drm/xe: Force write completion of MI_STORE_DATA_IMM

With Force write completion unset there is no guarantees of when the
write will be globally visible what is not the behavior wanted.

Fixes: 9c57bc0 ("drm/xe/lnl: Drop force_probe requirement")
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241217160732.46280-1-jose.souza@intel.com

Author: zehortigoza

drivers/gpu/drm/xe/instructions/xe_mi_commands.h

@@ -33,12 +33,13 @@
 #define MI_TOPOLOGY_FILTER		__MI_INSTR(0xD)
 #define MI_FORCE_WAKEUP			__MI_INSTR(0x1D)
 
-#define MI_STORE_DATA_IMM		__MI_INSTR(0x20)
-#define   MI_SDI_GGTT			REG_BIT(22)
-#define   MI_SDI_LEN_DW			GENMASK(9, 0)
-#define   MI_SDI_NUM_DW(x)		REG_FIELD_PREP(MI_SDI_LEN_DW, (x) + 3 - 2)
-#define   MI_SDI_NUM_QW(x)		(REG_FIELD_PREP(MI_SDI_LEN_DW, 2 * (x) + 3 - 2) | \
-					 REG_BIT(21))
+#define MI_STORE_DATA_IMM			__MI_INSTR(0x20)
+#define   MI_SDI_GGTT				REG_BIT(22)
+#define   MI_FORCE_WRITE_COMPLETION_CHECK	REG_BIT(10)
+#define   MI_SDI_LEN_DW				GENMASK(9, 0)
+#define   MI_SDI_NUM_DW(x)			REG_FIELD_PREP(MI_SDI_LEN_DW, (x) + 3 - 2)
+#define   MI_SDI_NUM_QW(x)			(REG_FIELD_PREP(MI_SDI_LEN_DW, 2 * (x) + 3 - 2) | \
+						 REG_BIT(21))
 
 #define MI_LOAD_REGISTER_IMM		__MI_INSTR(0x22)
 #define   MI_LRI_LRM_CS_MMIO		REG_BIT(19)

drivers/gpu/drm/xe/xe_migrate.c

@@ -581,7 +581,9 @@ static void emit_pte(struct xe_migrate *m,
 	while (ptes) {
 		u32 chunk = min(MAX_PTE_PER_SDI, ptes);
 
-		bb->cs[bb->len++] = MI_STORE_DATA_IMM | MI_SDI_NUM_QW(chunk);
+		bb->cs[bb->len++] = MI_STORE_DATA_IMM |
+				    MI_FORCE_WRITE_COMPLETION_CHECK |
+				    MI_SDI_NUM_QW(chunk);
 		bb->cs[bb->len++] = ofs;
 		bb->cs[bb->len++] = 0;
 
@@ -1223,7 +1225,9 @@ static void write_pgtable(struct xe_tile *tile, struct xe_bb *bb, u64 ppgtt_ofs,
 		if (!(bb->len & 1))
 			bb->cs[bb->len++] = MI_NOOP;
 
-		bb->cs[bb->len++] = MI_STORE_DATA_IMM | MI_SDI_NUM_QW(chunk);
+		bb->cs[bb->len++] = MI_STORE_DATA_IMM |
+				    MI_FORCE_WRITE_COMPLETION_CHECK |
+				    MI_SDI_NUM_QW(chunk);
 		bb->cs[bb->len++] = lower_32_bits(addr);
 		bb->cs[bb->len++] = upper_32_bits(addr);
 		if (pt_op->bind)
@@ -1388,7 +1392,8 @@ __xe_migrate_update_pgtables(struct xe_migrate *m,
 			u32 idx = 0;
 
 			bb->cs[bb->len++] = MI_STORE_DATA_IMM |
-				MI_SDI_NUM_QW(chunk);
+					    MI_FORCE_WRITE_COMPLETION_CHECK |
+					    MI_SDI_NUM_QW(chunk);
 			bb->cs[bb->len++] = ofs;
 			bb->cs[bb->len++] = 0; /* upper_32_bits */
 

drivers/gpu/drm/xe/xe_oa.c

@@ -690,7 +690,9 @@ static void xe_oa_store_flex(struct xe_oa_stream *stream, struct xe_lrc *lrc,
 	u32 offset = xe_bo_ggtt_addr(lrc->bo);
 
 	do {
-		bb->cs[bb->len++] = MI_STORE_DATA_IMM | MI_SDI_GGTT | MI_SDI_NUM_DW(1);
+		bb->cs[bb->len++] = MI_STORE_DATA_IMM | MI_SDI_GGTT |
+				    MI_FORCE_WRITE_COMPLETION_CHECK |
+				    MI_SDI_NUM_DW(1);
 		bb->cs[bb->len++] = offset + flex->offset * sizeof(u32);
 		bb->cs[bb->len++] = 0;
 		bb->cs[bb->len++] = flex->value;

drivers/gpu/drm/xe/xe_ring_ops.c

@@ -72,7 +72,8 @@ static int emit_user_interrupt(u32 *dw, int i)
 
 static int emit_store_imm_ggtt(u32 addr, u32 value, u32 *dw, int i)
 {
-	dw[i++] = MI_STORE_DATA_IMM | MI_SDI_GGTT | MI_SDI_NUM_DW(1);
+	dw[i++] = MI_STORE_DATA_IMM | MI_SDI_GGTT |
+		  MI_FORCE_WRITE_COMPLETION_CHECK | MI_SDI_NUM_DW(1);
 	dw[i++] = addr;
 	dw[i++] = 0;
 	dw[i++] = value;
@@ -162,7 +163,8 @@ static int emit_pipe_invalidate(u32 mask_flags, bool invalidate_tlb, u32 *dw,
 static int emit_store_imm_ppgtt_posted(u64 addr, u64 value,
 				       u32 *dw, int i)
 {
-	dw[i++] = MI_STORE_DATA_IMM | MI_SDI_NUM_QW(1);
+	dw[i++] = MI_STORE_DATA_IMM | MI_FORCE_WRITE_COMPLETION_CHECK |
+		  MI_SDI_NUM_QW(1);
 	dw[i++] = lower_32_bits(addr);
 	dw[i++] = upper_32_bits(addr);
 	dw[i++] = lower_32_bits(value);