Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Tom de Vries <tdevries@suse.de>
To: gdb-patches@sourceware.org
Subject: [PATCH v4 03/11] [gdb/tui] Disassemble less often in tui_find_disassembly_address
Date: Wed,  7 Oct 2026 11:40:49 +0200	[thread overview]
Message-ID: <20261007094057.1811886-4-tdevries@suse.de> (raw)
In-Reply-To: <20261007094057.1811886-1-tdevries@suse.de>

In tui_find_disassembly_address we have the following code:
...
  CORE_ADDR new_low = pc;
  do
    {
      /* The previous value of NEW_LOW so we know if the new value is
	 different or not.  */
      CORE_ADDR prev_low = new_low;

      /* Find an address from which we can start disassembling.  */
      new_low = tui_find_backward_disassembly_start_address (new_low);

      /* Disassemble forward.  */
      next_addr = tui_disassemble (gdbarch, asm_lines, new_low, max_lines);
      if (asm_lines.empty ())
	break;
      last_addr = asm_lines.back ().addr;

      if (new_low == prev_low)
	break;
...

Say we enter this code with some pc, and have pc == prev_low == new_low before
the call to tui_find_backward_disassembly_start_address.

If tui_find_backward_disassembly_start_address also returns pc, there's no
point in trying to disassemble starting from pc.

Fix this by moving this check to just after
tui_find_backward_disassembly_start_address:
...
      new_low = tui_find_backward_disassembly_start_address (new_low);
      if (new_low == prev_low)
	break;
...

At the point of the break, asm_lines either is:
- empty because it hasn't been set yet, or
- the result of disassembling starting at prev_low, which also happens to be
  new_low.

While I think that's fine, I think it's clearer if we don't carry over
asm_lines from a previous iteration, and do asm_lines.clear () at the start of
the loop.
---
 gdb/tui/tui-disasm.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/gdb/tui/tui-disasm.c b/gdb/tui/tui-disasm.c
index 18439580ffb..a6cf12ba9c2 100644
--- a/gdb/tui/tui-disasm.c
+++ b/gdb/tui/tui-disasm.c
@@ -237,12 +237,16 @@ tui_find_disassembly_address (struct gdbarch *gdbarch, CORE_ADDR pc, int from)
   CORE_ADDR new_low = pc;
   do
     {
+      asm_lines.clear ();
+
       /* The previous value of NEW_LOW so we know if the new value is
 	 different or not.  */
       CORE_ADDR prev_low = new_low;
 
       /* Find an address from which we can start disassembling.  */
       new_low = tui_find_backward_disassembly_start_address (new_low);
+      if (new_low == prev_low)
+	break;
 
       /* Disassemble forward.  */
       next_addr = tui_disassemble (gdbarch, asm_lines, new_low, max_lines);
@@ -250,9 +254,6 @@ tui_find_disassembly_address (struct gdbarch *gdbarch, CORE_ADDR pc, int from)
 	break;
       last_addr = asm_lines.back ().addr;
 
-      if (new_low == prev_low)
-	break;
-
       /* If disassembling from the current value of NEW_LOW reached PC
 	 (or went past it) then this would do as a starting point if we
 	 can't find anything better, so remember it.  */
-- 
2.51.0


  parent reply	other threads:[~2026-10-07  9:41 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-07  9:40 [PATCH v4 00/11] [gdb/tui] Some section hole handling fixes Tom de Vries
2026-10-07  9:40 ` [PATCH v4 01/11] [gdb/tui] Use early exit in tui_find_disassembly_address Tom de Vries
2026-10-07  9:40 ` [PATCH v4 02/11] [gdb/tui] Make low_prev loop-local " Tom de Vries
2026-10-07  9:40 ` Tom de Vries [this message]
2026-10-07  9:40 ` [PATCH v4 04/11] [gdb/tui] Fix buglet " Tom de Vries
2026-10-07  9:40 ` [PATCH v4 05/11] [gdb] Factor out update_section_map Tom de Vries
2026-10-07  9:40 ` [PATCH v4 06/11] [gdb] Add section_addrmap Tom de Vries
2026-10-07  9:40 ` [PATCH v4 07/11] [gdb] Extend addrmap::find with low/high results Tom de Vries
2026-10-07  9:40 ` [PATCH v4 08/11] [gdb] Make gdb_disassembly return number of insns Tom de Vries
2026-10-07  9:40 ` [PATCH v4 09/11] [gdb] Handle sections when disassembling Tom de Vries
2026-10-07  9:40 ` [PATCH v4 10/11] [gdb] Don't stop disassembly at section hole Tom de Vries
2026-10-07  9:40 ` [PATCH v4 11/11] [RFC, gdb/tui] Don't scroll into non-code sections Tom de Vries

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20261007094057.1811886-4-tdevries@suse.de \
    --to=tdevries@suse.de \
    --cc=gdb-patches@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox