MIPS: Add missing EHB in mtc0 -> mfc0 sequence.

Add a missing EHB (Execution Hazard Barrier) in mtc0 -> mfc0 sequence.
Without this execution hazard barrier it's possible for the value read
back from the KScratch register to be the value from before the mtc0.

Reproducible on P5600 & P6600.

The hazard is documented in the MIPS Architecture Reference Manual Vol.
III: MIPS32/microMIPS32 Privileged Resource Architecture (MD00088), rev
6.03 table 8.1 which includes:

   Producer | Consumer | Hazard
  ----------|----------|----------------------------
   mtc0     | mfc0     | any coprocessor 0 register
Signed-off-by: Dmitry Korotin <dkorotin@wavecomp.com>
[paul.burton@mips.com:
  - Commit message tweaks.
  - Add Fixes tags.
  - Mark for stable back to v3.15 where P5600 support was introduced.]
Signed-off-by: Paul Burton <paul.burton@mips.com>
Fixes: 3d8bfdd0 ("MIPS: Use C0_KScratch (if present) to hold PGD pointer.")
Fixes: 829dcc0a ("MIPS: Add MIPS P5600 probe support")
Cc: linux-mips@vger.kernel.org
Cc: stable@vger.kernel.org # v3.15+

arch/mips/mm/tlbex.c

@@ -391,6 +391,7 @@ static struct work_registers build_get_work_registers(u32 **p)
 static void build_restore_work_registers(u32 **p)
 {
 	if (scratch_reg >= 0) {
+		uasm_i_ehb(p);
 		UASM_i_MFC0(p, 1, c0_kscratch(), scratch_reg);
 		return;
 	}
@@ -668,10 +669,12 @@ static void build_restore_pagemask(u32 **p, struct uasm_reloc **r,
 			uasm_i_mtc0(p, 0, C0_PAGEMASK);
 			uasm_il_b(p, r, lid);
 		}
-		if (scratch_reg >= 0)
+		if (scratch_reg >= 0) {
+			uasm_i_ehb(p);
 			UASM_i_MFC0(p, 1, c0_kscratch(), scratch_reg);
-		else
+		} else {
 			UASM_i_LW(p, 1, scratchpad_offset(0), 0);
+		}
 	} else {
 		/* Reset default page size */
 		if (PM_DEFAULT_MASK >> 16) {
@@ -938,10 +941,12 @@ build_get_pgd_vmalloc64(u32 **p, struct uasm_label **l, struct uasm_reloc **r,
 		uasm_i_jr(p, ptr);
 
 		if (mode == refill_scratch) {
-			if (scratch_reg >= 0)
+			if (scratch_reg >= 0) {
+				uasm_i_ehb(p);
 				UASM_i_MFC0(p, 1, c0_kscratch(), scratch_reg);
-			else
+			} else {
 				UASM_i_LW(p, 1, scratchpad_offset(0), 0);
+			}
 		} else {
 			uasm_i_nop(p);
 		}
@@ -1258,6 +1263,7 @@ build_fast_tlb_refill_handler (u32 **p, struct uasm_label **l,
 	UASM_i_MTC0(p, odd, C0_ENTRYLO1); /* load it */
 
 	if (c0_scratch_reg >= 0) {
+		uasm_i_ehb(p);
 		UASM_i_MFC0(p, scratch, c0_kscratch(), c0_scratch_reg);
 		build_tlb_write_entry(p, l, r, tlb_random);
 		uasm_l_leave(l, *p);
@@ -1603,15 +1609,17 @@ static void build_setup_pgd(void)
 		uasm_i_dinsm(&p, a0, 0, 29, 64 - 29);
 		uasm_l_tlbl_goaround1(&l, p);
 		UASM_i_SLL(&p, a0, a0, 11);
-		uasm_i_jr(&p, 31);
 		UASM_i_MTC0(&p, a0, C0_CONTEXT);
+		uasm_i_jr(&p, 31);
+		uasm_i_ehb(&p);
 	} else {
 		/* PGD in c0_KScratch */
-		uasm_i_jr(&p, 31);
 		if (cpu_has_ldpte)
 			UASM_i_MTC0(&p, a0, C0_PWBASE);
 		else
 			UASM_i_MTC0(&p, a0, c0_kscratch(), pgd_reg);
+		uasm_i_jr(&p, 31);
+		uasm_i_ehb(&p);
 	}
 #else
 #ifdef CONFIG_SMP
@@ -1625,13 +1633,16 @@ static void build_setup_pgd(void)
 	UASM_i_LA_mostly(&p, a2, pgdc);
 	UASM_i_SW(&p, a0, uasm_rel_lo(pgdc), a2);
 #endif /* SMP */
-	uasm_i_jr(&p, 31);
 
 	/* if pgd_reg is allocated, save PGD also to scratch register */
-	if (pgd_reg != -1)
+	if (pgd_reg != -1) {
 		UASM_i_MTC0(&p, a0, c0_kscratch(), pgd_reg);
-	else
+		uasm_i_jr(&p, 31);
+		uasm_i_ehb(&p);
+	} else {
+		uasm_i_jr(&p, 31);
 		uasm_i_nop(&p);
+	}
 #endif
 	if (p >= (u32 *)tlbmiss_handler_setup_pgd_end)
 		panic("tlbmiss_handler_setup_pgd space exceeded");