Merge branch 'net_proc_perf'

Jia He says:

====================
Reduce cache miss for snmp_fold_field

In a PowerPc server with large cpu number(160), besides commit
a3a773726c9f ("net: Optimize snmp stat aggregation by walking all
the percpu data at once"), I watched several other snmp_fold_field
callsites which would cause high cache miss rate.

test source code:
================
My simple test case, which read from the procfs items endlessly:
/***********************************************************/
int main(int argc, char **argv)
{
        int i;
        int fd = -1 ;
        int rdsize = 0;
        char buf[LINELEN+1];

        buf[LINELEN] = 0;
        memset(buf,0,LINELEN);

        if(1 >= argc) {
                printf("file name empty\n");
                return -1;
        }

        fd = open(argv[1], O_RDWR, 0644);
        if(0 > fd){
                printf("open error\n");
                return -2;
        }

        for(i=0;i<0xffffffff;i++) {
                while(0 < (rdsize = read(fd,buf,LINELEN))){
                        //nothing here
                }

                lseek(fd, 0, SEEK_SET);
        }

        close(fd);
        return 0;
}
/**********************************************************/

compile and run:
================
gcc test.c -o test

perf stat -d -e cache-misses ./test /proc/net/snmp
perf stat -d -e cache-misses ./test /proc/net/snmp6
perf stat -d -e cache-misses ./test /proc/net/sctp/snmp
perf stat -d -e cache-misses ./test /proc/net/xfrm_stat

before the patch set:
====================
 Performance counter stats for 'system wide':

         355911097      cache-misses                                                 [40.08%]
        2356829300      L1-dcache-loads                                              [60.04%]
         355642645      L1-dcache-load-misses     #   15.09% of all L1-dcache hits   [60.02%]
         346544541      LLC-loads                                                    [59.97%]
            389763      LLC-load-misses           #    0.11% of all LL-cache hits    [40.02%]

       6.245162638 seconds time elapsed

After the patch set:
===================
 Performance counter stats for 'system wide':

         194992476      cache-misses                                                 [40.03%]
        6718051877      L1-dcache-loads                                              [60.07%]
         194871921      L1-dcache-load-misses     #    2.90% of all L1-dcache hits   [60.11%]
         187632232      LLC-loads                                                    [60.04%]
            464466      LLC-load-misses           #    0.25% of all LL-cache hits    [39.89%]

       6.868422769 seconds time elapsed
The cache-miss rate can be reduced from 15% to 2.9%

changelog
=========
v6:
- correct v5
v5:
- order local variables from longest to shortest line
v4:
- move memset into one block of if statement in snmp6_seq_show_item
- remove the changes in netstat_seq_show considerred the stack usage is too large
v3:
- introduce generic interface (suggested by Marcelo Ricardo Leitner)
- use max_t instead of self defined macro (suggested by David Miller)
v2:
- fix bug in udplite statistics.
- snmp_seq_show is split into 2 parts
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

2 files changed, 28 insertions, 14 deletions

diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 2f1f5d439788..35d4baa55c9d 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -4961,18 +4961,18 @@ static inline size_t inet6_if_nlmsg_size(void)
 }
 
 static inline void __snmp6_fill_statsdev(u64 *stats, atomic_long_t *mib,
-				      int items, int bytes)
+					int bytes)
 {
 	int i;
-	int pad = bytes - sizeof(u64) * items;
+	int pad = bytes - sizeof(u64) * ICMP6_MIB_MAX;
 	BUG_ON(pad < 0);
 
 	/* Use put_unaligned() because stats may not be aligned for u64. */
-	put_unaligned(items, &stats[0]);
-	for (i = 1; i < items; i++)
+	put_unaligned(ICMP6_MIB_MAX, &stats[0]);
+	for (i = 1; i < ICMP6_MIB_MAX; i++)
 		put_unaligned(atomic_long_read(&mib[i]), &stats[i]);
 
-	memset(&stats[items], 0, pad);
+	memset(&stats[ICMP6_MIB_MAX], 0, pad);
 }
 
 static inline void __snmp6_fill_stats64(u64 *stats, void __percpu *mib,
@@ -5005,7 +5005,7 @@ static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
 				     offsetof(struct ipstats_mib, syncp));
 		break;
 	case IFLA_INET6_ICMP6STATS:
-		__snmp6_fill_statsdev(stats, idev->stats.icmpv6dev->mibs, ICMP6_MIB_MAX, bytes);
+		__snmp6_fill_statsdev(stats, idev->stats.icmpv6dev->mibs, bytes);
 		break;
 	}
 }
diff --git a/net/ipv6/proc.c b/net/ipv6/proc.c
index 679253d0af84..cc8e3ae9ca73 100644
--- a/net/ipv6/proc.c
+++ b/net/ipv6/proc.c
@@ -30,6 +30,11 @@
 #include <net/transp_v6.h>
 #include <net/ipv6.h>
 
+#define MAX4(a, b, c, d) \
+	max_t(u32, max_t(u32, a, b), max_t(u32, c, d))
+#define SNMP_MIB_MAX MAX4(UDP_MIB_MAX, TCP_MIB_MAX, \
+			IPSTATS_MIB_MAX, ICMP_MIB_MAX)
+
 static int sockstat6_seq_show(struct seq_file *seq, void *v)
 {
 	struct net *net = seq->private;
@@ -191,25 +196,34 @@ static void snmp6_seq_show_item(struct seq_file *seq, void __percpu *pcpumib,
 				atomic_long_t *smib,
 				const struct snmp_mib *itemlist)
 {
+	unsigned long buff[SNMP_MIB_MAX];
 	int i;
-	unsigned long val;
 
-	for (i = 0; itemlist[i].name; i++) {
-		val = pcpumib ?
-			snmp_fold_field(pcpumib, itemlist[i].entry) :
-			atomic_long_read(smib + itemlist[i].entry);
-		seq_printf(seq, "%-32s\t%lu\n", itemlist[i].name, val);
+	if (pcpumib) {
+		memset(buff, 0, sizeof(unsigned long) * SNMP_MIB_MAX);
+
+		snmp_get_cpu_field_batch(buff, itemlist, pcpumib);
+		for (i = 0; itemlist[i].name; i++)
+			seq_printf(seq, "%-32s\t%lu\n",
+				   itemlist[i].name, buff[i]);
+	} else {
+		for (i = 0; itemlist[i].name; i++)
+			seq_printf(seq, "%-32s\t%lu\n", itemlist[i].name,
+				   atomic_long_read(smib + itemlist[i].entry));
 	}
 }
 
 static void snmp6_seq_show_item64(struct seq_file *seq, void __percpu *mib,
 				  const struct snmp_mib *itemlist, size_t syncpoff)
 {
+	u64 buff64[SNMP_MIB_MAX];
 	int i;
 
+	memset(buff64, 0, sizeof(unsigned long) * SNMP_MIB_MAX);
+
+	snmp_get_cpu_field64_batch(buff64, itemlist, mib, syncpoff);
 	for (i = 0; itemlist[i].name; i++)
-		seq_printf(seq, "%-32s\t%llu\n", itemlist[i].name,
-			   snmp_fold_field64(mib, itemlist[i].entry, syncpoff));
+		seq_printf(seq, "%-32s\t%llu\n", itemlist[i].name, buff64[i]);
 }
 
 static int snmp6_seq_show(struct seq_file *seq, void *v)