Re: kernel 2.4.37 usb acm hard lockup
==========================
Learned more after I sent the earlier email.
Failure *always* (N=4 see below) occurs during high speed
UPLOAD of image data from the USB scanner device.
When the application is trying to read about 7.5 Megabytes
of data from the scanner. (I added a zillion or two 'printk()'
statements in the acm driver)
Let me clearly state my statistics: Each time, I added more
printk() statements - etc - always it failed in the same area.
I now have a more exact sequence, which is:
The ACM/TTY code is reading data, the "throttle()" call back
is made into the acm driver, activity pauses (as it should),
then the "acm un throttle" call is made - and it hangs.
My printk() @ the very end of the acm UN-throttle function is
*exactly* the last thing that is printed on the console - consistently.
=======================================
Background:
Problem originally occurs in a highly customized/stripped down
RH-9 - 2.4.20-8 system, which is even worse. That's *MY* problem.
I can duplicate on RH-9 straight from CDs, and using that system,
with the 2.4.37 kernel I can continue to duplicate it.
I will surely help fix 2.4.37 (head of 2.4) because it helps me fix my
nightmare and backport what I need to 2.4.20-8.
Others (before me) back ported a number of fixes - etc - from
2.6<something>
and screamed for help at this point, with this bug. I stepped in cause I do
embedded linux (blackfin + arm) and other kernels (lots of JTAG stuff).
=======================================
Pete> On i386/x86_64 a watchdog facility exists to help with cases like
this.
Pete> Read /..../linux-2.4.32/Documentation/nmi_watchdog.txt
Thanks - that is on tomorrow AM's list. (I also will turn on every other
debug
feature, ie: spin lock debug, etc, in the AM)
=======================================
Oliver> Can you get anything from alt-sysrq-t ?
Not sure about that - Forgot about "magic sysreq" thing.
The "magic sysreq thing" will be turned on in the AM :-)
Thanks for the tip.
=======================================
-Duane.
--
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
