[lldb-dev] [Bug 20298] New: LLDB should have a crash handler to assist in reproducing crashes

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Mon Jul 14 12:47:40 PDT 2014


http://llvm.org/bugs/show_bug.cgi?id=20298

            Bug ID: 20298
           Summary: LLDB should have a crash handler to assist in
                    reproducing crashes
           Product: lldb
           Version: unspecified
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: All Bugs
          Assignee: lldb-dev at cs.uiuc.edu
          Reporter: rnk at google.com
                CC: chandlerc at gmail.com
    Classification: Unclassified

This is a relatively big feature request that I don't think will happen soon,
but it's a cool idea some I'm writing it down.

Taking inspiration from Clang, LLDB should have a crash handler that makes it
easy for users to file bug reports with reproductions.  The way this works in
Clang is that the driver process notices when the -cc1 process crashes and
automatically re-launches the -cc1 in pre-process only mode with
-frewrite-includes.  In LLDB, this might take the form of:

1. Recording all of the object files that were scanned for symbol information.
2. Recording all the commands the user wrote.
3. Make a tarball of all the files.
4. Make a shell script to unpack the tarball and launch LLDB on the user's
transcript.

I feel like there are some performance issues that would need to be worked out,
but it's an interesting idea.  For example, we should probably only do this if
the object files are small enough to reasonably package up.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-dev/attachments/20140714/56dc9ebd/attachment.html>


More information about the lldb-dev mailing list