Re: [PATCH v4 0/7] Thread handle to thread info mapping

[Kevin Buettner <kevinb at redhat dot com>]

Ping.

Any (further) comments on this patch set?

Kevin

On Wed, 16 Aug 2017 09:25:42 -0700
Kevin Buettner <kevinb@redhat.com> wrote:

> This patch set introduces support for mapping thread handles to the
> thread_info structs which GDB uses to keep track of threads in the
> inferiors which it's debugging.  I need this in order to find the GDB
> thread which corresponds to a saved thread handle (e.g.  pthread_t)
> within an implementation of a thread library built atop pthreads.  The
> mechanism is general enough however to support other thread handle
> representations as well.
> 
> Simon reviewed the non-documentation parts of the v3 version of this
> patch set.  I think I have addressed all of his concerns from that
> review.
> 
> Eli approved the two documentation patches.  There have been no
> changes from v3 to v4, but I'm including them here for completeness.
> 
> Part 1 introduces a target method which maps a thread handle to
> the corresponding internal GDB thread object, i.e. something of type
> `struct thread_info *'.  An implementation of this new method is
> provided for the Linux thread target.
> 
> Part 2 adds a python interface for the mechanism introduced in part 1.
> 
> Part 3 is a documentation patch.  As noted above, Eli has already
> approved this patch which is unchanged from the v3 patch set.
> 
> Part 4 adds a test case.  I've incorporated all of Simon's suggestions
> from his review of the v3 patch.  I've also added three new tests
> which check the behavior of thread_from_thread_handle when passing
> a bad thread handle.  After making this change, I found a problem
> in part 6.
> 
> Part 5 is a bug fix for a problem discovered while working on part 6.
> 
> Part 6 adds support for remote targets.  In addition to making several
> v3 changes requested by Simon, I adjusted remote_thread_handle_to_thread_info
> so that would throw an exception (via a call to error()) when the handle
> sizes are mismatched.
> 
> Part 7 is a documentation patch for the remote protocol changes that
> were implemented in part 6.  This patch is unchanged from the v3 version
> and has already been approved by Eli.
> 
> Kevin