>>>>> "rd" == RD Thrush <rd@thrush.com> writes:
rd> For almost 2 weeks I have had both amd64 and i386 kernels hang. The
rd> problem first occurred while updating -current ports since the libc
rd> bump. I've found the hang is triggered more quickly by building the
rd> eclipse-sdk port. The hang has occurred on 3 different machines.
rd> Entering ddb always show one process WAITing on "no_fltram5"
rd> (usually javadoc). The hung machines are usually pingable, KB leds
rd> always work and ddb can be entered from the keyboard. A full
rd> problem report, kernel/5777 [1], including dmesgs and ddb report was
rd> filed on 3/23/2008. I have since updated that report with details
rd> about the subsequent hangs found on a Sempron and amd64 laptop.

rd> Sorry to trouble tech@ but no response to the bug report or the 3/27
rd> report on misc@ prompted me to make sure the report hasn't fallen
rd> through the cracks. I suspect getting 4.3 out the door is occupying
rd> serious developer time so no panic.

rd> Let me know if there is further information that would be helpful to
rd> analyze / solve this problem.

rd> [1] <http://cvs.openbsd.org/cgi-bin/query-pr-wrapper?full=yes&textonly=yes&numbers=5777>

I have been able to recreate the kernel hang on a 4th machine, soekris
5501, running (GENERIC) #736: Fri Apr 4 17:28:40 MDT 2008.

The "WAIT no_fltram5" hang has now occurred on 4 machines running
{amd64,i386}. It has occurred with both GENERIC and GENERIC.MP
kernels.

The dmesg and ddb session have been added to the problem report. I
also have the associated crash dump. If a developer is interested, I
can make any or all of the crash dumps available (8 in total).

OpenBSD shouldn't be susceptible to this problem. Can someone help to
further analyze or resolve the problem?