linux-kernel - [tip: x86/core] objtool: Fix weak hole vs prefix symbol

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>] [day] [month] [year] [list]

Date:   Sat, 05 Nov 2022 10:36:26 -0000
From:   "tip-bot2 for Peter Zijlstra" <tip-bot2@...utronix.de>
To:     linux-tip-commits@...r.kernel.org
Cc:     Borislav Petkov <bp@...en8.de>, kernel test robot <lkp@...el.com>,
        "Peter Zijlstra (Intel)" <peterz@...radead.org>, x86@...nel.org,
        linux-kernel@...r.kernel.org
Subject: [tip: x86/core] objtool: Fix weak hole vs prefix symbol

The following commit has been merged into the x86/core branch of tip:

Commit-ID:     023f2340f053537cce170c31c430b0886c6f07ca
Gitweb:        https://git.kernel.org/tip/023f2340f053537cce170c31c430b0886c6f07ca
Author:        Peter Zijlstra <peterz@...radead.org>
AuthorDate:    Thu, 03 Nov 2022 20:57:51 +01:00
Committer:     Peter Zijlstra <peterz@...radead.org>
CommitterDate: Sat, 05 Nov 2022 11:28:02 +01:00

objtool: Fix weak hole vs prefix symbol

Boris (and the robot) reported that objtool grew a new complaint about
unreachable instructions. Upon inspection it was immediately clear
the __weak zombie instructions struck again.

For the unweary, the linker will simply remove the symbol for
overriden __weak symbols but leave the instructions in place, creating
unreachable instructions -- and objtool likes to report these.

Commit 4adb23686795 ("objtool: Ignore extra-symbol code") was supposed
to have dealt with that, but the new commit 9f2899fe36a6 ("objtool:
Add option to generate prefix symbols") subtly broke that logic by
created unvisited symbols.

Fixes: 9f2899fe36a6 ("objtool: Add option to generate prefix symbols")
Reported-by: Borislav Petkov <bp@...en8.de>
Reported-by: kernel test robot <lkp@...el.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
---
 tools/objtool/check.c | 22 +++++++++++++++++++++-
 1 file changed, 21 insertions(+), 1 deletion(-)

diff --git a/tools/objtool/check.c b/tools/objtool/check.c
index 55066c4..4f1a738 100644
--- a/tools/objtool/check.c
+++ b/tools/objtool/check.c
@@ -4053,8 +4053,28 @@ static int add_prefix_symbol(struct objtool_file *file, struct symbol *func,
 
 		offset = func->offset - prev->offset;
 		if (offset >= opts.prefix) {
-			if (offset == opts.prefix)
+			if (offset == opts.prefix) {
+				/*
+				 * Since the sec->symbol_list is ordered by
+				 * offset (see elf_add_symbol()) the added
+				 * symbol will not be seen by the iteration in
+				 * validate_section().
+				 *
+				 * Hence the lack of list_for_each_entry_safe()
+				 * there.
+				 *
+				 * The direct concequence is that prefix symbols
+				 * don't get visited (because pointless), except
+				 * for the logic in ignore_unreachable_insn()
+				 * that needs the terminating insn to be visited
+				 * otherwise it will report the hole.
+				 *
+				 * Hence mark the first instruction of the
+				 * prefix symbol as visisted.
+				 */
+				prev->visited |= VISITED_BRANCH;
 				elf_create_prefix_symbol(file->elf, func, opts.prefix);
+			}
 			break;
 		}
 		insn = prev;