valgrind if I could  
reproduce it ...

Cheers,
Simon

        
On Nov 30, 2009, at 11:01 , Richard R. Liu wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Richard R. Liu
    

    

      
      #
    

  


  

      
Hi Simon,

I think I've found the problem:  filehash needs to be rebuilt.

The data that I've been using the problematic expressions on resides  
in a filehash database.  Having reduced the problem to one particular  
data frame of two columns and 1000 rows, I just copied it to its own  
database, zipped it, and attached it to an email containing  
instructions on how to reproduce the problem, when I decided I should  
test whether you could indeed reproduce the problem by following my  
instructions.  And, of course, everything worked just as I had not  
expected, i.e., there was no bus error.

So I copied the data frame from the small database that I was going to  
send to you to the original, and there the problem occurred.  I could  
not reproduce the problem, however, if I copied the original data  
frame to my workspace and forced R to actually instantiate the copy  
before running the problematic script.  So the problem seemed to have  
something to do with filehash and the size of the database.  I also  
recalled that I had no problem with the script if I ran R in 32-bit  
mode.  I therefore tried reinstalling the filehash binary, but that  
did not solve the problem.  Then, I installed it from the source.   
That evidently is what was required.

So I guess it only needs a rebuild so that other folks do not  
experience the same problem.

Regards,
Richard

Richard R. Liu
Dittingerstr. 33
CH-4053 Basel
Switzerland

Tel.:  +41 61 331 10 47
Mobil: +41 79 708 67 66
Email:  richard.liu at pueo-owl.ch

        
On Nov 30, 2009, at 17:18 , Simon Urbanek wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2429 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-mac/attachments/20091130/9b01647b/attachment-0001.bin>