FS#64413 - [dhcpcd] route socket overflowed - learning interface state (core-dump)

Attached to Project: Arch Linux
Opened by james stronz (comrumino) - Wednesday, 06 November 2019, 18:40 GMT
Last edited by Antonio Rojas (arojas) - Thursday, 07 November 2019, 19:51 GMT
Task Type Bug Report
Category Packages: Core
Status Closed
Assigned To No-one
Architecture All
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 2
Private No

Details

Description:
To quote https://roy.marples.name/archives/dhcpcd-discuss/0002662.html

| The problem seems to be that after the link local address is added, the kernel eventually sends you a RTM_NEWADDR event.
| Upon receipt of the event you add the IPv6 address to your state->addrs list. After it gets removed the kernel eventually
| sends a RTM_DELADDR and you remove it again from the list.
|
| The problem is that those events arrive delayed, and we are hitting a situation in which the RTM_NEWADDR event is
| not delivered until after the address has been deleted. And RTM_DELADDR is even lagging further behind.
|
| Since state->addr still has the address in the list, ipv6_tryaddlinklocal() thinks there is already a link local address,
| so does no attempt to add one.


Additional info:
* 8.1.1
* log attached
* master should have fix, bug thread is discussed https://roy.marples.name/archives/dhcpcd-discuss/0002662.html
Steps to reproduce:
* wait for vmware guest and docker monster implode
This task depends upon

Closed by  Antonio Rojas (arojas)
Thursday, 07 November 2019, 19:51 GMT
Reason for closing:  Fixed
Additional comments about closing:  dhcpcd 8.1.1-2

Loading...