On Fri, 10 Dec 2010, ENRIQUE ARIZON BENITO wrote:
> On Thu, Dec 9, 2010 at 11:30 PM, Matthew Dharm
> <mdharm-usb@xxxxxxxxxxxxxxxxxx> wrote:
> > On Thu, Dec 09, 2010 at 10:45:13PM +0100, ENRIQUE ARIZON BENITO wrote:
> >> A simple script like:
> >>
> >> echo "1" > /dev/ttyUSB0
> >>
> >> stopped to work after upgrading my system to a custom compiled version
> >> of kernel 2.6.35.
> >>
> >> There are no errors in syslog or dmesg and the pl2303 loads without
> >> complains, but strace shows that
> >> open("/dev/ttyUSB0",...) blocks forever.
> >>
> >> I guess it must be some wrong parameter in the compilation, but I'm
> >> blind trying to find the solution.
> >
> > Have you used stty to check the port mode settings? Â It may be blocking for
> > carrier detect....
> >
> > Matt
> >
>
> Yes, I did. In fact the setting as displayed from the output command
> "stty -a -F /dev/ttyUSB0" are equals both for the working kernel and
> the new (non working) one.
That means nothing, since the kernel's behavior in this regard has
changed. Older versions of the kernel were buggy; they would not block
for a handshaking signal when they should have. That's probably why
your test didn't hang under the working kernel.
To prevent waiting for a handshake signal, you can do
stty -F /dev/ttyUSB0 clocal -crtscts
before running the test. But a better idea is to avoid doing regular
I/O to a serial port and instead use a program designed for that task,
like minicom.
On Fri, 10 Dec 2010, ENRIQUE ARIZON BENITO wrote:
> Solved!
>
> Udev was loading both uhci an ohci even if the system was not using ohci at all.
That would not cause any problems. The ohci-hcd driver would simply be
inactive; it would not interfere with uhci-hcd.
> "rmmoding" both on them as well as "pl2303" and "modprobing" uchi and
> pl2303 (in that order), the blocking dissapeared.
I think you are fooling yourself. The port settings are much more
likely to be the cause of the problem. Reloading pl2303 probably
caused the settings to change.
> ohci was also beeing loaded by the old 2.6.30 kernel. Still it didn't
> affect the normal behaviour. That makes me suspect there could be
> another bug somewhere else.
Not a bug, just the system operating the way it is supposed to.
Alan Stern
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
