[PATCH][RFC] usbatm.[ch]: cleanup and OAM F5 loopback
castet.matthieu at free.fr
Sun Mar 13 17:21:35 EST 2005
Roman Kagan wrote:
> On Sun, Mar 13, 2005 at 12:31:59PM +0100, matthieu castet wrote:
>>I don't know, but what is strange is that if I force my modem to
>>synchronize to a lower speed 6500kbits/s, and if I only download there
>>aren't any error. If I start uploading there are errors.
>>In 7000kits/s there are always errors even if I download only.
> Are you doing your tests in 2.6.10 or 2.6.11? There've been a number of
> changes in usb hcd drivers which might affect performance.
2.6.11 but I had the same behaviour with older kernel.
> Besides, can you please add printing of urb status to
> usbatm_complete_receive and usbatm_complete_send, to see if the errors
> happen during transfer and are reported by the hcd but ignored by
> usbatm, or on the contrary the modem happily sends garbage?
I wasn't able to reproduce crc errors, but I have "bogus pdu_length" and
all the status are correct.
"bogus pdu_length" indicate it miss data in the atm trame ?
> BTW how does firmware swapping interfere with the data transfers? I
> can't imagine the modem reloading a different part of firmware for every
> transmission or reception without influencing the performance.
The modem don't seem to swap in operationnal mode, only in
initialisation(rebooting, synchronisation, ...) + I don't know.
More information about the Usbatm