Merge tag 'tsa_x86_bugs_for_6.16' into tip-x86-bugs

Pick up TSA changes from mainline so that attack vectors work can
continue ontop.

Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>

Author: bp3tk0v

Documentation/ABI/testing/sysfs-devices-system-cpu

@@ -584,6 +584,7 @@ What:		/sys/devices/system/cpu/vulnerabilities
 		/sys/devices/system/cpu/vulnerabilities/spectre_v1
 		/sys/devices/system/cpu/vulnerabilities/spectre_v2
 		/sys/devices/system/cpu/vulnerabilities/srbds
+		/sys/devices/system/cpu/vulnerabilities/tsa
 		/sys/devices/system/cpu/vulnerabilities/tsx_async_abort
 Date:		January 2018
 Contact:	Linux kernel mailing list <linux-kernel@vger.kernel.org>

Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst

@@ -157,9 +157,7 @@ This is achieved by using the otherwise unused and obsolete VERW instruction in
 combination with a microcode update. The microcode clears the affected CPU
 buffers when the VERW instruction is executed.
 
-Kernel reuses the MDS function to invoke the buffer clearing:
-
-	mds_clear_cpu_buffers()
+Kernel does the buffer clearing with x86_clear_cpu_buffers().
 
 On MDS affected CPUs, the kernel already invokes CPU buffer clear on
 kernel/userspace, hypervisor/guest and C-state (idle) transitions. No

Documentation/admin-guide/kernel-parameters.txt

@@ -7488,6 +7488,19 @@
 			having this key zero'ed is acceptable. E.g. in testing
 			scenarios.
 
+	tsa=		[X86] Control mitigation for Transient Scheduler
+			Attacks on AMD CPUs. Search the following in your
+			favourite search engine for more details:
+
+			"Technical guidance for mitigating transient scheduler
+			attacks".
+
+			off		- disable the mitigation
+			on		- enable the mitigation (default)
+			user		- mitigate only user/kernel transitions
+			vm		- mitigate only guest/host transitions
+
+
 	tsc=		Disable clocksource stability checks for TSC.
 			Format: <string>
 			[x86] reliable: mark tsc clocksource as reliable, this

Documentation/arch/x86/mds.rst

@@ -93,7 +93,7 @@ enters a C-state.
 
 The kernel provides a function to invoke the buffer clearing:
 
-    mds_clear_cpu_buffers()
+    x86_clear_cpu_buffers()
 
 Also macro CLEAR_CPU_BUFFERS can be used in ASM late in exit-to-user path.
 Other than CFLAGS.ZF, this macro doesn't clobber any registers.
@@ -185,9 +185,9 @@ Mitigation points
    idle clearing would be a window dressing exercise and is therefore not
    activated.
 
-   The invocation is controlled by the static key mds_idle_clear which is
-   switched depending on the chosen mitigation mode and the SMT state of
-   the system.
+   The invocation is controlled by the static key cpu_buf_idle_clear which is
+   switched depending on the chosen mitigation mode and the SMT state of the
+   system.
 
    The buffer clear is only invoked before entering the C-State to prevent
    that stale data from the idling CPU from spilling to the Hyper-Thread

arch/x86/Kconfig

@@ -2695,6 +2695,15 @@ config MITIGATION_ITS
 	  disabled, mitigation cannot be enabled via cmdline.
 	  See <file:Documentation/admin-guide/hw-vuln/indirect-target-selection.rst>
 
+config MITIGATION_TSA
+	bool "Mitigate Transient Scheduler Attacks"
+	depends on CPU_SUP_AMD
+	default y
+	help
+	  Enable mitigation for Transient Scheduler Attacks. TSA is a hardware
+	  security vulnerability on AMD CPUs which can lead to forwarding of
+	  invalid info to subsequent instructions and thus can affect their
+	  timing and thereby cause a leakage.
 endif
 
 config ARCH_HAS_ADD_PAGES

arch/x86/entry/entry.S

@@ -36,20 +36,20 @@ EXPORT_SYMBOL_GPL(write_ibpb);
 
 /*
  * Define the VERW operand that is disguised as entry code so that
- * it can be referenced with KPTI enabled. This ensure VERW can be
+ * it can be referenced with KPTI enabled. This ensures VERW can be
  * used late in exit-to-user path after page tables are switched.
  */
 .pushsection .entry.text, "ax"
 
 .align L1_CACHE_BYTES, 0xcc
-SYM_CODE_START_NOALIGN(mds_verw_sel)
+SYM_CODE_START_NOALIGN(x86_verw_sel)
 	UNWIND_HINT_UNDEFINED
 	ANNOTATE_NOENDBR
 	.word __KERNEL_DS
 .align L1_CACHE_BYTES, 0xcc
-SYM_CODE_END(mds_verw_sel);
+SYM_CODE_END(x86_verw_sel);
 /* For KVM */
-EXPORT_SYMBOL_GPL(mds_verw_sel);
+EXPORT_SYMBOL_GPL(x86_verw_sel);
 
 .popsection
 

arch/x86/include/asm/cpufeatures.h

@@ -456,6 +456,7 @@
 #define X86_FEATURE_NO_NESTED_DATA_BP	(20*32+ 0) /* No Nested Data Breakpoints */
 #define X86_FEATURE_WRMSR_XX_BASE_NS	(20*32+ 1) /* WRMSR to {FS,GS,KERNEL_GS}_BASE is non-serializing */
 #define X86_FEATURE_LFENCE_RDTSC	(20*32+ 2) /* LFENCE always serializing / synchronizes RDTSC */
+#define X86_FEATURE_VERW_CLEAR		(20*32+ 5) /* The memory form of VERW mitigates TSA */
 #define X86_FEATURE_NULL_SEL_CLR_BASE	(20*32+ 6) /* Null Selector Clears Base */
 #define X86_FEATURE_AUTOIBRS		(20*32+ 8) /* Automatic IBRS */
 #define X86_FEATURE_NO_SMM_CTL_MSR	(20*32+ 9) /* SMM_CTL MSR is not present */
@@ -487,6 +488,9 @@
 #define X86_FEATURE_PREFER_YMM		(21*32+ 8) /* Avoid ZMM registers due to downclocking */
 #define X86_FEATURE_APX			(21*32+ 9) /* Advanced Performance Extensions */
 #define X86_FEATURE_INDIRECT_THUNK_ITS	(21*32+10) /* Use thunk for indirect branches in lower half of cacheline */
+#define X86_FEATURE_TSA_SQ_NO		(21*32+11) /* AMD CPU not vulnerable to TSA-SQ */
+#define X86_FEATURE_TSA_L1_NO		(21*32+12) /* AMD CPU not vulnerable to TSA-L1 */
+#define X86_FEATURE_CLEAR_CPU_BUF_VM	(21*32+13) /* Clear CPU buffers using VERW before VMRUN */
 
 /*
  * BUG word(s)
@@ -542,5 +546,5 @@
 #define X86_BUG_OLD_MICROCODE		X86_BUG( 1*32+ 6) /* "old_microcode" CPU has old microcode, it is surely vulnerable to something */
 #define X86_BUG_ITS			X86_BUG( 1*32+ 7) /* "its" CPU is affected by Indirect Target Selection */
 #define X86_BUG_ITS_NATIVE_ONLY		X86_BUG( 1*32+ 8) /* "its_native_only" CPU is affected by ITS, VMX is not affected */
-
+#define X86_BUG_TSA			X86_BUG( 1*32+ 9) /* "tsa" CPU is affected by Transient Scheduler Attacks */
 #endif /* _ASM_X86_CPUFEATURES_H */

arch/x86/include/asm/irqflags.h

@@ -44,13 +44,13 @@ static __always_inline void native_irq_enable(void)
 
 static __always_inline void native_safe_halt(void)
 {
-	mds_idle_clear_cpu_buffers();
+	x86_idle_clear_cpu_buffers();
 	asm volatile("sti; hlt": : :"memory");
 }
 
 static __always_inline void native_halt(void)
 {
-	mds_idle_clear_cpu_buffers();
+	x86_idle_clear_cpu_buffers();
 	asm volatile("hlt": : :"memory");
 }
 

arch/x86/include/asm/kvm_host.h

@@ -764,6 +764,7 @@ enum kvm_only_cpuid_leafs {
 	CPUID_8000_0022_EAX,
 	CPUID_7_2_EDX,
 	CPUID_24_0_EBX,
+	CPUID_8000_0021_ECX,
 	NR_KVM_CPU_CAPS,
 
 	NKVMCAPINTS = NR_KVM_CPU_CAPS - NCAPINTS,

arch/x86/include/asm/mwait.h

@@ -43,8 +43,6 @@ static __always_inline void __monitorx(const void *eax, u32 ecx, u32 edx)
 
 static __always_inline void __mwait(u32 eax, u32 ecx)
 {
-	mds_idle_clear_cpu_buffers();
-
 	/*
 	 * Use the instruction mnemonic with implicit operands, as the LLVM
 	 * assembler fails to assemble the mnemonic with explicit operands:
@@ -80,7 +78,7 @@ static __always_inline void __mwait(u32 eax, u32 ecx)
  */
 static __always_inline void __mwaitx(u32 eax, u32 ebx, u32 ecx)
 {
-	/* No MDS buffer clear as this is AMD/HYGON only */
+	/* No need for TSA buffer clearing on AMD */
 
 	/* "mwaitx %eax, %ebx, %ecx" */
 	asm volatile(".byte 0x0f, 0x01, 0xfb"
@@ -98,7 +96,6 @@ static __always_inline void __mwaitx(u32 eax, u32 ebx, u32 ecx)
  */
 static __always_inline void __sti_mwait(u32 eax, u32 ecx)
 {
-	mds_idle_clear_cpu_buffers();
 
 	asm volatile("sti; mwait" :: "a" (eax), "c" (ecx));
 }
@@ -115,21 +112,29 @@ static __always_inline void __sti_mwait(u32 eax, u32 ecx)
  */
 static __always_inline void mwait_idle_with_hints(u32 eax, u32 ecx)
 {
+	if (need_resched())
+		return;
+
+	x86_idle_clear_cpu_buffers();
+
 	if (static_cpu_has_bug(X86_BUG_MONITOR) || !current_set_polling_and_test()) {
 		const void *addr = &current_thread_info()->flags;
 
 		alternative_input("", "clflush (%[addr])", X86_BUG_CLFLUSH_MONITOR, [addr] "a" (addr));
 		__monitor(addr, 0, 0);
 
-		if (!need_resched()) {
-			if (ecx & 1) {
-				__mwait(eax, ecx);
-			} else {
-				__sti_mwait(eax, ecx);
-				raw_local_irq_disable();
-			}
+		if (need_resched())
+			goto out;
+
+		if (ecx & 1) {
+			__mwait(eax, ecx);
+		} else {
+			__sti_mwait(eax, ecx);
+			raw_local_irq_disable();
 		}
 	}
+
+out:
 	current_clr_polling();
 }