LoongArch: mm: Fix huge page entry update for virtual machine

In virtual machine (guest mode), the tlbwr instruction can not write the
last entry of MTLB, so we need to make it non-present by invtlb and then
write it by tlbfill. This also simplify the whole logic.
Signed-off-by: Rui Wang <wangrui@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>

arch/loongarch/mm/tlbex.S

@@ -10,6 +10,8 @@
 #include <asm/regdef.h>
 #include <asm/stackframe.h>
 
+#define INVTLB_ADDR_GFALSE_AND_ASID	5
+
 #define PTRS_PER_PGD_BITS	(PAGE_SHIFT - 3)
 #define PTRS_PER_PUD_BITS	(PAGE_SHIFT - 3)
 #define PTRS_PER_PMD_BITS	(PAGE_SHIFT - 3)
@@ -136,13 +138,10 @@ tlb_huge_update_load:
 	ori		t0, ra, _PAGE_VALID
 	st.d		t0, t1, 0
 #endif
-	tlbsrch
-	addu16i.d	t1, zero, -(CSR_TLBIDX_EHINV >> 16)
-	addi.d		ra, t1, 0
-	csrxchg		ra, t1, LOONGARCH_CSR_TLBIDX
-	tlbwr
-
-	csrxchg		zero, t1, LOONGARCH_CSR_TLBIDX
+	csrrd		ra, LOONGARCH_CSR_ASID
+	csrrd		t1, LOONGARCH_CSR_BADV
+	andi		ra, ra, CSR_ASID_ASID
+	invtlb		INVTLB_ADDR_GFALSE_AND_ASID, ra, t1
 
 	/*
 	 * A huge PTE describes an area the size of the
@@ -287,13 +286,11 @@ tlb_huge_update_store:
 	ori		t0, ra, (_PAGE_VALID | _PAGE_DIRTY | _PAGE_MODIFIED)
 	st.d		t0, t1, 0
 #endif
-	tlbsrch
-	addu16i.d	t1, zero, -(CSR_TLBIDX_EHINV >> 16)
-	addi.d		ra, t1, 0
-	csrxchg		ra, t1, LOONGARCH_CSR_TLBIDX
-	tlbwr
+	csrrd		ra, LOONGARCH_CSR_ASID
+	csrrd		t1, LOONGARCH_CSR_BADV
+	andi		ra, ra, CSR_ASID_ASID
+	invtlb		INVTLB_ADDR_GFALSE_AND_ASID, ra, t1
 
-	csrxchg		zero, t1, LOONGARCH_CSR_TLBIDX
 	/*
 	 * A huge PTE describes an area the size of the
 	 * configured huge page size. This is twice the
@@ -436,6 +433,11 @@ tlb_huge_update_modify:
 	ori		t0, ra, (_PAGE_VALID | _PAGE_DIRTY | _PAGE_MODIFIED)
 	st.d		t0, t1, 0
 #endif
+	csrrd		ra, LOONGARCH_CSR_ASID
+	csrrd		t1, LOONGARCH_CSR_BADV
+	andi		ra, ra, CSR_ASID_ASID
+	invtlb		INVTLB_ADDR_GFALSE_AND_ASID, ra, t1
+
 	/*
 	 * A huge PTE describes an area the size of the
 	 * configured huge page size. This is twice the
@@ -466,7 +468,7 @@ tlb_huge_update_modify:
 	addu16i.d	t1, zero, (PS_HUGE_SIZE << (CSR_TLBIDX_PS_SHIFT - 16))
 	csrxchg		t1, t0, LOONGARCH_CSR_TLBIDX
 
-	tlbwr
+	tlbfill
 
 	/* Reset default page size */
 	addu16i.d	t0, zero, (CSR_TLBIDX_PS >> 16)