* [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
@ 2012-11-15 23:41 Eric Dumazet
2012-11-15 23:55 ` Joe Perches
2012-11-16 6:39 ` David Miller
0 siblings, 2 replies; 8+ messages in thread
From: Eric Dumazet @ 2012-11-15 23:41 UTC (permalink / raw)
To: David Miller; +Cc: netdev, Julien Tinnes
From: Eric Dumazet <edumazet@google.com>
order-5 allocations can fail with current kernels, we should
try to reduce allocation sizes to allow network namespace
creation.
Reported-by: Julien Tinnes <jln@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
net/ipv4/tcp_metrics.c | 13 ++++++++-----
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index 53bc584..15e93c4 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -1030,14 +1030,17 @@ static int __net_init tcp_net_metrics_init(struct net *net)
else
slots = 8 * 1024;
}
-
+retry:
net->ipv4.tcp_metrics_hash_log = order_base_2(slots);
size = sizeof(struct tcpm_hash_bucket) << net->ipv4.tcp_metrics_hash_log;
- net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL);
- if (!net->ipv4.tcp_metrics_hash)
- return -ENOMEM;
-
+ net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL | __GFP_NOWARN);
+ if (!net->ipv4.tcp_metrics_hash) {
+ if (slots <= 16)
+ return -ENOMEM;
+ slots >>= 1;
+ goto retry;
+ }
return 0;
}
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-15 23:41 [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures Eric Dumazet
@ 2012-11-15 23:55 ` Joe Perches
2012-11-16 0:47 ` Eric Dumazet
2012-11-16 6:39 ` David Miller
1 sibling, 1 reply; 8+ messages in thread
From: Joe Perches @ 2012-11-15 23:55 UTC (permalink / raw)
To: Eric Dumazet; +Cc: David Miller, netdev, Julien Tinnes
On Thu, 2012-11-15 at 15:41 -0800, Eric Dumazet wrote:
> From: Eric Dumazet <edumazet@google.com>
>
> order-5 allocations can fail with current kernels, we should
> try to reduce allocation sizes to allow network namespace
> creation.
>
> Reported-by: Julien Tinnes <jln@google.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> ---
> net/ipv4/tcp_metrics.c | 13 ++++++++-----
> 1 file changed, 8 insertions(+), 5 deletions(-)
>
> diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
> index 53bc584..15e93c4 100644
> --- a/net/ipv4/tcp_metrics.c
> +++ b/net/ipv4/tcp_metrics.c
> @@ -1030,14 +1030,17 @@ static int __net_init tcp_net_metrics_init(struct net *net)
> else
> slots = 8 * 1024;
> }
> -
> +retry:
> net->ipv4.tcp_metrics_hash_log = order_base_2(slots);
> size = sizeof(struct tcpm_hash_bucket) << net->ipv4.tcp_metrics_hash_log;
>
> - net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL);
> - if (!net->ipv4.tcp_metrics_hash)
> - return -ENOMEM;
> -
> + net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL | __GFP_NOWARN);
> + if (!net->ipv4.tcp_metrics_hash) {
> + if (slots <= 16)
> + return -ENOMEM;
maybe readd the warning for OOM reporting here?
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-15 23:55 ` Joe Perches
@ 2012-11-16 0:47 ` Eric Dumazet
0 siblings, 0 replies; 8+ messages in thread
From: Eric Dumazet @ 2012-11-16 0:47 UTC (permalink / raw)
To: Joe Perches; +Cc: David Miller, netdev, Julien Tinnes
On Thu, 2012-11-15 at 15:55 -0800, Joe Perches wrote:
> maybe readd the warning for OOM reporting here?
>
Storing tcp metrics is a best effort, this can be silently ignored.
And order-0 are going to work, unless machine is in critical state,
that gives enough room.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-15 23:41 [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures Eric Dumazet
2012-11-15 23:55 ` Joe Perches
@ 2012-11-16 6:39 ` David Miller
2012-11-16 15:31 ` Eric Dumazet
1 sibling, 1 reply; 8+ messages in thread
From: David Miller @ 2012-11-16 6:39 UTC (permalink / raw)
To: eric.dumazet; +Cc: netdev, jln
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 15 Nov 2012 15:41:04 -0800
> From: Eric Dumazet <edumazet@google.com>
>
> order-5 allocations can fail with current kernels, we should
> try to reduce allocation sizes to allow network namespace
> creation.
>
> Reported-by: Julien Tinnes <jln@google.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
Indeed, this has to be done better.
But this kind of retry solution results in non-deterministic behavior.
Yes the tcp metrics cache is best effort, but it's size can influence
behavior in a substantial way depending upon the workload.
I would suggest that we instead use different limits, ones which the
page allocator will satisfy for us always with GFP_KERNEL.
1) include linux/mmzone.h
2) Make the two limits based upon PAGE_ALLOC_COSTLY_ORDER.
That is, make the larger table size PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER
and the smaller one PAGE_SIZE << (PAGE_ALLOC_COSTLY_ORDER - 1).
How about something like this?
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index 53bc584..d4b2d42 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -1,7 +1,6 @@
#include <linux/rcupdate.h>
#include <linux/spinlock.h>
#include <linux/jiffies.h>
-#include <linux/bootmem.h>
#include <linux/module.h>
#include <linux/cache.h>
#include <linux/slab.h>
@@ -9,6 +8,7 @@
#include <linux/tcp.h>
#include <linux/hash.h>
#include <linux/tcp_metrics.h>
+#include <linux/mmzone.h>
#include <net/inet_connection_sock.h>
#include <net/net_namespace.h>
@@ -1025,10 +1025,12 @@ static int __net_init tcp_net_metrics_init(struct net *net)
slots = tcpmhash_entries;
if (!slots) {
- if (totalram_pages >= 128 * 1024)
- slots = 16 * 1024;
- else
- slots = 8 * 1024;
+ int order = PAGE_ALLOC_COSTLY_ORDER;
+
+ if (totalram_pages < 128 * 1024)
+ order--;
+ slots = (PAGE_SIZE << order) /
+ sizeof(struct tcpm_hash_bucket);
}
net->ipv4.tcp_metrics_hash_log = order_base_2(slots);
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-16 6:39 ` David Miller
@ 2012-11-16 15:31 ` Eric Dumazet
2012-11-16 18:37 ` David Miller
2012-11-16 18:51 ` Julien Tinnes
0 siblings, 2 replies; 8+ messages in thread
From: Eric Dumazet @ 2012-11-16 15:31 UTC (permalink / raw)
To: David Miller; +Cc: netdev, jln
On Fri, 2012-11-16 at 01:39 -0500, David Miller wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Thu, 15 Nov 2012 15:41:04 -0800
>
> > From: Eric Dumazet <edumazet@google.com>
> >
> > order-5 allocations can fail with current kernels, we should
> > try to reduce allocation sizes to allow network namespace
> > creation.
> >
> > Reported-by: Julien Tinnes <jln@google.com>
> > Signed-off-by: Eric Dumazet <edumazet@google.com>
>
> Indeed, this has to be done better.
>
> But this kind of retry solution results in non-deterministic behavior.
> Yes the tcp metrics cache is best effort, but it's size can influence
> behavior in a substantial way depending upon the workload.
>
> I would suggest that we instead use different limits, ones which the
> page allocator will satisfy for us always with GFP_KERNEL.
>
> 1) include linux/mmzone.h
>
> 2) Make the two limits based upon PAGE_ALLOC_COSTLY_ORDER.
>
> That is, make the larger table size PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER
> and the smaller one PAGE_SIZE << (PAGE_ALLOC_COSTLY_ORDER - 1).
Well, we dont really know what the size needs to be, and your proposal
reduces the size by a 4 factor, even for the initial namespace.
Julien report was about Chrome browser own netns, on a suspend/resume
cycle (or something like that)
If size can influence behavior, we could try a vmalloc() if kmalloc()
fails...
Thanks
[PATCH v3] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
order-5 allocations can fail with current kernels, we should
try vmalloc() as well.
Reported-by: Julien Tinnes <jln@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
net/ipv4/tcp_metrics.c | 12 +++++++++---
1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index 53bc584..f696d7c 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -1,7 +1,6 @@
#include <linux/rcupdate.h>
#include <linux/spinlock.h>
#include <linux/jiffies.h>
-#include <linux/bootmem.h>
#include <linux/module.h>
#include <linux/cache.h>
#include <linux/slab.h>
@@ -9,6 +8,7 @@
#include <linux/tcp.h>
#include <linux/hash.h>
#include <linux/tcp_metrics.h>
+#include <linux/vmalloc.h>
#include <net/inet_connection_sock.h>
#include <net/net_namespace.h>
@@ -1034,7 +1034,10 @@ static int __net_init tcp_net_metrics_init(struct net *net)
net->ipv4.tcp_metrics_hash_log = order_base_2(slots);
size = sizeof(struct tcpm_hash_bucket) << net->ipv4.tcp_metrics_hash_log;
- net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL);
+ net->ipv4.tcp_metrics_hash = kzalloc(size, GFP_KERNEL | __GFP_NOWARN);
+ if (!net->ipv4.tcp_metrics_hash)
+ net->ipv4.tcp_metrics_hash = vzalloc(size);
+
if (!net->ipv4.tcp_metrics_hash)
return -ENOMEM;
@@ -1055,7 +1058,10 @@ static void __net_exit tcp_net_metrics_exit(struct net *net)
tm = next;
}
}
- kfree(net->ipv4.tcp_metrics_hash);
+ if (is_vmalloc_addr(net->ipv4.tcp_metrics_hash))
+ vfree(net->ipv4.tcp_metrics_hash);
+ else
+ kfree(net->ipv4.tcp_metrics_hash);
}
static __net_initdata struct pernet_operations tcp_net_metrics_ops = {
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-16 15:31 ` Eric Dumazet
@ 2012-11-16 18:37 ` David Miller
2012-11-16 18:51 ` Julien Tinnes
1 sibling, 0 replies; 8+ messages in thread
From: David Miller @ 2012-11-16 18:37 UTC (permalink / raw)
To: eric.dumazet; +Cc: netdev, jln
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Fri, 16 Nov 2012 07:31:53 -0800
> Well, we dont really know what the size needs to be, and your proposal
> reduces the size by a 4 factor, even for the initial namespace.
>
> Julien report was about Chrome browser own netns, on a suspend/resume
> cycle (or something like that)
>
> If size can influence behavior, we could try a vmalloc() if kmalloc()
> fails...
Agreed.
> [PATCH v3] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
>
> order-5 allocations can fail with current kernels, we should
> try vmalloc() as well.
>
> Reported-by: Julien Tinnes <jln@google.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
This looks great, applied, thanks.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-16 15:31 ` Eric Dumazet
2012-11-16 18:37 ` David Miller
@ 2012-11-16 18:51 ` Julien Tinnes
2012-11-16 19:08 ` Eric Dumazet
1 sibling, 1 reply; 8+ messages in thread
From: Julien Tinnes @ 2012-11-16 18:51 UTC (permalink / raw)
To: Eric Dumazet; +Cc: David Miller, netdev
On Fri, Nov 16, 2012 at 7:31 AM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> On Fri, 2012-11-16 at 01:39 -0500, David Miller wrote:
>> From: Eric Dumazet <eric.dumazet@gmail.com>
>> Date: Thu, 15 Nov 2012 15:41:04 -0800
>>
>> > From: Eric Dumazet <edumazet@google.com>
>> >
>> > order-5 allocations can fail with current kernels, we should
>> > try to reduce allocation sizes to allow network namespace
>> > creation.
>> >
>> > Reported-by: Julien Tinnes <jln@google.com>
>> > Signed-off-by: Eric Dumazet <edumazet@google.com>
>>
>> Indeed, this has to be done better.
>>
>> But this kind of retry solution results in non-deterministic behavior.
>> Yes the tcp metrics cache is best effort, but it's size can influence
>> behavior in a substantial way depending upon the workload.
>>
>> I would suggest that we instead use different limits, ones which the
>> page allocator will satisfy for us always with GFP_KERNEL.
>>
>> 1) include linux/mmzone.h
>>
>> 2) Make the two limits based upon PAGE_ALLOC_COSTLY_ORDER.
>>
>> That is, make the larger table size PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER
>> and the smaller one PAGE_SIZE << (PAGE_ALLOC_COSTLY_ORDER - 1).
>
> Well, we dont really know what the size needs to be, and your proposal
> reduces the size by a 4 factor, even for the initial namespace.
>
> Julien report was about Chrome browser own netns, on a suspend/resume
> cycle (or something like that)
It happens when users start Chrome. Chrome will create one new network
NS (for the sandbox).
This has been used for a few years now, but we had our first report in
January of this year and we've been getting a few reports very
recently at a rate that is starting to worry me (crbug.com/110756).
Thanks a lot for helping with this!
Julien
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures
2012-11-16 18:51 ` Julien Tinnes
@ 2012-11-16 19:08 ` Eric Dumazet
0 siblings, 0 replies; 8+ messages in thread
From: Eric Dumazet @ 2012-11-16 19:08 UTC (permalink / raw)
To: Julien Tinnes; +Cc: David Miller, netdev
On Fri, 2012-11-16 at 10:51 -0800, Julien Tinnes wrote:
> It happens when users start Chrome. Chrome will create one new network
> NS (for the sandbox).
>
> This has been used for a few years now, but we had our first report in
> January of this year and we've been getting a few reports very
> recently at a rate that is starting to worry me (crbug.com/110756).
>
> Thanks a lot for helping with this!
Thanks for bringing this issue to our attention !
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2012-11-16 19:08 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-11-15 23:41 [PATCH] tcp: handle tcp_net_metrics_init() order-5 memory allocation failures Eric Dumazet
2012-11-15 23:55 ` Joe Perches
2012-11-16 0:47 ` Eric Dumazet
2012-11-16 6:39 ` David Miller
2012-11-16 15:31 ` Eric Dumazet
2012-11-16 18:37 ` David Miller
2012-11-16 18:51 ` Julien Tinnes
2012-11-16 19:08 ` Eric Dumazet
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.