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: 9c57bc08652a ("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

diff --git a/drivers/gpu/drm/xe/instructions/xe_mi_commands.h b/drivers/gpu/drm/xe/instructions/xe_mi_commands.h
index 10ec2920d31b34cee9f7e175d4f9dcd16a9c7b20..f4ee910f09432c7fbcad300b5e73b91592cd4a77 100644 (file)
--- a/drivers/gpu/drm/xe/instructions/xe_mi_commands.h
+++ b/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)

diff --git a/drivers/gpu/drm/xe/xe_migrate.c b/drivers/gpu/drm/xe/xe_migrate.c
index 1b97d90aaddaf46103a5326f29671b5385d94008..8b32fad678782b5852e83b3b6de23f0e30992452 100644 (file)
--- a/drivers/gpu/drm/xe/xe_migrate.c
+++ b/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 */
 

diff --git a/drivers/gpu/drm/xe/xe_oa.c b/drivers/gpu/drm/xe/xe_oa.c
index 56bf375a9d4bc52faaa6fddc3cee34737afb6ef6..ae94490b0eac83881c4a5dad4edbd8aeeb1c8d9b 100644 (file)
--- a/drivers/gpu/drm/xe/xe_oa.c
+++ b/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;

diff --git a/drivers/gpu/drm/xe/xe_ring_ops.c b/drivers/gpu/drm/xe/xe_ring_ops.c
index 0be4f489d3e126a236b4d47070da7d19265e835e..3a75a08b6be92d10f567f8d2ca9af9c72523e75d 100644 (file)
--- a/drivers/gpu/drm/xe/xe_ring_ops.c
+++ b/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);