Bug 24949
| Summary: | mbuf memory leak in de0 driver | ||
|---|---|---|---|
| Product: | Base System | Reporter: | dirk.meyer <dirk.meyer> |
| Component: | i386 | Assignee: | freebsd-bugs (Nobody) <bugs> |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | ||
| Priority: | Normal | ||
| Version: | 4.2.-RELEASE | ||
| Hardware: | Any | ||
| OS: | Any | ||
State Changed From-To: open-> feedback The PR `netstat -m' output(s) suggest that there is no leak but only that something may be pre-allocating several mbufs, which isn't a big deal. The number of allocated mbufs and clusters first rises but then decreases, even though the peak clearly rises, but the system seems to stay steady from that point on. A `slow down' is mentionned, but it is difficult from the provided information to determine where the slow-down is occuring and certainly, we have not observed a similar thing here. The PR is not being closed just yet. I would like to urge the submitter to re-submit, if possible, some more information regarding the slow-down, assuming that it is still occuring and, if it is no longer occuring, to let us know so that we can close the PR. > that it is still occuring and, if it is no longer occuring, to let us know
> so that we can close the PR.
Test with 4.3 in progress, final information will follow in a few days.
kind regards Dirk
- Dirk Meyer, Im Grund 4, 34317 Habichtswald, Germany
> Test with 4.3 in progress, final information will follow in a few days.
The network slowdown did not oocur any more,
we can take this Problem as fixed and close the PR.
The new kernal was compiled with NMBCLUSTERS=8192 too.
kind regards Dirk
- Dirk Meyer, Im Grund 4, 34317 Habichtswald, Germany
State Changed From-To: feedback->closed Problem fixed. Submitter says it no longer happens. Thanks! |
Network operations start to slow down times 2 and more. Logs of mbuf may help, Lines with "0" omitted. ------------- $ netstat -m 101/432/4096 mbufs in use (current/peak/max): 68 mbufs allocated to data 33 mbufs allocated to packet headers 34/206/1024 mbuf clusters in use (current/peak/max) 520 Kbytes allocated to network (16% of mb_map in use) $ ifconfig de0 down $ netstat -m 2/432/4096 mbufs in use (current/peak/max): 1 mbufs allocated to data 1 mbufs allocated to packet headers 0/206/1024 mbuf clusters in use (current/peak/max) 520 Kbytes allocated to network (16% of mb_map in use) $ ifconfig de0 up $ netstat -m 34/432/4096 mbufs in use (current/peak/max): 33 mbufs allocated to data 1 mbufs allocated to packet headers 32/206/1024 mbuf clusters in use (current/peak/max) 520 Kbytes allocated to network (16% of mb_map in use) ------------- 319/1152/4096 mbufs in use (current/peak/max): 240 mbufs allocated to data 79 mbufs allocated to packet headers 111/422/1024 mbuf clusters in use (current/peak/max) 1132 Kbytes allocated to network (36% of mb_map in use) 156/1152/4096 mbufs in use (current/peak/max): 83 mbufs allocated to data 73 mbufs allocated to packet headers 73/422/1024 mbuf clusters in use (current/peak/max) 1132 Kbytes allocated to network (36% of mb_map in use) 188/1152/4096 mbufs in use (current/peak/max): 115 mbufs allocated to data 73 mbufs allocated to packet headers 105/422/1024 mbuf clusters in use (current/peak/max) 1132 Kbytes allocated to network (36% of mb_map in use) 60/1152/4096 mbufs in use (current/peak/max): 54 mbufs allocated to data 6 mbufs allocated to packet headers 53/422/1024 mbuf clusters in use (current/peak/max) 1132 Kbytes allocated to network (36% of mb_map in use) 92/1152/4096 mbufs in use (current/peak/max): 86 mbufs allocated to data 6 mbufs allocated to packet headers 85/422/1024 mbuf clusters in use (current/peak/max) 1132 Kbytes allocated to network (36% of mb_map in use) ------------- Fix: hack: ifconfig down / ifconfig up seems to keep the serrver useable How-To-Repeat: I have no idea what causes this. But another server with and xl0 interface works fine.