Core Issue
Whenever the Cisco Catalyst 2950 series switch is powered on or reloaded, the switch ports to which the 7905 series IP phones are connected enter into errdisable state, and the phones also reload and do not boot up properly. The phones reload frequently. They register to the Cisco CallManager only after the switch ports are configured with the shutdown command, followed by the no shutdown command.
The phones have the load CP79050101SCCP030404A.ZUP. The IP phones register to the Cisco CallManager version 3.3(2) with Service Pack B. The Cisco 2950 series Catalyst switch is running the 12.1(13)EA1b version of Cisco IOS Software.
Resolution
The problem lies with the Cisco IOS Software on the Catalyst 2950 series switch, not the IP phones. The Catalyst 2950 series switches with version 12.1(13)EA1 consider a port as loopback state and put it in errdisable state when it detects more than two collisions in two seconds with no packets received. The example of the loopback condition is when there is a bad network design and loop connected to the port, or collision condition beyond the port due to a bad cable. Very rarely does a switch detect loopback state with a hub when there is no bad design or bad cable.
For a workaround for this problem, configure the errdisable recovery cause all command or no errdisable detect cause all command on all the switch ports with this issue.