drm/amdkfd: Use SQC when TCP would fail in gfx9 context save.

When a wavefront raises TRAPSTS.XNACK_ERROR with STATUS.ALLOW_REPLAY=0
subsequent memory instructions have undefined behavior. In practice
SQC stores continue to work but TCP stores do not.

Context save is permitted to fail after XNACK error because the
wavefront will be halted and subsequently terminated. However the
debugger has an interest in retrieving the wavefront VGPR/LDS state.

Detect the out-of-spec case and use SQC stores during context save
in place of TCP stores.
Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -274,7 +274,7 @@ static const uint32_t cwsr_trap_gfx8_hex[] = {
 
 
 static const uint32_t cwsr_trap_gfx9_hex[] = {
-	0xbf820001, 0xbf82015e,
+	0xbf820001, 0xbf820241,
 	0xb8f8f802, 0x89788678,
 	0xb8fbf803, 0x866eff7b,
 	0x00000400, 0xbf85003b,
@@ -404,15 +404,57 @@ static const uint32_t cwsr_trap_gfx9_hex[] = {
 	0xbefe00c1, 0xbeff00c1,
 	0xbee80080, 0xbee90080,
 	0xbef600ff, 0x01000000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf85004d, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbf820008,
 	0xe0724000, 0x701d0000,
 	0xe0724100, 0x701d0100,
 	0xe0724200, 0x701d0200,
 	0xe0724300, 0x701d0300,
 	0xbefe00c1, 0xbeff00c1,
 	0xb8fb4306, 0x867bc17b,
-	0xbf84002c, 0xbf8a0000,
+	0xbf840063, 0xbf8a0000,
 	0x867aff6f, 0x04000000,
-	0xbf840028, 0x8e7b867b,
+	0xbf84005f, 0x8e7b867b,
 	0x8e7b827b, 0xbef6007b,
 	0xb8f02a05, 0x80708170,
 	0x8e708a70, 0xb8fa1605,
@@ -422,144 +464,215 @@ static const uint32_t cwsr_trap_gfx9_hex[] = {
 	0x01000000, 0xbefc0080,
 	0xd28c0002, 0x000100c1,
 	0xd28d0003, 0x000204c1,
-	0xd1060002, 0x00011103,
-	0x7e0602ff, 0x00000200,
-	0xbefc00ff, 0x00010000,
-	0xbe800077, 0x8677ff77,
-	0xff7fffff, 0x8777ff77,
-	0x00058000, 0xd8ec0000,
-	0x00000002, 0xbf8cc07f,
-	0xe0765000, 0x701d0002,
-	0x68040702, 0xd0c9006a,
-	0x0000f702, 0xbf87fff7,
-	0xbef70000, 0xbef000ff,
-	0x00000400, 0xbefe00c1,
-	0xbeff00c1, 0xb8fb2a05,
-	0x807b817b, 0x8e7b827b,
-	0x8e76887b, 0xbef600ff,
-	0x01000000, 0xbefc0084,
-	0xbf0a7b7c, 0xbf840015,
-	0xbf11017c, 0x807bff7b,
-	0x00001000, 0x7e000300,
-	0x7e020301, 0x7e040302,
-	0x7e060303, 0xe0724000,
-	0x701d0000, 0xe0724100,
-	0x701d0100, 0xe0724200,
-	0x701d0200, 0xe0724300,
-	0x701d0300, 0x807c847c,
-	0x8070ff70, 0x00000400,
-	0xbf0a7b7c, 0xbf85ffef,
-	0xbf9c0000, 0xbf8200da,
-	0xbef4007e, 0x8675ff7f,
-	0x0000ffff, 0x8775ff75,
-	0x00040000, 0xbef60080,
-	0xbef700ff, 0x00807fac,
-	0x866eff7f, 0x08000000,
-	0x8f6e836e, 0x87776e77,
-	0x866eff7f, 0x70000000,
-	0x8f6e816e, 0x87776e77,
-	0x866eff7f, 0x04000000,
-	0xbf84001e, 0xbefe00c1,
-	0xbeff00c1, 0xb8ef4306,
-	0x866fc16f, 0xbf840019,
-	0x8e6f866f, 0x8e6f826f,
-	0xbef6006f, 0xb8f82a05,
-	0x80788178, 0x8e788a78,
-	0xb8ee1605, 0x806e816e,
-	0x8e6e866e, 0x80786e78,
-	0x8078ff78, 0x00000080,
-	0xbef600ff, 0x01000000,
-	0xbefc0080, 0xe0510000,
-	0x781d0000, 0xe0510100,
-	0x781d0000, 0x807cff7c,
-	0x00000200, 0x8078ff78,
-	0x00000200, 0xbf0a6f7c,
-	0xbf85fff6, 0xbef80080,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850030, 0x24040682,
+	0xd86e4000, 0x00000002,
+	0xbf8cc07f, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x680404ff,
+	0x00000200, 0xd0c9006a,
+	0x0000f702, 0xbf87ffd2,
+	0xbf820015, 0xd1060002,
+	0x00011103, 0x7e0602ff,
+	0x00000200, 0xbefc00ff,
+	0x00010000, 0xbe800077,
+	0x8677ff77, 0xff7fffff,
+	0x8777ff77, 0x00058000,
+	0xd8ec0000, 0x00000002,
+	0xbf8cc07f, 0xe0765000,
+	0x701d0002, 0x68040702,
+	0xd0c9006a, 0x0000f702,
+	0xbf87fff7, 0xbef70000,
+	0xbef000ff, 0x00000400,
 	0xbefe00c1, 0xbeff00c1,
-	0xb8ef2a05, 0x806f816f,
-	0x8e6f826f, 0x8e76886f,
+	0xb8fb2a05, 0x807b817b,
+	0x8e7b827b, 0x8e76887b,
 	0xbef600ff, 0x01000000,
-	0xbeee0078, 0x8078ff78,
-	0x00000400, 0xbefc0084,
-	0xbf11087c, 0x806fff6f,
-	0x00008000, 0xe0524000,
-	0x781d0000, 0xe0524100,
-	0x781d0100, 0xe0524200,
-	0x781d0200, 0xe0524300,
-	0x781d0300, 0xbf8c0f70,
+	0xbefc0084, 0xbf0a7b7c,
+	0xbf84006d, 0xbf11017c,
+	0x807bff7b, 0x00001000,
+	0x867aff78, 0x00400000,
+	0xbf850003, 0xb8faf803,
+	0x897a7aff, 0x10000000,
+	0xbf850051, 0xbe840080,
+	0xd2890000, 0x00000900,
+	0x80048104, 0xd2890001,
+	0x00000900, 0x80048104,
+	0xd2890002, 0x00000900,
+	0x80048104, 0xd2890003,
+	0x00000900, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000901, 0x80048104,
+	0xd2890001, 0x00000901,
+	0x80048104, 0xd2890002,
+	0x00000901, 0x80048104,
+	0xd2890003, 0x00000901,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0xbe840080,
+	0xd2890000, 0x00000902,
+	0x80048104, 0xd2890001,
+	0x00000902, 0x80048104,
+	0xd2890002, 0x00000902,
+	0x80048104, 0xd2890003,
+	0x00000902, 0x80048104,
+	0xc069003a, 0x00000070,
+	0xbf8cc07f, 0x80709070,
+	0xbf06c004, 0xbf84ffee,
+	0xbe840080, 0xd2890000,
+	0x00000903, 0x80048104,
+	0xd2890001, 0x00000903,
+	0x80048104, 0xd2890002,
+	0x00000903, 0x80048104,
+	0xd2890003, 0x00000903,
+	0x80048104, 0xc069003a,
+	0x00000070, 0xbf8cc07f,
+	0x80709070, 0xbf06c004,
+	0xbf84ffee, 0x807c847c,
+	0xbf0a7b7c, 0xbf85ffb1,
+	0xbf9c0000, 0xbf820012,
 	0x7e000300, 0x7e020301,
 	0x7e040302, 0x7e060303,
-	0x807c847c, 0x8078ff78,
-	0x00000400, 0xbf0a6f7c,
-	0xbf85ffee, 0xbf9c0000,
-	0xe0524000, 0x6e1d0000,
-	0xe0524100, 0x6e1d0100,
-	0xe0524200, 0x6e1d0200,
-	0xe0524300, 0x6e1d0300,
+	0xe0724000, 0x701d0000,
+	0xe0724100, 0x701d0100,
+	0xe0724200, 0x701d0200,
+	0xe0724300, 0x701d0300,
+	0x807c847c, 0x8070ff70,
+	0x00000400, 0xbf0a7b7c,
+	0xbf85ffef, 0xbf9c0000,
+	0xbf8200da, 0xbef4007e,
+	0x8675ff7f, 0x0000ffff,
+	0x8775ff75, 0x00040000,
+	0xbef60080, 0xbef700ff,
+	0x00807fac, 0x866eff7f,
+	0x08000000, 0x8f6e836e,
+	0x87776e77, 0x866eff7f,
+	0x70000000, 0x8f6e816e,
+	0x87776e77, 0x866eff7f,
+	0x04000000, 0xbf84001e,
+	0xbefe00c1, 0xbeff00c1,
+	0xb8ef4306, 0x866fc16f,
+	0xbf840019, 0x8e6f866f,
+	0x8e6f826f, 0xbef6006f,
 	0xb8f82a05, 0x80788178,
 	0x8e788a78, 0xb8ee1605,
 	0x806e816e, 0x8e6e866e,
-	0x80786e78, 0x80f8c078,
-	0xb8ef1605, 0x806f816f,
-	0x8e6f846f, 0x8e76826f,
-	0xbef600ff, 0x01000000,
-	0xbefc006f, 0xc031003a,
-	0x00000078, 0x80f8c078,
-	0xbf8cc07f, 0x80fc907c,
-	0xbf800000, 0xbe802d00,
-	0xbe822d02, 0xbe842d04,
-	0xbe862d06, 0xbe882d08,
-	0xbe8a2d0a, 0xbe8c2d0c,
-	0xbe8e2d0e, 0xbf06807c,
-	0xbf84fff0, 0xb8f82a05,
+	0x80786e78, 0x8078ff78,
+	0x00000080, 0xbef600ff,
+	0x01000000, 0xbefc0080,
+	0xe0510000, 0x781d0000,
+	0xe0510100, 0x781d0000,
+	0x807cff7c, 0x00000200,
+	0x8078ff78, 0x00000200,
+	0xbf0a6f7c, 0xbf85fff6,
+	0xbef80080, 0xbefe00c1,
+	0xbeff00c1, 0xb8ef2a05,
+	0x806f816f, 0x8e6f826f,
+	0x8e76886f, 0xbef600ff,
+	0x01000000, 0xbeee0078,
+	0x8078ff78, 0x00000400,
+	0xbefc0084, 0xbf11087c,
+	0x806fff6f, 0x00008000,
+	0xe0524000, 0x781d0000,
+	0xe0524100, 0x781d0100,
+	0xe0524200, 0x781d0200,
+	0xe0524300, 0x781d0300,
+	0xbf8c0f70, 0x7e000300,
+	0x7e020301, 0x7e040302,
+	0x7e060303, 0x807c847c,
+	0x8078ff78, 0x00000400,
+	0xbf0a6f7c, 0xbf85ffee,
+	0xbf9c0000, 0xe0524000,
+	0x6e1d0000, 0xe0524100,
+	0x6e1d0100, 0xe0524200,
+	0x6e1d0200, 0xe0524300,
+	0x6e1d0300, 0xb8f82a05,
 	0x80788178, 0x8e788a78,
 	0xb8ee1605, 0x806e816e,
 	0x8e6e866e, 0x80786e78,
-	0xbef60084, 0xbef600ff,
-	0x01000000, 0xc0211bfa,
+	0x80f8c078, 0xb8ef1605,
+	0x806f816f, 0x8e6f846f,
+	0x8e76826f, 0xbef600ff,
+	0x01000000, 0xbefc006f,
+	0xc031003a, 0x00000078,
+	0x80f8c078, 0xbf8cc07f,
+	0x80fc907c, 0xbf800000,
+	0xbe802d00, 0xbe822d02,
+	0xbe842d04, 0xbe862d06,
+	0xbe882d08, 0xbe8a2d0a,
+	0xbe8c2d0c, 0xbe8e2d0e,
+	0xbf06807c, 0xbf84fff0,
+	0xb8f82a05, 0x80788178,
+	0x8e788a78, 0xb8ee1605,
+	0x806e816e, 0x8e6e866e,
+	0x80786e78, 0xbef60084,
+	0xbef600ff, 0x01000000,
+	0xc0211bfa, 0x00000078,
+	0x80788478, 0xc0211b3a,
 	0x00000078, 0x80788478,
-	0xc0211b3a, 0x00000078,
-	0x80788478, 0xc0211b7a,
+	0xc0211b7a, 0x00000078,
+	0x80788478, 0xc0211c3a,
 	0x00000078, 0x80788478,
-	0xc0211c3a, 0x00000078,
-	0x80788478, 0xc0211c7a,
+	0xc0211c7a, 0x00000078,
+	0x80788478, 0xc0211eba,
 	0x00000078, 0x80788478,
-	0xc0211eba, 0x00000078,
-	0x80788478, 0xc0211efa,
+	0xc0211efa, 0x00000078,
+	0x80788478, 0xc0211a3a,
 	0x00000078, 0x80788478,
-	0xc0211a3a, 0x00000078,
-	0x80788478, 0xc0211a7a,
+	0xc0211a7a, 0x00000078,
+	0x80788478, 0xc0211cfa,
 	0x00000078, 0x80788478,
-	0xc0211cfa, 0x00000078,
-	0x80788478, 0xbf8cc07f,
-	0xbefc006f, 0xbefe0070,
-	0xbeff0071, 0x866f7bff,
-	0x000003ff, 0xb96f4803,
-	0x866f7bff, 0xfffff800,
-	0x8f6f8b6f, 0xb96fa2c3,
-	0xb973f801, 0xb8ee2a05,
-	0x806e816e, 0x8e6e8a6e,
-	0xb8ef1605, 0x806f816f,
-	0x8e6f866f, 0x806e6f6e,
-	0x806e746e, 0x826f8075,
-	0x866fff6f, 0x0000ffff,
-	0xc00b1c37, 0x00000050,
-	0xc00b1d37, 0x00000060,
-	0xc0031e77, 0x00000074,
-	0xbf8cc07f, 0x866fff6d,
-	0xf8000000, 0x8f6f9b6f,
-	0x8e6f906f, 0xbeee0080,
-	0x876e6f6e, 0x866fff6d,
-	0x04000000, 0x8f6f9a6f,
-	0x8e6f8f6f, 0x876e6f6e,
-	0x866fff7a, 0x00800000,
-	0x8f6f976f, 0xb96ef807,
-	0x866dff6d, 0x0000ffff,
-	0x86fe7e7e, 0x86ea6a6a,
-	0x8f6e837a, 0xb96ee0c2,
-	0xbf800002, 0xb97a0002,
-	0xbf8a0000, 0x95806f6c,
-	0xbf810000, 0x00000000,
+	0xbf8cc07f, 0xbefc006f,
+	0xbefe0070, 0xbeff0071,
+	0x866f7bff, 0x000003ff,
+	0xb96f4803, 0x866f7bff,
+	0xfffff800, 0x8f6f8b6f,
+	0xb96fa2c3, 0xb973f801,
+	0xb8ee2a05, 0x806e816e,
+	0x8e6e8a6e, 0xb8ef1605,
+	0x806f816f, 0x8e6f866f,
+	0x806e6f6e, 0x806e746e,
+	0x826f8075, 0x866fff6f,
+	0x0000ffff, 0xc00b1c37,
+	0x00000050, 0xc00b1d37,
+	0x00000060, 0xc0031e77,
+	0x00000074, 0xbf8cc07f,
+	0x866fff6d, 0xf8000000,
+	0x8f6f9b6f, 0x8e6f906f,
+	0xbeee0080, 0x876e6f6e,
+	0x866fff6d, 0x04000000,
+	0x8f6f9a6f, 0x8e6f8f6f,
+	0x876e6f6e, 0x866fff7a,
+	0x00800000, 0x8f6f976f,
+	0xb96ef807, 0x866dff6d,
+	0x0000ffff, 0x86fe7e7e,
+	0x86ea6a6a, 0x8f6e837a,
+	0xb96ee0c2, 0xbf800002,
+	0xb97a0002, 0xbf8a0000,
+	0x95806f6c, 0xbf810000,
 };
 
 static const uint32_t cwsr_trap_gfx10_hex[] = {

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

@@ -94,6 +94,7 @@ var SGPR_SAVE_USE_SQC		    =	1		    //use SQC D$ to do the write
 var USE_MTBUF_INSTEAD_OF_MUBUF	    =	0		    //because TC EMU currently asserts on 0 of // overload DFMT field to carry 4 more bits of stride for MUBUF opcodes
 var SWIZZLE_EN			    =	0		    //whether we use swizzled buffer addressing
 var ACK_SQC_STORE		    =	1		    //workaround for suspected SQC store bug causing incorrect stores under concurrency
+var SAVE_AFTER_XNACK_ERROR	    =	1		    //workaround for TCP store failure after XNACK error when ALLOW_REPLAY=0, for debugger
 
 /**************************************************************************/
 /*			variables					  */
@@ -107,6 +108,7 @@ var SQ_WAVE_STATUS_PRE_SPI_PRIO_SHIFT   = 0
 var SQ_WAVE_STATUS_PRE_SPI_PRIO_SIZE    = 1
 var SQ_WAVE_STATUS_POST_SPI_PRIO_SHIFT  = 3
 var SQ_WAVE_STATUS_POST_SPI_PRIO_SIZE   = 29
+var SQ_WAVE_STATUS_ALLOW_REPLAY_MASK    = 0x400000
 
 var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT	= 12
 var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE	= 9
@@ -127,6 +129,7 @@ var SQ_WAVE_TRAPSTS_POST_SAVECTX_MASK	=   0xFFFFF800
 var SQ_WAVE_TRAPSTS_POST_SAVECTX_SHIFT	=   11
 var SQ_WAVE_TRAPSTS_POST_SAVECTX_SIZE	=   21
 var SQ_WAVE_TRAPSTS_ILLEGAL_INST_MASK	=   0x800
+var SQ_WAVE_TRAPSTS_XNACK_ERROR_MASK	=   0x10000000
 
 var SQ_WAVE_IB_STS_RCNT_SHIFT		=   16			//FIXME
 var SQ_WAVE_IB_STS_FIRST_REPLAY_SHIFT	=   15			//FIXME
@@ -584,6 +587,16 @@ if G8SR_VGPR_SR_IN_DWX4
 	s_and_b32 s_save_buf_rsrc1, s_save_buf_rsrc1, 0x0000FFFF   // reset const stride to 0
 	s_or_b32  s_save_buf_rsrc1, s_save_buf_rsrc1, S_SAVE_BUF_RSRC_WORD1_STRIDE  // reset const stride to 4 bytes
 else
+if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_FIRST_VGPRS_WITH_TCP
+
+	write_vgprs_to_mem_with_sqc(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+	s_branch L_SAVE_LDS
+
+L_SAVE_FIRST_VGPRS_WITH_TCP:
+end
+
 	buffer_store_dword v0, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1
 	buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256
 	buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset slc:1 glc:1  offset:256*2
@@ -683,6 +696,27 @@ elsif LDS_DMA_ENABLE==1 && UNROLL==1 // UNROOL	, has ichace miss
 else   // BUFFER_STORE
       v_mbcnt_lo_u32_b32 v2, 0xffffffff, 0x0
       v_mbcnt_hi_u32_b32 v3, 0xffffffff, v2	// tid
+
+if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_LDS_WITH_TCP
+
+	v_lshlrev_b32 v2, 2, v3
+L_SAVE_LDS_LOOP_SQC:
+	ds_read2_b32 v[0:1], v2 offset0:0 offset1:0x40
+	s_waitcnt lgkmcnt(0)
+
+	write_vgprs_to_mem_with_sqc(v0, 2, s_save_buf_rsrc0, s_save_mem_offset)
+
+	v_add_u32 v2, 0x200, v2
+	v_cmp_lt_u32 vcc[0:1], v2, s_save_alloc_size
+	s_cbranch_vccnz L_SAVE_LDS_LOOP_SQC
+
+	s_branch L_SAVE_LDS_DONE
+
+L_SAVE_LDS_WITH_TCP:
+end
+
       v_mul_i32_i24 v2, v3, 8	// tid*8
       v_mov_b32 v3, 256*2
       s_mov_b32 m0, 0x10000
@@ -769,6 +803,21 @@ else
     s_set_gpr_idx_on	m0, 0x1 //M0[7:0] = M0[7:0] and M0[15:12] = 0x1
     s_add_u32	    s_save_alloc_size, s_save_alloc_size, 0x1000		    //add 0x1000 since we compare m0 against it later
 
+if SAVE_AFTER_XNACK_ERROR
+	check_if_tcp_store_ok()
+	s_cbranch_scc1 L_SAVE_VGPR_LOOP
+
+L_SAVE_VGPR_LOOP_SQC:
+	write_vgprs_to_mem_with_sqc(v0, 4, s_save_buf_rsrc0, s_save_mem_offset)
+
+	s_add_u32 m0, m0, 4
+	s_cmp_lt_u32 m0, s_save_alloc_size
+	s_cbranch_scc1 L_SAVE_VGPR_LOOP_SQC
+
+	s_set_gpr_idx_off
+	s_branch L_SAVE_VGPR_END
+end
+
   L_SAVE_VGPR_LOOP:
     v_mov_b32	    v0, v0		//v0 = v[0+m0]
     v_mov_b32	    v1, v1		//v0 = v[0+m0]
@@ -1265,7 +1314,39 @@ function read_16sgpr_from_mem(s, s_rsrc, s_mem_offset)
     s_sub_u32	    s_mem_offset, s_mem_offset, 4*16
 end
 
+function check_if_tcp_store_ok
+	// If STATUS.ALLOW_REPLAY=0 and TRAPSTS.XNACK_ERROR=1 then TCP stores will fail.
+	s_and_b32 s_save_tmp, s_save_status, SQ_WAVE_STATUS_ALLOW_REPLAY_MASK
+	s_cbranch_scc1 L_TCP_STORE_CHECK_DONE
+
+	s_getreg_b32 s_save_tmp, hwreg(HW_REG_TRAPSTS)
+	s_andn2_b32 s_save_tmp, SQ_WAVE_TRAPSTS_XNACK_ERROR_MASK, s_save_tmp
+
+L_TCP_STORE_CHECK_DONE:
+end
+
+function write_vgpr_to_mem_with_sqc(v, s_rsrc, s_mem_offset)
+	s_mov_b32 s4, 0
 
+L_WRITE_VGPR_LANE_LOOP:
+	for var lane = 0; lane < 4; ++ lane
+		v_readlane_b32 s[lane], v, s4
+		s_add_u32 s4, s4, 1
+	end
+
+	s_buffer_store_dwordx4 s[0:3], s_rsrc, s_mem_offset glc:1
+	ack_sqc_store_workaround()
+
+	s_add_u32 s_mem_offset, s_mem_offset, 0x10
+	s_cmp_eq_u32 s4, 0x40
+	s_cbranch_scc0 L_WRITE_VGPR_LANE_LOOP
+end
+
+function write_vgprs_to_mem_with_sqc(v, n_vgprs, s_rsrc, s_mem_offset)
+	for var vgpr = 0; vgpr < n_vgprs; ++ vgpr
+		write_vgpr_to_mem_with_sqc(v[vgpr], s_rsrc, s_mem_offset)
+	end
+end
 
 function get_lds_size_bytes(s_lds_size_byte)
     // SQ LDS granularity is 64DW, while PGM_RSRC2.lds_size is in granularity 128DW