Sharyathi Nagesh wrote:

Dave
        Please go through my observation.
>
> That makes the most sense at this point. This is what I was
> planning for vm_init():
>
>
> @@ -694,7 +708,8 @@ vm_init(void)
>                         vt->PG_slab = 10;
>                 else if (THIS_KERNEL_VERSION >= LINUX(2,6,0))
>                         vt->PG_slab = 7;
> -       }
> +       } else if (THIS_KERNEL_VERSION >= LINUX(2,6,0))
> +               vt->PG_slab = 7;
> }
here for the dump I was looking into vt->PG_slab needs to be set to 10
and not 7. as slab value was 1024 in dump_mem_map() function.

What kernel version are you testing with?

> /*
>
> >
> >
> >
> > o The output is breaking as totalhigh_pages is coming out zero,
> > So part of code in function dump_kmeminfo()
> > pct = freehighmem_pages ?
> >                         (freehighmem_pages * 100)/totalhigh_pages :
> > 0;
> > Cause Arithmetic calculation error.
> >
> > Still looking into the problem.
> >
> I can't reproduce this one. In your case though, with
> about 500MB, both "freehighmem_pages" and "totalhigh_pages"
> should both be zero, leading to a "pct" setting of 0.
> Are you showing "freehighmem_pages" as non-zero?
>
> Dave
I could see totalhigh_pages coming out zero which was the right thing
(As it was x86_64 machine) but freehighmem_pages came out as a non zero
quantity.
In function :dump_free_pages_zones_v2(). I could see vt->nr_zones as 4
which I feel is incorrect and hence it entered the block
        if (i == ZONE_HIGHMEM) {
                                readmem(node_zones+
                                       OFFSET(zone_free_pages),
                                             KVADDR, &value,
sizeof(ulong),
                                             "node_zones free_pages",
                                             FAULT_ON_ERROR);
                                      sum += value;
                                }
and sum was updated with a value and hence freehighmem_pages causing
Arithmetic exception

Ah -- the value of ZONE_HIGHMEM can no longer be fixed since
it's been changed from 3 to 4. It should be based upon
the array size of the pglist_data.node_zones[] minus 1.
Since we already save this array index at init-time,
I'll make ZONE_HIGHMEM a variable in the vm_table.
The other two definitions (ZONE_DMA and ZONE_LOW) aren't
used).

Thanks,
Dave