drm/amdkfd: Enable SQ watchpoint for gfx10

There are new control registers introduced in gfx10 used to configure
hardware watchpoints triggered by SMEM instructions:
SQ_WATCH{0,1,2,3}_{CNTL_ADDR_HI,ADDR_L}.

Those registers work in a similar way as the TCP_WATCH* registers
currently used for gfx9 and above.

This patch adds support to program the SQ_WATCH registers for gfx10.

The SQ_WATCH?_CNTL.MASK field has one bit more than
TCP_WATCH?_CNTL.MASK, so SQ watchpoints can have a finer granularity
than TCP_WATCH watchpoints.  In this patch, we keep the capabilities
advertised to the debugger unchanged
(HSA_DBG_WATCH_ADDR_MASK_*_BIT_GFX10) as this reflects what both TCP and
SQ watchpoints can do and both watchpoints are configured together.
Signed-off-by: Lancelot SIX <lancelot.six@amd.com>
Reviewed-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c

@@ -881,6 +881,7 @@ uint32_t kgd_gfx_v10_set_wave_launch_mode(struct amdgpu_device *adev,
 }
 
 #define TCP_WATCH_STRIDE (mmTCP_WATCH1_ADDR_H - mmTCP_WATCH0_ADDR_H)
+#define SQ_WATCH_STRIDE (mmSQ_WATCH1_ADDR_H - mmSQ_WATCH0_ADDR_H)
 uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev,
 					uint64_t watch_address,
 					uint32_t watch_address_mask,
@@ -889,55 +890,93 @@ uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev,
 					uint32_t debug_vmid,
 					uint32_t inst)
 {
+	/* SQ_WATCH?_ADDR_* and TCP_WATCH?_ADDR_* are programmed with the
+	 * same values.
+	 */
 	uint32_t watch_address_high;
 	uint32_t watch_address_low;
-	uint32_t watch_address_cntl;
-
-	watch_address_cntl = 0;
+	uint32_t tcp_watch_address_cntl;
+	uint32_t sq_watch_address_cntl;
 
 	watch_address_low = lower_32_bits(watch_address);
 	watch_address_high = upper_32_bits(watch_address) & 0xffff;
 
-	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+	tcp_watch_address_cntl = 0;
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
 			TCP_WATCH0_CNTL,
 			VMID,
 			debug_vmid);
-	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
 			TCP_WATCH0_CNTL,
 			MODE,
 			watch_mode);
-	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
 			TCP_WATCH0_CNTL,
 			MASK,
 			watch_address_mask >> 7);
 
+	sq_watch_address_cntl = 0;
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			VMID,
+			debug_vmid);
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 6);
+
 	/* Turning off this watch point until we set all the registers */
-	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
 			TCP_WATCH0_CNTL,
 			VALID,
 			0);
-
 	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
 			(watch_id * TCP_WATCH_STRIDE)),
-			watch_address_cntl);
+			tcp_watch_address_cntl);
+
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			VALID,
+			0);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			sq_watch_address_cntl);
 
+	/* Program {TCP,SQ}_WATCH?_ADDR* */
 	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) +
 			(watch_id * TCP_WATCH_STRIDE)),
 			watch_address_high);
-
 	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_L) +
 			(watch_id * TCP_WATCH_STRIDE)),
 			watch_address_low);
 
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_ADDR_H) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			watch_address_high);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_ADDR_L) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			watch_address_low);
+
 	/* Enable the watch point */
-	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+	tcp_watch_address_cntl = REG_SET_FIELD(tcp_watch_address_cntl,
 			TCP_WATCH0_CNTL,
 			VALID,
 			1);
-
 	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
 			(watch_id * TCP_WATCH_STRIDE)),
-			watch_address_cntl);
+			tcp_watch_address_cntl);
+
+	sq_watch_address_cntl = REG_SET_FIELD(sq_watch_address_cntl,
+			SQ_WATCH0_CNTL,
+			VALID,
+			1);
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			sq_watch_address_cntl);
 
 	return 0;
 }
@@ -953,8 +992,14 @@ uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev,
 			(watch_id * TCP_WATCH_STRIDE)),
 			watch_address_cntl);
 
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmSQ_WATCH0_CNTL) +
+			(watch_id * SQ_WATCH_STRIDE)),
+			watch_address_cntl);
+
 	return 0;
 }
+#undef TCP_WATCH_STRIDE
+#undef SQ_WATCH_STRIDE
 
 
 /* kgd_gfx_v10_get_iq_wait_times: Returns the mmCP_IQ_WAIT_TIME1/2 values