[binutils-gdb] [LynxOS] GDBserver crash debugging threaded program

[Joel Brobecker <brobecke at sourceware dot org>]

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=80d82c196402f6a61aa84452104b9aaed364eb42

commit 80d82c196402f6a61aa84452104b9aaed364eb42
Author: Joel Brobecker <brobecker@adacore.com>
Date:   Mon Nov 23 09:56:23 2015 -0800

    [LynxOS] GDBserver crash debugging threaded program
    
    This crash is observable by debugging a threaded program on LynxOS.
    On the GDB side, this is what we would see:
    
        % gdb q
        (gdb) target remote machine:4444
        (gdb) break q.adb:6
        (gdb) cont
        [gdb hits breakpoint]
        (gdb) cont
        Remote connection closed    <<<--- expected: [Inferior 1 (Remote target) exited normally]
    
    On the gdbserver side, which was launched as usual:
    
        % gdbserver --once :4444 q
        Segmentation fault (core dumped)
    
    Ooops!
    
    The problem happens while GDB is trying to handle the thread termination
    event of the thread that hit the breakpoint. It started happening after
    the following change was made:
    
        commit 96e7a1eb6d09fda9e22e112e35e7d0085a8f4fd0
        Date:   Fri Oct 16 11:08:38 2015 -0400
        Subject: gdbserver: Reset current_thread when the thread is removed.
    
        Reset current_thread and make sure 'remove_process' is used
        after all associated threads have been removed first.
    
    More precisely:
    
      . GDBserver receives the execution-resume order;
    
      . lynx-low resumes it succesfully, and then relies on lynx_wait_1
        to wait for the next event;
    
      . We quickly receive one, which lynx_wait_1 analyzes to be
        a "thread exit" event, and therefore does...
    
              case SIGTHREADEXIT:
                remove_thread (find_thread_ptid (new_ptid));
                lynx_continue (new_ptid);
                goto retry;
    
        => remove_thread causes current_thread to be set to NULL...
           (that's the recent change mentioned above)
    
        => ... which causes problems during lynx_continue, because
           it calls lynx_resume, which calls regcache_invalidate,
           which unfortunately assumes that CURRENT_THREAD is not NULL:
    
            void
            regcache_invalidate (void)
            {
              /* Only update the threads of the current process.  */
    SEGV!-->  int pid = ptid_get_pid (current_thread->entry.id);
    
              find_inferior (&all_threads, regcache_invalidate_one, &pid);
            }
    
    Since the problem at hand is caused by trying to figure out which
    inferior to reset the regcache for, and since lynx_resume actually
    had that info, this patch fixes the problem by introducing a new
    routine called regcache_invalidate_pid, which invalidates the cache
    of the given pid; and then modifies lynx_resume use that new routine
    rather than relying on regcache_invalidate to invalidate the regcache
    of the expected inferior.
    
    gdb/gdbserver/ChangeLog:
    
            * regcache.h (regcache_invalidate_pid): Add declaration.
            * regcache.c (regcache_invalidate_pid): New function, extracted
            from regcache_invalidate.
            (regcache_invalidate): Reimplement using regcache_invalidate_pid.
            Add trivial documentation comment.
            * lynx-low.c: Use regcache_invalidate_pid instead of
            regcache_invalidate.

Diff:
---
 gdb/gdbserver/ChangeLog  | 10 ++++++++++
 gdb/gdbserver/lynx-low.c |  2 +-
 gdb/gdbserver/regcache.c | 12 +++++++++++-
 gdb/gdbserver/regcache.h |  4 ++++
 4 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/gdb/gdbserver/ChangeLog b/gdb/gdbserver/ChangeLog
index 7e9d695..e265798 100644
--- a/gdb/gdbserver/ChangeLog
+++ b/gdb/gdbserver/ChangeLog
@@ -1,5 +1,15 @@
 2015-11-23  Joel Brobecker  <brobecker@adacore.com>
 
+	* regcache.h (regcache_invalidate_pid): Add declaration.
+	* regcache.c (regcache_invalidate_pid): New function, extracted
+	from regcache_invalidate.
+	(regcache_invalidate): Reimplement using regcache_invalidate_pid.
+	Add trivial documentation comment.
+	* lynx-low.c: Use regcache_invalidate_pid instead of
+	regcache_invalidate.
+
+2015-11-23  Joel Brobecker  <brobecker@adacore.com>
+
 	* configure.ac: Do not call AC_CHECK_TYPES for Elf32_auxv_t
 	and Elf64_auxv_t if the target is Android.
 
diff --git a/gdb/gdbserver/lynx-low.c b/gdb/gdbserver/lynx-low.c
index 0582399..36ab0a3 100644
--- a/gdb/gdbserver/lynx-low.c
+++ b/gdb/gdbserver/lynx-low.c
@@ -343,7 +343,7 @@ lynx_resume (struct thread_resume *resume_info, size_t n)
   if (ptid_equal (ptid, minus_one_ptid))
     ptid = thread_to_gdb_id (current_thread);
 
-  regcache_invalidate ();
+  regcache_invalidate_pid (ptid_get_pid (ptid));
 
   errno = 0;
   lynx_ptrace (request, ptid, 1, signal, 0);
diff --git a/gdb/gdbserver/regcache.c b/gdb/gdbserver/regcache.c
index e11b173..b9311fe 100644
--- a/gdb/gdbserver/regcache.c
+++ b/gdb/gdbserver/regcache.c
@@ -107,13 +107,23 @@ regcache_invalidate_one (struct inferior_list_entry *entry,
   return 0;
 }
 
+/* See regcache.h.  */
+
+void
+regcache_invalidate_pid (int pid)
+{
+  find_inferior (&all_threads, regcache_invalidate_one, &pid);
+}
+
+/* See regcache.h.  */
+
 void
 regcache_invalidate (void)
 {
   /* Only update the threads of the current process.  */
   int pid = ptid_get_pid (current_thread->entry.id);
 
-  find_inferior (&all_threads, regcache_invalidate_one, &pid);
+  regcache_invalidate_pid (pid);
 }
 
 #endif
diff --git a/gdb/gdbserver/regcache.h b/gdb/gdbserver/regcache.h
index d2d9fc7..a0be95e 100644
--- a/gdb/gdbserver/regcache.h
+++ b/gdb/gdbserver/regcache.h
@@ -74,6 +74,10 @@ void free_register_cache (struct regcache *regcache);
 
 void regcache_invalidate_thread (struct thread_info *);
 
+/* Invalidate cached registers for all threads of the given process.  */
+
+void regcache_invalidate_pid (int pid);
+
 /* Invalidate cached registers for all threads of the current
    process.  */