Re: loop-AES 1.4d crypto swap => __alloc_pages: 0-order allocationfailed (gfp=0x20/0) from c882256e

[Jari Ruusu <jari.ruusu@pp.inet.fi>]

Hank Leininger wrote:
> This seems to help, but I still get oops's.  The "losetup; mke2fs" test
> of a 23GB partition now gets further than it did before--instead of
> dying while inode tables are written, it dies while writing superblocks
> / FS accounting information.  Interestingly this time it did *not* log a
> '0-order allocation' error first.  The oops was very similar, but not
> identical; here it is:

Can you try this patch then, against original v1.4d

    cd loop-AES-v1.4d
    patch -p0 < filename
    make

Also, add this line to your /etc/modules.conf

    options loop lo_prealloc=150

Regards,
Jari Ruusu <jari.ruusu@pp.inet.fi>


--- loop.c-2.4.diff.old	Sun Sep  2 21:33:01 2001
+++ loop.c-2.4.diff	Wed Sep 26 01:16:00 2001
@@ -261,12 +261,12 @@
 +        memset(bh, 0, sizeof(struct buffer_head));
 +        bh->b_page = p;
 +    } else {
-+        bh = kmem_cache_alloc(bh_cachep, SLAB_ATOMIC);
++        bh = 0; /* kmem_cache_alloc(bh_cachep, SLAB_ATOMIC); */
 +        if(!bh) {
 +            return((struct buffer_head *)0);
 +        }
 +        memset(bh, 0, sizeof(struct buffer_head));
-+        bh->b_page = alloc_page(GFP_ATOMIC);
++        bh->b_page = 0; /* alloc_page(GFP_ATOMIC); */
 +        if(!bh->b_page) {
 +            kmem_cache_free(bh_cachep, bh);
 +            return((struct buffer_head *)0);


Linux-crypto:  cryptography in and on the Linux system
Archive:       http://mail.nl.linux.org/linux-crypto/