bin/167103: dtrace(1) generates core dump trying to build perl with dtrace enabled
Mark Johnston
markjdb at gmail.com
Wed Jan 9 00:30:01 UTC 2013
The following reply was made to PR bin/167103; it has been noted by GNATS.
From: Mark Johnston <markjdb at gmail.com>
To: bug-followup at FreeBSD.org, swills at FreeBSD.org
Cc:
Subject: Re: bin/167103: dtrace(1) generates core dump trying to build perl
with dtrace enabled
Date: Tue, 8 Jan 2013 19:27:29 -0500
I've spent some more time digging into this and found a simple way to
reproduce it. The problem only seems to occur when a probe is located in
a static function. Some comments in dt_link.c indicate that there's some
special handling that's needed in this case, but I still don't quite
understand what's causing the problem.
I've placed a simple provider definition and sample program that
reproduce the issue here:
http://people.freebsd.org/~markj/dtrace/bin167103/
When the repro program is built, dtrace(1) will segfault when processing
the object file. When the patch attached to the PR is applied, dtrace
will just exit with a generic error:
dtrace -G -s provider.d repro.o
provider: failed to link script provider: an error was encountered while processing repro.o
*** [beforelinking] Error code 1
I'm still working on figuring this out.
Thanks,
-Mark
More information about the freebsd-bugs
mailing list