On Tue, Apr 01, 2003 at 09:05:16AM -0800, Greg KH wrote: > On Tue, Apr 01, 2003 at 05:24:11PM +0100, Nick Craig-Wood wrote: > > I tried 2.5.66 and yes it no longer Oopses on close! I tried it 100 > > times on USA-49W and USA-49WLC just to be sure. > > Good :) > > > Is it easy to break out a patch for this for 2.4? I'd willingly test > > it for you. > > Heh, it isn't a simple patch, sorry. I'll try to get some time to work > on it, but help is appreciated.
Great! Whenever you are ready! > > While testing on 2.5 I tried to use multiple ports at once with > > cambert. The problem is that only the first port transfers data > > properly - the second and subsequent ports transfer very little data. > > mostly. (I did see it once transfer an outrageously large amount like > > 100MB in a second but I haven't managed to reproduce this!) > > > > Once cambert has got going it just does non-blocking reads and writes > > into all the sockets it has open. Eg from strace... > > What is cambert doing? It sets up the ttys into the mode suggested on the command line, sets them non-blocking then launches into an infinite read() and write() loop. For each port... It generates some random data if it hasn't got any It attempts to write() that data to the port. If the write() succeeds it decrements the amount of random data its got It then attempts to read data from the port. If it gets some and it is already synced up then it continues. > Any chance to see the source? I can send you the source I expect, but I don't think I can send it to the world! I'll try tonight to write a simple test prog which demonstrates the problem. > Could we just be flooding the first port and not getting a chance to > send any data out the second? Do you mean user space or kernel space here? cambert itself just tries to write() stuff to each port in turn and when that succeeds it generates a little more random data and write()s that in turn. > What causes the test to end, one port finishing? The test runs until CTRL-C > Does slower baud rates work ok? It still goes wrong at 1200 baud, so no. -- Nick Craig-Wood [EMAIL PROTECTED] ------------------------------------------------------- This SF.net email is sponsored by: ValueWeb: Dedicated Hosting for just $79/mo with 500 GB of bandwidth! No other company gives more support or power for your dedicated server http://click.atdmt.com/AFF/go/sdnxxaff00300020aff/direct/01/ _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
