Created attachment 2221 [details] The output of running BB1.16.1 manually NOTE: This problem occurs in version 1.17.0 I tried to run MicroCore mc3.0rc1 (http://tinycorelinux.com/), That is runnimg BusyBox 1.17.0 On boot it runs '/bin/udhcpc -b -i eth0 -h box -p /var/run/udhcpc.etho.pid' and fails I tried to mount the CD of and old release (mc3.0alpha5), running BB1.16.1, and run it manually, and it worked. (see attachment1 [details]) I then tried vesion 1.17.0 manually, but again it failed (see attachment 2) I then ran wireshark, and found, that the 2 packages were not identical. It is some flags that are not identical (see attachment 3 [details] and 4) The distribution is running under VMWare on Win 7
Created attachment 2227 [details] The output of running BB1.17.1 manually
Created attachment 2233 [details] Wireshark-dump of udhcpc 1.16.1
(In reply to comment #2) > Created attachment 2233 [details] > Wireshark-dump of udhcpc 1.16.1 And the dump of version 1.17.0 is available at: http://batnas.dk/BB1.17.0.pcap
Here is 1.16.1 request packet: 19:51:48.642325 IP (tos 0x0, ttl 64, id 0, offset 0, flags [none], proto: UDP (17), length: 576) 0.0.0.0.bootpc > 255.255.255.255.bootps: [udp sum ok] BOOTP/DHCP, Request from 00:0c:29:25:6b:ae (oui Unknown), length: 548, xid:0xa11106b, flags: [Broadcast] (0x8000) Client Ethernet Address: 00:0c:29:25:6b:ae (oui Unknown) Vendor-rfc1048: DHCP:DISCOVER CID:[ether]00:0c:29:25:6b:ae HN:"box" VC:"udhcp 1.16.1" MSZ:576 PR:SM+DG+NS+HN+DN+BR+NTP 0x0000: 4500 0240 0000 0000 4011 78ae 0000 0000 E..@....@.x..... 0x0010: ffff ffff 0044 0043 022c 576d 0101 0600 .....D.C.,Wm.... 0x0020: 0a11 106b 0000 8000 0000 0000 0000 0000 ...k............ 0x0030: 0000 0000 0000 0000 000c 2925 6bae 0000 ..........)%k... 0x0040: 0000 0000 0000 0000 0000 0000 0000 0000 ................ ... 0x00f0: 0000 0000 0000 0000 0000 0000 0000 0000 ................ 0x0100: 0000 0000 0000 0000 6382 5363 3501 013d ........c.Sc5..= 0x0110: 0701 000c 2925 6bae 0c03 626f 783c 0c75 ....)%k...box<.u 0x0120: 6468 6370 2031 2e31 362e 3139 0202 4037 dhcp.1.16.19..@7 0x0130: 0701 0306 0c0f 1c2a ff00 0000 0000 0000 .......*........ 0x0140: 0000 0000 0000 0000 0000 0000 0000 0000 ................ ... 0x0230: 0000 0000 0000 0000 0000 0000 0000 0000 ................ and 1.17.0 request packet: 19:52:46.847263 IP (tos 0x0, ttl 64, id 0, offset 0, flags [none], proto: UDP (17), length: 313) 0.0.0.0.bootpc > 255.255.255.255.bootps: [udp sum ok] BOOTP/DHCP, Request from 00:0c:29:25:6b:ae (oui Unknown), length: 285, xid:0x4cbb4e39, flags: [none] (0x0000) Client Ethernet Address: 00:0c:29:25:6b:ae (oui Unknown) Vendor-rfc1048: DHCP:DISCOVER CID:[ether]00:0c:29:25:6b:ae HN:"box" VC:"udhcp 1.17.0" MSZ:576 PR:SM+DG+NS+HN+DN+BR+NTP 0x0000: 4500 0139 0000 0000 4011 79b5 0000 0000 E..9....@.y..... 0x0010: ffff ffff 0044 0043 0125 5903 0101 0600 .....D.C.%Y..... 0x0020: 4cbb 4e39 0000 0000 0000 0000 0000 0000 L.N9............ 0x0030: 0000 0000 0000 0000 000c 2925 6bae 0000 ..........)%k... 0x0040: 0000 0000 0000 0000 0000 0000 0000 0000 ................ ... 0x00f0: 0000 0000 0000 0000 0000 0000 0000 0000 ................ 0x0100: 0000 0000 0000 0000 6382 5363 3501 013d ........c.Sc5..= 0x0110: 0701 000c 2925 6bae 0c03 626f 783c 0c75 ....)%k...box<.u 0x0120: 6468 6370 2031 2e31 372e 3039 0202 4037 dhcp.1.17.09..@7 0x0130: 0701 0306 0c0f 1c2a ff .......*. As you see, it is shorter. But this should be ok - Windows also uses unpadded packets. Ideas so far: * old packet has bcast flag in dhcp header. (1) Why? and (2) why new one doesn't? * maybe your server insists on packets rounded up to 2 or 4 byte boundary in size?
* Old packet has bcast flag in dhcp header - That's really interesting. How it ended up set? There is no code in dhcpc to set it. No way. Maybe tinycorelinux use patched busybox source? * Maybe your server insists on packets rounded up to 2 or 4 byte boundary in size? - You can test this theory by making hostname longer. Instead of -h box, try -h box1, then -h box12345. If one of those work, then we have this bug nailed.
Any news?
Closing for the lack of feedback from reporter