Re: [RFC] solution for the inet_ntoa problem, buffer allocator

From: Olaf Titz (olaf@bigred.inka.de)
Date: Sat Jul 08 2000 - 05:51:39 EST


> > sometimes use sort of a round-robbin buffer, shared by several
> > functions. This way, I can call the same function several times
> > without allocating memory, and the number of calls is just limited
> > by the total size. Here comes an example (from scratch, completely
> > untested).
> >
> > Any comments ?
>
> Not threadsafe.

This one is threadsafe (in the limit given by the number of buffers,
you probably want to configure that depending on the number of CPUs).

#define NTOABUFS 8
static char ntoabuf[NTOABUFS][16];
static int ntoaptr=0;
spinlock_t cipe_ntoalock=SPIN_LOCK_UNLOCKED;

const char *cipe_ntoa(const __u32 addr)
{
    const unsigned char *x=(const unsigned char *)&addr;
    char *p;
    int b, i;
    unsigned long flags;

    spin_lock_irqsave(&cipe_ntoalock, flags);
    b=ntoaptr;
    if (++b>=NTOABUFS)
        b=0;
    ntoaptr=b;
    spin_unlock_irqrestore(&cipe_ntoalock, flags);

    p=ntoabuf[b];
    for (i=0; i<4; ++i) {
        int k=x[i]/100;
        if (k)
            *p++=k+'0';
        k=(x[i]/10)%10;
        if (k)
            *p++=k+'0';
        *p++=(x[i]%10)+'0';
        if (i<3)
            *p++='.';
    }
    *p='\0';
    return ntoabuf[b];
}

Olaf

PS. Why does gcc (2.95.2) say "variable defined but not used" for
cipe_ntoalock when I make it static?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sat Jul 15 2000 - 21:00:08 EST