| Summary: | busybox: udhcpc misbehavior when "ip=dhcp" | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Jesse Zhang <sen.zhang> |
| Component: | core | Assignee: | Jesse Zhang <sen.zhang> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium | CC: | jessica.zhang, liang.li2, meta.mr.watcher, meta.watcher, richard.purdie |
| Version: | 1.3 | ||
| Target Milestone: | 1.3 M3 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | --- | |
Fixed in master: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=a4b1e348484b74d055b8906413892789d3452f4a verified on mpc8315e-rdb board with nightly build 20120822-1 ,when change command line to "ip=dhcp",no such error. git info:bb0aab79e488a99cb1b43f72f0e8ddbe39d97519 git branch :poky/master |
I recently found that if "ip=dhcp" is specified in the kernel cmdline, there would be some strange behaviour of udhcpc. 1. udhcpc is started two times during bootup. Starting Bootlog daemon: bootlogd. Configuring network interfaces... udhcpc (v1.19.4) started Sending discover... Sending select for 128.224.162.132... Lease of 128.224.162.132 obtained, lease time 172800 /etc/udhcpc.d/50default: Adding DNS 128.224.160.11 /etc/udhcpc.d/50default: Adding DNS 147.11.100.30 /etc/udhcpc.d/50default: Adding DNS 147.11.1.11 done. Starting portmap daemon... [...] Starting domain name service: named. udhcpc (v1.19.4) started Sending discover... Sending select for 128.224.162.132... Lease of 128.224.162.132 obtained, lease time 172800 RTNETLINK answers: File exists /etc/udhcpc.d/50default: Adding DNS 128.224.160.11 /etc/udhcpc.d/50default: Adding DNS 147.11.100.30 /etc/udhcpc.d/50default: Adding DNS 147.11.1.11 creating NFS state directory: done 2. udhcpc is *started* when system is being shutdown. Stopping domain name service: namedrndc: neither /etc/bind/rndc.conf nor /etc/bind/rndc.key wd . udhcpc (v1.19.4) started Sending discover... Sending select for 128.224.162.132... Lease of 128.224.162.132 obtained, lease time 172800 RTNETLINK answers: File exists /etc/udhcpc.d/50default: Adding DNS 128.224.160.11 /etc/udhcpc.d/50default: Adding DNS 147.11.100.30 /etc/udhcpc.d/50default: Adding DNS 147.11.1.11 Stopping system message bus: dbus. stopping statd: done The reason is busybox installs a busybox-udhcpc init script, which would start udhcpc when "ip=dhcp" is specified, but udhcpc is also invoked by the networking service (as specified through /etc/network/interfaces). That's why udhcpc is started twice. And since the init script doesn't actually support "stop" actions, it will start udhcpc again during system shutdown. The obvious fix is to remove the busybox-udhcpc init script; all configs should go through /etc/network/interfaces. P.S. it only happens for non-qemu targets since they have different configs. I used an intel emenlow.