[PATCH] drm/amdkfd: Move gfx12 trap handler to separate file
Lancelot SIX
Lancelot.Six at amd.com
Fri Oct 4 09:42:44 UTC 2024
Hi Jay,
Thanks for doing this. I have minor comments inlined below. Overall,
this new version assembles to the same binary, so no functional change
expected.
Best,
Lancelot.
On 03/10/2024 20:31, Jay Cornwall wrote:
> gfx12 derivatives will have substantially different trap handler
> implementations from gfx10/gfx11. Add a separate source file for
> gfx12+ and remove unneeded conditional code.
>
> No functional change.
>
> Signed-off-by: Jay Cornwall <jay.cornwall at amd.com>
> Cc: Lancelot Six <lancelot.six at amd.com>
> Cc: Jonathan Kim <jonathan.kim at amd.com>
> ---
> .../amd/amdkfd/cwsr_trap_handler_gfx10.asm | 202 +--
> .../amd/amdkfd/cwsr_trap_handler_gfx12.asm | 1128 +++++++++++++++++
> 2 files changed, 1129 insertions(+), 201 deletions(-)
> create mode 100644 drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
> index 44772eec9ef4..96fbb16ceb21 100644
> --- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
> +++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm
> @@ -34,41 +34,24 @@
> * cpp -DASIC_FAMILY=CHIP_PLUM_BONITO cwsr_trap_handler_gfx10.asm -P -o gfx11.sp3
> * sp3 gfx11.sp3 -hex gfx11.hex
> *
> - * gfx12:
> - * cpp -DASIC_FAMILY=CHIP_GFX12 cwsr_trap_handler_gfx10.asm -P -o gfx12.sp3
> - * sp3 gfx12.sp3 -hex gfx12.hex
> */
>
> #define CHIP_NAVI10 26
> #define CHIP_SIENNA_CICHLID 30
> #define CHIP_PLUM_BONITO 36
> -#define CHIP_GFX12 37
>
> #define NO_SQC_STORE (ASIC_FAMILY >= CHIP_SIENNA_CICHLID)
> #define HAVE_XNACK (ASIC_FAMILY < CHIP_SIENNA_CICHLID)
> #define HAVE_SENDMSG_RTN (ASIC_FAMILY >= CHIP_PLUM_BONITO)
> #define HAVE_BUFFER_LDS_LOAD (ASIC_FAMILY < CHIP_PLUM_BONITO)
> -#define SW_SA_TRAP (ASIC_FAMILY >= CHIP_PLUM_BONITO && ASIC_FAMILY < CHIP_GFX12)
> +#define SW_SA_TRAP (ASIC_FAMILY == CHIP_PLUM_BONITO)
> #define SAVE_AFTER_XNACK_ERROR (HAVE_XNACK && !NO_SQC_STORE) // workaround for TCP store failure after XNACK error when ALLOW_REPLAY=0, for debugger
> #define SINGLE_STEP_MISSED_WORKAROUND 1 //workaround for lost MODE.DEBUG_EN exception when SAVECTX raised
>
> -#if ASIC_FAMILY < CHIP_GFX12
> #define S_COHERENCE glc:1
> #define V_COHERENCE slc:1 glc:1
> #define S_WAITCNT_0 s_waitcnt 0
> -#else
> -#define S_COHERENCE scope:SCOPE_SYS
> -#define V_COHERENCE scope:SCOPE_SYS
> -#define S_WAITCNT_0 s_wait_idle
> -
> -#define HW_REG_SHADER_FLAT_SCRATCH_LO HW_REG_WAVE_SCRATCH_BASE_LO
> -#define HW_REG_SHADER_FLAT_SCRATCH_HI HW_REG_WAVE_SCRATCH_BASE_HI
> -#define HW_REG_GPR_ALLOC HW_REG_WAVE_GPR_ALLOC
> -#define HW_REG_LDS_ALLOC HW_REG_WAVE_LDS_ALLOC
> -#define HW_REG_MODE HW_REG_WAVE_MODE
> -#endif
>
> -#if ASIC_FAMILY < CHIP_GFX12
> var SQ_WAVE_STATUS_SPI_PRIO_MASK = 0x00000006
> var SQ_WAVE_STATUS_HALT_MASK = 0x2000
> var SQ_WAVE_STATUS_ECC_ERR_MASK = 0x20000
> @@ -81,21 +64,6 @@ var S_STATUS_ALWAYS_CLEAR_MASK = SQ_WAVE_STATUS_SPI_PRIO_MASK|SQ_WAVE_STATUS_E
> var S_STATUS_HALT_MASK = SQ_WAVE_STATUS_HALT_MASK
> var S_SAVE_PC_HI_TRAP_ID_MASK = 0x00FF0000
> var S_SAVE_PC_HI_HT_MASK = 0x01000000
> -#else
> -var SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK = 0x4
> -var SQ_WAVE_STATE_PRIV_SCC_SHIFT = 9
> -var SQ_WAVE_STATE_PRIV_SYS_PRIO_MASK = 0xC00
> -var SQ_WAVE_STATE_PRIV_HALT_MASK = 0x4000
> -var SQ_WAVE_STATE_PRIV_POISON_ERR_MASK = 0x8000
> -var SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT = 15
> -var SQ_WAVE_STATUS_WAVE64_SHIFT = 29
> -var SQ_WAVE_STATUS_WAVE64_SIZE = 1
> -var SQ_WAVE_LDS_ALLOC_GRANULARITY = 9
> -var S_STATUS_HWREG = HW_REG_WAVE_STATE_PRIV
> -var S_STATUS_ALWAYS_CLEAR_MASK = SQ_WAVE_STATE_PRIV_SYS_PRIO_MASK|SQ_WAVE_STATE_PRIV_POISON_ERR_MASK
> -var S_STATUS_HALT_MASK = SQ_WAVE_STATE_PRIV_HALT_MASK
> -var S_SAVE_PC_HI_TRAP_ID_MASK = 0xF0000000
> -#endif
>
> var SQ_WAVE_STATUS_NO_VGPRS_SHIFT = 24
> var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT = 12
> @@ -110,7 +78,6 @@ var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT = 8
> var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT = 12
> #endif
>
> -#if ASIC_FAMILY < CHIP_GFX12
> var SQ_WAVE_TRAPSTS_SAVECTX_MASK = 0x400
> var SQ_WAVE_TRAPSTS_EXCP_MASK = 0x1FF
> var SQ_WAVE_TRAPSTS_SAVECTX_SHIFT = 10
> @@ -161,39 +128,6 @@ var S_TRAPSTS_RESTORE_PART_3_SIZE = 32 - S_TRAPSTS_RESTORE_PART_3_SHIFT
> var S_TRAPSTS_HWREG = HW_REG_TRAPSTS
> var S_TRAPSTS_SAVE_CONTEXT_MASK = SQ_WAVE_TRAPSTS_SAVECTX_MASK
> var S_TRAPSTS_SAVE_CONTEXT_SHIFT = SQ_WAVE_TRAPSTS_SAVECTX_SHIFT
> -#else
> -var SQ_WAVE_EXCP_FLAG_PRIV_ADDR_WATCH_MASK = 0xF
> -var SQ_WAVE_EXCP_FLAG_PRIV_MEM_VIOL_MASK = 0x10
> -var SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT = 5
> -var SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK = 0x20
> -var SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_MASK = 0x40
> -var SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT = 6
> -var SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK = 0x80
> -var SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT = 7
> -var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_MASK = 0x100
> -var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT = 8
> -var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_END_MASK = 0x200
> -var SQ_WAVE_EXCP_FLAG_PRIV_TRAP_AFTER_INST_MASK = 0x800
> -var SQ_WAVE_TRAP_CTRL_ADDR_WATCH_MASK = 0x80
> -var SQ_WAVE_TRAP_CTRL_TRAP_AFTER_INST_MASK = 0x200
> -
> -var S_TRAPSTS_HWREG = HW_REG_WAVE_EXCP_FLAG_PRIV
> -var S_TRAPSTS_SAVE_CONTEXT_MASK = SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK
> -var S_TRAPSTS_SAVE_CONTEXT_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT
> -var S_TRAPSTS_NON_MASKABLE_EXCP_MASK = SQ_WAVE_EXCP_FLAG_PRIV_MEM_VIOL_MASK |\
> - SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_MASK |\
> - SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK |\
> - SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_MASK |\
> - SQ_WAVE_EXCP_FLAG_PRIV_WAVE_END_MASK |\
> - SQ_WAVE_EXCP_FLAG_PRIV_TRAP_AFTER_INST_MASK
> -var S_TRAPSTS_RESTORE_PART_1_SIZE = SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT
> -var S_TRAPSTS_RESTORE_PART_2_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT
> -var S_TRAPSTS_RESTORE_PART_2_SIZE = SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT - SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT
> -var S_TRAPSTS_RESTORE_PART_3_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT
> -var S_TRAPSTS_RESTORE_PART_3_SIZE = 32 - S_TRAPSTS_RESTORE_PART_3_SHIFT
> -var BARRIER_STATE_SIGNAL_OFFSET = 16
> -var BARRIER_STATE_VALID_OFFSET = 0
> -#endif
>
> // bits [31:24] unused by SPI debug data
> var TTMP11_SAVE_REPLAY_W64H_SHIFT = 31
> @@ -305,11 +239,7 @@ L_TRAP_NO_BARRIER:
>
> L_HALTED:
> // Host trap may occur while wave is halted.
> -#if ASIC_FAMILY < CHIP_GFX12
> s_and_b32 ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
> -#else
> - s_and_b32 ttmp2, s_save_trapsts, SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK
> -#endif
> s_cbranch_scc1 L_FETCH_2ND_TRAP
>
> L_CHECK_SAVE:
> @@ -336,7 +266,6 @@ L_NOT_HALTED:
> // Check for maskable exceptions in trapsts.excp and trapsts.excp_hi.
> // Maskable exceptions only cause the wave to enter the trap handler if
> // their respective bit in mode.excp_en is set.
> -#if ASIC_FAMILY < CHIP_GFX12
> s_and_b32 ttmp2, s_save_trapsts, SQ_WAVE_TRAPSTS_EXCP_MASK|SQ_WAVE_TRAPSTS_EXCP_HI_MASK
> s_cbranch_scc0 L_CHECK_TRAP_ID
>
> @@ -349,17 +278,6 @@ L_NOT_ADDR_WATCH:
> s_lshl_b32 ttmp2, ttmp2, SQ_WAVE_MODE_EXCP_EN_SHIFT
> s_and_b32 ttmp2, ttmp2, ttmp3
> s_cbranch_scc1 L_FETCH_2ND_TRAP
> -#else
> - s_getreg_b32 ttmp2, hwreg(HW_REG_WAVE_EXCP_FLAG_USER)
> - s_and_b32 ttmp3, s_save_trapsts, SQ_WAVE_EXCP_FLAG_PRIV_ADDR_WATCH_MASK
> - s_cbranch_scc0 L_NOT_ADDR_WATCH
> - s_or_b32 ttmp2, ttmp2, SQ_WAVE_TRAP_CTRL_ADDR_WATCH_MASK
> -
> -L_NOT_ADDR_WATCH:
> - s_getreg_b32 ttmp3, hwreg(HW_REG_WAVE_TRAP_CTRL)
> - s_and_b32 ttmp2, ttmp3, ttmp2
> - s_cbranch_scc1 L_FETCH_2ND_TRAP
> -#endif
>
> L_CHECK_TRAP_ID:
> // Check trap_id != 0
> @@ -369,13 +287,8 @@ L_CHECK_TRAP_ID:
> #if SINGLE_STEP_MISSED_WORKAROUND
> // Prioritize single step exception over context save.
> // Second-level trap will halt wave and RFE, re-entering for SAVECTX.
> -#if ASIC_FAMILY < CHIP_GFX12
> s_getreg_b32 ttmp2, hwreg(HW_REG_MODE)
> s_and_b32 ttmp2, ttmp2, SQ_WAVE_MODE_DEBUG_EN_MASK
> -#else
> - // WAVE_TRAP_CTRL is already in ttmp3.
> - s_and_b32 ttmp3, ttmp3, SQ_WAVE_TRAP_CTRL_TRAP_AFTER_INST_MASK
> -#endif
> s_cbranch_scc1 L_FETCH_2ND_TRAP
> #endif
>
> @@ -425,12 +338,7 @@ L_NO_NEXT_TRAP:
> s_cbranch_scc1 L_TRAP_CASE
>
> // Host trap will not cause trap re-entry.
> -#if ASIC_FAMILY < CHIP_GFX12
> s_and_b32 ttmp2, s_save_pc_hi, S_SAVE_PC_HI_HT_MASK
> -#else
> - s_getreg_b32 ttmp2, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
> - s_and_b32 ttmp2, ttmp2, SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK
> -#endif
> s_cbranch_scc1 L_EXIT_TRAP
> s_or_b32 s_save_status, s_save_status, S_STATUS_HALT_MASK
>
> @@ -457,16 +365,7 @@ L_EXIT_TRAP:
> 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
>
> -#if ASIC_FAMILY < CHIP_GFX12
> s_setreg_b32 hwreg(S_STATUS_HWREG), s_save_status
> -#else
> - // STATE_PRIV.BARRIER_COMPLETE may have changed since we read it.
> - // Only restore fields which the trap handler changes.
> - s_lshr_b32 s_save_status, s_save_status, SQ_WAVE_STATE_PRIV_SCC_SHIFT
> - s_setreg_b32 hwreg(S_STATUS_HWREG, SQ_WAVE_STATE_PRIV_SCC_SHIFT, \
> - SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT - SQ_WAVE_STATE_PRIV_SCC_SHIFT + 1), s_save_status
> -#endif
> -
> s_rfe_b64 [ttmp0, ttmp1]
>
> L_SAVE:
> @@ -478,14 +377,6 @@ L_SAVE:
> s_endpgm
> L_HAVE_VGPRS:
> #endif
> -#if ASIC_FAMILY >= CHIP_GFX12
> - s_getreg_b32 s_save_tmp, hwreg(HW_REG_WAVE_STATUS)
> - s_bitcmp1_b32 s_save_tmp, SQ_WAVE_STATUS_NO_VGPRS_SHIFT
> - s_cbranch_scc0 L_HAVE_VGPRS
> - s_endpgm
> -L_HAVE_VGPRS:
> -#endif
> -
> s_and_b32 s_save_pc_hi, s_save_pc_hi, 0x0000ffff //pc[47:32]
> s_mov_b32 s_save_tmp, 0
> s_setreg_b32 hwreg(S_TRAPSTS_HWREG, S_TRAPSTS_SAVE_CONTEXT_SHIFT, 1), s_save_tmp //clear saveCtx bit
> @@ -671,19 +562,6 @@ L_SAVE_HWREG:
> s_mov_b32 m0, 0x0 //Next lane of v2 to write to
> #endif
>
> -#if ASIC_FAMILY >= CHIP_GFX12
> - // Ensure no further changes to barrier or LDS state.
> - // STATE_PRIV.BARRIER_COMPLETE may change up to this point.
> - s_barrier_signal -2
> - s_barrier_wait -2
> -
> - // Re-read final state of BARRIER_COMPLETE field for save.
> - s_getreg_b32 s_save_tmp, hwreg(S_STATUS_HWREG)
> - s_and_b32 s_save_tmp, s_save_tmp, SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK
> - s_andn2_b32 s_save_status, s_save_status, SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK
> - s_or_b32 s_save_status, s_save_status, s_save_tmp
> -#endif
> -
> write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
> write_hwreg_to_mem(s_save_pc_lo, s_save_buf_rsrc0, s_save_mem_offset)
> s_andn2_b32 s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
> @@ -707,21 +585,6 @@ L_SAVE_HWREG:
> s_getreg_b32 s_save_m0, hwreg(HW_REG_SHADER_FLAT_SCRATCH_HI)
> write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
>
> -#if ASIC_FAMILY >= CHIP_GFX12
> - s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_EXCP_FLAG_USER)
> - write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
> -
> - s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_TRAP_CTRL)
> - write_hwreg_to_mem(s_save_m0, s_save_buf_rsrc0, s_save_mem_offset)
> -
> - s_getreg_b32 s_save_tmp, hwreg(HW_REG_WAVE_STATUS)
> - write_hwreg_to_mem(s_save_tmp, s_save_buf_rsrc0, s_save_mem_offset)
> -
> - s_get_barrier_state s_save_tmp, -1
> - s_wait_kmcnt (0)
> - write_hwreg_to_mem(s_save_tmp, s_save_buf_rsrc0, s_save_mem_offset)
> -#endif
> -
> #if NO_SQC_STORE
> // Write HWREGs with 16 VGPR lanes. TTMPs occupy space after this.
> s_mov_b32 exec_lo, 0xFFFF
> @@ -814,9 +677,7 @@ L_SAVE_LDS_NORMAL:
> s_and_b32 s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF //lds_size is zero?
> s_cbranch_scc0 L_SAVE_LDS_DONE //no lds used? jump to L_SAVE_DONE
>
> -#if ASIC_FAMILY < CHIP_GFX12
> s_barrier //LDS is used? wait for other waves in the same TG
> -#endif
> s_and_b32 s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
> s_cbranch_scc0 L_SAVE_LDS_DONE
>
> @@ -1081,11 +942,6 @@ L_RESTORE:
> s_mov_b32 s_restore_buf_rsrc2, 0 //NUM_RECORDS initial value = 0 (in bytes)
> s_mov_b32 s_restore_buf_rsrc3, S_RESTORE_BUF_RSRC_WORD3_MISC
>
> -#if ASIC_FAMILY >= CHIP_GFX12
> - // Save s_restore_spi_init_hi for later use.
> - s_mov_b32 s_restore_spi_init_hi_save, s_restore_spi_init_hi
> -#endif
> -
> //determine it is wave32 or wave64
> get_wave_size2(s_restore_size)
>
> @@ -1320,9 +1176,7 @@ L_RESTORE_SGPR:
> // s_barrier with MODE.DEBUG_EN=1, STATUS.PRIV=1 incorrectly asserts debug exception.
> // Clear DEBUG_EN before and restore MODE after the barrier.
> s_setreg_imm32_b32 hwreg(HW_REG_MODE), 0
> -#if ASIC_FAMILY < CHIP_GFX12
> s_barrier //barrier to ensure the readiness of LDS before access attemps from any other wave in the same TG
> -#endif
>
> /* restore HW registers */
> L_RESTORE_HWREG:
> @@ -1334,11 +1188,6 @@ L_RESTORE_HWREG:
>
> s_mov_b32 s_restore_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
>
> -#if ASIC_FAMILY >= CHIP_GFX12
> - // Restore s_restore_spi_init_hi before the saved value gets clobbered.
> - s_mov_b32 s_restore_spi_init_hi, s_restore_spi_init_hi_save
> -#endif
> -
> read_hwreg_from_mem(s_restore_m0, s_restore_buf_rsrc0, s_restore_mem_offset)
> read_hwreg_from_mem(s_restore_pc_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
> read_hwreg_from_mem(s_restore_pc_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
> @@ -1358,44 +1207,6 @@ L_RESTORE_HWREG:
>
> s_setreg_b32 hwreg(HW_REG_SHADER_FLAT_SCRATCH_HI), s_restore_flat_scratch
>
> -#if ASIC_FAMILY >= CHIP_GFX12
> - read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
> - S_WAITCNT_0
> - s_setreg_b32 hwreg(HW_REG_WAVE_EXCP_FLAG_USER), s_restore_tmp
> -
> - read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
> - S_WAITCNT_0
> - s_setreg_b32 hwreg(HW_REG_WAVE_TRAP_CTRL), s_restore_tmp
> -
> - // Only the first wave needs to restore the workgroup barrier.
> - s_and_b32 s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
> - s_cbranch_scc0 L_SKIP_BARRIER_RESTORE
> -
> - // Skip over WAVE_STATUS, since there is no state to restore from it
> - s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 4
> -
> - read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
> - S_WAITCNT_0
> -
> - s_bitcmp1_b32 s_restore_tmp, BARRIER_STATE_VALID_OFFSET
> - s_cbranch_scc0 L_SKIP_BARRIER_RESTORE
> -
> - // extract the saved signal count from s_restore_tmp
> - s_lshr_b32 s_restore_tmp, s_restore_tmp, BARRIER_STATE_SIGNAL_OFFSET
> -
> - // We need to call s_barrier_signal repeatedly to restore the signal
> - // count of the work group barrier. The member count is already
> - // initialized with the number of waves in the work group.
> -L_BARRIER_RESTORE_LOOP:
> - s_and_b32 s_restore_tmp, s_restore_tmp, s_restore_tmp
> - s_cbranch_scc0 L_SKIP_BARRIER_RESTORE
> - s_barrier_signal -1
> - s_add_i32 s_restore_tmp, s_restore_tmp, -1
> - s_branch L_BARRIER_RESTORE_LOOP
> -
> -L_SKIP_BARRIER_RESTORE:
> -#endif
> -
> s_mov_b32 m0, s_restore_m0
> s_mov_b32 exec_lo, s_restore_exec_lo
> s_mov_b32 exec_hi, s_restore_exec_hi
> @@ -1453,13 +1264,6 @@ L_RETURN_WITHOUT_PRIV:
>
> s_setreg_b32 hwreg(S_STATUS_HWREG), s_restore_status // SCC is included, which is changed by previous salu
>
> -#if ASIC_FAMILY >= CHIP_GFX12
> - // Make barrier and LDS state visible to all waves in the group.
> - // STATE_PRIV.BARRIER_COMPLETE may change after this point.
> - s_barrier_signal -2
> - s_barrier_wait -2
> -#endif
> -
> s_rfe_b64 s_restore_pc_lo //Return to the main shader program and resume execution
>
> L_END_PGM:
> @@ -1598,11 +1402,7 @@ function get_hwreg_size_bytes
> end
>
> function get_wave_size2(s_reg)
> -#if ASIC_FAMILY < CHIP_GFX12
> s_getreg_b32 s_reg, hwreg(HW_REG_IB_STS2,SQ_WAVE_IB_STS2_WAVE64_SHIFT,SQ_WAVE_IB_STS2_WAVE64_SIZE)
> -#else
> - s_getreg_b32 s_reg, hwreg(HW_REG_WAVE_STATUS,SQ_WAVE_STATUS_WAVE64_SHIFT,SQ_WAVE_STATUS_WAVE64_SIZE)
> -#endif
> s_lshl_b32 s_reg, s_reg, S_WAVE_SIZE
> end
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm
> new file mode 100644
> index 000000000000..6630519892fc
> --- /dev/null
> +++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx12.asm
> @@ -0,0 +1,1128 @@
> +/*
> + * Copyright 2024 Advanced Micro Devices, Inc.
I am not really sure bout the year policy in the kernel, but all the
content here is dated from before 2024. The vast majority is taken form
the cwsr_trap_handler_gfx10.asm file (copyright started in 2018), and
even the gfx12 specific part have been introduced starting 2023.
> + *
> + * Permission is hereby granted, free of charge, to any person obtaining a
> + * copy of this software and associated documentation files (the "Software"),
> + * to deal in the Software without restriction, including without limitation
> + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
> + * and/or sell copies of the Software, and to permit persons to whom the
> + * Software is furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice shall be included in
> + * all copies or substantial portions of the Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
> + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
> + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
> + * OTHER DEALINGS IN THE SOFTWARE.
> + */
> +
> +/* To compile this assembly code:
> + *
> + * gfx12:
> + * cpp -DASIC_FAMILY=CHIP_GFX12 cwsr_trap_handler_gfx12.asm -P -o gfx12.sp3
> + * sp3 gfx12.sp3 -hex gfx12.hex
> + */
> +
> +#define CHIP_GFX12 37
> +
> +#define SINGLE_STEP_MISSED_WORKAROUND 1 //workaround for lost MODE.DEBUG_EN exception when SAVECTX raised
MODE.DEBUG_EN is really TRAP_CTRL.TRAP_AFTER_INST in gfx12.
> +
> +var SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK = 0x4
> +var SQ_WAVE_STATE_PRIV_SCC_SHIFT = 9
> +var SQ_WAVE_STATE_PRIV_SYS_PRIO_MASK = 0xC00
> +var SQ_WAVE_STATE_PRIV_HALT_MASK = 0x4000
> +var SQ_WAVE_STATE_PRIV_POISON_ERR_MASK = 0x8000
> +var SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT = 15
> +var SQ_WAVE_STATUS_WAVE64_SHIFT = 29
> +var SQ_WAVE_STATUS_WAVE64_SIZE = 1
> +var SQ_WAVE_STATUS_NO_VGPRS_SHIFT = 24
> +var SQ_WAVE_STATE_PRIV_ALWAYS_CLEAR_MASK = SQ_WAVE_STATE_PRIV_SYS_PRIO_MASK|SQ_WAVE_STATE_PRIV_POISON_ERR_MASK
> +var S_SAVE_PC_HI_TRAP_ID_MASK = 0xF0000000
> +
> +var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT = 12
> +var SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE = 9
> +var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE = 8
> +var SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT = 12
> +var SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT = 24
> +var SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE = 4
> +var SQ_WAVE_LDS_ALLOC_GRANULARITY = 9
> +
> +var SQ_WAVE_EXCP_FLAG_PRIV_ADDR_WATCH_MASK = 0xF
> +var SQ_WAVE_EXCP_FLAG_PRIV_MEM_VIOL_MASK = 0x10
> +var SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT = 5
> +var SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK = 0x20
> +var SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_MASK = 0x40
> +var SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT = 6
> +var SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK = 0x80
> +var SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT = 7
> +var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_MASK = 0x100
> +var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT = 8
> +var SQ_WAVE_EXCP_FLAG_PRIV_WAVE_END_MASK = 0x200
> +var SQ_WAVE_EXCP_FLAG_PRIV_TRAP_AFTER_INST_MASK = 0x800
> +var SQ_WAVE_TRAP_CTRL_ADDR_WATCH_MASK = 0x80
> +var SQ_WAVE_TRAP_CTRL_TRAP_AFTER_INST_MASK = 0x200
> +
> +var SQ_WAVE_EXCP_FLAG_PRIV_NON_MASKABLE_EXCP_MASK= SQ_WAVE_EXCP_FLAG_PRIV_MEM_VIOL_MASK |\
> + SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_MASK |\
> + SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK |\
> + SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_MASK |\
> + SQ_WAVE_EXCP_FLAG_PRIV_WAVE_END_MASK |\
> + SQ_WAVE_EXCP_FLAG_PRIV_TRAP_AFTER_INST_MASK
> +var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_1_SIZE = SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT
> +var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT
> +var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SIZE = SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_SHIFT - SQ_WAVE_EXCP_FLAG_PRIV_ILLEGAL_INST_SHIFT
> +var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT = SQ_WAVE_EXCP_FLAG_PRIV_WAVE_START_SHIFT
> +var SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SIZE = 32 - SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT
> +var BARRIER_STATE_SIGNAL_OFFSET = 16
> +var BARRIER_STATE_VALID_OFFSET = 0
> +
> +var TTMP11_DEBUG_TRAP_ENABLED_SHIFT = 23
> +var TTMP11_DEBUG_TRAP_ENABLED_MASK = 0x800000
> +
> +// SQ_SEL_X/Y/Z/W, BUF_NUM_FORMAT_FLOAT, (0 for MUBUF stride[17:14]
> +// when ADD_TID_ENABLE and BUF_DATA_FORMAT_32 for MTBUF), ADD_TID_ENABLE
> +var S_SAVE_BUF_RSRC_WORD1_STRIDE = 0x00040000
> +var S_SAVE_BUF_RSRC_WORD3_MISC = 0x10807FAC
> +var S_SAVE_SPI_INIT_FIRST_WAVE_MASK = 0x04000000
> +var S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT = 26
> +
> +var S_SAVE_PC_HI_FIRST_WAVE_MASK = 0x80000000
> +var S_SAVE_PC_HI_FIRST_WAVE_SHIFT = 31
> +
> +var s_sgpr_save_num = 108
> +
> +var s_save_spi_init_lo = exec_lo
> +var s_save_spi_init_hi = exec_hi
> +var s_save_pc_lo = ttmp0
> +var s_save_pc_hi = ttmp1
> +var s_save_exec_lo = ttmp2
> +var s_save_exec_hi = ttmp3
> +var s_save_state_priv = ttmp12
> +var s_save_excp_flag_priv = ttmp15
> +var s_save_xnack_mask = s_save_excp_flag_priv
> +var s_wave_size = ttmp7
> +var s_save_buf_rsrc0 = ttmp8
> +var s_save_buf_rsrc1 = ttmp9
> +var s_save_buf_rsrc2 = ttmp10
> +var s_save_buf_rsrc3 = ttmp11
> +var s_save_mem_offset = ttmp4
> +var s_save_alloc_size = s_save_excp_flag_priv
> +var s_save_tmp = ttmp14
> +var s_save_m0 = ttmp5
> +var s_save_ttmps_lo = s_save_tmp
> +var s_save_ttmps_hi = s_save_excp_flag_priv
> +
> +var S_RESTORE_BUF_RSRC_WORD1_STRIDE = S_SAVE_BUF_RSRC_WORD1_STRIDE
> +var S_RESTORE_BUF_RSRC_WORD3_MISC = S_SAVE_BUF_RSRC_WORD3_MISC
> +
> +var S_RESTORE_SPI_INIT_FIRST_WAVE_MASK = 0x04000000
> +var S_RESTORE_SPI_INIT_FIRST_WAVE_SHIFT = 26
> +var S_WAVE_SIZE = 25
> +
> +var s_restore_spi_init_lo = exec_lo
> +var s_restore_spi_init_hi = exec_hi
> +var s_restore_mem_offset = ttmp12
> +var s_restore_alloc_size = ttmp3
> +var s_restore_tmp = ttmp2
> +var s_restore_mem_offset_save = s_restore_tmp
> +var s_restore_m0 = s_restore_alloc_size
> +var s_restore_mode = ttmp7
> +var s_restore_flat_scratch = s_restore_tmp
> +var s_restore_pc_lo = ttmp0
> +var s_restore_pc_hi = ttmp1
> +var s_restore_exec_lo = ttmp4
> +var s_restore_exec_hi = ttmp5
> +var s_restore_state_priv = ttmp14
> +var s_restore_excp_flag_priv = ttmp15
> +var s_restore_xnack_mask = ttmp13
> +var s_restore_buf_rsrc0 = ttmp8
> +var s_restore_buf_rsrc1 = ttmp9
> +var s_restore_buf_rsrc2 = ttmp10
> +var s_restore_buf_rsrc3 = ttmp11
> +var s_restore_size = ttmp6
> +var s_restore_ttmps_lo = s_restore_tmp
> +var s_restore_ttmps_hi = s_restore_alloc_size
> +var s_restore_spi_init_hi_save = s_restore_exec_hi
> +
> +shader main
> + asic(DEFAULT)
> + type(CS)
> + wave_size(32)
> +
> + s_branch L_SKIP_RESTORE //NOT restore. might be a regular trap or save
> +
> +L_JUMP_TO_RESTORE:
> + s_branch L_RESTORE
> +
> +L_SKIP_RESTORE:
> + s_getreg_b32 s_save_state_priv, hwreg(HW_REG_WAVE_STATE_PRIV) //save STATUS since we will change SCC
> +
> + // Clear SPI_PRIO: do not save with elevated priority.
> + // Clear ECC_ERR: prevents SQC store and triggers FATAL_HALT if setreg'd.
> + s_andn2_b32 s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_ALWAYS_CLEAR_MASK
> +
> + s_getreg_b32 s_save_excp_flag_priv, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
> +
> + s_and_b32 ttmp2, s_save_state_priv, SQ_WAVE_STATE_PRIV_HALT_MASK
> + s_cbranch_scc0 L_NOT_HALTED
> +
> +L_HALTED:
> + // Host trap may occur while wave is halted.
> + s_and_b32 ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK
> + s_cbranch_scc1 L_FETCH_2ND_TRAP
> +
> +L_CHECK_SAVE:
> + s_and_b32 ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK
> + s_cbranch_scc1 L_SAVE
> +
> + // Wave is halted but neither host trap nor SAVECTX is raised.
> + // Caused by instruction fetch memory violation.
> + // Spin wait until context saved to prevent interrupt storm.
> + s_sleep 0x10
> + s_getreg_b32 s_save_excp_flag_priv, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
> + s_branch L_CHECK_SAVE
> +
> +L_NOT_HALTED:
> + // Let second-level handle non-SAVECTX exception or trap.
> + // Any concurrent SAVECTX will be handled upon re-entry once halted.
> +
> + // Check non-maskable exceptions. memory_violation, illegal_instruction
> + // and xnack_error exceptions always cause the wave to enter the trap
> + // handler.
> + s_and_b32 ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_NON_MASKABLE_EXCP_MASK
> + s_cbranch_scc1 L_FETCH_2ND_TRAP
> +
> + // Check for maskable exceptions in trapsts.excp and trapsts.excp_hi.
> + // Maskable exceptions only cause the wave to enter the trap handler if
> + // their respective bit in mode.excp_en is set.
> + s_getreg_b32 ttmp2, hwreg(HW_REG_WAVE_EXCP_FLAG_USER)
> + s_and_b32 ttmp3, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_ADDR_WATCH_MASK
> + s_cbranch_scc0 L_NOT_ADDR_WATCH
> + s_or_b32 ttmp2, ttmp2, SQ_WAVE_TRAP_CTRL_ADDR_WATCH_MASK
> +
> +L_NOT_ADDR_WATCH:
> + s_getreg_b32 ttmp3, hwreg(HW_REG_WAVE_TRAP_CTRL)
> + s_and_b32 ttmp2, ttmp3, ttmp2
> + s_cbranch_scc1 L_FETCH_2ND_TRAP
> +
> +L_CHECK_TRAP_ID:
> + // Check trap_id != 0
> + s_and_b32 ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
> + s_cbranch_scc1 L_FETCH_2ND_TRAP
> +
> +#if SINGLE_STEP_MISSED_WORKAROUND
> + // Prioritize single step exception over context save.
> + // Second-level trap will halt wave and RFE, re-entering for SAVECTX.
> + // WAVE_TRAP_CTRL is already in ttmp3.
> + s_and_b32 ttmp3, ttmp3, SQ_WAVE_TRAP_CTRL_TRAP_AFTER_INST_MASK
> + s_cbranch_scc1 L_FETCH_2ND_TRAP
> +#endif
> +
> + s_and_b32 ttmp2, s_save_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_MASK
> + s_cbranch_scc1 L_SAVE
> +
> +L_FETCH_2ND_TRAP:
> + // Read second-level TBA/TMA from first-level TMA and jump if available.
> + // ttmp[2:5] and ttmp12 can be used (others hold SPI-initialized debug data)
> + // ttmp12 holds SQ_WAVE_STATUS
> + s_sendmsg_rtn_b64 [ttmp14, ttmp15], sendmsg(MSG_RTN_GET_TMA)
> + s_wait_idle
> + s_lshl_b64 [ttmp14, ttmp15], [ttmp14, ttmp15], 0x8
> +
> + s_bitcmp1_b32 ttmp15, 0xF
> + s_cbranch_scc0 L_NO_SIGN_EXTEND_TMA
> + s_or_b32 ttmp15, ttmp15, 0xFFFF0000
> +L_NO_SIGN_EXTEND_TMA:
> +
> + s_load_dword ttmp2, [ttmp14, ttmp15], 0x10 scope:SCOPE_SYS // debug trap enabled flag
> + s_wait_idle
> + s_lshl_b32 ttmp2, ttmp2, TTMP11_DEBUG_TRAP_ENABLED_SHIFT
> + s_andn2_b32 ttmp11, ttmp11, TTMP11_DEBUG_TRAP_ENABLED_MASK
> + s_or_b32 ttmp11, ttmp11, ttmp2
> +
> + s_load_dwordx2 [ttmp2, ttmp3], [ttmp14, ttmp15], 0x0 scope:SCOPE_SYS // second-level TBA
> + s_wait_idle
> + s_load_dwordx2 [ttmp14, ttmp15], [ttmp14, ttmp15], 0x8 scope:SCOPE_SYS // second-level TMA
> + s_wait_idle
> +
> + s_and_b64 [ttmp2, ttmp3], [ttmp2, ttmp3], [ttmp2, ttmp3]
> + s_cbranch_scc0 L_NO_NEXT_TRAP // second-level trap handler not been set
> + s_setpc_b64 [ttmp2, ttmp3] // jump to second-level trap handler
> +
> +L_NO_NEXT_TRAP:
> + // If not caused by trap then halt wave to prevent re-entry.
> + s_and_b32 ttmp2, s_save_pc_hi, S_SAVE_PC_HI_TRAP_ID_MASK
> + s_cbranch_scc1 L_TRAP_CASE
> +
> + // Host trap will not cause trap re-entry.
> + s_getreg_b32 ttmp2, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
> + s_and_b32 ttmp2, ttmp2, SQ_WAVE_EXCP_FLAG_PRIV_HOST_TRAP_MASK
> + s_cbranch_scc1 L_EXIT_TRAP
> + s_or_b32 s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_HALT_MASK
> +
> + // If the PC points to S_ENDPGM then context save will fail if STATUS.HALT is set.
STATE_PRIV.HALT. Also, I am not sure this part is still required for
gfx12. That being said, this has nothing to do with the rework, so
removing is would be outside of the scope of this patch.
> + // Rewind the PC to prevent this from occurring.
> + s_sub_u32 ttmp0, ttmp0, 0x8
> + s_subb_u32 ttmp1, ttmp1, 0x0
> +
> + s_branch L_EXIT_TRAP
> +
> +L_TRAP_CASE:
> + // Advance past trap instruction to prevent re-entry.
> + s_add_u32 ttmp0, ttmp0, 0x4
> + s_addc_u32 ttmp1, ttmp1, 0x0
> +
> +L_EXIT_TRAP:
> + s_and_b32 ttmp1, ttmp1, 0xFFFF
> +
> + // Restore SQ_WAVE_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
> +
> + // STATE_PRIV.BARRIER_COMPLETE may have changed since we read it.
> + // Only restore fields which the trap handler changes.
> + s_lshr_b32 s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_SCC_SHIFT
> + s_setreg_b32 hwreg(HW_REG_WAVE_STATE_PRIV, SQ_WAVE_STATE_PRIV_SCC_SHIFT, \
> + SQ_WAVE_STATE_PRIV_POISON_ERR_SHIFT - SQ_WAVE_STATE_PRIV_SCC_SHIFT + 1), s_save_state_priv
> +
> + s_rfe_b64 [ttmp0, ttmp1]
> +
> +L_SAVE:
> + // If VGPRs have been deallocated then terminate the wavefront.
> + // It has no remaining program to run and cannot save without VGPRs.
> + s_getreg_b32 s_save_tmp, hwreg(HW_REG_WAVE_STATUS)
> + s_bitcmp1_b32 s_save_tmp, SQ_WAVE_STATUS_NO_VGPRS_SHIFT
> + s_cbranch_scc0 L_HAVE_VGPRS
> + s_endpgm
> +L_HAVE_VGPRS:
> +
> + s_and_b32 s_save_pc_hi, s_save_pc_hi, 0x0000ffff //pc[47:32]
> + s_mov_b32 s_save_tmp, 0
> + s_setreg_b32 hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, SQ_WAVE_EXCP_FLAG_PRIV_SAVE_CONTEXT_SHIFT, 1), s_save_tmp //clear saveCtx bit
> +
> + /* inform SPI the readiness and wait for SPI's go signal */
> + s_mov_b32 s_save_exec_lo, exec_lo //save EXEC and use EXEC for the go signal from SPI
> + s_mov_b32 s_save_exec_hi, exec_hi
> + s_mov_b64 exec, 0x0 //clear EXEC to get ready to receive
> +
> + s_sendmsg_rtn_b64 [exec_lo, exec_hi], sendmsg(MSG_RTN_SAVE_WAVE)
> + s_wait_idle
> +
> + // Save first_wave flag so we can clear high bits of save address.
> + s_and_b32 s_save_tmp, s_save_spi_init_hi, S_SAVE_SPI_INIT_FIRST_WAVE_MASK
> + s_lshl_b32 s_save_tmp, s_save_tmp, (S_SAVE_PC_HI_FIRST_WAVE_SHIFT - S_SAVE_SPI_INIT_FIRST_WAVE_SHIFT)
> + s_or_b32 s_save_pc_hi, s_save_pc_hi, s_save_tmp
> +
> + // Trap temporaries must be saved via VGPR but all VGPRs are in use.
> + // There is no ttmp space to hold the resource constant for VGPR save.
> + // Save v0 by itself since it requires only two SGPRs.
> + s_mov_b32 s_save_ttmps_lo, exec_lo
> + s_and_b32 s_save_ttmps_hi, exec_hi, 0xFFFF
> + s_mov_b32 exec_lo, 0xFFFFFFFF
> + s_mov_b32 exec_hi, 0xFFFFFFFF
> + global_store_dword_addtid v0, [s_save_ttmps_lo, s_save_ttmps_hi] scope:SCOPE_SYS
> + v_mov_b32 v0, 0x0
> + s_mov_b32 exec_lo, s_save_ttmps_lo
> + s_mov_b32 exec_hi, s_save_ttmps_hi
> +
> + // Save trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
> + // ttmp SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)+0x40
> + get_wave_size2(s_save_ttmps_hi)
> + get_vgpr_size_bytes(s_save_ttmps_lo, s_save_ttmps_hi)
> + get_svgpr_size_bytes(s_save_ttmps_hi)
> + s_add_u32 s_save_ttmps_lo, s_save_ttmps_lo, s_save_ttmps_hi
> + s_and_b32 s_save_ttmps_hi, s_save_spi_init_hi, 0xFFFF
> + s_add_u32 s_save_ttmps_lo, s_save_ttmps_lo, get_sgpr_size_bytes()
> + s_add_u32 s_save_ttmps_lo, s_save_ttmps_lo, s_save_spi_init_lo
> + s_addc_u32 s_save_ttmps_hi, s_save_ttmps_hi, 0x0
> +
> + v_writelane_b32 v0, ttmp4, 0x4
> + v_writelane_b32 v0, ttmp5, 0x5
> + v_writelane_b32 v0, ttmp6, 0x6
> + v_writelane_b32 v0, ttmp7, 0x7
> + v_writelane_b32 v0, ttmp8, 0x8
> + v_writelane_b32 v0, ttmp9, 0x9
> + v_writelane_b32 v0, ttmp10, 0xA
> + v_writelane_b32 v0, ttmp11, 0xB
> + v_writelane_b32 v0, ttmp13, 0xD
> + v_writelane_b32 v0, exec_lo, 0xE
> + v_writelane_b32 v0, exec_hi, 0xF
> +
> + s_mov_b32 exec_lo, 0x3FFF
> + s_mov_b32 exec_hi, 0x0
> + global_store_dword_addtid v0, [s_save_ttmps_lo, s_save_ttmps_hi] offset:0x40 scope:SCOPE_SYS
> + v_readlane_b32 ttmp14, v0, 0xE
> + v_readlane_b32 ttmp15, v0, 0xF
> + s_mov_b32 exec_lo, ttmp14
> + s_mov_b32 exec_hi, ttmp15
> +
> + /* setup Resource Contants */
> + s_mov_b32 s_save_buf_rsrc0, s_save_spi_init_lo //base_addr_lo
> + s_and_b32 s_save_buf_rsrc1, s_save_spi_init_hi, 0x0000FFFF //base_addr_hi
> + s_or_b32 s_save_buf_rsrc1, s_save_buf_rsrc1, S_SAVE_BUF_RSRC_WORD1_STRIDE
> + s_mov_b32 s_save_buf_rsrc2, 0 //NUM_RECORDS initial value = 0 (in bytes) although not neccessarily inited
> + s_mov_b32 s_save_buf_rsrc3, S_SAVE_BUF_RSRC_WORD3_MISC
> +
> + s_mov_b32 s_save_m0, m0
> +
> + /* global mem offset */
> + s_mov_b32 s_save_mem_offset, 0x0
> + get_wave_size2(s_wave_size)
> +
> + /* save first 4 VGPRs, needed for SGPR save */
> + s_mov_b32 exec_lo, 0xFFFFFFFF //need every thread from now on
> + s_lshr_b32 m0, s_wave_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_ENABLE_SAVE_4VGPR_EXEC_HI
> + s_mov_b32 exec_hi, 0x00000000
> + s_branch L_SAVE_4VGPR_WAVE32
> +L_ENABLE_SAVE_4VGPR_EXEC_HI:
> + s_mov_b32 exec_hi, 0xFFFFFFFF
> + s_branch L_SAVE_4VGPR_WAVE64
> +L_SAVE_4VGPR_WAVE32:
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // VGPR Allocated in 4-GPR granularity
> +
> + buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*2
> + buffer_store_dword v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*3
> + s_branch L_SAVE_HWREG
> +
> +L_SAVE_4VGPR_WAVE64:
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // VGPR Allocated in 4-GPR granularity
> +
> + buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*2
> + buffer_store_dword v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*3
> +
> + /* save HW registers */
> +
> +L_SAVE_HWREG:
> + // HWREG SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)
> + get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
> + get_svgpr_size_bytes(s_save_tmp)
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, s_save_tmp
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, get_sgpr_size_bytes()
> +
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + v_mov_b32 v0, 0x0 //Offset[31:0] from buffer resource
> + v_mov_b32 v1, 0x0 //Offset[63:32] from buffer resource
> + v_mov_b32 v2, 0x0 //Set of SGPRs for TCP store
> + s_mov_b32 m0, 0x0 //Next lane of v2 to write to
> +
> + // Ensure no further changes to barrier or LDS state.
> + // STATE_PRIV.BARRIER_COMPLETE may change up to this point.
> + s_barrier_signal -2
> + s_barrier_wait -2
> +
> + // Re-read final state of BARRIER_COMPLETE field for save.
> + s_getreg_b32 s_save_tmp, hwreg(HW_REG_WAVE_STATE_PRIV)
> + s_and_b32 s_save_tmp, s_save_tmp, SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK
> + s_andn2_b32 s_save_state_priv, s_save_state_priv, SQ_WAVE_STATE_PRIV_BARRIER_COMPLETE_MASK
> + s_or_b32 s_save_state_priv, s_save_state_priv, s_save_tmp
> +
> + write_hwreg_to_v2(s_save_m0)
> + write_hwreg_to_v2(s_save_pc_lo)
> + s_andn2_b32 s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
> + write_hwreg_to_v2(s_save_tmp)
> + write_hwreg_to_v2(s_save_exec_lo)
> + write_hwreg_to_v2(s_save_exec_hi)
> + write_hwreg_to_v2(s_save_state_priv)
> +
> + s_getreg_b32 s_save_tmp, hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV)
> + write_hwreg_to_v2(s_save_tmp)
> +
> + write_hwreg_to_v2(s_save_xnack_mask)
> +
> + s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_MODE)
> + write_hwreg_to_v2(s_save_m0)
> +
> + s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_SCRATCH_BASE_LO)
> + write_hwreg_to_v2(s_save_m0)
> +
> + s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_SCRATCH_BASE_HI)
> + write_hwreg_to_v2(s_save_m0)
> +
> + s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_EXCP_FLAG_USER)
> + write_hwreg_to_v2(s_save_m0)
> +
> + s_getreg_b32 s_save_m0, hwreg(HW_REG_WAVE_TRAP_CTRL)
> + write_hwreg_to_v2(s_save_m0)
> +
> + s_getreg_b32 s_save_tmp, hwreg(HW_REG_WAVE_STATUS)
> + write_hwreg_to_v2(s_save_tmp)
> +
> + s_get_barrier_state s_save_tmp, -1
> + s_wait_kmcnt (0)
> + write_hwreg_to_v2(s_save_tmp)
> +
> + // Write HWREGs with 16 VGPR lanes. TTMPs occupy space after this.
> + s_mov_b32 exec_lo, 0xFFFF
> + s_mov_b32 exec_hi, 0x0
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> +
> + // Write SGPRs with 32 VGPR lanes. This works in wave32 and wave64 mode.
> + s_mov_b32 exec_lo, 0xFFFFFFFF
> +
> + /* save SGPRs */
> + // Save SGPR before LDS save, then the s0 to s4 can be used during LDS save...
> +
> + // SGPR SR memory offset : size(VGPR)+size(SVGPR)
> + get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
> + get_svgpr_size_bytes(s_save_tmp)
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, s_save_tmp
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + s_mov_b32 ttmp13, 0x0 //next VGPR lane to copy SGPR into
> +
> + s_mov_b32 m0, 0x0 //SGPR initial index value =0
> + s_nop 0x0 //Manually inserted wait states
> +L_SAVE_SGPR_LOOP:
> + // SGPR is allocated in 16 SGPR granularity
> + s_movrels_b64 s0, s0 //s0 = s[0+m0], s1 = s[1+m0]
> + s_movrels_b64 s2, s2 //s2 = s[2+m0], s3 = s[3+m0]
> + s_movrels_b64 s4, s4 //s4 = s[4+m0], s5 = s[5+m0]
> + s_movrels_b64 s6, s6 //s6 = s[6+m0], s7 = s[7+m0]
> + s_movrels_b64 s8, s8 //s8 = s[8+m0], s9 = s[9+m0]
> + s_movrels_b64 s10, s10 //s10 = s[10+m0], s11 = s[11+m0]
> + s_movrels_b64 s12, s12 //s12 = s[12+m0], s13 = s[13+m0]
> + s_movrels_b64 s14, s14 //s14 = s[14+m0], s15 = s[15+m0]
> +
> + write_16sgpr_to_v2(s0)
> +
> + s_cmp_eq_u32 ttmp13, 0x20 //have 32 VGPR lanes filled?
> + s_cbranch_scc0 L_SAVE_SGPR_SKIP_TCP_STORE
> +
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, 0x80
> + s_mov_b32 ttmp13, 0x0
> + v_mov_b32 v2, 0x0
> +L_SAVE_SGPR_SKIP_TCP_STORE:
> +
> + s_add_u32 m0, m0, 16 //next sgpr index
> + s_cmp_lt_u32 m0, 96 //scc = (m0 < first 96 SGPR) ? 1 : 0
> + s_cbranch_scc1 L_SAVE_SGPR_LOOP //first 96 SGPR save is complete?
> +
> + //save the rest 12 SGPR
> + s_movrels_b64 s0, s0 //s0 = s[0+m0], s1 = s[1+m0]
> + s_movrels_b64 s2, s2 //s2 = s[2+m0], s3 = s[3+m0]
> + s_movrels_b64 s4, s4 //s4 = s[4+m0], s5 = s[5+m0]
> + s_movrels_b64 s6, s6 //s6 = s[6+m0], s7 = s[7+m0]
> + s_movrels_b64 s8, s8 //s8 = s[8+m0], s9 = s[9+m0]
> + s_movrels_b64 s10, s10 //s10 = s[10+m0], s11 = s[11+m0]
> + write_12sgpr_to_v2(s0)
> +
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> +
> + /* save LDS */
> +
> +L_SAVE_LDS:
> + // Change EXEC to all threads...
> + s_mov_b32 exec_lo, 0xFFFFFFFF //need every thread from now on
> + s_lshr_b32 m0, s_wave_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_ENABLE_SAVE_LDS_EXEC_HI
> + s_mov_b32 exec_hi, 0x00000000
> + s_branch L_SAVE_LDS_NORMAL
> +L_ENABLE_SAVE_LDS_EXEC_HI:
> + s_mov_b32 exec_hi, 0xFFFFFFFF
> +L_SAVE_LDS_NORMAL:
> + s_getreg_b32 s_save_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)
> + s_and_b32 s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF //lds_size is zero?
> + s_cbranch_scc0 L_SAVE_LDS_DONE //no lds used? jump to L_SAVE_DONE
> +
> + s_and_b32 s_save_tmp, s_save_pc_hi, S_SAVE_PC_HI_FIRST_WAVE_MASK
> + s_cbranch_scc0 L_SAVE_LDS_DONE
> +
> + // first wave do LDS save;
> +
> + s_lshl_b32 s_save_alloc_size, s_save_alloc_size, SQ_WAVE_LDS_ALLOC_GRANULARITY
> + s_mov_b32 s_save_buf_rsrc2, s_save_alloc_size //NUM_RECORDS in bytes
> +
> + // LDS at offset: size(VGPR)+size(SVGPR)+SIZE(SGPR)+SIZE(HWREG)
> + //
> + get_vgpr_size_bytes(s_save_mem_offset, s_wave_size)
> + get_svgpr_size_bytes(s_save_tmp)
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, s_save_tmp
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, get_sgpr_size_bytes()
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, get_hwreg_size_bytes()
> +
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + //load 0~63*4(byte address) to vgpr v0
> + v_mbcnt_lo_u32_b32 v0, -1, 0
> + v_mbcnt_hi_u32_b32 v0, -1, v0
> + v_mul_u32_u24 v0, 4, v0
> +
> + s_lshr_b32 m0, s_wave_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_mov_b32 m0, 0x0
> + s_cbranch_scc1 L_SAVE_LDS_W64
> +
> +L_SAVE_LDS_W32:
> + s_mov_b32 s3, 128
> + s_nop 0
> + s_nop 0
> + s_nop 0
> +L_SAVE_LDS_LOOP_W32:
> + ds_read_b32 v1, v0
> + s_wait_idle
> + buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> +
> + s_add_u32 m0, m0, s3 //every buffer_store_lds does 128 bytes
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, s3
> + v_add_nc_u32 v0, v0, 128 //mem offset increased by 128 bytes
> + s_cmp_lt_u32 m0, s_save_alloc_size //scc=(m0 < s_save_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_SAVE_LDS_LOOP_W32 //LDS save is complete?
> +
> + s_branch L_SAVE_LDS_DONE
> +
> +L_SAVE_LDS_W64:
> + s_mov_b32 s3, 256
> + s_nop 0
> + s_nop 0
> + s_nop 0
> +L_SAVE_LDS_LOOP_W64:
> + ds_read_b32 v1, v0
> + s_wait_idle
> + buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> +
> + s_add_u32 m0, m0, s3 //every buffer_store_lds does 256 bytes
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, s3
> + v_add_nc_u32 v0, v0, 256 //mem offset increased by 256 bytes
> + s_cmp_lt_u32 m0, s_save_alloc_size //scc=(m0 < s_save_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_SAVE_LDS_LOOP_W64 //LDS save is complete?
> +
> +L_SAVE_LDS_DONE:
> + /* save VGPRs - set the Rest VGPRs */
> +L_SAVE_VGPR:
> + // VGPR SR memory offset: 0
> + s_mov_b32 exec_lo, 0xFFFFFFFF //need every thread from now on
> + s_lshr_b32 m0, s_wave_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_ENABLE_SAVE_VGPR_EXEC_HI
> + s_mov_b32 s_save_mem_offset, (0+128*4) // for the rest VGPRs
> + s_mov_b32 exec_hi, 0x00000000
> + s_branch L_SAVE_VGPR_NORMAL
> +L_ENABLE_SAVE_VGPR_EXEC_HI:
> + s_mov_b32 s_save_mem_offset, (0+256*4) // for the rest VGPRs
> + s_mov_b32 exec_hi, 0xFFFFFFFF
> +L_SAVE_VGPR_NORMAL:
> + s_getreg_b32 s_save_alloc_size, hwreg(HW_REG_WAVE_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
> + s_add_u32 s_save_alloc_size, s_save_alloc_size, 1
> + s_lshl_b32 s_save_alloc_size, s_save_alloc_size, 2 //Number of VGPRs = (vgpr_size + 1) * 4 (non-zero value)
> + //determine it is wave32 or wave64
> + s_lshr_b32 m0, s_wave_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_SAVE_VGPR_WAVE64
> +
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // VGPR Allocated in 4-GPR granularity
> +
> + // VGPR store using dw burst
> + s_mov_b32 m0, 0x4 //VGPR initial index value =4
> + s_cmp_lt_u32 m0, s_save_alloc_size
> + s_cbranch_scc0 L_SAVE_VGPR_END
> +
> +L_SAVE_VGPR_W32_LOOP:
> + v_movrels_b32 v0, v0 //v0 = v[0+m0]
> + v_movrels_b32 v1, v1 //v1 = v[1+m0]
> + v_movrels_b32 v2, v2 //v2 = v[2+m0]
> + v_movrels_b32 v3, v3 //v3 = v[3+m0]
> +
> + buffer_store_dword v0, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> + buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*2
> + buffer_store_dword v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:128*3
> +
> + s_add_u32 m0, m0, 4 //next vgpr index
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, 128*4 //every buffer_store_dword does 128 bytes
> + s_cmp_lt_u32 m0, s_save_alloc_size //scc = (m0 < s_save_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_SAVE_VGPR_W32_LOOP //VGPR save is complete?
> +
> + s_branch L_SAVE_VGPR_END
> +
> +L_SAVE_VGPR_WAVE64:
> + s_mov_b32 s_save_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // VGPR store using dw burst
> + s_mov_b32 m0, 0x4 //VGPR initial index value =4
> + s_cmp_lt_u32 m0, s_save_alloc_size
> + s_cbranch_scc0 L_SAVE_SHARED_VGPR
> +
> +L_SAVE_VGPR_W64_LOOP:
> + v_movrels_b32 v0, v0 //v0 = v[0+m0]
> + v_movrels_b32 v1, v1 //v1 = v[1+m0]
> + v_movrels_b32 v2, v2 //v2 = v[2+m0]
> + v_movrels_b32 v3, v3 //v3 = v[3+m0]
> +
> + buffer_store_dword v0, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> + buffer_store_dword v1, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256
> + buffer_store_dword v2, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*2
> + buffer_store_dword v3, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS offset:256*3
> +
> + s_add_u32 m0, m0, 4 //next vgpr index
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, 256*4 //every buffer_store_dword does 256 bytes
> + s_cmp_lt_u32 m0, s_save_alloc_size //scc = (m0 < s_save_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_SAVE_VGPR_W64_LOOP //VGPR save is complete?
> +
> +L_SAVE_SHARED_VGPR:
> + //Below part will be the save shared vgpr part (new for gfx10)
This comment is not really relevant for gfx12.
> + s_getreg_b32 s_save_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)
> + s_and_b32 s_save_alloc_size, s_save_alloc_size, 0xFFFFFFFF //shared_vgpr_size is zero?
> + s_cbranch_scc0 L_SAVE_VGPR_END //no shared_vgpr used? jump to L_SAVE_LDS
> + s_lshl_b32 s_save_alloc_size, s_save_alloc_size, 3 //Number of SHARED_VGPRs = shared_vgpr_size * 8 (non-zero value)
> + //m0 now has the value of normal vgpr count, just add the m0 with shared_vgpr count to get the total count.
> + //save shared_vgpr will start from the index of m0
> + s_add_u32 s_save_alloc_size, s_save_alloc_size, m0
> + s_mov_b32 exec_lo, 0xFFFFFFFF
> + s_mov_b32 exec_hi, 0x00000000
> +
> +L_SAVE_SHARED_VGPR_WAVE64_LOOP:
> + v_movrels_b32 v0, v0 //v0 = v[0+m0]
> + buffer_store_dword v0, v0, s_save_buf_rsrc0, s_save_mem_offset scope:SCOPE_SYS
> + s_add_u32 m0, m0, 1 //next vgpr index
> + s_add_u32 s_save_mem_offset, s_save_mem_offset, 128
> + s_cmp_lt_u32 m0, s_save_alloc_size //scc = (m0 < s_save_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_SAVE_SHARED_VGPR_WAVE64_LOOP //SHARED_VGPR save is complete?
> +
> +L_SAVE_VGPR_END:
> + s_branch L_END_PGM
> +
> +L_RESTORE:
> + /* Setup Resource Contants */
> + s_mov_b32 s_restore_buf_rsrc0, s_restore_spi_init_lo //base_addr_lo
> + s_and_b32 s_restore_buf_rsrc1, s_restore_spi_init_hi, 0x0000FFFF //base_addr_hi
> + s_or_b32 s_restore_buf_rsrc1, s_restore_buf_rsrc1, S_RESTORE_BUF_RSRC_WORD1_STRIDE
> + s_mov_b32 s_restore_buf_rsrc2, 0 //NUM_RECORDS initial value = 0 (in bytes)
> + s_mov_b32 s_restore_buf_rsrc3, S_RESTORE_BUF_RSRC_WORD3_MISC
> +
> + // Save s_restore_spi_init_hi for later use.
> + s_mov_b32 s_restore_spi_init_hi_save, s_restore_spi_init_hi
> +
> + //determine it is wave32 or wave64
> + get_wave_size2(s_restore_size)
> +
> + s_and_b32 s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
> + s_cbranch_scc0 L_RESTORE_VGPR
> +
> + /* restore LDS */
> +L_RESTORE_LDS:
> + s_mov_b32 exec_lo, 0xFFFFFFFF //need every thread from now on
> + s_lshr_b32 m0, s_restore_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_ENABLE_RESTORE_LDS_EXEC_HI
> + s_mov_b32 exec_hi, 0x00000000
> + s_branch L_RESTORE_LDS_NORMAL
> +L_ENABLE_RESTORE_LDS_EXEC_HI:
> + s_mov_b32 exec_hi, 0xFFFFFFFF
> +L_RESTORE_LDS_NORMAL:
> + s_getreg_b32 s_restore_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_LDS_SIZE_SIZE)
> + s_and_b32 s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF //lds_size is zero?
> + s_cbranch_scc0 L_RESTORE_VGPR //no lds used? jump to L_RESTORE_VGPR
> + s_lshl_b32 s_restore_alloc_size, s_restore_alloc_size, SQ_WAVE_LDS_ALLOC_GRANULARITY
> + s_mov_b32 s_restore_buf_rsrc2, s_restore_alloc_size //NUM_RECORDS in bytes
> +
> + // LDS at offset: size(VGPR)+size(SVGPR)+SIZE(SGPR)+SIZE(HWREG)
> + //
> + get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
> + get_svgpr_size_bytes(s_restore_tmp)
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, get_hwreg_size_bytes()
> +
> + s_mov_b32 s_restore_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + s_lshr_b32 m0, s_restore_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_mov_b32 m0, 0x0
> + s_cbranch_scc1 L_RESTORE_LDS_LOOP_W64
> +
> +L_RESTORE_LDS_LOOP_W32:
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset
> + s_wait_idle
> + ds_store_addtid_b32 v0
> + s_add_u32 m0, m0, 128 // 128 DW
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 128 //mem offset increased by 128DW
> + s_cmp_lt_u32 m0, s_restore_alloc_size //scc=(m0 < s_restore_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_RESTORE_LDS_LOOP_W32 //LDS restore is complete?
> + s_branch L_RESTORE_VGPR
> +
> +L_RESTORE_LDS_LOOP_W64:
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset
> + s_wait_idle
> + ds_store_addtid_b32 v0
> + s_add_u32 m0, m0, 256 // 256 DW
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 256 //mem offset increased by 256DW
> + s_cmp_lt_u32 m0, s_restore_alloc_size //scc=(m0 < s_restore_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_RESTORE_LDS_LOOP_W64 //LDS restore is complete?
> +
> + /* restore VGPRs */
> +L_RESTORE_VGPR:
> + // VGPR SR memory offset : 0
> + s_mov_b32 s_restore_mem_offset, 0x0
> + s_mov_b32 exec_lo, 0xFFFFFFFF //need every thread from now on
> + s_lshr_b32 m0, s_restore_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_ENABLE_RESTORE_VGPR_EXEC_HI
> + s_mov_b32 exec_hi, 0x00000000
> + s_branch L_RESTORE_VGPR_NORMAL
> +L_ENABLE_RESTORE_VGPR_EXEC_HI:
> + s_mov_b32 exec_hi, 0xFFFFFFFF
> +L_RESTORE_VGPR_NORMAL:
> + s_getreg_b32 s_restore_alloc_size, hwreg(HW_REG_WAVE_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
> + s_add_u32 s_restore_alloc_size, s_restore_alloc_size, 1
> + s_lshl_b32 s_restore_alloc_size, s_restore_alloc_size, 2 //Number of VGPRs = (vgpr_size + 1) * 4 (non-zero value)
> + //determine it is wave32 or wave64
> + s_lshr_b32 m0, s_restore_size, S_WAVE_SIZE
> + s_and_b32 m0, m0, 1
> + s_cmp_eq_u32 m0, 1
> + s_cbranch_scc1 L_RESTORE_VGPR_WAVE64
> +
> + s_mov_b32 s_restore_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // VGPR load using dw burst
> + s_mov_b32 s_restore_mem_offset_save, s_restore_mem_offset // restore start with v1, v0 will be the last
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 128*4
> + s_mov_b32 m0, 4 //VGPR initial index value = 4
> + s_cmp_lt_u32 m0, s_restore_alloc_size
> + s_cbranch_scc0 L_RESTORE_SGPR
> +
> +L_RESTORE_VGPR_WAVE32_LOOP:
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS
> + buffer_load_dword v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:128
> + buffer_load_dword v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:128*2
> + buffer_load_dword v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:128*3
> + s_wait_idle
> + v_movreld_b32 v0, v0 //v[0+m0] = v0
> + v_movreld_b32 v1, v1
> + v_movreld_b32 v2, v2
> + v_movreld_b32 v3, v3
> + s_add_u32 m0, m0, 4 //next vgpr index
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 128*4 //every buffer_load_dword does 128 bytes
> + s_cmp_lt_u32 m0, s_restore_alloc_size //scc = (m0 < s_restore_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_RESTORE_VGPR_WAVE32_LOOP //VGPR restore (except v0) is complete?
> +
> + /* VGPR restore on v0 */
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS
> + buffer_load_dword v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:128
> + buffer_load_dword v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:128*2
> + buffer_load_dword v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:128*3
> + s_wait_idle
> +
> + s_branch L_RESTORE_SGPR
> +
> +L_RESTORE_VGPR_WAVE64:
> + s_mov_b32 s_restore_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // VGPR load using dw burst
> + s_mov_b32 s_restore_mem_offset_save, s_restore_mem_offset // restore start with v4, v0 will be the last
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 256*4
> + s_mov_b32 m0, 4 //VGPR initial index value = 4
> + s_cmp_lt_u32 m0, s_restore_alloc_size
> + s_cbranch_scc0 L_RESTORE_SHARED_VGPR
> +
> +L_RESTORE_VGPR_WAVE64_LOOP:
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS
> + buffer_load_dword v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:256
> + buffer_load_dword v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:256*2
> + buffer_load_dword v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS offset:256*3
> + s_wait_idle
> + v_movreld_b32 v0, v0 //v[0+m0] = v0
> + v_movreld_b32 v1, v1
> + v_movreld_b32 v2, v2
> + v_movreld_b32 v3, v3
> + s_add_u32 m0, m0, 4 //next vgpr index
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 256*4 //every buffer_load_dword does 256 bytes
> + s_cmp_lt_u32 m0, s_restore_alloc_size //scc = (m0 < s_restore_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_RESTORE_VGPR_WAVE64_LOOP //VGPR restore (except v0) is complete?
> +
> +L_RESTORE_SHARED_VGPR:
> + //Below part will be the restore shared vgpr part (new for gfx10)
This comment is not really relevant for gfx12.
> + s_getreg_b32 s_restore_alloc_size, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE) //shared_vgpr_size
> + s_and_b32 s_restore_alloc_size, s_restore_alloc_size, 0xFFFFFFFF //shared_vgpr_size is zero?
> + s_cbranch_scc0 L_RESTORE_V0 //no shared_vgpr used?
> + s_lshl_b32 s_restore_alloc_size, s_restore_alloc_size, 3 //Number of SHARED_VGPRs = shared_vgpr_size * 8 (non-zero value)
> + //m0 now has the value of normal vgpr count, just add the m0 with shared_vgpr count to get the total count.
> + //restore shared_vgpr will start from the index of m0
> + s_add_u32 s_restore_alloc_size, s_restore_alloc_size, m0
> + s_mov_b32 exec_lo, 0xFFFFFFFF
> + s_mov_b32 exec_hi, 0x00000000
> +L_RESTORE_SHARED_VGPR_WAVE64_LOOP:
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset scope:SCOPE_SYS
> + s_wait_idle
> + v_movreld_b32 v0, v0 //v[0+m0] = v0
> + s_add_u32 m0, m0, 1 //next vgpr index
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 128
> + s_cmp_lt_u32 m0, s_restore_alloc_size //scc = (m0 < s_restore_alloc_size) ? 1 : 0
> + s_cbranch_scc1 L_RESTORE_SHARED_VGPR_WAVE64_LOOP //VGPR restore (except v0) is complete?
> +
> + s_mov_b32 exec_hi, 0xFFFFFFFF //restore back exec_hi before restoring V0!!
> +
> + /* VGPR restore on v0 */
> +L_RESTORE_V0:
> + buffer_load_dword v0, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS
> + buffer_load_dword v1, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:256
> + buffer_load_dword v2, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:256*2
> + buffer_load_dword v3, v0, s_restore_buf_rsrc0, s_restore_mem_offset_save scope:SCOPE_SYS offset:256*3
> + s_wait_idle
> +
> + /* restore SGPRs */
> + //will be 2+8+16*6
> + // SGPR SR memory offset : size(VGPR)+size(SVGPR)
> +L_RESTORE_SGPR:
> + get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
> + get_svgpr_size_bytes(s_restore_tmp)
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
> + s_sub_u32 s_restore_mem_offset, s_restore_mem_offset, 20*4 //s108~s127 is not saved
> +
> + s_mov_b32 s_restore_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + s_mov_b32 m0, s_sgpr_save_num
> +
> + read_4sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> +
> + s_sub_u32 m0, m0, 4 // Restore from S[0] to S[104]
> + s_nop 0 // hazard SALU M0=> S_MOVREL
> +
> + s_movreld_b64 s0, s0 //s[0+m0] = s0
> + s_movreld_b64 s2, s2
> +
> + read_8sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> +
> + s_sub_u32 m0, m0, 8 // Restore from S[0] to S[96]
> + s_nop 0 // hazard SALU M0=> S_MOVREL
> +
> + s_movreld_b64 s0, s0 //s[0+m0] = s0
> + s_movreld_b64 s2, s2
> + s_movreld_b64 s4, s4
> + s_movreld_b64 s6, s6
> +
> + L_RESTORE_SGPR_LOOP:
> + read_16sgpr_from_mem(s0, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> +
> + s_sub_u32 m0, m0, 16 // Restore from S[n] to S[0]
> + s_nop 0 // hazard SALU M0=> S_MOVREL
> +
> + s_movreld_b64 s0, s0 //s[0+m0] = s0
> + s_movreld_b64 s2, s2
> + s_movreld_b64 s4, s4
> + s_movreld_b64 s6, s6
> + s_movreld_b64 s8, s8
> + s_movreld_b64 s10, s10
> + s_movreld_b64 s12, s12
> + s_movreld_b64 s14, s14
> +
> + s_cmp_eq_u32 m0, 0 //scc = (m0 < s_sgpr_save_num) ? 1 : 0
> + s_cbranch_scc0 L_RESTORE_SGPR_LOOP
> +
> + // s_barrier with MODE.DEBUG_EN=1, STATUS.PRIV=1 incorrectly asserts debug exception.
> + // Clear DEBUG_EN before and restore MODE after the barrier.
> + s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE), 0
MODE.DEBUG_EN should be STATE_PRIV.TRAP_AFTER_INST. And because of
that, I guess this insn is not really relevant for gfx12. Might be
addressed in another commit, as the trap handler should assemble
identically before and after this commit.
> +
> + /* restore HW registers */
> +L_RESTORE_HWREG:
> + // HWREG SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)
> + get_vgpr_size_bytes(s_restore_mem_offset, s_restore_size)
> + get_svgpr_size_bytes(s_restore_tmp)
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, s_restore_tmp
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, get_sgpr_size_bytes()
> +
> + s_mov_b32 s_restore_buf_rsrc2, 0x1000000 //NUM_RECORDS in bytes
> +
> + // Restore s_restore_spi_init_hi before the saved value gets clobbered.
> + s_mov_b32 s_restore_spi_init_hi, s_restore_spi_init_hi_save
> +
> + read_hwreg_from_mem(s_restore_m0, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_pc_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_pc_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_exec_lo, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_exec_hi, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_state_priv, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_excp_flag_priv, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_xnack_mask, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_mode, s_restore_buf_rsrc0, s_restore_mem_offset)
> + read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> +
> + s_setreg_b32 hwreg(HW_REG_WAVE_SCRATCH_BASE_LO), s_restore_flat_scratch
> +
> + read_hwreg_from_mem(s_restore_flat_scratch, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> +
> + s_setreg_b32 hwreg(HW_REG_WAVE_SCRATCH_BASE_HI), s_restore_flat_scratch
> +
> + read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> + s_setreg_b32 hwreg(HW_REG_WAVE_EXCP_FLAG_USER), s_restore_tmp
> +
> + read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> + s_setreg_b32 hwreg(HW_REG_WAVE_TRAP_CTRL), s_restore_tmp
> +
> + // Only the first wave needs to restore the workgroup barrier.
> + s_and_b32 s_restore_tmp, s_restore_spi_init_hi, S_RESTORE_SPI_INIT_FIRST_WAVE_MASK
> + s_cbranch_scc0 L_SKIP_BARRIER_RESTORE
> +
> + // Skip over WAVE_STATUS, since there is no state to restore from it
> + s_add_u32 s_restore_mem_offset, s_restore_mem_offset, 4
> +
> + read_hwreg_from_mem(s_restore_tmp, s_restore_buf_rsrc0, s_restore_mem_offset)
> + s_wait_idle
> +
> + s_bitcmp1_b32 s_restore_tmp, BARRIER_STATE_VALID_OFFSET
> + s_cbranch_scc0 L_SKIP_BARRIER_RESTORE
> +
> + // extract the saved signal count from s_restore_tmp
> + s_lshr_b32 s_restore_tmp, s_restore_tmp, BARRIER_STATE_SIGNAL_OFFSET
> +
> + // We need to call s_barrier_signal repeatedly to restore the signal
> + // count of the work group barrier. The member count is already
> + // initialized with the number of waves in the work group.
> +L_BARRIER_RESTORE_LOOP:
> + s_and_b32 s_restore_tmp, s_restore_tmp, s_restore_tmp
> + s_cbranch_scc0 L_SKIP_BARRIER_RESTORE
> + s_barrier_signal -1
> + s_add_i32 s_restore_tmp, s_restore_tmp, -1
> + s_branch L_BARRIER_RESTORE_LOOP
> +
> +L_SKIP_BARRIER_RESTORE:
> +
> + s_mov_b32 m0, s_restore_m0
> + s_mov_b32 exec_lo, s_restore_exec_lo
> + s_mov_b32 exec_hi, s_restore_exec_hi
> +
> + // {TRAPSTS/EXCP_FLAG_PRIV}.SAVE_CONTEXT and HOST_TRAP may have changed.
Remove TRAPSTS.
> + // Only restore the other fields to avoid clobbering them.
> + s_setreg_b32 hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, 0, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_1_SIZE), s_restore_excp_flag_priv
> + s_lshr_b32 s_restore_excp_flag_priv, s_restore_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT
> + s_setreg_b32 hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SIZE), s_restore_excp_flag_priv
> + s_lshr_b32 s_restore_excp_flag_priv, s_restore_excp_flag_priv, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT - SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_2_SHIFT
> + s_setreg_b32 hwreg(HW_REG_WAVE_EXCP_FLAG_PRIV, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SHIFT, SQ_WAVE_EXCP_FLAG_PRIV_RESTORE_PART_3_SIZE), s_restore_excp_flag_priv
> +
> + s_setreg_b32 hwreg(HW_REG_WAVE_MODE), s_restore_mode
> +
> + // Restore trap temporaries 4-11, 13 initialized by SPI debug dispatch logic
> + // ttmp SR memory offset : size(VGPR)+size(SVGPR)+size(SGPR)+0x40
> + get_vgpr_size_bytes(s_restore_ttmps_lo, s_restore_size)
> + get_svgpr_size_bytes(s_restore_ttmps_hi)
> + s_add_u32 s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_ttmps_hi
> + s_add_u32 s_restore_ttmps_lo, s_restore_ttmps_lo, get_sgpr_size_bytes()
> + s_add_u32 s_restore_ttmps_lo, s_restore_ttmps_lo, s_restore_buf_rsrc0
> + s_addc_u32 s_restore_ttmps_hi, s_restore_buf_rsrc1, 0x0
> + s_and_b32 s_restore_ttmps_hi, s_restore_ttmps_hi, 0xFFFF
> + s_load_dwordx4 [ttmp4, ttmp5, ttmp6, ttmp7], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x50 scope:SCOPE_SYS
> + s_load_dwordx4 [ttmp8, ttmp9, ttmp10, ttmp11], [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x60 scope:SCOPE_SYS
> + s_load_dword ttmp13, [s_restore_ttmps_lo, s_restore_ttmps_hi], 0x74 scope:SCOPE_SYS
> + s_wait_idle
> +
> + 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_WAVE_STATE_PRIV), s_restore_state_priv // SCC is included, which is changed by previous salu
> +
> + // Make barrier and LDS state visible to all waves in the group.
> + // STATE_PRIV.BARRIER_COMPLETE may change after this point.
> + s_barrier_signal -2
> + s_barrier_wait -2
> +
> + s_rfe_b64 s_restore_pc_lo //Return to the main shader program and resume execution
> +
> +L_END_PGM:
> + s_endpgm_saved
> +end
> +
> +function write_hwreg_to_v2(s)
> + // Copy into VGPR for later TCP store.
> + v_writelane_b32 v2, s, m0
> + s_add_u32 m0, m0, 0x1
> +end
> +
> +
> +function write_16sgpr_to_v2(s)
> + // Copy into VGPR for later TCP store.
> + for var sgpr_idx = 0; sgpr_idx < 16; sgpr_idx ++
> + v_writelane_b32 v2, s[sgpr_idx], ttmp13
> + s_add_u32 ttmp13, ttmp13, 0x1
> + end
> +end
> +
> +function write_12sgpr_to_v2(s)
> + // Copy into VGPR for later TCP store.
> + for var sgpr_idx = 0; sgpr_idx < 12; sgpr_idx ++
> + v_writelane_b32 v2, s[sgpr_idx], ttmp13
> + s_add_u32 ttmp13, ttmp13, 0x1
> + end
> +end
> +
> +function read_hwreg_from_mem(s, s_rsrc, s_mem_offset)
> + s_buffer_load_dword s, s_rsrc, s_mem_offset scope:SCOPE_SYS
> + s_add_u32 s_mem_offset, s_mem_offset, 4
> +end
> +
> +function read_16sgpr_from_mem(s, s_rsrc, s_mem_offset)
> + s_sub_u32 s_mem_offset, s_mem_offset, 4*16
> + s_buffer_load_dwordx16 s, s_rsrc, s_mem_offset scope:SCOPE_SYS
> +end
> +
> +function read_8sgpr_from_mem(s, s_rsrc, s_mem_offset)
> + s_sub_u32 s_mem_offset, s_mem_offset, 4*8
> + s_buffer_load_dwordx8 s, s_rsrc, s_mem_offset scope:SCOPE_SYS
> +end
> +
> +function read_4sgpr_from_mem(s, s_rsrc, s_mem_offset)
> + s_sub_u32 s_mem_offset, s_mem_offset, 4*4
> + s_buffer_load_dwordx4 s, s_rsrc, s_mem_offset scope:SCOPE_SYS
> +end
> +
> +function get_vgpr_size_bytes(s_vgpr_size_byte, s_size)
> + s_getreg_b32 s_vgpr_size_byte, hwreg(HW_REG_WAVE_GPR_ALLOC,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SHIFT,SQ_WAVE_GPR_ALLOC_VGPR_SIZE_SIZE)
> + s_add_u32 s_vgpr_size_byte, s_vgpr_size_byte, 1
> + s_bitcmp1_b32 s_size, S_WAVE_SIZE
> + s_cbranch_scc1 L_ENABLE_SHIFT_W64
> + s_lshl_b32 s_vgpr_size_byte, s_vgpr_size_byte, (2+7) //Number of VGPRs = (vgpr_size + 1) * 4 * 32 * 4 (non-zero value)
> + s_branch L_SHIFT_DONE
> +L_ENABLE_SHIFT_W64:
> + s_lshl_b32 s_vgpr_size_byte, s_vgpr_size_byte, (2+8) //Number of VGPRs = (vgpr_size + 1) * 4 * 64 * 4 (non-zero value)
> +L_SHIFT_DONE:
> +end
> +
> +function get_svgpr_size_bytes(s_svgpr_size_byte)
> + s_getreg_b32 s_svgpr_size_byte, hwreg(HW_REG_WAVE_LDS_ALLOC,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SHIFT,SQ_WAVE_LDS_ALLOC_VGPR_SHARED_SIZE_SIZE)
> + s_lshl_b32 s_svgpr_size_byte, s_svgpr_size_byte, (3+7)
> +end
> +
> +function get_sgpr_size_bytes
> + return 512
> +end
> +
> +function get_hwreg_size_bytes
> + return 128
> +end
> +
> +function get_wave_size2(s_reg)
> + s_getreg_b32 s_reg, hwreg(HW_REG_WAVE_STATUS,SQ_WAVE_STATUS_WAVE64_SHIFT,SQ_WAVE_STATUS_WAVE64_SIZE)
> + s_lshl_b32 s_reg, s_reg, S_WAVE_SIZE
> +end
More information about the amd-gfx
mailing list