[PATCH 5.4 05/16] drm/etnaviv: fix TS cache flushing on GPUs with BLT engine

[Greg Kroah-Hartman]

From: Lucas Stach <l.stach@xxxxxxxxxxxxxx>

commit f232d9ec029ce3e2543b05213e2979e01e503408 upstream.

As seen in the Vivante kernel driver, most GPUs with the BLT engine have
a broken TS cache flush. The workaround is to temporarily set the BLT
command to CLEAR_IMAGE, without actually executing the clear. Apparently
this state change is enough to trigger the required TS cache flush. As
the BLT engine is completely asychronous, we also need a few more stall
states to synchronize the flush with the frontend.

Root-caused-by: Jonathan Marek <jonathan@xxxxxxxx>
Signed-off-by: Lucas Stach <l.stach@xxxxxxxxxxxxxx>
Cc: Walter Lozano <walter.lozano@xxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>

---
 drivers/gpu/drm/etnaviv/etnaviv_buffer.c |   60 ++++++++++++++++++++++++++++---
 drivers/gpu/drm/etnaviv/state_blt.xml.h  |    2 +
 2 files changed, 57 insertions(+), 5 deletions(-)

--- a/drivers/gpu/drm/etnaviv/etnaviv_buffer.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_buffer.c
@@ -12,6 +12,7 @@
 
 #include "common.xml.h"
 #include "state.xml.h"
+#include "state_blt.xml.h"
 #include "state_hi.xml.h"
 #include "state_3d.xml.h"
 #include "cmdstream.xml.h"
@@ -233,6 +234,8 @@ void etnaviv_buffer_end(struct etnaviv_g
 	struct etnaviv_cmdbuf *buffer = &gpu->buffer;
 	unsigned int waitlink_offset = buffer->user_size - 16;
 	u32 link_target, flush = 0;
+	bool has_blt = !!(gpu->identity.minor_features5 &
+			  chipMinorFeatures5_BLT_ENGINE);
 
 	lockdep_assert_held(&gpu->lock);
 
@@ -248,16 +251,38 @@ void etnaviv_buffer_end(struct etnaviv_g
 	if (flush) {
 		unsigned int dwords = 7;
 
+		if (has_blt)
+			dwords += 10;
+
 		link_target = etnaviv_buffer_reserve(gpu, buffer, dwords);
 
 		CMD_SEM(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_PE);
 		CMD_STALL(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_PE);
+		if (has_blt) {
+			CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x1);
+			CMD_SEM(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_BLT);
+			CMD_STALL(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_BLT);
+			CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x0);
+		}
 		CMD_LOAD_STATE(buffer, VIVS_GL_FLUSH_CACHE, flush);
-		if (gpu->exec_state == ETNA_PIPE_3D)
-			CMD_LOAD_STATE(buffer, VIVS_TS_FLUSH_CACHE,
-				       VIVS_TS_FLUSH_CACHE_FLUSH);
+		if (gpu->exec_state == ETNA_PIPE_3D) {
+			if (has_blt) {
+				CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x1);
+				CMD_LOAD_STATE(buffer, VIVS_BLT_SET_COMMAND, 0x1);
+				CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x0);
+			} else {
+				CMD_LOAD_STATE(buffer, VIVS_TS_FLUSH_CACHE,
+					       VIVS_TS_FLUSH_CACHE_FLUSH);
+			}
+		}
 		CMD_SEM(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_PE);
 		CMD_STALL(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_PE);
+		if (has_blt) {
+			CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x1);
+			CMD_SEM(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_BLT);
+			CMD_STALL(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_BLT);
+			CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x0);
+		}
 		CMD_END(buffer);
 
 		etnaviv_buffer_replace_wait(buffer, waitlink_offset,
@@ -323,6 +348,8 @@ void etnaviv_buffer_queue(struct etnaviv
 	bool switch_mmu_context = gpu->mmu_context != mmu_context;
 	unsigned int new_flush_seq = READ_ONCE(gpu->mmu_context->flush_seq);
 	bool need_flush = switch_mmu_context || gpu->flush_seq != new_flush_seq;
+	bool has_blt = !!(gpu->identity.minor_features5 &
+			  chipMinorFeatures5_BLT_ENGINE);
 
 	lockdep_assert_held(&gpu->lock);
 
@@ -433,6 +460,15 @@ void etnaviv_buffer_queue(struct etnaviv
 	 * 2 semaphore stall + 1 event + 1 wait + 1 link.
 	 */
 	return_dwords = 7;
+
+	/*
+	 * When the BLT engine is present we need 6 more dwords in the return
+	 * target: 3 enable/flush/disable + 4 enable/semaphore stall/disable,
+	 * but we don't need the normal TS flush state.
+	 */
+	if (has_blt)
+		return_dwords += 6;
+
 	return_target = etnaviv_buffer_reserve(gpu, buffer, return_dwords);
 	CMD_LINK(cmdbuf, return_dwords, return_target);
 
@@ -447,11 +483,25 @@ void etnaviv_buffer_queue(struct etnaviv
 		CMD_LOAD_STATE(buffer, VIVS_GL_FLUSH_CACHE,
 				       VIVS_GL_FLUSH_CACHE_DEPTH |
 				       VIVS_GL_FLUSH_CACHE_COLOR);
-		CMD_LOAD_STATE(buffer, VIVS_TS_FLUSH_CACHE,
-				       VIVS_TS_FLUSH_CACHE_FLUSH);
+		if (has_blt) {
+			CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x1);
+			CMD_LOAD_STATE(buffer, VIVS_BLT_SET_COMMAND, 0x1);
+			CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x0);
+		} else {
+			CMD_LOAD_STATE(buffer, VIVS_TS_FLUSH_CACHE,
+					       VIVS_TS_FLUSH_CACHE_FLUSH);
+		}
 	}
 	CMD_SEM(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_PE);
 	CMD_STALL(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_PE);
+
+	if (has_blt) {
+		CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x1);
+		CMD_SEM(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_BLT);
+		CMD_STALL(buffer, SYNC_RECIPIENT_FE, SYNC_RECIPIENT_BLT);
+		CMD_LOAD_STATE(buffer, VIVS_BLT_ENABLE, 0x0);
+	}
+
 	CMD_LOAD_STATE(buffer, VIVS_GL_EVENT, VIVS_GL_EVENT_EVENT_ID(event) |
 		       VIVS_GL_EVENT_FROM_PE);
 	CMD_WAIT(buffer);
--- a/drivers/gpu/drm/etnaviv/state_blt.xml.h
+++ b/drivers/gpu/drm/etnaviv/state_blt.xml.h
@@ -46,6 +46,8 @@ DEALINGS IN THE SOFTWARE.
 
 /* This is a cut-down version of the state_blt.xml.h file */
 
+#define VIVS_BLT_SET_COMMAND					0x000140ac
+
 #define VIVS_BLT_ENABLE						0x000140b8
 #define VIVS_BLT_ENABLE_ENABLE					0x00000001