powerpc/64s: Add support for RFI flush of L1-D cache

On some CPUs we can prevent the Meltdown vulnerability by flushing the
L1-D cache on exit from kernel to user mode, and from hypervisor to
guest.

This is known to be the case on at least Power7, Power8 and Power9. At
this time we do not know the status of the vulnerability on other CPUs
such as the 970 (Apple G5), pasemi CPUs (AmigaOne X1000) or Freescale
CPUs. As more information comes to light we can enable this, or other
mechanisms on those CPUs.

The vulnerability occurs when the load of an architecturally
inaccessible memory region (eg. userspace load of kernel memory) is
speculatively executed to the point where its result can influence the
address of a subsequent speculatively executed load.

In order for that to happen, the first load must hit in the L1,
because before the load is sent to the L2 the permission check is
performed. Therefore if no kernel addresses hit in the L1 the
vulnerability can not occur. We can ensure that is the case by
flushing the L1 whenever we return to userspace. Similarly for
hypervisor vs guest.

In order to flush the L1-D cache on exit, we add a section of nops at
each (h)rfi location that returns to a lower privileged context, and
patch that with some sequence. Newer firmwares are able to advertise
to us that there is a special nop instruction that flushes the L1-D.
If we do not see that advertised, we fall back to doing a displacement
flush in software.

For guest kernels we support migration between some CPU versions, and
different CPUs may use different flush instructions. So that we are
prepared to migrate to a machine with a different flush instruction
activated, we may have to patch more than one flush instruction at
boot if the hypervisor tells us to.

In the end this patch is mostly the work of Nicholas Piggin and
Michael Ellerman. However a cast of thousands contributed to analysis
of the issue, earlier versions of the patch, back ports testing etc.
Many thanks to all of them.
Tested-by: Jon Masters <jcm@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

arch/powerpc/include/asm/exception-64s.h

@@ -74,34 +74,58 @@
  */
 #define EX_R3		EX_DAR
 
-/* Macros for annotating the expected destination of (h)rfid */
+/*
+ * Macros for annotating the expected destination of (h)rfid
+ *
+ * The nop instructions allow us to insert one or more instructions to flush the
+ * L1-D cache when returning to userspace or a guest.
+ */
+#define RFI_FLUSH_SLOT							\
+	RFI_FLUSH_FIXUP_SECTION;					\
+	nop;								\
+	nop;								\
+	nop
 
 #define RFI_TO_KERNEL							\
 	rfid
 
 #define RFI_TO_USER							\
-	rfid
+	RFI_FLUSH_SLOT;							\
+	rfid;								\
+	b	rfi_flush_fallback
 
 #define RFI_TO_USER_OR_KERNEL						\
-	rfid
+	RFI_FLUSH_SLOT;							\
+	rfid;								\
+	b	rfi_flush_fallback
 
 #define RFI_TO_GUEST							\
-	rfid
+	RFI_FLUSH_SLOT;							\
+	rfid;								\
+	b	rfi_flush_fallback
 
 #define HRFI_TO_KERNEL							\
 	hrfid
 
 #define HRFI_TO_USER							\
-	hrfid
+	RFI_FLUSH_SLOT;							\
+	hrfid;								\
+	b	hrfi_flush_fallback
 
 #define HRFI_TO_USER_OR_KERNEL						\
-	hrfid
+	RFI_FLUSH_SLOT;							\
+	hrfid;								\
+	b	hrfi_flush_fallback
 
 #define HRFI_TO_GUEST							\
-	hrfid
+	RFI_FLUSH_SLOT;							\
+	hrfid;								\
+	b	hrfi_flush_fallback
 
 #define HRFI_TO_UNKNOWN							\
-	hrfid
+	RFI_FLUSH_SLOT;							\
+	hrfid;								\
+	b	hrfi_flush_fallback
 
 #ifdef CONFIG_RELOCATABLE
 #define __EXCEPTION_RELON_PROLOG_PSERIES_1(label, h)			\

arch/powerpc/include/asm/feature-fixups.h

@@ -187,7 +187,20 @@ label##3:					       	\
 	FTR_ENTRY_OFFSET label##1b-label##3b;		\
 	.popsection;
 
+#define RFI_FLUSH_FIXUP_SECTION				\
+951:							\
+	.pushsection __rfi_flush_fixup,"a";		\
+	.align 2;					\
+952:							\
+	FTR_ENTRY_OFFSET 951b-952b;			\
+	.popsection;
+
+
 #ifndef __ASSEMBLY__
+#include <linux/types.h>
+
+extern long __start___rfi_flush_fixup, __stop___rfi_flush_fixup;
+
 void apply_feature_fixups(void);
 void setup_feature_keys(void);
 #endif

arch/powerpc/include/asm/paca.h

@@ -232,6 +232,16 @@ struct paca_struct {
 	struct sibling_subcore_state *sibling_subcore_state;
 #endif
 #endif
+#ifdef CONFIG_PPC_BOOK3S_64
+	/*
+	 * rfi fallback flush must be in its own cacheline to prevent
+	 * other paca data leaking into the L1d
+	 */
+	u64 exrfi[EX_SIZE] __aligned(0x80);
+	void *rfi_flush_fallback_area;
+	u64 l1d_flush_congruence;
+	u64 l1d_flush_sets;
+#endif
 };
 
 extern void copy_mm_to_paca(struct mm_struct *mm);

arch/powerpc/include/asm/setup.h

@@ -39,6 +39,19 @@ static inline void pseries_big_endian_exceptions(void) {}
 static inline void pseries_little_endian_exceptions(void) {}
 #endif /* CONFIG_PPC_PSERIES */
 
+void rfi_flush_enable(bool enable);
+
+/* These are bit flags */
+enum l1d_flush_type {
+	L1D_FLUSH_NONE		= 0x1,
+	L1D_FLUSH_FALLBACK	= 0x2,
+	L1D_FLUSH_ORI		= 0x4,
+	L1D_FLUSH_MTTRIG	= 0x8,
+};
+
+void __init setup_rfi_flush(enum l1d_flush_type, bool enable);
+void do_rfi_flush_fixups(enum l1d_flush_type types);
+
 #endif /* !__ASSEMBLY__ */
 
 #endif	/* _ASM_POWERPC_SETUP_H */

arch/powerpc/kernel/asm-offsets.c

@@ -237,6 +237,11 @@ int main(void)
 	OFFSET(PACA_NMI_EMERG_SP, paca_struct, nmi_emergency_sp);
 	OFFSET(PACA_IN_MCE, paca_struct, in_mce);
 	OFFSET(PACA_IN_NMI, paca_struct, in_nmi);
+	OFFSET(PACA_RFI_FLUSH_FALLBACK_AREA, paca_struct, rfi_flush_fallback_area);
+	OFFSET(PACA_EXRFI, paca_struct, exrfi);
+	OFFSET(PACA_L1D_FLUSH_CONGRUENCE, paca_struct, l1d_flush_congruence);
+	OFFSET(PACA_L1D_FLUSH_SETS, paca_struct, l1d_flush_sets);
+
 #endif
 	OFFSET(PACAHWCPUID, paca_struct, hw_cpu_id);
 	OFFSET(PACAKEXECSTATE, paca_struct, kexec_state);

arch/powerpc/kernel/exceptions-64s.S

@@ -1449,6 +1449,90 @@ masked_##_H##interrupt:					\
 	b	.;					\
 	MASKED_DEC_HANDLER(_H)
 
+TRAMP_REAL_BEGIN(rfi_flush_fallback)
+	SET_SCRATCH0(r13);
+	GET_PACA(r13);
+	std	r9,PACA_EXRFI+EX_R9(r13)
+	std	r10,PACA_EXRFI+EX_R10(r13)
+	std	r11,PACA_EXRFI+EX_R11(r13)
+	std	r12,PACA_EXRFI+EX_R12(r13)
+	std	r8,PACA_EXRFI+EX_R13(r13)
+	mfctr	r9
+	ld	r10,PACA_RFI_FLUSH_FALLBACK_AREA(r13)
+	ld	r11,PACA_L1D_FLUSH_SETS(r13)
+	ld	r12,PACA_L1D_FLUSH_CONGRUENCE(r13)
+	/*
+	 * The load adresses are at staggered offsets within cachelines,
+	 * which suits some pipelines better (on others it should not
+	 * hurt).
+	 */
+	addi	r12,r12,8
+	mtctr	r11
+	DCBT_STOP_ALL_STREAM_IDS(r11) /* Stop prefetch streams */
+
+	/* order ld/st prior to dcbt stop all streams with flushing */
+	sync
+1:	li	r8,0
+	.rept	8 /* 8-way set associative */
+	ldx	r11,r10,r8
+	add	r8,r8,r12
+	xor	r11,r11,r11	// Ensure r11 is 0 even if fallback area is not
+	add	r8,r8,r11	// Add 0, this creates a dependency on the ldx
+	.endr
+	addi	r10,r10,128 /* 128 byte cache line */
+	bdnz	1b
+
+	mtctr	r9
+	ld	r9,PACA_EXRFI+EX_R9(r13)
+	ld	r10,PACA_EXRFI+EX_R10(r13)
+	ld	r11,PACA_EXRFI+EX_R11(r13)
+	ld	r12,PACA_EXRFI+EX_R12(r13)
+	ld	r8,PACA_EXRFI+EX_R13(r13)
+	GET_SCRATCH0(r13);
+	rfid
+
+TRAMP_REAL_BEGIN(hrfi_flush_fallback)
+	SET_SCRATCH0(r13);
+	GET_PACA(r13);
+	std	r9,PACA_EXRFI+EX_R9(r13)
+	std	r10,PACA_EXRFI+EX_R10(r13)
+	std	r11,PACA_EXRFI+EX_R11(r13)
+	std	r12,PACA_EXRFI+EX_R12(r13)
+	std	r8,PACA_EXRFI+EX_R13(r13)
+	mfctr	r9
+	ld	r10,PACA_RFI_FLUSH_FALLBACK_AREA(r13)
+	ld	r11,PACA_L1D_FLUSH_SETS(r13)
+	ld	r12,PACA_L1D_FLUSH_CONGRUENCE(r13)
+	/*
+	 * The load adresses are at staggered offsets within cachelines,
+	 * which suits some pipelines better (on others it should not
+	 * hurt).
+	 */
+	addi	r12,r12,8
+	mtctr	r11
+	DCBT_STOP_ALL_STREAM_IDS(r11) /* Stop prefetch streams */
+
+	/* order ld/st prior to dcbt stop all streams with flushing */
+	sync
+1:	li	r8,0
+	.rept	8 /* 8-way set associative */
+	ldx	r11,r10,r8
+	add	r8,r8,r12
+	xor	r11,r11,r11	// Ensure r11 is 0 even if fallback area is not
+	add	r8,r8,r11	// Add 0, this creates a dependency on the ldx
+	.endr
+	addi	r10,r10,128 /* 128 byte cache line */
+	bdnz	1b
+
+	mtctr	r9
+	ld	r9,PACA_EXRFI+EX_R9(r13)
+	ld	r10,PACA_EXRFI+EX_R10(r13)
+	ld	r11,PACA_EXRFI+EX_R11(r13)
+	ld	r12,PACA_EXRFI+EX_R12(r13)
+	ld	r8,PACA_EXRFI+EX_R13(r13)
+	GET_SCRATCH0(r13);
+	hrfid
+
 /*
  * Real mode exceptions actually use this too, but alternate
  * instruction code patches (which end up in the common .text area)

arch/powerpc/kernel/setup_64.c

@@ -801,3 +801,82 @@ static int __init disable_hardlockup_detector(void)
 	return 0;
 }
 early_initcall(disable_hardlockup_detector);
+
+#ifdef CONFIG_PPC_BOOK3S_64
+static enum l1d_flush_type enabled_flush_types;
+static void *l1d_flush_fallback_area;
+bool rfi_flush;
+
+static void do_nothing(void *unused)
+{
+	/*
+	 * We don't need to do the flush explicitly, just enter+exit kernel is
+	 * sufficient, the RFI exit handlers will do the right thing.
+	 */
+}
+
+void rfi_flush_enable(bool enable)
+{
+	if (rfi_flush == enable)
+		return;
+
+	if (enable) {
+		do_rfi_flush_fixups(enabled_flush_types);
+		on_each_cpu(do_nothing, NULL, 1);
+	} else
+		do_rfi_flush_fixups(L1D_FLUSH_NONE);
+
+	rfi_flush = enable;
+}
+
+static void init_fallback_flush(void)
+{
+	u64 l1d_size, limit;
+	int cpu;
+
+	l1d_size = ppc64_caches.l1d.size;
+	limit = min(safe_stack_limit(), ppc64_rma_size);
+
+	/*
+	 * Align to L1d size, and size it at 2x L1d size, to catch possible
+	 * hardware prefetch runoff. We don't have a recipe for load patterns to
+	 * reliably avoid the prefetcher.
+	 */
+	l1d_flush_fallback_area = __va(memblock_alloc_base(l1d_size * 2, l1d_size, limit));
+	memset(l1d_flush_fallback_area, 0, l1d_size * 2);
+
+	for_each_possible_cpu(cpu) {
+		/*
+		 * The fallback flush is currently coded for 8-way
+		 * associativity. Different associativity is possible, but it
+		 * will be treated as 8-way and may not evict the lines as
+		 * effectively.
+		 *
+		 * 128 byte lines are mandatory.
+		 */
+		u64 c = l1d_size / 8;
+
+		paca[cpu].rfi_flush_fallback_area = l1d_flush_fallback_area;
+		paca[cpu].l1d_flush_congruence = c;
+		paca[cpu].l1d_flush_sets = c / 128;
+	}
+}
+
+void __init setup_rfi_flush(enum l1d_flush_type types, bool enable)
+{
+	if (types & L1D_FLUSH_FALLBACK) {
+		pr_info("rfi-flush: Using fallback displacement flush\n");
+		init_fallback_flush();
+	}
+
+	if (types & L1D_FLUSH_ORI)
+		pr_info("rfi-flush: Using ori type flush\n");
+
+	if (types & L1D_FLUSH_MTTRIG)
+		pr_info("rfi-flush: Using mttrig type flush\n");
+
+	enabled_flush_types = types;
+
+	rfi_flush_enable(enable);
+}
+#endif /* CONFIG_PPC_BOOK3S_64 */

arch/powerpc/kernel/vmlinux.lds.S

@@ -132,6 +132,15 @@ SECTIONS
 	/* Read-only data */
 	RO_DATA(PAGE_SIZE)
 
+#ifdef CONFIG_PPC64
+	. = ALIGN(8);
+	__rfi_flush_fixup : AT(ADDR(__rfi_flush_fixup) - LOAD_OFFSET) {
+		__start___rfi_flush_fixup = .;
+		*(__rfi_flush_fixup)
+		__stop___rfi_flush_fixup = .;
+	}
+#endif
+
 	EXCEPTION_TABLE(0)
 
 	NOTES :kernel :notes

arch/powerpc/lib/feature-fixups.c

@@ -116,6 +116,47 @@ void do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end)
 	}
 }
 
+#ifdef CONFIG_PPC_BOOK3S_64
+void do_rfi_flush_fixups(enum l1d_flush_type types)
+{
+	unsigned int instrs[3], *dest;
+	long *start, *end;
+	int i;
+
+	start = PTRRELOC(&__start___rfi_flush_fixup),
+	end = PTRRELOC(&__stop___rfi_flush_fixup);
+
+	instrs[0] = 0x60000000; /* nop */
+	instrs[1] = 0x60000000; /* nop */
+	instrs[2] = 0x60000000; /* nop */
+
+	if (types & L1D_FLUSH_FALLBACK)
+		/* b .+16 to fallback flush */
+		instrs[0] = 0x48000010;
+
+	i = 0;
+	if (types & L1D_FLUSH_ORI) {
+		instrs[i++] = 0x63ff0000; /* ori 31,31,0 speculation barrier */
+		instrs[i++] = 0x63de0000; /* ori 30,30,0 L1d flush*/
+	}
+
+	if (types & L1D_FLUSH_MTTRIG)
+		instrs[i++] = 0x7c12dba6; /* mtspr TRIG2,r0 (SPR #882) */
+
+	for (i = 0; start < end; start++, i++) {
+		dest = (void *)start + *start;
+
+		pr_devel("patching dest %lx\n", (unsigned long)dest);
+
+		patch_instruction(dest, instrs[0]);
+		patch_instruction(dest + 1, instrs[1]);
+		patch_instruction(dest + 2, instrs[2]);
+	}
+
+	printk(KERN_DEBUG "rfi-flush: patched %d locations\n", i);
+}
+#endif /* CONFIG_PPC_BOOK3S_64 */
+
 void do_lwsync_fixups(unsigned long value, void *fixup_start, void *fixup_end)
 {
 	long *start, *end;