linux-stable.git/include/net/snmp.h, branch linux-3.1.y

snmp: reduce percpu needs by 50%

2011-06-11T23:23:59+00:00

SNMP mibs use two percpu arrays, one used in BH context, another in USER
context. With increasing number of cpus in machines, and fact that ipv6
uses per network device ipstats_mib, this is consuming a lot of memory
if many network devices are registered.

commit be281e554e2a (ipv6: reduce per device ICMP mib sizes) shrinked
percpu needs for ipv6, but we can reduce memory use a bit more.

With recent percpu infrastructure (irqsafe_cpu_inc() ...), we no longer
need this BH/USER separation since we can update counters in a single
x86 instruction, regardless of the BH/USER context.

Other arches than x86 might need to disable irq in their
irqsafe_cpu_inc() implementation : If this happens to be a problem, we
can make SNMP_ARRAY_SZ arch dependent, but a previous poll
( https://lkml.org/lkml/2011/3/17/174 ) to arch maintainers did not
raise strong opposition.

Only on 32bit arches, we need to disable BH for 64bit counters updates
done from USER context (currently used for IP MIB)

This also reduces vmlinux size :

1) x86_64 build
$ size vmlinux.before vmlinux.after
   text	   data	    bss	    dec	    hex	filename
7853650	1293772	1896448	11043870	 a8841e	vmlinux.before
7850578	1293772	1896448	11040798	 a8781e	vmlinux.after

2) i386  build
$ size vmlinux.before vmlinux.afterpatch
   text	   data	    bss	    dec	    hex	filename
6039335	 635076	3670016	10344427	 9dd7eb	vmlinux.before
6037342	 635076	3670016	10342434	 9dd022	vmlinux.afterpatch

Signed-off-by: Eric Dumazet 
CC: Andi Kleen 
CC: Ingo Molnar 
CC: Tejun Heo 
CC: Christoph Lameter 
CC: Benjamin Herrenschmidt

ipv6: reduce per device ICMP mib sizes

2011-05-19T20:21:22+00:00

ipv6 has per device ICMP SNMP counters, taking too much space because
they use percpu storage.

needed size per device is :
(512+4)*sizeof(long)*number_of_possible_cpus*2

On a 32bit kernel, 16 possible cpus, this wastes more than 64kbytes of
memory per ipv6 enabled network device, taken in vmalloc pool.

Since ICMP messages are rare, just use shared counters (atomic_long_t)

Per network space ICMP counters are still using percpu memory, we might
also convert them to shared counters in a future patch.

Signed-off-by: Eric Dumazet 
CC: Denys Fedoryshchenko 
Signed-off-by: David S. Miller

snmp: SNMP_UPD_PO_STATS_BH() always called from softirq

2011-03-22T01:12:54+00:00

We dont need to test if we run from softirq context, we definitely are.

This saves few instructions in ip_rcv() & ip_rcv_finish()

Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller

net: kill unused macros from head file

2010-12-02T21:27:33+00:00

These macros have been defined for several years since v2.6.12-rc2（tracing by git）,
but never be used. So remove them.

Signed-off-by: Shan Wei 
Signed-off-by: David S. Miller

net: snmp: fix the wrong ICMP_MIB_MAX value

2010-12-02T21:27:31+00:00

__ICMP_MIB_MAX is equal to the total number of icmp mib,
So no need to add 1. This wastes 4/8 bytes memory.

Change it to be same as ICMP6_MIB_MAX, TCP_MIB_MAX, UDP_MIB_MAX.

Signed-off-by: Shan Wei 
Signed-off-by: David S. Miller

snmp: 64bit ipstats_mib for all arches

2010-06-30T20:31:19+00:00

/proc/net/snmp and /proc/net/netstat expose SNMP counters.

Width of these counters is either 32 or 64 bits, depending on the size
of "unsigned long" in kernel.

This means user program parsing these files must already be prepared to
deal with 64bit values, regardless of user program being 32 or 64 bit.

This patch introduces 64bit snmp values for IPSTAT mib, where some
counters can wrap pretty fast if they are 32bit wide.

# netstat -s|egrep "InOctets|OutOctets"
    InOctets: 244068329096
    OutOctets: 244069348848

Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller

snmp: fix SNMP_ADD_STATS()

2010-06-23T18:48:19+00:00

commit aa2ea0586d9d (tcp: fix outsegs stat for TSO segments) incorrectly
assumed SNMP_ADD_STATS() was used from BH context.

Fix this using mib[!in_softirq()] instead of mib[0]

Signed-off-by: Eric Dumazet 
CC: Tom Herbert 
Signed-off-by: David S. Miller

tcp: fix outsegs stat for TSO segments

2010-04-22T23:00:00+00:00

Account for TSO segments of an skb in TCP_MIB_OUTSEGS counter.  Without
doing this, the counter can be off by orders of magnitude from the
actual number of segments sent.

Signed-off-by: Tom Herbert 
Signed-off-by: David S. Miller

net: snmp mib cleanup

2010-03-22T01:34:16+00:00

There is no point to align or pad mibs to cache lines, they are per cpu
allocated with a 8 bytes alignment anyway.
This wastes space for no gain. This patch removes __SNMP_MIB_ALIGN__

Since SNMP mibs contain "unsigned long" fields only, we can relax the
allocation alignment from "unsigned long long" to "unsigned long"

Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller

percpu: add __percpu sparse annotations to net

2010-02-17T07:05:38+00:00

Add __percpu sparse annotations to net.

These annotations are to make sparse consider percpu variables to be
in a different address space and warn if accessed without going
through percpu accessors.  This patch doesn't affect normal builds.

The macro and type tricks around snmp stats make things a bit
interesting.  DEFINE/DECLARE_SNMP_STAT() macros mark the target field
as __percpu and SNMP_UPD_PO_STATS() macro is updated accordingly.  All
snmp_mib_*() users which used to cast the argument to (void **) are
updated to cast it to (void __percpu **).

Signed-off-by: Tejun Heo 
Acked-by: David S. Miller 
Cc: Patrick McHardy 
Cc: Arnaldo Carvalho de Melo 
Cc: Vlad Yasevich 
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller