Hello List,
On Sat, 18 Sep 2004 08:38:23 +0900, adsynori <[EMAIL PROTECTED]> wrote:
Hi David Brownell
though you name David Brownell I thought you could possibly address me with this help, so I tested it...
According to bus timing, Download BINGO version and refer to hc_isp116x.c
READ_REG** WRITE_REG**
You find ndelay(600); Please attach it in your code and try. (At first reduce printk)
Reading this and the code I thought that would definitely be the answer to my problems. But I tried that and there was no difference to the 'no-printk' behaviour: It didn' t work and I got these timeouts. I set the delay to 2000 and later to 4000 but that didn' t work, also.
I wanted to try the BINGO version itself, but I vouldn't compile it yet.
by the way if you try other device (mouse,keyboard,flash stick...)? or if you see descriptor of device?
I don't have very much USB devices. As I can say, it's only the bluetooth
connection that does not work.
Well, I have had an USB disk connected. Reading was OK, writing often took
very long (aprox. 2 min for a few kbytes of a textfile).
Perhaps thats the same error, but I don' t know. I get no error messages
doing that.
Mouse and Keyboard are shown correctly in lsusb (I have no screen for testing).
The bluetooth dongles seem to work at first. Configuration with 'hciconfig'
works and 'l2ping' (ping on bluetooth l2cap layer level) does work.
But then a connection (for exymple) with 'pand' that creates a network device
you can configure with 'ifconfig bnep0' does not work. Well the connection
is shown as established on both sides, but doing a tcp/ip ping gives zero
packets received. And afterwards I get timeouts doing other bluetooth commands
as 'hciconfig'. Oh, and removing the hc_isp116x module afterwards I sometimes
get a segmentaition fault.
After this I do a reboot.
After this reboot 'hciconfig' works, but I have to do a 'hciconfig hci0 reset'
to make 'l2ping' work.
This 'pand' connection works perfectly with the debug message in hci_submit_urb
from hc_simple.c. I get no timeouts or errors afterwards.
On Fri, 17 Sep 2004 09:28:29 -0700, David Brownell <[EMAIL PROTECTED]> wrote:
Maybe not -- but that's a nice precise clue to have.
But I have absolutely no idea what to do.
And if it's a timing problem with these debug messages, how could this be
fixed anyway?
The problem is evidently that adding a delay in that routine makes the timings work. It's your job to find out why. What are the relevant timings? Does replacing the printk with mdelay(1) make it work? mdelay(10)? 100? 1000? 5000?
ok, I tested:
800 - tcp/ip ping works in one direction only
400 - works in both directions, but 30% packet loss
200 - works good (slow)
100 - works good (a bit faster)
50 - works good (a bit faster)
25 - works good (a bit faster)
13 - works good (a bit faster)
7 - works good (a bit faster)
4 - works good (a bit faster) I tested at least 5000 tcp packets
3 - worked in my first test, didn't in a second test, worked in a third test,
but only for approx. 1400 tcp packets
2 - did not work, errors as without any printk or delay
So, it works with mdelay(4)
What to do next?
I wonder why the ndelay changes from BINGO didn't work.
I thought the delays there would be as good as in the hci_submit_urb function.
That's why I tested ndelay(2000) and ndelay(4000). I thought 600 would perhaps
not be enough.
I also wonder why 'l2ping' works.
Any ideas?
-- Regards, Christoph
------------------------------------------------------- This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 Project Admins to receive an Apple iPod Mini FREE for your judgement on who ports your project to Linux PPC the best. Sponsored by IBM. Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
