add ramdisk sizing,

[mothran]

ecfs SEGV's when the dump has more than 4 gigs of memory mapped in, and in the case of a large heap of OOM condition ecfs fails to collect the process. Maybe add the physical max mem to be the max size of the ramdisk to enable collection of any possible process.

[mothran]

I was informed that some systems more recently (such as ec2: http://www.rackspace.com/knowledge_center/article/swap-space-on-cloud-servers) do not have swaps configured by default. Maybe if the core is going to be larger the addressable space the ramdisk can be bypassed and direct disk writes can be done.

[elfmaster]

That's an astute point. I have this about 70% fixed, I added a function:

int inquire_meminfo(void)
{
FILE *fp;
char *s1 = alloca(64);
char *s2 = alloca(64);
size_t kbytes;
size_t gbytes;

    fp = fopen("/proc/meminfo", "r");
    if (fp == NULL) {       
            perror("fopen");
            return -1;
    }
    fscanf(fp, "%s %u %s", s1, &kbytes, s2);
    fclose(fp);
    if (kbytes < 1048576)
            return 0;
    gbytes = kbytes / 1024 / 1024;
    return gbytes;

}

Basically first checks to see if there's less than 1GB of ram, if so then it returns 0-- that way I can say hey lets not even use a ramdisk. If there's 1GB or more of ram, then we return that amount in gigabytes

I haven't yet added anything to equate how large the core will be, and if its larger than max ram then use real disk. But I will, thanks.