[Prism54-devel] GW3887 STALL endpoint problem

[Sebastien B]

Hello,
My WG111v2 stalls endpoint 0x81 right after the MGMT readback. The last MGMT 
transfer is OK, the STALL concerns the URB which comes right after. 
Despite that, the device seems to be still responsive to requests on endpoint 
0x01 ; sending a LED control packet works.
NDISWrapper log is at http://yoshiyo.ath.cx/seb/download/wg111_258.log.bz2. I 
send the same thing in the last MGMT frame.
Does anyone have a clue about this ?
I bet the problem's the same with the Linux driver, patch 98 still crashes my 
USB controller ; I think the GW3887 stalls the endpoint just like it does 
under FreeBSD, but the Linux driver is unable to cope with that and messes 
something up with USB...
By the way, with firmware 2.5.6, and probably later versions (first generation 
devices), endpoints 0x02 and 0x82 seem unused ; perform all transfers on 
endpoints 0x01 and 0x81 instead, just like with 3887-based devices. My 
Gigaset works with this with my BSD driver.
NDISWrapper log with that firmware is at 
http://yoshiyo.ath.cx/seb/download/gigaset54_256.log.bz2
One last thing, has anyone ever tried to get information from disassembling 
the SoftMAC firmware itself ? It is compressed, as the "PACKPACKPACK" string 
suggests, but examining it reveals that the uncompressed firmware should 
contain lots of strings which should help reverse engineering a lot, there 
are fragments of them which the compression algorithm let slip. And that's 
probably self-extracting software, the first instruction is a correct ARM 
opcode.
Regards,
Sebastien