From: Doug Gilmore <Doug.Gilmore@imgtec.com>
To: Simon Marchi <simon.marchi@polymtl.ca>
Cc: Luis Machado <lgustavo@codesourcery.com>, <gdb-patches@sourceware.org>
Subject: Re: [PATCH] Fix PR 21337 v2: segfault when re-reading symbols with remote debugging.
Date: Wed, 10 May 2017 23:26:00 -0000 [thread overview]
Message-ID: <28fcce08-cab6-1e63-80d7-1d61c688cc10@imgtec.com> (raw)
In-Reply-To: <f431263b-02f6-67f8-29df-753f367c2ebf@imgtec.com>
On 04/29/2017 10:14 PM, Doug Gilmore wrote:
>
> >>> ...
> >> I'll need to take a look. Last time I tried I it was more difficult
> >> to expose the problem on the native build of GDB.
> >
> > reread_symbols is called when using the run (run_command_1), attach (attach_post_wait which then calls setup_inferior) and load (load_command) commands. So maybe something like this would reproduce it?
> >
> > - compile test program
> > - launch gdb with test program
> > - touch test program
> > - run
> >
> > Simon
> >
> Hi Simon,
>
> What I meant was that when I previously did tests with the native
> build, for some reason the freed data was not being overwritten, or
> possible written with the same data, at the time read_symbols was
> called in reread_symbols. Thus problem wasn't exposed before the
> objfiles_changed is eventually called in reread_symbols.
>
> I just did a test with a native build of gdb (7.9.1) and the problem
> was exposed, so chances are it will be also be exposed with a ToT
> build.
>
> Doug
>
It tooks some experimentation, but I found a quick, and it appears
reliable, way to reproduce the problem natively.
I have attached a new patch with commit with a detailed explanation
for the commit log entry.
OK to apply?
Thanks,
Doug
Fix PR 21337: segfault when re-reading symbols.
Fix issue exposed by commit 3e29f34.
The basic issue is that section data referenced through an objfile
pointer can also be referenced via the program-space data pointer,
although via a separate mapping mechanism, which is set up by
update_section_map. Thus once section data attached to an objfile
pointer is released, the section map associated with the program-space
data pointer must be marked dirty to ensure that update_section_map is
called to prevent stale data being referenced. For the matter at hand
this marking is being done via a call to objfiles_changed.
Before commit 3e29f34 objfiles_changed could be called after all of
the objfile pointers were processed in reread_symbols since section
data references via the program-space data pointer would not occur in
the calls of read_symbols performed by reread_symbols.
With commit 3e29f34 MIPS target specific calls to find_pc_section were
added to the code for DWARF information processing, which is called
via read_symbols. Thus in reread_symbols the call to objfiles_changed
needs to be called before calling read_symbols, otherwise stale
section data can be referenced.
Thanks to Luis Machado for providing text for the main comment
associated with the change.
gdb/
2017-??-?? Doug Gilmore <Doug.Gilmore@Doug.Gilmore@imgtec.com>
* symfile.c (reread_symbols): Fix PR 21337.
gdb/testsuite
2017-??-?? Doug Gilmore <Doug Gilmore@rDoug.Gilmore@imgtec.com>
PR gdb/r21337
* gdb.base/pr21337.c: New file.
* gdb.base/pr21337.exp: New file.
* gdb.base/pr21337.gdb: New file.
---
gdb/symfile.c | 20 +++++++++++--
gdb/testsuite/gdb.base/pr21337.c | 4 +++
gdb/testsuite/gdb.base/pr21337.exp | 57 ++++++++++++++++++++++++++++++++++++++
gdb/testsuite/gdb.base/pr21337.gdb | 13 +++++++++
4 files changed, 91 insertions(+), 3 deletions(-)
create mode 100644 gdb/testsuite/gdb.base/pr21337.c
create mode 100644 gdb/testsuite/gdb.base/pr21337.exp
create mode 100644 gdb/testsuite/gdb.base/pr21337.gdb
diff --git a/gdb/symfile.c b/gdb/symfile.c
index 846aabe..0492f56 100644
--- a/gdb/symfile.c
+++ b/gdb/symfile.c
@@ -2576,6 +2576,9 @@ reread_symbols (void)
/* Free the obstacks for non-reusable objfiles. */
psymbol_bcache_free (objfile->psymbol_cache);
objfile->psymbol_cache = psymbol_bcache_init ();
+
+ /* NB: after this call to obstack_free, objfiles_changed
+ will need to be called (see discussion below). */
obstack_free (&objfile->objfile_obstack, 0);
objfile->sections = NULL;
objfile->compunit_symtabs = NULL;
@@ -2628,6 +2631,20 @@ reread_symbols (void)
clear_complaints (&symfile_complaints, 1, 1);
objfile->flags &= ~OBJF_PSYMTABS_READ;
+
+ /* We are about to read new symbols and potentially also DWARF
+ information. Some targets may want to pass addresses read from
+ DWARF DIE's through an adjustment function before saving them, like
+ MIPS, which may call into "find_pc_section". When called, that
+ function will make use of per-objfile program space data.
+
+ Since we discarded our section information above, we have dangling
+ pointers in the per-objfile program space data structure. Force
+ GDB to update the section mapping information by letting it know
+ the objfile has changed, making the dangling pointers point to
+ correct data again. */
+ objfiles_changed ();
+
read_symbols (objfile, 0);
if (!objfile_has_symbols (objfile))
@@ -2660,9 +2677,6 @@ reread_symbols (void)
if (!new_objfiles.empty ())
{
- /* Notify objfiles that we've modified objfile sections. */
- objfiles_changed ();
-
clear_symtab_users (0);
/* clear_objfile_data for each objfile was called before freeing it and
diff --git a/gdb/testsuite/gdb.base/pr21337.c b/gdb/testsuite/gdb.base/pr21337.c
new file mode 100644
index 0000000..f8b643a
--- /dev/null
+++ b/gdb/testsuite/gdb.base/pr21337.c
@@ -0,0 +1,4 @@
+int main()
+{
+ return 0;
+}
diff --git a/gdb/testsuite/gdb.base/pr21337.exp b/gdb/testsuite/gdb.base/pr21337.exp
new file mode 100644
index 0000000..d7718b4
--- /dev/null
+++ b/gdb/testsuite/gdb.base/pr21337.exp
@@ -0,0 +1,57 @@
+# Copyright 1998-2017 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+
+set prototypes 1
+
+# build the test case
+
+set testfile "pr21337"
+set srcfile ${testfile}.c
+# Cygwin needs $EXEEXT.
+set binfile [standard_output_file ${testfile}$EXEEXT]
+# set binfile ${testfile}
+
+set gdbfile [standard_output_file ${testfile}.gdb]
+
+if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
+ untested "failed to compile first testcase"
+ return -1
+}
+
+# Using the source command to read commands from a file is important,
+# otherwise section data is freed and reallocated using the same
+# memory locations and the bug is not exposed.
+
+set ifd [open "$srcdir/$subdir/${testfile}.gdb" r]
+set ofd [open $gdbfile w]
+while {[gets $ifd line] >= 0} {
+ regsub $testfile $line $binfile tmp
+ puts $ofd $tmp
+}
+close $ifd
+close $ofd
+
+gdb_start
+
+if [is_remote target] {
+ unsupported $test
+} else {
+ gdb_test "source $gdbfile" ".*source-command-completed.*" "source $testfile.gdb"
+ gdb_test "source $gdbfile" ".*source-command-completed.*" "source $testfile.gdb"
+}
+
+# End of tests.
+
+return 0
diff --git a/gdb/testsuite/gdb.base/pr21337.gdb b/gdb/testsuite/gdb.base/pr21337.gdb
new file mode 100644
index 0000000..42fac26
--- /dev/null
+++ b/gdb/testsuite/gdb.base/pr21337.gdb
@@ -0,0 +1,13 @@
+file pr21337
+shell sleep 1; touch pr21337
+run
+file
+file pr21337
+shell sleep 1; touch pr21337
+run
+file
+file pr21337
+shell sleep 1; touch pr21337
+run
+file
+p "source-command-completed"
--
1.9.1
next prev parent reply other threads:[~2017-05-10 23:26 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-03-31 23:04 [PATCH] [mips] Fix PR 21337 v1: " Doug Gilmore
2017-04-10 16:01 ` Doug Gilmore
2017-04-12 18:52 ` Luis Machado
2017-04-12 21:42 ` Doug Gilmore
2017-04-13 18:56 ` [PATCH] Fix PR 21337 v2: " Doug Gilmore
2017-04-14 15:33 ` Luis Machado
2017-04-22 2:15 ` Simon Marchi
2017-04-25 18:16 ` Doug Gilmore
2017-04-27 19:46 ` Simon Marchi
2017-04-28 23:44 ` Doug Gilmore
2017-04-29 1:13 ` Luis Machado
2017-04-29 1:42 ` Simon Marchi
2017-04-29 17:12 ` Doug Gilmore
2017-04-29 23:26 ` Simon Marchi
2017-04-30 5:14 ` Doug Gilmore
2017-05-10 23:26 ` Doug Gilmore [this message]
2017-05-12 3:29 ` Simon Marchi
2017-05-12 19:24 ` Doug Gilmore
2017-05-16 23:11 ` [PATCH] Fix PR 21337 (v3): " Doug Gilmore
2017-06-06 16:08 ` [PING][PATCH] " Doug Gilmore
2017-06-23 18:20 ` [PING^2][PATCH] " Doug Gilmore
2017-06-25 11:25 ` [PATCH] " Simon Marchi
2017-06-27 17:29 ` [PATCH] Fix PR 21337 (v4): " Doug Gilmore
2017-06-27 21:35 ` Doug Gilmore
2017-06-28 2:01 ` Maciej W. Rozycki
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=28fcce08-cab6-1e63-80d7-1d61c688cc10@imgtec.com \
--to=doug.gilmore@imgtec.com \
--cc=gdb-patches@sourceware.org \
--cc=lgustavo@codesourcery.com \
--cc=simon.marchi@polymtl.ca \
/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