drm/amdkfd: Fix CP soft hang on APUs

The problem happens on Raven and Carrizo. The context save handler
should not clear the high bits of PC_HI before extracting the bits
of IB_STS.

The bug is not relevant to VEGA10 until we enable demand paging.
Signed-off-by: Jay Cornwall <Jay.Cornwall@amd.com>
Signed-off-by: Yong Zhao <yong.zhao@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h

@@ -253,7 +253,6 @@ static const uint32_t cwsr_trap_gfx8_hex[] = {
 	0x00000072, 0x80728472,
 	0xc0211b7c, 0x00000072,
 	0x80728472, 0xbf8c007f,
-	0x8671ff71, 0x0000ffff,
 	0xbefc0073, 0xbefe006e,
 	0xbeff006f, 0x867375ff,
 	0x000003ff, 0xb9734803,
@@ -267,6 +266,7 @@ static const uint32_t cwsr_trap_gfx8_hex[] = {
 	0x8e738f73, 0x87767376,
 	0x8673ff74, 0x00800000,
 	0x8f739773, 0xb976f807,
+	0x8671ff71, 0x0000ffff,
 	0x86fe7e7e, 0x86ea6a6a,
 	0xb974f802, 0xbf8a0000,
 	0x95807370, 0xbf810000,
@@ -530,7 +530,6 @@ static const uint32_t cwsr_trap_gfx9_hex[] = {
 	0x00000078, 0x80788478,
 	0xc0211cfa, 0x00000078,
 	0x80788478, 0xbf8cc07f,
-	0x866dff6d, 0x0000ffff,
 	0xbefc006f, 0xbefe007a,
 	0xbeff007b, 0x866f71ff,
 	0x000003ff, 0xb96f4803,
@@ -554,6 +553,7 @@ static const uint32_t cwsr_trap_gfx9_hex[] = {
 	0x8e6f8f6f, 0x876e6f6e,
 	0x866fff70, 0x00800000,
 	0x8f6f976f, 0xb96ef807,
+	0x866dff6d, 0x0000ffff,
 	0x86fe7e7e, 0x86ea6a6a,
 	0xb970f802, 0xbf8a0000,
 	0x95806f6c, 0xbf810000,

drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx8.asm

@@ -1015,8 +1015,6 @@ end
 
     s_waitcnt       lgkmcnt(0)                                                                                      //from now on, it is safe to restore STATUS and IB_STS
 
-    s_and_b32 s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff      //pc[47:32]        //Do it here in order not to affect STATUS
-
     //for normal save & restore, the saved PC points to the next inst to execute, no adjustment needs to be made, otherwise:
     if ((EMU_RUN_HACK) && (!EMU_RUN_HACK_RESTORE_NORMAL))
         s_add_u32 s_restore_pc_lo, s_restore_pc_lo, 8            //pc[31:0]+8     //two back-to-back s_trap are used (first for save and second for restore)
@@ -1052,6 +1050,7 @@ end
     s_lshr_b32      s_restore_m0, s_restore_m0, SQ_WAVE_STATUS_INST_ATC_SHIFT
     s_setreg_b32    hwreg(HW_REG_IB_STS),   s_restore_tmp
 
+    s_and_b32 s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff      //pc[47:32]        //Do it here in order not to affect STATUS
     s_and_b64    exec, exec, exec  // Restore STATUS.EXECZ, not writable by s_setreg_b32
     s_and_b64    vcc, vcc, vcc  // Restore STATUS.VCCZ, not writable by s_setreg_b32
     s_setreg_b32    hwreg(HW_REG_STATUS),   s_restore_status     // SCC is included, which is changed by previous salu

drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx9.asm

@@ -1067,8 +1067,6 @@ end
 
     s_waitcnt	    lgkmcnt(0)											    //from now on, it is safe to restore STATUS and IB_STS
 
-    s_and_b32 s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff	//pc[47:32]	   //Do it here in order not to affect STATUS
-
     //for normal save & restore, the saved PC points to the next inst to execute, no adjustment needs to be made, otherwise:
     if ((EMU_RUN_HACK) && (!EMU_RUN_HACK_RESTORE_NORMAL))
 	s_add_u32 s_restore_pc_lo, s_restore_pc_lo, 8		 //pc[31:0]+8	  //two back-to-back s_trap are used (first for save and second for restore)
@@ -1119,6 +1117,7 @@ end
     s_lshr_b32	    s_restore_m0, s_restore_m0, SQ_WAVE_STATUS_INST_ATC_SHIFT
     s_setreg_b32    hwreg(HW_REG_IB_STS),   s_restore_tmp
 
+    s_and_b32 s_restore_pc_hi, s_restore_pc_hi, 0x0000ffff	//pc[47:32]	   //Do it here in order not to affect STATUS
     s_and_b64	 exec, exec, exec  // Restore STATUS.EXECZ, not writable by s_setreg_b32
     s_and_b64	 vcc, vcc, vcc	// Restore STATUS.VCCZ, not writable by s_setreg_b32
     s_setreg_b32    hwreg(HW_REG_STATUS),   s_restore_status	 // SCC is included, which is changed by previous salu