lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <16afb602640ef43b7782087d6cca17bf6fc13603.1457502970.git.jpoimboe@redhat.com>
Date:	Wed,  9 Mar 2016 00:06:51 -0600
From:	Josh Poimboeuf <jpoimboe@...hat.com>
To:	Ingo Molnar <mingo@...nel.org>
Cc:	Arnaldo Carvalho de Melo <acme@...radead.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	"H. Peter Anvin" <hpa@...or.com>, x86@...nel.org,
	linux-kernel@...r.kernel.org, live-patching@...r.kernel.org,
	Michal Marek <mmarek@...e.cz>,
	Peter Zijlstra <peterz@...radead.org>,
	Andy Lutomirski <luto@...nel.org>,
	Borislav Petkov <bp@...en8.de>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Andi Kleen <andi@...stfloor.org>,
	Pedro Alves <palves@...hat.com>,
	Namhyung Kim <namhyung@...il.com>,
	Bernd Petrovitsch <bernd@...rovitsch.priv.at>,
	Chris J Arges <chris.j.arges@...onical.com>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Jiri Slaby <jslaby@...e.cz>,
	Arnaldo Carvalho de Melo <acme@...nel.org>
Subject: [PATCH 01/11] objtool: Prevent infinite recursion in noreturn detection

Ingo reported an infinite loop in objtool with a certain randconfig [1].
With the given config, two functions in crypto/ablkcipher.o contained
sibling calls to each other, which threw the recursive call in
dead_end_function() for a loop (literally!).

Split the noreturn detection into two passes.  In the first pass, check
for return instructions.  In the second pass, do the potentially
recursive sibling call check.  In most cases, the first pass will be
good enough.  In the rare case where a second pass is needed, recursion
should hopefully no longer be possible.

[1] https://lkml.kernel.org/r/20160308154909.GA20956@gmail.com

Reported-by: Ingo Molnar <mingo@...nel.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@...hat.com>
---
 tools/objtool/builtin-check.c | 24 ++++++++++++++++++++----
 1 file changed, 20 insertions(+), 4 deletions(-)

diff --git a/tools/objtool/builtin-check.c b/tools/objtool/builtin-check.c
index f7e0eba..80d9ed9 100644
--- a/tools/objtool/builtin-check.c
+++ b/tools/objtool/builtin-check.c
@@ -125,7 +125,7 @@ static bool ignore_func(struct objtool_file *file, struct symbol *func)
 static bool dead_end_function(struct objtool_file *file, struct symbol *func)
 {
 	int i;
-	struct instruction *insn;
+	struct instruction *insn, *func_insn;
 	bool empty = true;
 
 	/*
@@ -154,10 +154,11 @@ static bool dead_end_function(struct objtool_file *file, struct symbol *func)
 	if (!func->sec)
 		return false;
 
-	insn = find_instruction(file, func->sec, func->offset);
-	if (!insn)
+	func_insn = find_instruction(file, func->sec, func->offset);
+	if (!func_insn)
 		return false;
 
+	insn = func_insn;
 	list_for_each_entry_from(insn, &file->insns, list) {
 		if (insn->sec != func->sec ||
 		    insn->offset >= func->offset + func->len)
@@ -167,6 +168,21 @@ static bool dead_end_function(struct objtool_file *file, struct symbol *func)
 
 		if (insn->type == INSN_RETURN)
 			return false;
+	}
+
+	if (empty)
+		return false;
+
+	/*
+	 * A function can have a sibling call instead of a return.  In that
+	 * case, the function's dead-end status depends on whether the target
+	 * of the sibling call returns.
+	 */
+	insn = func_insn;
+	list_for_each_entry_from(insn, &file->insns, list) {
+		if (insn->sec != func->sec ||
+		    insn->offset >= func->offset + func->len)
+			break;
 
 		if (insn->type == INSN_JUMP_UNCONDITIONAL) {
 			struct instruction *dest = insn->jump_dest;
@@ -194,7 +210,7 @@ static bool dead_end_function(struct objtool_file *file, struct symbol *func)
 			return false;
 	}
 
-	return !empty;
+	return true;
 }
 
 /*
-- 
2.4.3

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ