[Trac #1026] SCP failures across multi-hop paths

[Zarro Boogs per Child]

#1026: SCP failures across multi-hop paths
----------------------+-----------------------------------------------------
 Reporter:  jcardona  |       Owner:  jcardona 
     Type:  defect    |      Status:  new      
 Priority:  normal    |   Milestone:  Untriaged
Component:  wireless  |    Keywords:           
----------------------+-----------------------------------------------------
 '''Originally reported by mbletsas'''

 After the TCP flow control fix in the driver, I can replicate reliably the
 SCP failures.
 They basically happen in multihop situations (direct routes seem to work).

 {{{
 [root at mshy13 ~]# scp
 kernel-2.6.21-20070305.2.olpc.fb4edc7fa9b1ce8.i586.rpm root at mshy10:
 kernel-2.6.21-20070305.2.olpc.fb4edc7fa9b1ce8.i586.rpm
 0%    0     0.0KB/s   --:-- ETA
 Received disconnect from 10.2.45.238: 2: Corrupted MAC on input. lost
 connection


 [root at mshy13 ~]# arp mshy10
 Address                  HWtype  HWaddress           Flags Mask
 Iface
 mshy10                   ether   00:17:C4:02:2D:EE   C
 msh0
 [root at mshy13 ~]# ./fwt_list.sh msh0 | grep -i 00:17:C4:02:2D:EE
 00:17:c4:01:19:2a  00:17:c4:02:2d:ee    59    0   4110      0      2
 8   170726710    0   74
 00:17:c4:01:19:85  00:17:c4:02:2d:ee    59    0   2420      0      2
 8   171138290    0   75
 00:17:c4:02:2d:ee  00:17:c4:02:2d:ee    46    0   2310      0      1
 9   171232180    0   75
 00:17:c4:02:2d:ee  00:17:c4:02:30:11    26    0   2315      0      2
 8   171243200    0   52
 00:17:c4:02:2d:ee  00:17:c4:01:18:87    39    0   2306      0      3
 7   171221170    0   52
 00:17:c4:02:2d:ee  00:17:c4:01:19:2a    59    0   2241      0      2
 8   171062570    0   74
 00:17:c4:02:2d:ee  00:17:c4:02:2f:af    52    0   1899      0      4
 6   170286300    0   59
 00:17:c4:02:2d:ee  00:17:c4:02:2f:65    41    0   2279      0      2
 8   171156730    0   58
 00:17:c4:02:2d:ee  00:17:c4:01:18:88    52    0   2267      0      4
 6   171127120    0   16
 00:17:c4:02:2d:ee  00:17:c4:02:2e:fe    39    0   2263      0      3
 7   171116200    0   52
 00:17:c4:02:2d:ee  00:17:c4:02:33:32    54    0   2267      0      3
 7   171127070    0   65
 00:17:c4:02:2d:ee  00:17:c4:01:17:ec    39    0   2248      0      3
 7   171083860    0   62
 00:17:c4:02:2d:ee  00:17:c4:02:30:10    39    0   2267      0      3
 7   171127160    0   15
 00:17:c4:02:2d:ee  00:17:c4:01:19:85    59    0   2267      0      2
 8   171127060    0   78
 00:17:c4:02:2d:ee  00:17:c4:02:31:ce    39    0   2236      0      3
 7   171050760    0   60
 00:17:c4:02:2d:ee  00:17:c4:02:2d:ee    46    1   1070   2228      1
 0   171039940    0   75
 00:17:c4:02:2d:ee  00:17:c4:02:30:11    26    1   1087   2276      2
 0   171150240    0   52
 00:17:c4:02:2d:ee  00:17:c4:01:18:88    52    1   1051   2204      4
 0   171014210    0   15
 00:17:c4:02:2d:ee  00:17:c4:02:2f:65    26    1   1059   2210      2
 0   171014710    0   58
 00:17:c4:02:2d:ee  00:17:c4:02:2e:fe    26    1   1080   2244      2
 0   171071720    0   52
 }}}

 Not all of the machines in the testbed have been updated to the 2.6.21
 kernel, however that shouldn't play any role at intermediate nodes.

-- 
Ticket URL: <http://dev.laptop.org/ticket/1026>
One Laptop Per Child <http://laptop.org/>