[lustre-devel] [PATCH 3/5] lustre: change lnet_ipaddr_enumerate() to use for_each_netdev()

Doug Oucharek doucharek at cray.com
Wed Jul 25 15:39:12 PDT 2018


Reviewed-by: Doug Oucharek <dougso at me.com<mailto:dougso at me.com>>

Note:
I did not think your changes would collide with Multi-Rail.  Unfortunately, it looks like this one does.  This config code was changed quite a bit by the Multi-Rail feature.  However, it is ok as landing this change now will prompt the Multi-Rail changes to utilize the better way of getting interface information.  Just a heads up to Multi-Rail porters.

Doug

On Jul 24, 2018, at 4:07 PM, NeilBrown <neilb at suse.com<mailto:neilb at suse.com>> wrote:

for_each_netdev() is a more direct interface than
lnet_ipif_enumerate(), so use it instead.  Also get
address and 'up' status directly from the device.

This means we need to possible re-allocate the storage
space if there are lots of IP addresses.

However there is no need to resize the allocation down if we
over-allocated.  This is only used once, and is freed soon
after it is allocated, so that is a false optimization.

Signed-off-by: NeilBrown <neilb at suse.com<mailto:neilb at suse.com>>
---
drivers/staging/lustre/lnet/lnet/config.c |   78 +++++++++++++----------------
1 file changed, 35 insertions(+), 43 deletions(-)

diff --git a/drivers/staging/lustre/lnet/lnet/config.c b/drivers/staging/lustre/lnet/lnet/config.c
index 55ecc1998b7e..136905db2746 100644
--- a/drivers/staging/lustre/lnet/lnet/config.c
+++ b/drivers/staging/lustre/lnet/lnet/config.c
@@ -36,6 +36,7 @@
#include <net/net_namespace.h>
#include <linux/ctype.h>
#include <linux/lnet/lib-lnet.h>
+#include <linux/inetdevice.h>

struct lnet_text_buf {    /* tmp struct for parsing routes */
struct list_head ltb_list; /* stash on lists */
@@ -1134,66 +1135,57 @@ lnet_match_networks(char **networksp, char *ip2nets, __u32 *ipaddrs, int nip)
static int
lnet_ipaddr_enumerate(__u32 **ipaddrsp)
{
- int up;
- __u32 netmask;
+ struct net_device *dev;
__u32 *ipaddrs;
- __u32 *ipaddrs2;
+ int nalloc = 64;
int nip;
- char **ifnames;
- int nif = lnet_ipif_enumerate(&ifnames);
- int i;
- int rc;

- if (nif <= 0)
- return nif;
-
- ipaddrs = kcalloc(nif, sizeof(*ipaddrs), GFP_KERNEL);
+ ipaddrs = kcalloc(nalloc, sizeof(*ipaddrs), GFP_KERNEL);
if (!ipaddrs) {
- CERROR("Can't allocate ipaddrs[%d]\n", nif);
- lnet_ipif_free_enumeration(ifnames, nif);
+ CERROR("Can't allocate ipaddrs[%d]\n", nalloc);
return -ENOMEM;
}

- for (i = nip = 0; i < nif; i++) {
- if (!strcmp(ifnames[i], "lo"))
+ rtnl_lock();
+ for_each_netdev(&init_net, dev) {
+ struct in_device *in_dev;
+
+ if (strcmp(dev->name, "lo") == 0)
continue;

- rc = lnet_ipif_query(ifnames[i], &up, &ipaddrs[nip], &netmask);
- if (rc) {
- CWARN("Can't query interface %s: %d\n",
-      ifnames[i], rc);
+ if (!(dev_get_flags(dev) & IFF_UP)) {
+ CWARN("Ignoring interface %s: it's down\n", dev->name);
continue;
}
-
- if (!up) {
- CWARN("Ignoring interface %s: it's down\n",
-      ifnames[i]);
+ in_dev = __in_dev_get_rtnl(dev);
+ if (!in_dev) {
+ CWARN("Interface %s has no IPv4 status.\n", dev->name);
continue;
}

- nip++;
- }
-
- lnet_ipif_free_enumeration(ifnames, nif);
-
- if (nip == nif) {
- *ipaddrsp = ipaddrs;
- } else {
- if (nip > 0) {
- ipaddrs2 = kcalloc(nip, sizeof(*ipaddrs2),
-   GFP_KERNEL);
- if (!ipaddrs2) {
- CERROR("Can't allocate ipaddrs[%d]\n", nip);
- nip = -ENOMEM;
- } else {
- memcpy(ipaddrs2, ipaddrs,
-       nip * sizeof(*ipaddrs));
- *ipaddrsp = ipaddrs2;
- rc = nip;
+ if (nip >= nalloc) {
+ __u32 *ipaddrs2;
+ nalloc += nalloc;
+ ipaddrs2 = krealloc(ipaddrs, nalloc * sizeof(*ipaddrs2),
+    GFP_KERNEL);
+ if (ipaddrs2 == NULL) {
+ kfree(ipaddrs);
+ CERROR("Can't allocate ipaddrs[%d]\n", nalloc);
+ return -ENOMEM;
}
+ ipaddrs = ipaddrs2;
}
- kfree(ipaddrs);
+
+ for_primary_ifa(in_dev)
+ if (strcmp(ifa->ifa_label, dev->name) == 0) {
+ ipaddrs[nip++] = ifa->ifa_local;
+ break;
+ }
+ endfor_ifa(in_dev);
}
+ rtnl_unlock();
+
+ *ipaddrsp = ipaddrs;
return nip;
}



_______________________________________________
lustre-devel mailing list
lustre-devel at lists.lustre.org<mailto:lustre-devel at lists.lustre.org>
http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-devel-lustre.org/attachments/20180725/e7ceb368/attachment-0001.html>


More information about the lustre-devel mailing list