Re: tracking shared dirty pages in a mmap file

[Neil Horman <nhorman@tuxdriver.com>]

On Sun, Aug 27, 2006 at 12:37:52PM +1000, Kandan Venkataraman wrote:
> Hi
> I want to track any pages that have been written to a mmap file using a
> segv handler.
> 
> I have given an example program below to illustrate my idea.
> 
> 
> What I want to know is whether the modifications to the mmap file will
> be guaranteed to be unaffected by the interruption of the signal
> handler, i.e. would the resultant changes in the mmap file be as if the
> mmap file had write access set.
> 
> 
> It can also be assumped that only the current process would be writing
> to the mmap file.
> 
> 
> Thanks
> Kandan
> 
Can't you accomplish the same goal, in a much more portable way using inotify?

Neil

> 
> #include <stdio.h>
> #include <stdlib.h>
> #include <sys/types.h>
> #include <unistd.h>
> #include <sys/wait.h>
> #include <sys/stat.h>
> #include <sys/mman.h>
> #include <fcntl.h>
> #include <string.h>
> #include <assert.h>
> #include <signal.h>
> 
> 
> typedef struct TwoInts_tag {
>    int i;
>    int j;
> 
> 
> 
> } TwoInts;
> 
> 
> const int maxPages = 10;
> 
> char* start = 0;
> int fd;
> TwoInts *array = 0;
> int pageSize;
> int max;
> int elemsPerPage;
> 
> 
> void memory_fault(int signum, siginfo_t *info, void *ptr)
> {
>    char *faultAddr  = (char *)info->si_addr;
> 
> 
>    /* we only handle the faults within our known range , anything else
> is really an illegal access */
>    if (faultAddr >= start && faultAddr < start + (pageSize * maxPages))
> {
> 
> 
>       int faultPageIdx = (faultAddr - start)/pageSize;
> 
> 
>       fprintf(stderr, "fault at page index %d\n", faultPageIdx);
> 
> 
>       if (mprotect(start + (faultPageIdx * pageSize), pageSize,
> PROT_READ | PROT_WRITE)) {
>          perror("mprotect failed");
>          abort();
>       }
>    }
>    else
>       abort();
> 
> 
> 
> }
> 
> 
> int main()
> {
>    struct sigaction act;
> 
>    act.sa_sigaction = memory_fault;
>    act.sa_flags = SA_SIGINFO;
>    sigemptyset (&act.sa_mask);
> 
> 
>    int status = sigaction (SIGSEGV, &act, NULL);
>    if (status) {
>       perror ("sigaction");
>       exit (1);
>    }
> 
> 
>    pageSize = getpagesize();
> 
> 
>    elemsPerPage = pageSize/sizeof(TwoInts);
> 
> 
>    max = elemsPerPage * maxPages;
> 
> 
>    fprintf(stderr, "max elements is %d\n", max);
> 
> 
>    assert(pageSize % sizeof(TwoInts) == 0);
> 
> 
>    /* open/create the file */
>    if ((fd = open ("kandan", O_RDWR | O_CREAT | O_TRUNC, S_IRWXU)) < 0)
> {
>       fprintf(stderr, "can't create file for writing");
>       exit(1);
>    }
> 
> 
>    /* set the lenght of file */
>    if (ftruncate(fd, maxPages * pageSize) == -1) {
>       fprintf(stderr, "error on ftruncate\n");
>       exit(1);
>    }
> 
> 
>    /* mmap the file with read access*/
> 
> 
>    if ((start = mmap(0, maxPages * pageSize, PROT_READ, MAP_SHARED, fd,
> 0)) == MAP_FAILED) {
> 
> 
>       perror("mmap error");
>       exit(1);
>    }
> 
> 
>    array = (TwoInts *)start;
> 
> 
>    array[0].i = 5;
> 
> 
>    fprintf(stderr, "value = %d\n", array[0].i);
> 
> 
>    array[0].i = 7;
> 
> 
>    fprintf(stderr, "value = %d\n", array[0].i);
> 
> 
>    array[1].i = 9;
> 
> 
>    fprintf(stderr, "value = %d\n", array[1].i);
> 
> 
>    array[elemsPerPage].i = 14;
> 
> 
>    fprintf(stderr, "value = %d\n", array[elemsPerPage].i);
> 
> 
>    close(fd);
> 
> 
>    return 0;
> 
> 
> }
> -
> To unsubscribe from this list: send the line "unsubscribe 
> linux-c-programming" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 
/***************************************************
 *Neil Horman
 *Software Engineer
 *gpg keyid: 1024D / 0x92A74FA1 - http://pgp.mit.edu
 ***************************************************/