| Summary: | DAHLIN-00343: Dynamic span modules fail to decrement reference counter for the underlying VLAN interface when unloading | ||
| Reporter: | Assen Totin (assen) | Labels: | |
| Date Opened: | 2014-12-10 07:12:13.000-0600 | Date Closed: | 2019-06-06 07:06:42 |
| Priority: | Minor | Regression? | No |
| Status: | Closed/Complete | Components: | dahdi_dynamic dahdi_dynamic_eth |
| Versions: | 2.10.0 | Frequency of Occurrence | Constant |
| Related Issues: | |||
| Environment: | Linux 2.6 | Attachments: | ( 0) 0001-dahdi_dynamic-Release-reference-count-on-network-dev.patch |
| Description: | Scope: the issue is present only when a dynamic span is configured on top of a IEEE 802.1q interface (VLAN interface).
Effect: when a dynamic span is configured on top of a VLAN interface (e.g., eth0.N) and is later removed, unloading the modules (dahdi_dynamic_eth, dahdi_dynamic) does not decrement the kernel reference counter for the underlying VLAN interface. As a result, the kernel believes the underlying network interface is still in use, refusing to remove it upon network shutdown. The latter means that clean power-off of the Linux OS become impossible and power cycle is almost the only remaining option (especially for remote machines and VMs). Notes: The effect is present only on VLAN interfaces and is not present on regular interfaces (e.g., eth0). Tested only on Linux; because Linux handles network devices differently than most other UNIX, it is likely limited to Linux. Testbed: confirmed on CentOS 5 and 6 VM in ESXi 5.5 with either VMXNET3 or emulated Intel EEPRO 1000 network adapters. | ||
| Comments: | By: Shaun Ruffell (sruffell) 2014-12-10 08:47:01.805-0600 Thanks for the report. Odd that what you see is only seen with VLAN interfaces, but it definitely looks like there was an increment of the netdev reference count when the dynamic_spans were created without a subsequent decrement on destruction. I've attached the untested [^0001-dahdi_dynamic-Release-reference-count-on-network-dev.patch] which I believe should fix this problem for you if you wanted to try it out. Otherwise it might be awhile before a test can be setup for this specifically. By: Assen Totin (assen) 2014-12-19 16:59:48.892-0600 Hi, Shaun! Thanks for your prompt reply. Managed to test today on a production machine. Works great, no more problems with network restarts. To me the issue is solved - feel free to close the ticket as your discretion. WWell, Assen By: Keith Morgan (kmorgan) 2019-06-06 07:06:42.398-0500 Patch was applied some time ago. | ||