Подтвердить что ты не робот

Использование R_alloc в C

У меня есть код C, который я вызываю из R:

.C("giveProb",as.double(2),as.double(2),as.double(c(0,1,0,1,1,0,1,0)))

Я бы назвал это связкой (сотни тысяч или миллионы) раз (с разными третьими аргументами), и он отлично работает, когда я помещаю его в цикл for примерно в 100 раз, но что-то выше этого сбоя R.

У меня такое чувство, что проблема памяти с использованием R_alloc. У меня есть шесть массивов, выделенных в C, например:

newCoefArray = (double *)R_alloc(1,curSize * sizeof(double));

Но в соответствии с R-руководством "Написание R-расширений":

Эта память берется из кучи и освобождается в конце вызова .C,.Call или .External.

который я понял, чтобы память была освобождена во время каждой итерации цикла. Но в следующем предложении:

Пользователи также могут управлять им, отмечая текущую позицию при вызове vmaxget и освобождая память, выделенную впоследствии вызовом vmaxset. Это рекомендуется только для экспертов.

Как человек, который нигде рядом с экспертом, я надеялся, что кто-то здесь может помочь. Код C в целом ниже.

#include <R.h>
#include <Rmath.h>
#include <stdio.h>
void giveProb(double *k, double *q,double *order){
double  curSize;
double  tmpSize;
double  *newCoefArray;
double  *oldCoefArray;
double  *newAArray;
double  *oldAArray;
double  *newBArray;
double  *oldBArray;
int     position=0;

long int factorial(int n){
    if(n==0||n==1){
        return(1);
    }
    int tmp=1,i=1;
    while(i<=n){
        tmp=tmp*i;
        i++;
    }
    return(tmp);
} 

void expander(double a, double b,double c,double d,double coeff){
    double leadingTerm=beta(a,b);
    int bb=b; 
    double index[bb], sumLeaders[bb];
    for(int i=0;i<bb;i++){
        index[i]=a+i;
        sumLeaders[i]=factorial(a+b-1)/(factorial(index[i])*factorial(a+b-1-index[i]));
        newCoefArray[i+position]=coeff*leadingTerm*sumLeaders[i];
        newAArray[i+position]=index[i]+c+1;
        newBArray[i+position]=a+b+d-index[i];
    }
    position=position+bb;
    curSize=position;
}

void separator(double e, double f){
    double a, b, coeff;
    for(int i=0; i<tmpSize; i++){
        coeff=oldCoefArray[i];
        a=oldAArray[i];
        b=oldBArray[i];
        expander(a,b,e,f,coeff);
    }
}

void condenser(){
    tmpSize=0;
    for(int i=1; i<curSize; i++){
        for(int j=0; j<i; j++){
            if(newAArray[j]==newAArray[i]){
                newCoefArray[j]=newCoefArray[j]+newCoefArray[i];
                newCoefArray[i]=0;
            }
        }
    }
    for(int i=0; i<curSize; i++){
        tmpSize=tmpSize+(newCoefArray[i]!=0);
    }
    oldCoefArray =(double *) R_alloc(1,tmpSize * sizeof(double));
    oldAArray = (double *)R_alloc(1,tmpSize * sizeof(double));
    oldBArray = (double *)R_alloc(1,tmpSize * sizeof(double));
    for(int i=0; i<tmpSize; i++){
        oldCoefArray[i]=newCoefArray[i];
        oldAArray[i]=newAArray[i];
        oldBArray[i]=newBArray[i];
    }
    curSize=tmpSize;    
}

  long double coefficient=1;
  for(int i=0;i<*k;i++){
    coefficient=coefficient*factorial(*k)/(factorial(i)*factorial(*k-i-1));
  }
  for(int i=0;i<*q;i++){
    coefficient=coefficient*factorial(*q)/(factorial(i)*factorial(*q-i-1));
  }

  double numObs=*k+*q;
  double out=0;
  curSize=order[1]+1;

  newCoefArray = (double *)R_alloc(1,curSize * sizeof(double));
  newAArray = (double *)R_alloc(1,curSize * sizeof(double));
  newBArray = (double *)R_alloc(1,curSize * sizeof(double));

  expander(order[0]+1,order[1]+1,order[2],order[3],coefficient);

  oldCoefArray = (double *)R_alloc(1,curSize * sizeof(double));
  oldAArray = (double *)R_alloc(1,curSize * sizeof(double));
  oldBArray = (double *)R_alloc(1,curSize * sizeof(double));

  for(int i=0;i<curSize; i++){
    oldCoefArray[i]=newCoefArray[i];
    oldAArray[i]=newAArray[i];
    oldBArray[i]=newBArray[i];
  }

  for(int i=4;i<2*numObs;i+=2){ 
    position=0;
    tmpSize=curSize;
    separator(order[i],order[i+1]);
    condenser();
  }
  position=0;
  for(int i=0;i<curSize;i++){
    out=out+newCoefArray[i]*beta(newAArray[i],newBArray[i]);
  }
  *k=out;

}

UPDATE: Используя предложение в комментарии ниже, я получаю следующее (что подтверждает то, что я думал раньше, не так ли?):

R -d valgrind -f test_script.R   == 11131 == Memcheck, детектор ошибок памяти   == 11131 == Авторское право (C) 2002-2010 и GNU GPL'd от Julian Seward et al.   == 11131 == Использование Valgrind-3.6.0 и LibVEX; повторить с -h для информации об авторских правах   == 11131 == Команда:/usr/lib64/R/bin/exec/R -f test_script.R   == 11131 ==

R version 2.15.1 (2012-06-22) -- "Roasted Marshmallows"
Copyright (C) 2012 The R Foundation for Statistical Computing
ISBN 3-900051-07-0
Platform: x86_64-redhat-linux-gnu (64-bit)

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.

==11131== Conditional jump or move depends on uninitialised value(s)
==11131==    at 0x3A6A685F80: __GI___strcasecmp_l (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A61FF24: __gconv_open (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A62D3B7: _nl_find_msg (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A62DB83: __dcigettext (in /lib64/libc-2.12.so)
==11131==    by 0x3A6C3BD2DF: ??? (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x3A6C313FC8: setup_Rmainloop (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x3A6C315278: Rf_mainloop (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x40084A: main (in /usr/lib64/R/bin/exec/R)
==11131==
==11131== Use of uninitialised value of size 8
==11131==    at 0x3A6A6863A4: __GI___strcasecmp_l (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A61FF24: __gconv_open (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A62D3B7: _nl_find_msg (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A62DB83: __dcigettext (in /lib64/libc-2.12.so)
==11131==    by 0x3A6C3BD2DF: ??? (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x3A6C313FC8: setup_Rmainloop (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x3A6C315278: Rf_mainloop (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x40084A: main (in /usr/lib64/R/bin/exec/R)
==11131==
==11131== Use of uninitialised value of size 8
==11131==    at 0x3A6A6863A8: __GI___strcasecmp_l (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A61FF24: __gconv_open (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A62D3B7: _nl_find_msg (in /lib64/libc-2.12.so)
==11131==    by 0x3A6A62DB83: __dcigettext (in /lib64/libc-2.12.so)
==11131==    by 0x3A6C3BD2DF: ??? (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x3A6C313FC8: setup_Rmainloop (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x3A6C315278: Rf_mainloop (in /usr/lib64/R/lib/libR.so)
==11131==    by 0x40084A: main (in /usr/lib64/R/bin/exec/R)
==11131==
  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.

[Previously saved workspace restored]

> dyn.load("SchWolfenew.so")
> for(i in 1:1000){
+ .C("giveProb",as.double(2),as.double(2),as.double(c(0,1,0,1,1,0,1,0)))
+ }

==29371== Invalid read of size 1
==29371==    at 0x3A6C31A2E9: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CBC9: Rf_cons (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2D2B74: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DD121: Rf_eval (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DF830: Rf_applyClosure (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C32B828: Rf_usemethod (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C32BAE7: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2D290B: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DD121: Rf_eval (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DF830: Rf_applyClosure (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DD3F7: Rf_eval (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DEF4F: ??? (in /usr/lib64/R/lib/libR.so)
==29371==  Address 0x3ff0000000000003 is not stack'd, malloc'd or (recently) free'd
==29371==

 *** caught segfault ***
address (nil), cause 'unknown'
==29371== Invalid read of size 1
==29371==    at 0x3A6C31AF0B: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CBC9: Rf_cons (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CC71: Rf_allocList (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2C8CD4: R_GetTraceback (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C313472: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6B20F4FF: ??? (in /lib64/libpthread-2.12.so)
==29371==    by 0x3A6C31A2E8: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CBC9: Rf_cons (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2D2B74: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DD121: Rf_eval (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DF830: Rf_applyClosure (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C32B828: Rf_usemethod (in /usr/lib64/R/lib/libR.so)
==29371==  Address 0x4020000000000003 is not stack'd, malloc'd or (recently) free'd
==29371==
==29371==
==29371== Process terminating with default action of signal 11 (SIGSEGV)
==29371==  General Protection Fault
==29371==    at 0x3A6C31AF0B: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CBC9: Rf_cons (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CC71: Rf_allocList (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2C8CD4: R_GetTraceback (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C313472: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6B20F4FF: ??? (in /lib64/libpthread-2.12.so)
==29371==    by 0x3A6C31A2E8: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C31CBC9: Rf_cons (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2D2B74: ??? (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DD121: Rf_eval (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C2DF830: Rf_applyClosure (in /usr/lib64/R/lib/libR.so)
==29371==    by 0x3A6C32B828: Rf_usemethod (in /usr/lib64/R/lib/libR.so)
==29371==
==29371== HEAP SUMMARY:
==29371==     in use at exit: 29,307,724 bytes in 12,896 blocks
==29371==   total heap usage: 28,845 allocs, 15,949 frees, 48,495,252 bytes allocated
==29371==
==29371== LEAK SUMMARY:
==29371==    definitely lost: 0 bytes in 0 blocks
==29371==    indirectly lost: 0 bytes in 0 blocks
==29371==      possibly lost: 0 bytes in 0 blocks
==29371==    still reachable: 29,307,724 bytes in 12,896 blocks
==29371==         suppressed: 0 bytes in 0 blocks
==29371== Rerun with --leak-check=full to see details of leaked memory
==29371==
==29371== For counts of detected and suppressed errors, rerun with: -v
==29371== Use --track-origins=yes to see where uninitialised values come from
==29371== ERROR SUMMARY: 5 errors from 5 contexts (suppressed: 21 from 9)
Segmentation fault (core dumped)
4b9b3361

Ответ 1

Я поместил ваш C-код в файл memory.c и запустил R CMD SHLIB memory.c. Здесь мой тест script

dyn.load("/tmp/memory.so")
set.seed(123L)
while (TRUE)
    .C("giveProb",as.double(2),as.double(2), sample(c(0, 1), 8, TRUE))

И R -d valgrind -f test_script.R говорит

> dyn.load("/tmp/memory.so")
> set.seed(123L)
> while (TRUE)
+     .C("giveProb",as.double(2),as.double(2), sample(c(0, 1), 8, TRUE))
==3461== Invalid write of size 8
==3461==    at 0xBF29D78: expander.4631 (memory.c:35)
==3461==    by 0xBF29EBB: separator.4643 (memory.c:49)
==3461==    by 0xBF29AAB: giveProb (memory.c:108)
==3461==    by 0x4EEAF49: do_dotCode (dotcode.c:1689)
==3461==    by 0x4F1F4E2: Rf_eval (eval.c:493)
==3461==    by 0x4F218F6: do_for (eval.c:1310)
==3461==    by 0x4F1F2E8: Rf_eval (eval.c:467)
==3461==    by 0x4F6B5FB: Rf_ReplIteration (main.c:256)
==3461==    by 0x4F6B7B2: R_ReplConsole (main.c:305)
==3461==    by 0x4F6D022: run_Rmainloop (main.c:987)
==3461==    by 0x4F6D037: Rf_mainloop (main.c:994)
==3461==    by 0x400845: main (Rmain.c:32)

(плюс многое другое). строка 35 памяти.

newCoefArray[i+position]=coeff*leadingTerm*sumLeaders[i];

поэтому я + позиция больше, чем выделенный массив.