riscv: Add ERRATA_MIPS_P8700_WFI to replace WFI with mips.pause

The MIPS P8700 has bugs with the WFI instruction. A WFI instruction
after a mispredicted branch can result in erroneous address
translation on the MIPS P8700. MIPS has not published an erratum
for this.

The workaround uses the RISC-V alternatives framework to patch all WFI
instructions with the MIPS P8700 pause opcode (0x00501013)
at runtime when running on P8700 hardware.

Two call sites are patched:
 - arch/riscv/kernel/head.S: secondary hart parking loop
 - arch/riscv/include/asm/processor.h: wait_for_interrupt()

Signed-off-by: Djordje Todorovic <djordje.todorovic@htecgroup.com>
Signed-off-by: Aleksa Paunovic <aleksa.paunovic@htecgroup.com>
Link: https://patch.msgid.link/20260917-p8700-wfi-v3-1-ada4b0584712@htecgroup.com
[pjw@kernel.org: updated to apply]
Signed-off-by: Paul Walmsley <pjw@kernel.org>

3 files changed, 42 insertions, 3 deletions

diff --git a/arch/riscv/include/asm/errata_list.h b/arch/riscv/include/asm/errata_list.h
index 6694b5ccdcf8..6fdb445c0f45 100644
--- a/arch/riscv/include/asm/errata_list.h
+++ b/arch/riscv/include/asm/errata_list.h
@@ -25,6 +25,16 @@ ALTERNATIVE(__stringify(RISCV_PTR do_page_fault),			\
 	    __stringify(RISCV_PTR sifive_cip_453_page_fault_trp),	\
 	    SIFIVE_VENDOR_ID, ERRATA_SIFIVE_CIP_453,			\
 	    CONFIG_ERRATA_SIFIVE_CIP_453)
+
+#ifdef CONFIG_ERRATA_MIPS_P8700_WFI
+#define ALT_WFI								\
+ALTERNATIVE("wfi; .rept 7; nop; .endr;",				\
+	".rept 8; .4byte 0x00501013; .endr;", MIPS_VENDOR_ID,		\
+	ERRATA_MIPS_P8700_WFI, CONFIG_ERRATA_MIPS_P8700_WFI)
+#else
+#define ALT_WFI	wfi
+#endif
+
 #else /* !__ASSEMBLER__ */
 
 #define ALT_SFENCE_VMA_ASID(asid)					\
@@ -53,6 +63,33 @@ asm(ALTERNATIVE(	\
 	: /* no inputs */	\
 	: "memory")
 
+#ifdef CONFIG_ERRATA_MIPS_P8700_WFI
+/*
+ * A WFI instruction after a mispredicted branch can result in erroneous address translation
+ * on the MIPS P8700.
+ * The number of MIPS_PAUSE instructions required was determined by measuring
+ * Coremark performance on a one-core, two-hart configuration. Eight PAUSE instructions matched
+ * the latency of a single WFI instruction.
+ * Usually, the maximum number of cycles per PAUSE instruction is 63. Exceptionally, if the hart's
+ * LR reservation bit is set, the maximum number of cycles is extended to 255
+ * (and the hart is woken up once it is cleared). This case should not affect the patched sites.
+ */
+#define ALT_RISCV_WFI()										\
+asm volatile(ALTERNATIVE(									\
+		"wfi\n" /* Original RISC-V wfi insn */						\
+		__nops(7),									\
+		".rept 8;" MIPS_PAUSE ".endr;\n", /* Replacement: mips.pause for P8700 */	\
+		MIPS_VENDOR_ID, /* Vendor ID to match */					\
+		ERRATA_MIPS_P8700_WFI, /* patch_id */						\
+		CONFIG_ERRATA_MIPS_P8700_WFI)							\
+		: /* no outputs */								\
+		: /* no inputs */								\
+		: "memory")
+#else
+#define ALT_RISCV_WFI()		\
+	__asm__ __volatile__ ("wfi")
+#endif
+
 /*
  * _val is marked as "will be overwritten", so need to set it to 0
  * in the default case.
diff --git a/arch/riscv/include/asm/errata_list_vendors.h b/arch/riscv/include/asm/errata_list_vendors.h
index c62a82f60468..02b5ab0d23a8 100644
--- a/arch/riscv/include/asm/errata_list_vendors.h
+++ b/arch/riscv/include/asm/errata_list_vendors.h
@@ -23,8 +23,9 @@
 #endif
 
 #ifdef CONFIG_ERRATA_MIPS
-#define	ERRATA_MIPS_P8700_PAUSE_OPCODE 0
-#define	ERRATA_MIPS_NUMBER 1
+#define ERRATA_MIPS_P8700_PAUSE_OPCODE 0
+#define ERRATA_MIPS_P8700_WFI 1
+#define ERRATA_MIPS_NUMBER 2
 #endif
 
 #endif /* ASM_ERRATA_LIST_VENDORS_H */
diff --git a/arch/riscv/include/asm/processor.h b/arch/riscv/include/asm/processor.h
index 815715c67f94..3686f871c8ed 100644
--- a/arch/riscv/include/asm/processor.h
+++ b/arch/riscv/include/asm/processor.h
@@ -17,6 +17,7 @@
 #include <asm/alternative-macros.h>
 #include <asm/hwcap.h>
 #include <asm/usercfi.h>
+#include <asm/errata_list.h>
 
 #define arch_get_mmap_end(addr, len, flags)			\
 ({								\
@@ -187,7 +188,7 @@ extern unsigned long __get_wchan(struct task_struct *p);
 
 static inline void wait_for_interrupt(void)
 {
-	__asm__ __volatile__ ("wfi");
+	ALT_RISCV_WFI();
 }
 
 extern phys_addr_t dma32_phys_limit;