[2.4.9 SMP] alloc_pages failed

[2.4.9 SMP] alloc_pages failed

[Igor Mozetic]

My box is dual Xeon 550, Intel C440GX+, 2GB RAM, AIC7XXX.
Last night I got log full of the following:

Sep  2 04:00:46 jerolim kernel: __alloc_pages: 0-order allocation failed.
Sep  2 04:00:47 jerolim last message repeated 208 times
Sep  2 04:00:47 jerolim kernel: ed.
...
Sep  2 04:01:29 jerolim kernel: eth0: can't fill rx buffer (force 0)!
Sep  2 04:01:29 jerolim kernel: eth0: Tx ring dump,  Tx queue 4082238 / 4082238:
Sep  2 04:01:29 jerolim kernel: eth0:     0 200ca000.
...

-- Versions installed: (if some fields are empty or looks
-- unusual then possibly you have very old versions)
Linux jerolim 2.4.9 #1 SMP Fri Aug 17 11:33:44 CEST 2001 i686 unknown
Kernel modules         2.4.6
Gnu C                  2.95.2
Binutils               2.9.5.0.37
Linux C Library        2.1.3
Dynamic linker         ldd: version 1.9.11
Procps                 2.0.6
Mount                  2.10s
Net-tools              2.05
Console-tools          0.2.3
Sh-utils               2.0
Modules Loaded         

No crash or obvious problems, four long jobs are still running OK.
Full details if needed, please CC to me.

-Igor Mozetic

Re: [2.4.9 SMP] alloc_pages failed

[Daniel Phillips]

On September 2, 2001 03:14 pm, Igor Mozetic wrote:
> My box is dual Xeon 550, Intel C440GX+, 2GB RAM, AIC7XXX.
> Last night I got log full of the following:
> 
> Sep  2 04:00:46 jerolim kernel: __alloc_pages: 0-order allocation failed.
> Sep  2 04:00:47 jerolim last message repeated 208 times
> Sep  2 04:00:47 jerolim kernel: ed.

Could you please apply this patch so we can see which kind of allocation is
failing (patch -p0):

--- ../2.4.9.clean/mm/page_alloc.c	Thu Aug 16 12:43:02 2001
+++ ./mm/page_alloc.c	Wed Aug 29 23:47:39 2001
@@ -493,6 +493,9 @@
 		}
 
 		/* XXX: is pages_min/4 a good amount to reserve for this? */
+		if (z->free_pages < z->pages_min / 3 && (gfp_mask & __GFP_WAIT) &&
+				!(current->flags & PF_MEMALLOC))
+			continue;
 		if (z->free_pages < z->pages_min / 4 &&
 				!(current->flags & PF_MEMALLOC))
 			continue;


On the assumption it's an atomic allocation, chould you try this patch and 
see if it reduces the frequency:

--- 2.4.9.clean/mm/page_alloc.c	Thu Aug 16 12:43:02 2001
+++ 2.4.9/mm/page_alloc.c	Mon Aug 20 22:05:40 2001
@@ -502,7 +502,8 @@
 	}
 
 	/* No luck.. */
-	printk(KERN_ERR "__alloc_pages: %lu-order allocation failed.\n", order);
+	printk(KERN_ERR "__alloc_pages: %lu-order allocation failed 
(gfp=0x%x/%i).\n",
+		order, gfp_mask, !!(current->flags & PF_MEMALLOC));
 	return NULL;
 }
 

Re: [2.4.9 SMP] alloc_pages failed

[Alex Bligh - linux-kernel]

I think Daniel's two patches are the wrong way around. IE
the second (labelled as the first) is the instrumentation
to differentiate between different kinds of failure, and
the first (labelled as the second) is the attempt to
reserve extra pages for atomic allocations to reduce
0-order failures.

Alex Bligh

--On Sunday, 02 September, 2001 6:50 PM +0200 Daniel Phillips 
<phillips@bonn-fries.net> wrote:

> On September 2, 2001 03:14 pm, Igor Mozetic wrote:
>> My box is dual Xeon 550, Intel C440GX+, 2GB RAM, AIC7XXX.
>> Last night I got log full of the following:
>>
>> Sep  2 04:00:46 jerolim kernel: __alloc_pages: 0-order allocation failed.
>> Sep  2 04:00:47 jerolim last message repeated 208 times
>> Sep  2 04:00:47 jerolim kernel: ed.
>
> Could you please apply this patch so we can see which kind of allocation
> is failing (patch -p0):
>
> --- ../2.4.9.clean/mm/page_alloc.c	Thu Aug 16 12:43:02 2001
> +++ ./mm/page_alloc.c	Wed Aug 29 23:47:39 2001
> @@ -493,6 +493,9 @@
>  		}
>
>  		/* XXX: is pages_min/4 a good amount to reserve for this? */
> +		if (z->free_pages < z->pages_min / 3 && (gfp_mask & __GFP_WAIT) &&
> +				!(current->flags & PF_MEMALLOC))
> +			continue;
>  		if (z->free_pages < z->pages_min / 4 &&
>  				!(current->flags & PF_MEMALLOC))
>  			continue;
>
>
> On the assumption it's an atomic allocation, chould you try this patch
> and  see if it reduces the frequency:
>
> --- 2.4.9.clean/mm/page_alloc.c	Thu Aug 16 12:43:02 2001
> +++ 2.4.9/mm/page_alloc.c	Mon Aug 20 22:05:40 2001
> @@ -502,7 +502,8 @@
>  	}
>
>  	/* No luck.. */
> -	printk(KERN_ERR "__alloc_pages: %lu-order allocation failed.\n", order);
> +	printk(KERN_ERR "__alloc_pages: %lu-order allocation failed
> (gfp=0x%x/%i).\n",
> +		order, gfp_mask, !!(current->flags & PF_MEMALLOC));
>  	return NULL;
>  }
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>



--
Alex Bligh

Re: [2.4.9 SMP] alloc_pages failed

[Daniel Phillips]

On September 2, 2001 07:19 pm, Alex Bligh - linux-kernel wrote:
> I think Daniel's two patches are the wrong way around. IE
> the second (labelled as the first) is the instrumentation
> to differentiate between different kinds of failure, and
> the first (labelled as the second) is the attempt to
> reserve extra pages for atomic allocations to reduce
> 0-order failures.

Yes, indeed - cut & paste error.

--
Daniel

Re: [2.4.9 SMP] alloc_pages failed

[Daniel Phillips]

(reposted to correct cut & pasted errors)

On September 2, 2001 03:14 pm, Igor Mozetic wrote:
> My box is dual Xeon 550, Intel C440GX+, 2GB RAM, AIC7XXX.
> Last night I got log full of the following:
> 
> Sep  2 04:00:46 jerolim kernel: __alloc_pages: 0-order allocation failed.
> Sep  2 04:00:47 jerolim last message repeated 208 times
> Sep  2 04:00:47 jerolim kernel: ed.

Could you please apply this patch so we can see which kind of allocation is
failing (patch -p0):

--- 2.4.9.clean/mm/page_alloc.c	Thu Aug 16 12:43:02 2001
+++ 2.4.9/mm/page_alloc.c	Mon Aug 20 22:05:40 2001
@@ -502,7 +502,8 @@
 	}
 
 	/* No luck.. */
-	printk(KERN_ERR "__alloc_pages: %lu-order allocation failed.\n", order);
+	printk(KERN_ERR "__alloc_pages: %lu-order allocation failed (gfp=0x%x/%i).\n",
+		order, gfp_mask, !!(current->flags & PF_MEMALLOC));
 	return NULL;
 }
 

On the assumption it's an atomic allocation, chould you try this patch and 
see if it reduces the frequency:

--- ../2.4.9.clean/mm/page_alloc.c	Thu Aug 16 12:43:02 2001
+++ ./mm/page_alloc.c	Wed Aug 29 23:47:39 2001
@@ -493,6 +493,9 @@
 		}
 
 		/* XXX: is pages_min/4 a good amount to reserve for this? */
+		if (z->free_pages < z->pages_min / 3 && (gfp_mask & __GFP_WAIT) &&
+				!(current->flags & PF_MEMALLOC))
+			continue;
 		if (z->free_pages < z->pages_min / 4 &&
 				!(current->flags & PF_MEMALLOC))
 			continue;