Re: monit and grsecurity

[Martin Pala]

It seems good - thanks :)

What does the other developers think about it? I'm +1 to add it to 4.0 
(this is sort of bugfix => it will not break code freeze stage much).

Martin

Arkadiusz Miskiewicz wrote:

> On Monday 08 of September 2003 18:15, Arkadiusz Miskiewicz wrote:
>  
>
> > I suggest adding /proc/meminfo parsing (from ie. procps) to check how many
> > ram we have instead of stating /proc/kcore before 4.0 release (which I'm
> > willing to test btw).
> >    
> Here is patch:
>
> diff -urN monit-4.0.org/process/sysdep_LINUX.c 
> monit-4.0/process/sysdep_LINUX.c
> --- monit-4.0.org/process/sysdep_LINUX.c        2003-09-08 19:20:44.000000000 
> +0200
> +++ monit-4.0/process/sysdep_LINUX.c    2003-09-08 19:20:23.000000000 +0200
> @@ -84,24 +84,37 @@
>  *  @file
>  */
>
> -#define PAGE_TO_KBYTE_SHIFT PAGE_SHIFT-10
> +#define FREEMEM         "MemFree:"
> +#define FREESWAP        "SwapFree:"
>
> int init_process_info_sysdep(void) {
>
> -  struct stat buf;
> -  
> +  int memfd;
> +  char buf[1024], *ptr1, *ptr2;
>
> -  /* I hope this is okay hack to get the total memsize. (-: */
> +  if ((memfd = open("/proc/meminfo", O_RDONLY)) == -1) {
> +      return FALSE;
> +  }
>
> -  if ( stat("/proc/kcore", &buf) != 0 ) {
> +  /* read the file */
> +  if (read(mem, buf, sizeof(buf)) < 0) {
> +      close(memfd);
> +      return FALSE;
> +  }
>
> -    return FALSE;
> +  close(memfd);
> +  
> +  ptr1 = strstr(buf, FREEMEM);
> +  ptr2 = strstr(buf, FREESWAP);
>
> +  if (!ptr1 || !ptr2) {
> +      return FALSE;
>   }
>
> -  num_cpus= sysconf(_SC_NPROCESSORS_CONF);
> +  /* we only care about integer values */
> +  mem_kbyte_max = atoi(ptr1+strlen(FREEMEM)) /* + atoi(ptr2+strlen(FREESWAP)) 
> */ ;
>
> -  mem_kbyte_max = buf.st_size>>10;
> +  num_cpus= sysconf(_SC_NPROCESSORS_CONF);
>
>   return TRUE;
>
>
>  
>
> > Thanks!
> >    
>
>