Don't touch frame_info objects if frame cache was reinitialized

This fixes yet another bug exposed by ASAN + multi-target.exp

Running an Asan-enabled GDB against gdb.multi/multi-target.exp exposed
yet another latent GDB bug.  See here for the full log:

  https://sourceware.org/pipermail/gdb-patches/2020-July/170761.html

As Simon described, the problem is:

 - We create a new frame_info object in restore_selected_frame (by
   calling find_relative_frame)

 - The frame is allocated on the frame_cache_obstack

 - In frame_unwind_try_unwinder, we try to find an unwinder for that
   frame

 - While trying unwinders, memory read fails because the remote target
   closes, because of "monitor exit"

 - That calls reinit_frame_cache (as shown above), which resets
   frame_cache_obstack

 - When handling the exception in frame_unwind_try_unwinder, we try to
   set some things on the frame_info object (like *this_cache, which
   in fact tries to write into frame_info::prologue_cache), but the
   frame_info object is no more, it went away with the obstack.

Fix this by maintaining a frame cache generation counter.  Then in
exception handling code paths, don't touch frame objects if the
generation is not the same as it was on entry.

This commit generalizes the gdb.server/server-kill.exp testcase and
reuses it to test the scenario in question.  The new tests fail
without the GDB fix.

gdb/ChangeLog:

	* frame-unwind.c (frame_unwind_try_unwinder): On exception, don't
	touch THIS_CACHE/THIS_FRAME if the frame cache was cleared
	meanwhile.
	* frame.c (frame_cache_generation, get_frame_cache_generation):
	New.
	(reinit_frame_cache): Increment FRAME_CACHE_GENERATION.
	(get_prev_frame_if_no_cycle): On exception, don't touch
	PREV_FRAME/THIS_FRAME if the frame cache was cleared meanwhile.
	* frame.h (get_frame_cache_generation): Declare.

gdb/testsuite/ChangeLog:

	* gdb.server/server-kill.exp (prepare): New, factored out from the
	top level.
	(kill_server): New.
	(test_tstatus, test_unwind_nosyms, test_unwind_syms): New.
	(top level) : Call test_tstatus, test_unwind_nosyms, test_unwind_syms.

1 files changed, 90 insertions, 24 deletions

diff --git a/gdb/testsuite/gdb.server/server-kill.exp b/gdb/testsuite/gdb.server/server-kill.exp
index 0072b28eb97..37b42460730 100644
--- a/gdb/testsuite/gdb.server/server-kill.exp
+++ b/gdb/testsuite/gdb.server/server-kill.exp
@@ -15,6 +15,9 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
+# Check that GDB handles GDBserver disconnecting abruptly, in several
+# scenarios.
+
 load_lib gdbserver-support.exp
 
 standard_testfile
@@ -23,40 +26,103 @@ if {[skip_gdbserver_tests]} {
     return 0
 }
 
-if { [prepare_for_testing "failed to prepare" ${testfile}] } {
+if { [build_executable "failed to prepare" ${testfile}] } {
     return -1
 }
 
-# Make sure we're disconnected, in case we're testing with an
-# extended-remote board, therefore already connected.
-gdb_test "disconnect" ".*"
+# Spawn GDBserver, run to main, extract GDBserver's PID and save it in
+# the SERVER_PID global.
+
+proc prepare {} {
+    global binfile gdb_prompt srcfile decimal
+    global server_pid
+
+    clean_restart $binfile
+
+    # Make sure we're disconnected, in case we're testing with an
+    # extended-remote board, therefore already connected.
+    gdb_test "disconnect" ".*"
 
-gdbserver_run ""
+    gdbserver_run ""
 
-# Continue past server_pid assignment.
-gdb_breakpoint ${srcfile}:[gdb_get_line_number "i = 0;"]
-gdb_continue_to_breakpoint "after server_pid assignment"
+    # Continue past server_pid assignment.
+    gdb_breakpoint ${srcfile}:[gdb_get_line_number "i = 0;"]
+    gdb_continue_to_breakpoint "after server_pid assignment"
 
-# Get the pid of GDBServer.
-set test "p server_pid"
-gdb_test_multiple $test $test {
-    -re " = ($decimal)\r\n$gdb_prompt $" {
-	set server_pid $expect_out(1,string)
-	pass $test
+    # Get the pid of GDBServer.
+    set test "p server_pid"
+    set server_pid 0
+    gdb_test_multiple $test $test {
+	-re " = ($decimal)\r\n$gdb_prompt $" {
+	    set server_pid $expect_out(1,string)
+	    pass $test
+	}
     }
+
+    if {$server_pid == 0} {
+	return 0
+    }
+
+    return 1
 }
 
-if ![info exists server_pid] {
-    return -1
+# Kill GDBserver using the PID saved by prepare.
+
+proc kill_server {} {
+    global server_pid
+
+    remote_exec target "kill -9 $server_pid"
+}
+
+# Test issuing "tstatus" right after the connection is dropped.
+
+proc_with_prefix test_tstatus {} {
+    if ![prepare] {
+	return
+    }
+
+    kill_server
+
+    # Enable trace status packet which is disabled after the
+    # connection if the remote target doesn't support tracepoint at
+    # all.  Otherwise, no RSP packet is sent out.
+    gdb_test_no_output "set remote trace-status-packet on"
+
+    # Force GDB to talk with GDBserver, so that we can get the
+    # "connection closed" error.
+    gdb_test "tstatus" {Remote connection closed|Remote communication error\.  Target disconnected\.: Connection reset by peer\.}
+}
+
+# Test unwinding with no debug/unwind info, right after the connection
+# is dropped.
+
+proc_with_prefix test_unwind_nosyms {} {
+    if ![prepare] {
+	return
+    }
+
+    # Remove symbols, so that we try to unwind with one of the
+    # heuristic unwinders, and read memory from within its sniffer.
+    gdb_unload
+
+    kill_server
+
+    gdb_test "bt" "(Target disconnected|Remote connection closed|Remote communication error).*"
 }
 
-remote_exec target "kill -9 $server_pid"
+# Test unwinding with debug/unwind info, right after the connection is
+# dropped.
 
-# Enable trace status packet which is disabled after the connection
-# if the remote target doesn't support tracepoint at all.  Otherwise,
-# no RSP packet is sent out.
-gdb_test_no_output "set remote trace-status-packet on"
+proc_with_prefix test_unwind_syms {} {
+    if ![prepare] {
+	return
+    }
+
+    kill_server
+
+    gdb_test "bt" "(Target disconnected|Remote connection closed|Remote communication error).*"
+}
 
-# Force GDB to talk with GDBserver, so that we can get the
-# "connection closed" error.
-gdb_test "tstatus" {Remote connection closed|Remote communication error\.  Target disconnected\.: Connection reset by peer\.}
+test_tstatus
+test_unwind_nosyms
+test_unwind_syms