On Fri, Jun 13, 2014 at 08:46:59AM +0200, Michal Privoznik wrote:
The kernel's more broken than one would think. Various drivers report
various (usually spurious) values if the interface is down. While on
some we experience -EINVAL when read()-ing the speed sysfs file, with
other drivers we might get anything from 0 to UINT_MAX. If that's the
case it's better to not report link speed. Well, the interface is down
anyway.
Signed-off-by: Michal Privoznik <mprivozn@xxxxxxxxxx>
---
src/util/virnetdev.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/src/util/virnetdev.c b/src/util/virnetdev.c
index 6f3a202..80ef572 100644
--- a/src/util/virnetdev.c
+++ b/src/util/virnetdev.c
@@ -1843,7 +1843,7 @@ virNetDevGetLinkInfo(const char *ifname,
char *buf = NULL;
char *tmp;
int tmp_state;
- unsigned int tmp_speed;
+ unsigned int tmp_speed; /* virInterfaceState */
You probably wanted to put this comment next to the line with tmp_state and not tmp_speed.
if (virNetDevSysfsFile(&path, ifname, "operstate") < 0)
goto cleanup;
@@ -1875,6 +1875,16 @@ virNetDevGetLinkInfo(const char *ifname,
lnk->state = tmp_state;
+ /* Shortcut to avoid some kernel issues. If link is down (and possibly in
+ * other states too) several drivers report several values. While igb
+ * reports 65535, realtek goes with 10. To avoid muddying XML with insane
+ * values, don't report link speed */
+ if (lnk->state == VIR_INTERFACE_STATE_DOWN) {
+ lnk->speed = 0;
+ ret = 0;
+ goto cleanup;
+ }
+
Pity that's not standardized, but what you did makes more sense than anything else. If the link is down, then no need to report it.
VIR_FREE(path);
VIR_FREE(buf);
@@ -1884,6 +1894,7 @@ virNetDevGetLinkInfo(const char *ifname,
if (virFileReadAll(path, 1024, &buf) < 0) {
/* Some devices doesn't report speed, in which case we get EINVAL */
if (errno == EINVAL) {
+ lnk->speed = 0;
ret = 0;
goto cleanup;
}
--
1.8.5.5
ACK, Martin
Attachment:
signature.asc
Description: Digital signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list