public inbox for gdb-cvs@sourceware.org
help / color / mirror / Atom feed
From: Simon Marchi <simark@sourceware.org>
To: gdb-cvs@sourceware.org
Subject: [binutils-gdb] gdb: initialize ctf_context::builder in create_partial_symtab
Date: Thu, 31 Mar 2022 21:18:31 +0000 (GMT)	[thread overview]
Message-ID: <20220331211831.F226E3857034@sourceware.org> (raw)

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=59f837cb11e3b96c581fe23b79f6995b7c8177ee

commit 59f837cb11e3b96c581fe23b79f6995b7c8177ee
Author: Simon Marchi <simon.marchi@polymtl.ca>
Date:   Thu Mar 31 14:20:02 2022 -0400

    gdb: initialize ctf_context::builder in create_partial_symtab
    
    I built a random project with -gctf, in order to test the CTF support in
    GDB.  With my ASan/UBSan/etc-enabled build of GDB, I get:
    
        $ ./gdb --data-directory=data-directory /tmp/babeltrace-ctf/src/lib/.libs/libbabeltrace2.so.0.0.0
        ...
        Reading symbols from /tmp/babeltrace-ctf/src/lib/.libs/libbabeltrace2.so.0.0.0...
        /home/simark/src/binutils-gdb/gdb/ctfread.c:1545:31: runtime error: member call on misaligned address 0xbebebebebebebebe for type 'struct buildsym_compunit', which requires 8 byte alignment
        0xbebebebebebebebe: note: pointer points here
    
    The 0xbebebebebebebebe value is a sign that the ctf_context::builder
    field is uninitialized.  The problem probably goes under the radar if
    the field happens to be zero-initialized, because ctf_start_archive
    contains this code:
    
      if (ccx->builder == nullptr)
        {
          ccx->builder = new buildsym_compunit (of,
                          of->original_name, nullptr, language_c, 0);
    
    If the field was zero-initialized (by chance), this will create a new
    buildsym_compunit.  But if the field was purposely filled with random
    bytes by one of the sanitizers, we won't create a buildsym_compunit here
    and we'll continue with ccx->builder equal to 0xbebebebebebebebe.
    
    Fix this the easy way by initializing ccx->builder where the other
    ctf_context fields are initialized (yeah, this code could be made nicer
    C++, but I am going for the obvious fix here).
    
    With this patch, this passes cleanly on my system:
    
      $ make check TESTS="gdb.ctf/*.exp" RUNTESTFLAGS="CC_FOR_TARGET=/opt/gcc/git/bin/gcc"
      # of expected passes            40
    
    ... where /opt/gcc/git/bin/gcc is a gcc with CTF support, given my
    system gcc does not have it.
    
    Change-Id: Idea1b0cf3e3708b72ecb16b1b60222439160f9b9

Diff:
---
 gdb/ctfread.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/gdb/ctfread.c b/gdb/ctfread.c
index 585c664cce2..8da38cc70af 100644
--- a/gdb/ctfread.c
+++ b/gdb/ctfread.c
@@ -1449,6 +1449,7 @@ create_partial_symtab (const char *name,
   pst->context.of = objfile;
   pst->context.partial_symtabs = partial_symtabs;
   pst->context.pst = pst;
+  pst->context.builder = nullptr;
 
   return pst;
 }


                 reply	other threads:[~2022-03-31 21:18 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=20220331211831.F226E3857034@sourceware.org \
    --to=simark@sourceware.org \
    --cc=gdb-cvs@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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).