Blame SOURCES/gdb-rhbz1909902-frame_id_p-assert-2.patch

0a406a
From FEDORA_PATCHES Mon Sep 17 00:00:00 2001
0a406a
From: Pedro Alves <pedro@palves.net>
0a406a
Date: Sat, 31 Oct 2020 00:27:18 +0000
0a406a
Subject: gdb-rhbz1909902-frame_id_p-assert-2.patch
0a406a
0a406a
;; Backport patch #2 which fixes a frame_id_p assertion failure (RH BZ 1909902).
0a406a
0a406a
Fix frame cycle detection
0a406a
0a406a
The recent commit to make scoped_restore_current_thread's cdtors
0a406a
exception free regressed gdb.base/eh_return.exp:
0a406a
0a406a
  Breakpoint 1, 0x00000000004012bb in eh2 (gdb/frame.c:641: internal-error: frame_id get_frame_id(frame_info*): Assertion `stashed' failed.
0a406a
  A problem internal to GDB has been detected,
0a406a
  further debugging may prove unreliable.
0a406a
  Quit this debugging session? (y or n) FAIL: gdb.base/eh_return.exp: hit breakpoint (GDB internal error)
0a406a
0a406a
That testcase uses __builtin_eh_return and, before the regression, the
0a406a
backtrace at eh2 looked like this:
0a406a
0a406a
 (gdb) bt
0a406a
 #0  0x00000000004006eb in eh2 (p=0x4006ec <continuation>) at src/gdb/testsuite/gdb.base/eh_return.c:54
0a406a
 Backtrace stopped: previous frame identical to this frame (corrupt stack?)
0a406a
0a406a
That "previous frame identical to this frame" is caught by the cycle
0a406a
detection based on frame id.
0a406a
0a406a
The assertion failing is this one:
0a406a
0a406a
 638           /* Since this is the first frame in the chain, this should
0a406a
 639              always succeed.  */
0a406a
 640           bool stashed = frame_stash_add (fi);
0a406a
 641           gdb_assert (stashed);
0a406a
0a406a
originally added by
0a406a
0a406a
  commit f245535cf583ae4ca13b10d47b3c7d3334593ece
0a406a
  Author:     Pedro Alves <palves@redhat.com>
0a406a
  AuthorDate: Mon Sep 5 18:41:38 2016 +0100
0a406a
0a406a
      Fix PR19927: Avoid unwinder recursion if sniffer uses calls parse_and_eval
0a406a
0a406a
The assertion is failing because frame #1's frame id was stashed
0a406a
before the id of frame #0 is stashed.  The frame id of frame #1 was
0a406a
stashed here:
0a406a
0a406a
 (top-gdb) bt
0a406a
 #0  frame_stash_add (frame=0x1e24c90) at src/gdb/frame.c:276
0a406a
 #1  0x0000000000669c1b in get_prev_frame_if_no_cycle (this_frame=0x19f8370) at src/gdb/frame.c:2120
0a406a
 #2  0x000000000066a339 in get_prev_frame_always_1 (this_frame=0x19f8370) at src/gdb/frame.c:2303
0a406a
 #3  0x000000000066a360 in get_prev_frame_always (this_frame=0x19f8370) at src/gdb/frame.c:2319
0a406a
 #4  0x000000000066b56c in get_frame_unwind_stop_reason (frame=0x19f8370) at src/gdb/frame.c:3028
0a406a
 #5  0x000000000059f929 in dwarf2_frame_cfa (this_frame=0x19f8370) at src/gdb/dwarf2/frame.c:1462
0a406a
 #6  0x00000000005ce434 in dwarf_evaluate_loc_desc::get_frame_cfa (this=0x7fffffffc070) at src/gdb/dwarf2/loc.c:666
0a406a
 #7  0x00000000005989a9 in dwarf_expr_context::execute_stack_op (this=0x7fffffffc070, op_ptr=0x1b2a053 "\364\003", op_end=0x1b2a053 "\364\003") at src/gdb/dwarf2/expr.c:1161
0a406a
 #8  0x0000000000596af6 in dwarf_expr_context::eval (this=0x7fffffffc070, addr=0x1b2a052 "\234\364\003", len=1) at src/gdb/dwarf2/expr.c:303
0a406a
 #9  0x0000000000597b4e in dwarf_expr_context::execute_stack_op (this=0x7fffffffc070, op_ptr=0x1b2a063 "", op_end=0x1b2a063 "") at src/gdb/dwarf2/expr.c:865
0a406a
 #10 0x0000000000596af6 in dwarf_expr_context::eval (this=0x7fffffffc070, addr=0x1b2a061 "\221X", len=2) at src/gdb/dwarf2/expr.c:303
0a406a
 #11 0x00000000005c8b5a in dwarf2_evaluate_loc_desc_full (type=0x1b564d0, frame=0x19f8370, data=0x1b2a061 "\221X", size=2, per_cu=0x1b28760, per_objfile=0x1a84930, subobj_type=0x1b564d0, subobj_byte_offset=0) at src/gdb/dwarf2/loc.c:2260
0a406a
 #12 0x00000000005c9243 in dwarf2_evaluate_loc_desc (type=0x1b564d0, frame=0x19f8370, data=0x1b2a061 "\221X", size=2, per_cu=0x1b28760, per_objfile=0x1a84930) at src/gdb/dwarf2/loc.c:2444
0a406a
 #13 0x00000000005cb769 in locexpr_read_variable (symbol=0x1b59840, frame=0x19f8370) at src/gdb/dwarf2/loc.c:3687
0a406a
 #14 0x0000000000663137 in language_defn::read_var_value (this=0x122ea60 <c_language_defn>, var=0x1b59840, var_block=0x0, frame=0x19f8370) at src/gdb/findvar.c:618
0a406a
 #15 0x0000000000663c3b in read_var_value (var=0x1b59840, var_block=0x0, frame=0x19f8370) at src/gdb/findvar.c:822
0a406a
 #16 0x00000000008c7d9f in read_frame_arg (fp_opts=..., sym=0x1b59840, frame=0x19f8370, argp=0x7fffffffc470, entryargp=0x7fffffffc490) at src/gdb/stack.c:542
0a406a
 #17 0x00000000008c89cd in print_frame_args (fp_opts=..., func=0x1b597c0, frame=0x19f8370, num=-1, stream=0x1aba860) at src/gdb/stack.c:890
0a406a
 #18 0x00000000008c9bf8 in print_frame (fp_opts=..., frame=0x19f8370, print_level=0, print_what=SRC_AND_LOC, print_args=1, sal=...) at src/gdb/stack.c:1394
0a406a
 #19 0x00000000008c92b9 in print_frame_info (fp_opts=..., frame=0x19f8370, print_level=0, print_what=SRC_AND_LOC, print_args=1, set_current_sal=1) at src/gdb/stack.c:1119
0a406a
 #20 0x00000000008c75f0 in print_stack_frame (frame=0x19f8370, print_level=0, print_what=SRC_AND_LOC, set_current_sal=1) at src/gdb/stack.c:366
0a406a
 #21 0x000000000070250b in print_stop_location (ws=0x7fffffffc9e0) at src/gdb/infrun.c:8110
0a406a
 #22 0x0000000000702569 in print_stop_event (uiout=0x1a8b9e0, displays=true) at src/gdb/infrun.c:8126
0a406a
 #23 0x000000000096d04b in tui_on_normal_stop (bs=0x1bcd1c0, print_frame=1) at src/gdb/tui/tui-interp.c:98
0a406a
 ...
0a406a
0a406a
Before the commit to make scoped_restore_current_thread's cdtors
0a406a
exception free, scoped_restore_current_thread's dtor would call
0a406a
get_frame_id on the selected frame, and we use
0a406a
scoped_restore_current_thread pervasively.  That had the side effect
0a406a
of stashing the frame id of frame #0 before reaching the path shown in
0a406a
the backtrace.  I.e., the frame id of frame #0 happened to be stashed
0a406a
before the frame id of frame #1.  But that was by chance, not by
0a406a
design.
0a406a
0a406a
This commit:
0a406a
0a406a
  commit 256ae5dbc73d1348850f86ee77a0dc3b04bc7cc0
0a406a
  Author:     Kevin Buettner <kevinb@redhat.com>
0a406a
  AuthorDate: Mon Oct 31 12:47:42 2016 -0700
0a406a
0a406a
      Stash frame id of current frame before stashing frame id for previous frame
0a406a
0a406a
Fixed a similar problem, by making sure get_prev_frame computes the
0a406a
frame id of the current frame before unwinding the previous frame, so
0a406a
that the cycle detection works properly.  That fix misses the scenario
0a406a
we're now running against, because if you notice, the backtrace above
0a406a
shows that frame #4 calls get_prev_frame_always, not get_prev_frame.
0a406a
I.e., nothing is calling get_frame_id on the current frame.
0a406a
0a406a
The fix here is to move Kevin's fix down from get_prev_frame to
0a406a
get_prev_frame_always.  Or actually, a bit further down to
0a406a
get_prev_frame_always_1 -- note that inline_frame_this_id calls
0a406a
get_prev_frame_always, so we need to be careful to avoid recursion in
0a406a
that scenario.
0a406a
0a406a
gdb/ChangeLog:
0a406a
0a406a
	* frame.c (get_prev_frame): Move get_frame_id call from here ...
0a406a
	(get_prev_frame_always_1): ... to here.
0a406a
	* inline-frame.c (inline_frame_this_id): Mention
0a406a
	get_prev_frame_always_1 in comment.
0a406a
0a406a
Change-Id: Id960c98ab2d072c48a436c3eb160cc4b2a5cfd1d
0a406a
0a406a
diff --git a/gdb/frame.c b/gdb/frame.c
0a406a
--- a/gdb/frame.c
0a406a
+++ b/gdb/frame.c
0a406a
@@ -2133,6 +2133,23 @@ get_prev_frame_always_1 (struct frame_info *this_frame)
0a406a
   if (get_frame_type (this_frame) == INLINE_FRAME)
0a406a
     return get_prev_frame_if_no_cycle (this_frame);
0a406a
 
0a406a
+  /* If this_frame is the current frame, then compute and stash its
0a406a
+     frame id prior to fetching and computing the frame id of the
0a406a
+     previous frame.  Otherwise, the cycle detection code in
0a406a
+     get_prev_frame_if_no_cycle() will not work correctly.  When
0a406a
+     get_frame_id() is called later on, an assertion error will be
0a406a
+     triggered in the event of a cycle between the current frame and
0a406a
+     its previous frame.
0a406a
+
0a406a
+     Note we do this after the INLINE_FRAME check above.  That is
0a406a
+     because the inline frame's frame id computation needs to fetch
0a406a
+     the frame id of its previous real stack frame.  I.e., we need to
0a406a
+     avoid recursion in that case.  This is OK since we're sure the
0a406a
+     inline frame won't create a cycle with the real stack frame.  See
0a406a
+     inline_frame_this_id.  */
0a406a
+  if (this_frame->level == 0)
0a406a
+    get_frame_id (this_frame);
0a406a
+
0a406a
   /* Check that this frame is unwindable.  If it isn't, don't try to
0a406a
      unwind to the prev frame.  */
0a406a
   this_frame->stop_reason
0a406a
@@ -2410,16 +2427,6 @@ get_prev_frame (struct frame_info *this_frame)
0a406a
      something should be calling get_selected_frame() or
0a406a
      get_current_frame().  */
0a406a
   gdb_assert (this_frame != NULL);
0a406a
-  
0a406a
-  /* If this_frame is the current frame, then compute and stash
0a406a
-     its frame id prior to fetching and computing the frame id of the
0a406a
-     previous frame.  Otherwise, the cycle detection code in
0a406a
-     get_prev_frame_if_no_cycle() will not work correctly.  When
0a406a
-     get_frame_id() is called later on, an assertion error will
0a406a
-     be triggered in the event of a cycle between the current
0a406a
-     frame and its previous frame.  */
0a406a
-  if (this_frame->level == 0)
0a406a
-    get_frame_id (this_frame);
0a406a
 
0a406a
   frame_pc_p = get_frame_pc_if_available (this_frame, &frame_pc);
0a406a
 
0a406a
diff --git a/gdb/inline-frame.c b/gdb/inline-frame.c
0a406a
--- a/gdb/inline-frame.c
0a406a
+++ b/gdb/inline-frame.c
0a406a
@@ -161,7 +161,8 @@ inline_frame_this_id (struct frame_info *this_frame,
0a406a
      real frame's this_id method.  So we must call
0a406a
      get_prev_frame_always.  Because we are inlined into some
0a406a
      function, there must be previous frames, so this is safe - as
0a406a
-     long as we're careful not to create any cycles.  */
0a406a
+     long as we're careful not to create any cycles.  See related
0a406a
+     comments in get_prev_frame_always_1.  */
0a406a
   *this_id = get_frame_id (get_prev_frame_always (this_frame));
0a406a
 
0a406a
   /* We need a valid frame ID, so we need to be based on a valid