On 13.06.2014 12:25, Laine Stump wrote:
On 06/13/2014 12:34 PM, Michal Privoznik wrote:
The kernel's more broken than one would think. Various drivers report various (usually spurious) values if the interface is in other state than 'up' . 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 not up anyway.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> --- src/util/virnetdev.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-)
diff --git a/src/util/virnetdev.c b/src/util/virnetdev.c index 6f3a202..a551f98 100644 --- a/src/util/virnetdev.c +++ b/src/util/virnetdev.c @@ -1875,6 +1875,16 @@ virNetDevGetLinkInfo(const char *ifname,
lnk->state = tmp_state;
+ /* Shortcut to avoid some kernel issues. If link is not up several drivers + * report several misleading values. While igb reports 65535, realtek goes + * with 10. To avoid muddying XML with insane values, don't report link + * speed if that's the case. */ + if (lnk->state != VIR_INTERFACE_STATE_UP) { + lnk->speed = 0; + ret = 0; + goto cleanup; + } + VIR_FREE(path); VIR_FREE(buf);
@@ -1901,11 +1911,7 @@ virNetDevGetLinkInfo(const char *ifname, goto cleanup; }
- /* Workaround broken kernel API. If the link is unplugged then - * depending on the NIC driver, link speed can be reported as -1. - * However, the value is printed out as unsigned integer instead of - * signed one. Terrifying but true. */ - lnk->speed = (int) tmp_speed == -1 ? 0 : tmp_speed; + lnk->speed = tmp_speed;
ret = 0; cleanup:
ACK from me. Might want to give a short time for differing opinions (or just push this and change as required up until the next release :-)
Right. I've pushed this one as it enables testing for every one of us running libvirt from git instead of having to manually apply the patch. Michal