linux-kernel - [RFC PATCH v6 3/3] selftests/x86: sysret

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]

Date:   Thu, 26 Jan 2023 04:17:14 +0700
From:   Ammar Faizi <ammarfaizi2@...weeb.org>
To:     "H. Peter Anvin" <hpa@...or.com>, Xin Li <xin3.li@...el.com>
Cc:     Dave Hansen <dave.hansen@...el.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Andrew Cooper <Andrew.Cooper3@...rix.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Brian Gerst <brgerst@...il.com>,
        Borislav Petkov <bp@...en8.de>, Shuah Khan <shuah@...nel.org>,
        Ingo Molnar <mingo@...nel.org>,
        Andy Lutomirski <luto@...nel.org>,
        x86 Mailing List <x86@...nel.org>,
        Linux Kselftest Mailing List 
        <linux-kselftest@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: [RFC PATCH v6 3/3] selftests/x86: sysret_rip: Test opportunistic SYSRET

When run on a non-FRED system, test the opportunistic SYSRET fast-path.
Make sure the %rcx/%r11 clobbering behavior is consistent.

When run on a FRED system, test that %rcx/%r11 are preserved when
invoking syscall.

This is similar to what test_syscall_rcx_r11_consistent() is doing, but
with addition it's done via the SIGUSR2 signal handler.

Link: https://lore.kernel.org/lkml/8770815f-0f23-d0c5-e56a-d401827842c9@zytor.com
Suggested-by: H. Peter Anvin (Intel) <hpa@...or.com>
Signed-off-by: Ammar Faizi <ammarfaizi2@...weeb.org>
---

On Wed, 25 Jan 2023 00:39:26 -0800, "H. Peter Anvin" wrote:
> >    /* Set IP and CX to match so that SYSRET can happen. */
> >     ctx->uc_mcontext.gregs[REG_RIP] = rip;
> >     ctx->uc_mcontext.gregs[REG_RCX] = rip;
> 
> It would be interesting to have the syscall handler try both with and 
> without this (so it would end up doing both IRET and SYSCALL on legacy.) 
> Perhaps SIGUSR1 versus SIGUSR2...


 tools/testing/selftests/x86/sysret_rip.c | 37 ++++++++++++++++++++++++
 1 file changed, 37 insertions(+)

diff --git a/tools/testing/selftests/x86/sysret_rip.c b/tools/testing/selftests/x86/sysret_rip.c
index d45b7f0147cd25ad..a1e5ec6f08bcd523 100644
--- a/tools/testing/selftests/x86/sysret_rip.c
+++ b/tools/testing/selftests/x86/sysret_rip.c
@@ -275,6 +275,28 @@ static void test_syscall_rcx_r11_consistent(void)
 	do_syscall(__NR_getppid, 0, 0, 0, 0, 0, 0);
 }
 
+static unsigned long usr2_rcx;
+static unsigned long usr2_r11;
+
+static void sigusr2(int sig, siginfo_t *info, void *ctx_void)
+{
+	ucontext_t *ctx = (ucontext_t*)ctx_void;
+
+	usr2_r11 = ctx->uc_mcontext.gregs[REG_R11];
+	usr2_rcx = ctx->uc_mcontext.gregs[REG_RCX];
+
+	check_regs_result(ctx->uc_mcontext.gregs[REG_R11],
+			  ctx->uc_mcontext.gregs[REG_RCX],
+			  ctx->uc_mcontext.gregs[REG_RBX]);
+}
+
+static void test_sysret_consistent(void)
+{
+	printf("[RUN]\ttest_sysret_consistent\n");
+	__raise(SIGUSR2);
+	printf("[OK]\tRCX = %#lx;  R11 = %#lx\n", usr2_rcx, usr2_r11);
+}
+
 int main()
 {
 	int i;
@@ -292,6 +314,21 @@ int main()
 	for (i = 47; i < 64; i++)
 		test_sigreturn_to(1UL<<i);
 
+	/*
+	 *
+	 * When run on a non-FRED system, test the SYSRET path. Make
+	 * sure the %rcx/%r11 clobbering behavior is consistent.
+	 *
+	 * When run on a FRED system, test that %rcx/%r11 are preserved
+	 * when invoking syscall.
+	 *
+	 * This is similar to test_syscall_rcx_r11_consistent(), but via
+	 * a signal handler.
+	 */
+	sethandler(SIGUSR2, sigusr2, 0);
+	for (i = 0; i < 32; i++)
+		test_sysret_consistent();
+
 	clearhandler(SIGUSR1);
 
 	sethandler(SIGSEGV, sigsegv_for_fallthrough, 0);
-- 
Ammar Faizi