Re: 0-order allocation failed in 2.4.10-pre8

[Jens Axboe]

On Fri, Sep 14 2001, kelley eicher wrote:
> On Fri, 14 Sep 2001, J wrote:
> 
> > Hello,
> >     I am reporting the same problem that kelley eicher
> > has, "0-order allocation failed (gfp=0x70/1)", hopefully
> > my info is helpfull. I added BUG() after the printk's in
> > "mm/page_alloc.c:505". I am able to get this message
> > when ever I copy a large (2+ gig) file from one XFS filesystem
> > to another on the same disk controller. I have attached klogd -p output
> > and an output of /proc/slabinfo from <= 1 second before the
> > Oops. If anyone would like more info, or for me to apply dangerous
> > patches, or to shut up even, let me know.
> >
> > Jack
> >
> > Machine:
> > Kernel Version 2.4.10-pre8 with SGI-XFS patches and IBM-JFS patches.
> > compiled with egcs 2.91-66 ;) (Some more info @ http://whatevr.i2net.com)
> > SMP P3 Box with 2 Gig of memory
> 
> after spending a few dayz trying to figure out where this is happening, i
> noticed that the alloc_pages() errors only occur after used memory goes
> above 899MB. this is the limit of physical memory unless you enable the
> himem option. the machines i had been seeing this on all had 1G+ memory
> and i had enabled the 4G himem option on each of them. so turn that option
> off and you will no longer see alloc_pages errors. you'll have to suffer
> through only having 900MB of memory to play with though. ;>

Use the

*.kernel.org/pub/linux/kernel/people/axboe/patches/2.4.9/block-highmem-all

patch and you can use highmem without having to worry about failed
0-order bounce pages allocations.

-- 
Jens Axboe