public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Wang Bing <wangbing6@huawei.com>
To: <libc-alpha@sourceware.org>
Cc: <hjl.tools@gmail.com>, <nixiaoming@huawei.com>,
	<szabolcs.nagy@arm.com>, <wangbing6@huawei.com>,
	<yanhuijun@huawei.com>, <zhongjubin@huawei.com>
Subject: [PATCH v2] dlsym: Add RTLD_PROBE to dlsym only probe symbol without add dependency.
Date: Wed, 23 Nov 2022 10:21:37 +0800	[thread overview]
Message-ID: <20221123022137.65977-1-wangbing6@huawei.com> (raw)
In-Reply-To: <CAMe9rOo=AR4dj_O3LyUhVMzFWVGxj-6QjpuMBint0T_JhGLxvw@mail.gmail.com>

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="yes", Size: 4868 bytes --]

dlsym with RTLD_DEFAULT add dependency for target symbol, if program 
use dlsym only to detect if a symbol exist, and will not use it.

this operation will make unable to dlclose so file containing target 
symbol, add RTLD_PROBE to support symbol probe.

The reason why not useing struct link_map * as an argument to dlsym is 
that there may be more than one same-named symbol in different so 
files(provided by different vendors or components) in one process, and 
by -Wl,-z,now,only the first symbol is needed, and we cannot be sure 
which struct link_map* should be send to dlsym.

Signed-off-by: Wang Bing <wangbing6@huawei.com>

---
v2: fix code corruption
v1: https://public-inbox.org/libc-alpha/7cad26c73f5b4172bbcf570882bf9d3d@huawei.com/#R

---
 dlfcn/dlfcn.h              |  4 ++++
 elf/Makefile               |  1 +
 elf/dl-sym.c               | 13 +++++++++---
 elf/tst-dlsym-rtld-probe.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 65 insertions(+), 3 deletions(-)
 create mode 100644 elf/tst-dlsym-rtld-probe.c

diff --git a/dlfcn/dlfcn.h b/dlfcn/dlfcn.h
index 6f7cad8682..ab709883a6 100644
--- a/dlfcn/dlfcn.h
+++ b/dlfcn/dlfcn.h
@@ -49,6 +49,10 @@ typedef long int Lmid_t;
    is returned.  */
 #define RTLD_DEFAULT	((void *) 0)
 
+/* If only find sym in the global scope, but will not use it, do not
+   set sym dependency. */
+# define RTLD_PROBE    ((void *) -2l)
+
 __BEGIN_DECLS
 
 /* Open the shared object FILE and map it in; return a handle that can be
diff --git a/elf/Makefile b/elf/Makefile
index eca7b28ab5..f9fc9fbebb 100644
--- a/elf/Makefile
+++ b/elf/Makefile
@@ -472,6 +472,7 @@ tests += \
   unload7 \
   unload8 \
   valgrind-test \
+  tst-dlsym-rtld-probe \
   # tests
 tests-cxx = \
   tst-dlopen-nodelete-reloc \
diff --git a/elf/dl-sym.c b/elf/dl-sym.c
index b1cf42f36d..9fdf7fb036 100644
--- a/elf/dl-sym.c
+++ b/elf/dl-sym.c
@@ -92,10 +92,17 @@ do_sym (void *handle, const char *name, void *who,
   /* Link map of the caller if needed.  */
   struct link_map *match = NULL;
 
-  if (handle == RTLD_DEFAULT)
+  int def_flags;
+
+  if (handle == RTLD_DEFAULT || handle == RTLD_PROBE)
     {
       match = _dl_sym_find_caller_link_map (caller);
 
+      def_flags = flags
+      if (def_flags == RTLD_DEFAULT) {
+          def_flags |= DL_LOOKUP_ADD_DEPENDENCY;
+      }
+
       /* Search the global scope.  We have the simple case where
 	 we look up in the scope of an object which was part of
 	 the initial binary.  And then the more complex part
@@ -104,7 +111,7 @@ do_sym (void *handle, const char *name, void *who,
       if (RTLD_SINGLE_THREAD_P)
 	result = GLRO(dl_lookup_symbol_x) (name, match, &ref,
 					   match->l_scope, vers, 0,
-					   flags | DL_LOOKUP_ADD_DEPENDENCY,
+					   def_flags,
 					   NULL);
       else
 	{
@@ -113,7 +120,7 @@ do_sym (void *handle, const char *name, void *who,
 	  args.map = match;
 	  args.vers = vers;
 	  args.flags
-	    = flags | DL_LOOKUP_ADD_DEPENDENCY | DL_LOOKUP_GSCOPE_LOCK;
+	    = def_flags | DL_LOOKUP_GSCOPE_LOCK;
 	  args.refp = &ref;
 
 	  THREAD_GSCOPE_SET_FLAG ();
diff --git a/elf/tst-dlsym-rtld-probe.c b/elf/tst-dlsym-rtld-probe.c
new file mode 100644
index 0000000000..c72ceaa182
--- /dev/null
+++ b/elf/tst-dlsym-rtld-probe.c
@@ -0,0 +1,50 @@
+/* Test RTLD_PROBE for dlsym.
+   Copyright (C) 2022-2022 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <dlfcn.h>
+#include <gnu/lib-names.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <support/check.h>
+#include <support/xdlfcn.h>
+
+static int
+do_test (void)
+{
+  int *iptr;
+  int ret;
+  void *handle;
+
+  handle = dlopen (LIBM_SO, RTLD_LAZY);
+  TEST_VERIFY (handle == NULL);
+  iptr = (int *)dlsym (RTLD_PROBE, "finite");	// get sym but not call --detect if symbol exist
+  ret = dlclose (handle);
+  TEST_VERIFY (ret != 0);
+  ret = 0;
+
+  handle = dlopen (LIBM_SO, RTLD_LAZY);
+  TEST_VERIFY (handle == NULL);
+  iptr = (int *)dlsym (RTLD_DEFAULT, "finite"); // get sym and keep
+  ret = dlcose (handle);
+  TEST_VERIFY (ret == 0);
+  return 0;
+}
+
+
+#include <support/test-driver.c>
-- 
2.12.3


  reply	other threads:[~2022-11-23  2:21 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-22  2:30 [PATCH] " Wang Bing
2022-11-22  2:36 ` H.J. Lu
2022-11-22 16:35   ` H.J. Lu
2022-11-23  2:21     ` Wang Bing [this message]
2022-12-05 14:45       ` [PATCH v2] " Florian Weimer

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=20221123022137.65977-1-wangbing6@huawei.com \
    --to=wangbing6@huawei.com \
    --cc=hjl.tools@gmail.com \
    --cc=libc-alpha@sourceware.org \
    --cc=nixiaoming@huawei.com \
    --cc=szabolcs.nagy@arm.com \
    --cc=yanhuijun@huawei.com \
    --cc=zhongjubin@huawei.com \
    /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).