Ticket #2484 (closed defect: fixed)

Opened 9 years ago

Last modified 9 years ago

segv after codepage change

Reported by: gotar Owned by: angel_il
Priority: critical Milestone: 4.8.0-pre1
Component: mcedit Version: 4.7.5
Keywords: Cc:
Blocked By: Blocking:
Branch state: Votes for changeset: committed-master

Description

  1. mcedit file (with some 8-bit encoding, like latin2)
  2. type in some non-ascii letters (e.g. polish ąłśźż do harm while ćęńó are safe)
  3. change encoding to UTF-8 (alt-e, c)
  4. move cursor through these specified letters.

What seems weird is that not every non-7-bit letter causes segfault. Above ąłśźż are b1 b3 b6 bf bc in hex, while ćęńó are e6 ea f1 f3.

Moreover: consider ąłśóćźżłąśśą string - after switching to UTF-8 editor shows 10 dots (..........) while there was 12 letters. Having cursor at EOL left arrow moves it 3 cells left, next keypress does the same, third moves properly 1 cell, fourth breaks mcedit.

Apparently it tries to interpret bytes as unicode and seems to miss some sanity tests.

Change History

comment:1 Changed 9 years ago by angel_il

  • Owner set to angel_il
  • Status changed from new to accepted

comment:2 Changed 9 years ago by angel_il

try this patch

diff --git a/src/editor/edit.c b/src/editor/edit.c
index 494f015..ec7a922 100644
--- a/src/editor/edit.c
+++ b/src/editor/edit.c
@@ -1852,8 +1852,13 @@ edit_get_prev_utf (WEdit * edit, long byte_index, int *char_width)
     if (str != buf)
         str = g_utf8_find_prev_char (buf, str);

-    res = g_utf8_get_char_validated (str, -1);
-
+    if (str != NULL)
+        res = g_utf8_get_char_validated (str, -1);
+    else
+    {
+        *char_width = 0;
+        return 0;
+    }
     if (res < 0)
     {
         ch = *str;

comment:3 Changed 9 years ago by angel_il

every non-7-bit letter causes segfault

this my bug... i'm foget test 'str' on NULL before call 'g_utf8_get_char_validated'

comment:4 Changed 9 years ago by angel_il

  • severity changed from no branch to on review

branch: 2484_editor_get_prev_utf8_fix
changeset: http://www.midnight-commander.org/changeset/2484_editor_get_prev_utf8_fix

please review

PS: i think is "stable candidate"...

comment:5 Changed 9 years ago by andrew_b

  • Keywords stable-candidate added
  • Votes for changeset set to andrew_b
  • Milestone changed from 4.8 to 4.8.0-pre1

comment:6 Changed 9 years ago by slavazanko

  • Votes for changeset changed from andrew_b to andrew_b slavazanko
  • severity changed from on review to approved

comment:7 Changed 9 years ago by angel_il

  • Votes for changeset changed from andrew_b slavazanko to committed-master
  • severity changed from approved to merged

comment:8 Changed 9 years ago by angel_il

  • Status changed from accepted to testing
  • Keywords stable-candidate removed
  • Resolution set to fixed

cherry-picked: 4.7.0-stable (028c8dd12259a252a43ad2fa09a2f38a4f4cf718)
cherry-picked: 4.7.5-stable (9ff73b6d23f0f8b47d0dcbcffab0f64eeb99d3bf)

comment:9 Changed 9 years ago by angel_il

  • Status changed from testing to closed

cherry-picked: 4.7.0-stable (9258fa658d0f3da410493801b08aa164a78327b0)
cherry-picked: 4.7.5-stable (3b7477aae63360fc42bd1c5f831319cb654290a0)

Note: See TracTickets for help on using tickets.