linux-kernel - [PATCH 3/9] KVM: arm64: selftests: Enable single-step without a "full" ucall()

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: <20221119013450.2643007-4-seanjc@google.com>
Date:   Sat, 19 Nov 2022 01:34:44 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Yury Norov <yury.norov@...il.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Marc Zyngier <maz@...nel.org>,
        Paolo Bonzini <pbonzini@...hat.com>
Cc:     Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Rasmus Villemoes <linux@...musvillemoes.dk>,
        Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        James Morse <james.morse@....com>,
        Alexandru Elisei <alexandru.elisei@....com>,
        Suzuki K Poulose <suzuki.poulose@....com>,
        Oliver Upton <oliver.upton@...ux.dev>,
        linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, kvmarm@...ts.linux.dev,
        kvmarm@...ts.cs.columbia.edu, kvm@...r.kernel.org,
        Sean Christopherson <seanjc@...gle.com>
Subject: [PATCH 3/9] KVM: arm64: selftests: Enable single-step without a
 "full" ucall()

Add a new ucall hook, GUEST_UCALL_NONE(), to allow tests to make ucalls
without allocating a ucall struct, and use it to enable single-step
in ARM's debug-exceptions test.  Like the disable single-step path, the
enabling path also needs to ensure that no exclusive access sequences are
attempted after enabling single-step, as the exclusive monitor is cleared
on ERET from the debug exception taken to EL2.

The test currently "works" because clear_bit() isn't actually an atomic
operation... yet.

Signed-off-by: Sean Christopherson <seanjc@...gle.com>
---
 .../selftests/kvm/aarch64/debug-exceptions.c  | 21 ++++++++++---------
 .../selftests/kvm/include/ucall_common.h      |  8 +++++++
 2 files changed, 19 insertions(+), 10 deletions(-)

diff --git a/tools/testing/selftests/kvm/aarch64/debug-exceptions.c b/tools/testing/selftests/kvm/aarch64/debug-exceptions.c
index d86c4e4d1c82..c62ec4d7f6a3 100644
--- a/tools/testing/selftests/kvm/aarch64/debug-exceptions.c
+++ b/tools/testing/selftests/kvm/aarch64/debug-exceptions.c
@@ -239,10 +239,6 @@ static void guest_svc_handler(struct ex_regs *regs)
 	svc_addr = regs->pc;
 }
 
-enum single_step_op {
-	SINGLE_STEP_ENABLE = 0,
-};
-
 static void guest_code_ss(int test_cnt)
 {
 	uint64_t i;
@@ -253,8 +249,16 @@ static void guest_code_ss(int test_cnt)
 		w_bvr = i << 2;
 		w_wvr = i << 2;
 
-		/* Enable Single Step execution */
-		GUEST_SYNC(SINGLE_STEP_ENABLE);
+		/*
+		 * Enable Single Step execution.  Note!  This _must_ be a bare
+		 * ucall as the ucall() path uses atomic operations to manage
+		 * the ucall structures, and the built-in "atomics" are usually
+		 * implemented via exclusive access instructions.  The exlusive
+		 * monitor is cleared on ERET, and so taking debug exceptions
+		 * during a LDREX=>STREX sequence will prevent forward progress
+		 * and hang the guest/test.
+		 */
+		GUEST_UCALL_NONE();
 
 		/*
 		 * The userspace will verify that the pc is as expected during
@@ -356,12 +360,9 @@ void test_single_step_from_userspace(int test_cnt)
 				break;
 			}
 
-			TEST_ASSERT(cmd == UCALL_SYNC,
+			TEST_ASSERT(cmd == UCALL_NONE,
 				    "Unexpected ucall cmd 0x%lx", cmd);
 
-			TEST_ASSERT(uc.args[1] == SINGLE_STEP_ENABLE,
-				    "Unexpected ucall action 0x%lx", uc.args[1]);
-
 			debug.control = KVM_GUESTDBG_ENABLE |
 					KVM_GUESTDBG_SINGLESTEP;
 			ss_enable = true;
diff --git a/tools/testing/selftests/kvm/include/ucall_common.h b/tools/testing/selftests/kvm/include/ucall_common.h
index bdd373189a77..1a6aaef5ccae 100644
--- a/tools/testing/selftests/kvm/include/ucall_common.h
+++ b/tools/testing/selftests/kvm/include/ucall_common.h
@@ -35,6 +35,14 @@ void ucall(uint64_t cmd, int nargs, ...);
 uint64_t get_ucall(struct kvm_vcpu *vcpu, struct ucall *uc);
 void ucall_init(struct kvm_vm *vm, vm_paddr_t mmio_gpa);
 
+/*
+ * Perform userspace call without any associated data.  This bare call avoids
+ * allocating a ucall struct, which can be useful if the atomic operations in
+ * the full ucall() are problematic and/or unwanted.  Note, this will come out
+ * as UCALL_NONE on the backend.
+ */
+#define GUEST_UCALL_NONE()	ucall_arch_do_ucall((vm_vaddr_t)NULL)
+
 #define GUEST_SYNC_ARGS(stage, arg1, arg2, arg3, arg4)	\
 				ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)
 #define GUEST_SYNC(stage)	ucall(UCALL_SYNC, 2, "hello", stage)
-- 
2.38.1.584.g0f3c55d4c2-goog