[ltt-dev] tracepoints and small system stack

[Matt Schreiner]

I'm running RedHat 5.5 (2.6.18), and when I enable the trqacepoint callback for sched_switch, linux crashes if I put a printk in my callback function.  Storing information locally and printk'ing it later from an ioctl works fine.  Even calling get_cpu() in the callback causes a similar failure.  I understand the kernel has only a 4k stack, could I be running out?  How does LTTng deal with this, or is this problem not happening to other people?

- MSchreiner
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.casi.polymtl.ca/pipermail/lttng-dev/attachments/20101215/169783cc/attachment-0003.htm>