Re: [PATCH] Rewrite default_get_ada_task_ptid

[Tom Tromey <tromey@adacore.com>]

>>>>> "Tom" == Tom Tromey <tromey@adacore.com> writes:

Tom> Internal testing on a somewhat unusual configuration lead me to
Tom> examine default_get_ada_task_ptid.

Tom> This function computes a ptid in a way that, I think, maybe nothing
Tom> else in gdb uses, specifying both the LWP and the thread.  This in
Tom> turns means that Ada functionality like "info tasks" will never find
Tom> the thread corresponding to an Ada task.

Tom> This patch rewrites default_get_ada_task_ptid to use the most normal
Tom> form of ptid_t.  Redundant target-specific implementations are
Tom> removed.

I'm checking this in.

Tom> I couldn't figure out how to write a test case for this.  I had
Tom> believed that the problem arose from statically linking glibc, and
Tom> when the linux-thread-db was not pushed on the target stack; but more
Tom> testing shows that this is not in fact the problem, as
Tom> statically-linked executables work fine for me on my development
Tom> machine.  (If you're interested, you can try gdb.ada/tasks.exp with
Tom> static linking.)

It turns out that the problem was a missing backport of a patch to
handle the static linking case.  This caused gdb to call
default_get_ada_task_ptid rather than notice that the program in
question was using pthreads.

Tom