[Crash-utility] problem with crash on upstream kernel cores

[Dave Anderson]

Josef Whiter wrote:

> On Fri, Feb 23, 2007 at 04:19:37PM -0500, Dave Anderson wrote:
> > Josef Whiter wrote:
> >
> > > On Fri, Feb 23, 2007 at 04:00:09PM -0500, Dave Anderson wrote:
> > > > if you do this with your kernel:
> > > >
> > > >   # gdb vmlinux
> > > >   ...
> > > >   (gdb) ptype struct zone
> > > >
> > > > What do you see?
> > > >
> > >
> > > (gdb) ptype struct zone
> > > type = struct zone {
> > >     long unsigned int pages_min;
> > >     long unsigned int pages_low;
> > >     long unsigned int pages_high;
> > >     long unsigned int lowmem_reserve[3];
> > >     struct per_cpu_pageset pageset[32];
> > >     spinlock_t lock;
> > >     struct free_area free_area[11];
> > >     struct zone_padding _pad1_;
> > >     spinlock_t lru_lock;
> > >     struct list_head active_list;
> > >     struct list_head inactive_list;
> > >     long unsigned int nr_scan_active;
> > >     long unsigned int nr_scan_inactive;
> > >     long unsigned int pages_scanned;
> > >     int all_unreclaimable;
> > >     atomic_t reclaim_in_progress;
> > >     atomic_long_t vm_stat[14];
> > >     int prev_priority;
> > >     struct zone_padding _pad2_;
> > >     wait_queue_head_t *wait_table;
> > >     long unsigned int wait_table_hash_nr_entries;
> > >     long unsigned int wait_table_bits;
> > >     struct pglist_data *zone_pgdat;
> > >     long unsigned int zone_start_pfn;
> > >     long unsigned int spanned_pages;
> > >     long unsigned int present_pages;
> > >     const char *name;
> > > }
> > >
> > > It looks like free_pages has been removed altogether.
> >
> > That appears to be a very recent change -- 2.6.20.1 still has it:
> >
> > struct zone {
> >         /* Fields commonly accessed by the page allocator */
> >         unsigned long           free_pages;
> >         unsigned long           pages_min, pages_low, pages_high;
> >         ...
> >
> > What version are you using?  Are you running 2.6.21-rc1?
> >
>
> No, I pulled linus's git tree down maybe a week ago, so its close to -rc1.
>
> Josef

Ok -- it should be easy enough to fix, the free_pages count looks to
have been moved into the zone's embedded vm_stat[] array.

Thanks for reporting this.

Dave