Re: Big Swap...

[Robert Dinse]

On Sat, 15 Jan 2000, Eirik Fuller wrote:
> 
> I saw your thread in the Linux kernel archive at tux.org; perhaps the
> answer to your question can be found in mm/swapfile.c in the kernel
> source.  The sys_swapon function returns -EINVAL in a number of
> places; most of them seem to have calls to printk nearby.  If one of
> those printk calls appears in the code which is returning EINVAL on
> your system, the dmesg command should allow you to see it.  I found
> three such returns with no printk, all of them near the nr_badpages
> code.

     I'm a little confused actually.  The fix was to change:

        pagesize = getpagesize();

		-TO-

        pagesize = PAGE_SIZE;

     ... in mkswap.c.

     That did make it work, but it seems to me to be fundamentally wrong.  If
getpagesize() is returning the wrong value, it would seem to me that really
that is what needs to be fixed if it is really incorrect, it returns a page
size of 8192 as opposed to 4096, the value in PAGE_SIZE, and this is on a Sun
4/670MP.