[AMD Official Use Only - General]
Hi Christian,
The details as follows:
> 1. Use unmap_queue package to trigger preemption on gfx9
> Add trailing fence to track the preemption done.
On gfx9, there is no single package to complete the mcbp request in a single frame like gfx10 does.
To send preemption on gfx9, kmd needs to:
1. emit a trailing fence on gfx ring, do not update the wptr to cp.
2. emit a write_reg to reset mmCP_VMID_PREEMPT after the trailing fence.
3. send unmap_queue to kiq ring with field rb_wptr which is the offset of trailing fence on gfx ring.
When cp fw receives the unmap_queue in mec, it will:
1. Store mmCP_RB0_WPTR from rb_wptr to kick GFX RB off.
2. write mmCP_VMID_PREEMPT as 0xffff to request preemption on all vmids. Then wait on mmCP_VMID_PREEMPT to become 0x0 indicating the preemption is complete.
3. the rest of pipeline would do the preemption according to the mmCP_VMID_PREEMPT until it hits the trailing fence.
4. after the trailing fence is signaled, the write_reg to reset mmCP_VMID_PREEMPT unblocks the unmap_queue package to proceed.
The unmap_queue on gfx9 using rb_wptr is referred from the doc cp_packages_rn:
UNMAP_QUEUES
DW| Bits | Field | Description
4b | 19:0 | rb_wptr | If ((engine_sel = 4) and (action = 3)) then preempted GFX queue’s new RB
pointer.
2. Modify emit_ce_meta emit_de_meta functions
> for the resumed ibs.
For preemption enabled ibs, kmd add preamble ib(ce/de meta) to initialize csa data before send the main ib. The csa is used to save/restore ib execution infos when preemption/resubmit happens.
KMD is responsible to extract the content from CSA during re-submission of a previously pre-empted DMA frame.
The patch is to write csa data for resubmit ibs with previous preempted ib's csa.
Thanks,
Jiadong
-----Original Message-----
From: Christian König <ckoenig.leichtzumerken@xxxxxxxxx>
Sent: Friday, August 12, 2022 7:39 PM
To: Zhu, Jiadong <Jiadong.Zhu@xxxxxxx>; amd-gfx@xxxxxxxxxxxxxxxxxxxxx
Cc: Huang, Ray <Ray.Huang@xxxxxxx>; Liu, Aaron <Aaron.Liu@xxxxxxx>
Subject: Re: [PATCH] drm/amdgpu: modify mcbp implement for gfx9(v3)
[CAUTION: External Email]
Am 11.08.22 um 05:19 schrieb jiadong.zhu@xxxxxxx:
> From: "Jiadong.Zhu" <Jiadong.Zhu@xxxxxxx>
>
> 1. Use unmap_queue package to trigger preemption on gfx9
> Add trailing fence to track the preemption done.
> 2. Modify emit_ce_meta emit_de_meta functions
> for the resumed ibs.
>
> Signed-off-by: Jiadong.Zhu <Jiadong.Zhu@xxxxxxx>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 1 +
> drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 161 ++++++++++++++++++++---
> drivers/gpu/drm/amd/amdgpu/soc15d.h | 2 +
> 3 files changed, 143 insertions(+), 21 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> index 82c178a9033a..ca626f0ad7b1 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> @@ -59,6 +59,7 @@ enum amdgpu_ring_priority_level {
> #define AMDGPU_FENCE_FLAG_64BIT (1 << 0)
> #define AMDGPU_FENCE_FLAG_INT (1 << 1)
> #define AMDGPU_FENCE_FLAG_TC_WB_ONLY (1 << 2)
> +#define AMDGPU_FENCE_FLAG_EXEC (1 << 3)
Ok, that here needs much more explanation why you need it and how all this is supposed to work?
Regards,
Christian.
>
> #define to_amdgpu_ring(s) container_of((s), struct amdgpu_ring,
> sched)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> index 5332899642dc..887021fd56aa 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> @@ -751,7 +751,7 @@ static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev);
> static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
> struct amdgpu_cu_info *cu_info);
> static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device
> *adev); -static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring
> *ring);
> +static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool
> +resume);
> static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring);
> static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
> void *ras_error_status); @@
> -824,9 +824,10 @@ static void gfx_v9_0_kiq_unmap_queues(struct amdgpu_ring *kiq_ring,
>
> PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(ring->doorbell_index));
>
> if (action == PREEMPT_QUEUES_NO_UNMAP) {
> - amdgpu_ring_write(kiq_ring, lower_32_bits(gpu_addr));
> - amdgpu_ring_write(kiq_ring, upper_32_bits(gpu_addr));
> - amdgpu_ring_write(kiq_ring, seq);
> + amdgpu_ring_write(kiq_ring, lower_32_bits(ring->wptr & ring->buf_mask));
> + amdgpu_ring_write(kiq_ring, 0);
> + amdgpu_ring_write(kiq_ring, 0);
> +
> } else {
> amdgpu_ring_write(kiq_ring, 0);
> amdgpu_ring_write(kiq_ring, 0); @@ -5446,11 +5447,16 @@
> static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
>
> control |= ib->length_dw | (vmid << 24);
>
> - if (amdgpu_sriov_vf(ring->adev) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) {
> + if ((amdgpu_sriov_vf(ring->adev) || amdgpu_mcbp) && (ib->flags &
> + AMDGPU_IB_FLAG_PREEMPT)) {
> control |= INDIRECT_BUFFER_PRE_ENB(1);
>
> + if (flags & AMDGPU_IB_PREEMPTED)
> + control |= INDIRECT_BUFFER_PRE_RESUME(1);
> +
> if (!(ib->flags & AMDGPU_IB_FLAG_CE) && vmid)
> - gfx_v9_0_ring_emit_de_meta(ring);
> + gfx_v9_0_ring_emit_de_meta(ring,
> + (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ?
> + true : false);
> }
>
> amdgpu_ring_write(ring, header); @@ -5505,6 +5511,7 @@ static
> void gfx_v9_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
> bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
> bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
> bool writeback = flags & AMDGPU_FENCE_FLAG_TC_WB_ONLY;
> + bool exec = flags & AMDGPU_FENCE_FLAG_EXEC;
>
> /* RELEASE_MEM - flush caches, send int */
> amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 6)); @@
> -5515,6 +5522,7 @@ static void gfx_v9_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
> EOP_TC_WB_ACTION_EN |
> EOP_TC_MD_ACTION_EN)) |
>
> EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
> + (exec ? EOP_EXEC : 0x0) |
> EVENT_INDEX(5)));
> amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) |
> INT_SEL(int_sel ? 2 : 0));
>
> @@ -5620,33 +5628,135 @@ static void gfx_v9_ring_emit_sb(struct amdgpu_ring *ring)
> amdgpu_ring_write(ring, 0);
> }
>
> -static void gfx_v9_0_ring_emit_ce_meta(struct amdgpu_ring *ring)
> +static void gfx_v9_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool
> +resume)
> {
> + struct amdgpu_device *adev = ring->adev;
> struct v9_ce_ib_state ce_payload = {0};
> - uint64_t csa_addr;
> + uint64_t offset, ce_payload_gpu_addr;
> + void *ce_payload_cpu_addr;
> int cnt;
>
> cnt = (sizeof(ce_payload) >> 2) + 4 - 2;
> - csa_addr = amdgpu_csa_vaddr(ring->adev);
> +
> + if (ring->is_mes_queue) {
> + offset = offsetof(struct amdgpu_mes_ctx_meta_data,
> + gfx[0].gfx_meta_data) +
> + offsetof(struct v9_gfx_meta_data, ce_payload);
> + ce_payload_gpu_addr =
> + amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
> + ce_payload_cpu_addr =
> + amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
> + } else {
> + offset = offsetof(struct v9_gfx_meta_data, ce_payload);
> + ce_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
> + ce_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
> + }
>
> amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, cnt));
> amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(2) |
> WRITE_DATA_DST_SEL(8) |
> WR_CONFIRM) |
> WRITE_DATA_CACHE_POLICY(0));
> - amdgpu_ring_write(ring, lower_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, ce_payload)));
> - amdgpu_ring_write(ring, upper_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, ce_payload)));
> - amdgpu_ring_write_multiple(ring, (void *)&ce_payload, sizeof(ce_payload) >> 2);
> + amdgpu_ring_write(ring, lower_32_bits(ce_payload_gpu_addr));
> + amdgpu_ring_write(ring, upper_32_bits(ce_payload_gpu_addr));
> +
> + if (resume)
> + amdgpu_ring_write_multiple(ring, ce_payload_cpu_addr,
> + sizeof(ce_payload) >> 2);
> + else
> + amdgpu_ring_write_multiple(ring, (void *)&ce_payload,
> + sizeof(ce_payload) >> 2); }
> +
> +static int gfx_v9_0_ring_preempt_ib(struct amdgpu_ring *ring) {
> + int i, r = 0;
> + struct amdgpu_device *adev = ring->adev;
> + struct amdgpu_kiq *kiq = &adev->gfx.kiq;
> + struct amdgpu_ring *kiq_ring = &kiq->ring;
> + unsigned long flags;
> +
> + if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
> + return -EINVAL;
> +
> + spin_lock_irqsave(&kiq->ring_lock, flags);
> +
> + if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) {
> + spin_unlock_irqrestore(&kiq->ring_lock, flags);
> + return -ENOMEM;
> + }
> +
> + /* assert preemption condition */
> + amdgpu_ring_set_preempt_cond_exec(ring, false);
> +
> + ring->trail_seq += 1;
> + amdgpu_ring_alloc(ring, 13);
> + gfx_v9_0_ring_emit_fence(ring, ring->trail_fence_gpu_addr,
> + ring->trail_seq, AMDGPU_FENCE_FLAG_EXEC);
> + /*reset the CP_VMID_PREEMPT after trailing fence*/
> + amdgpu_ring_emit_wreg(ring,
> + SOC15_REG_OFFSET(GC, 0, mmCP_VMID_PREEMPT),
> + 0x0);
> +
> + /* assert IB preemption, emit the trailing fence */
> + kiq->pmf->kiq_unmap_queues(kiq_ring, ring, PREEMPT_QUEUES_NO_UNMAP,
> + ring->trail_fence_gpu_addr,
> + ring->trail_seq);
> +
> + amdgpu_ring_commit(kiq_ring);
> + spin_unlock_irqrestore(&kiq->ring_lock, flags);
> +
> + /* poll the trailing fence */
> + for (i = 0; i < adev->usec_timeout; i++) {
> + if (ring->trail_seq ==
> + le32_to_cpu(*(ring->trail_fence_cpu_addr)))
> + break;
> + udelay(1);
> + }
> +
> + if (i >= adev->usec_timeout) {
> + r = -EINVAL;
> + DRM_ERROR("ring %d failed to preempt ib\n", ring->idx);
> + }
> +
> + amdgpu_ring_commit(ring);
> +
> + /* deassert preemption condition */
> + amdgpu_ring_set_preempt_cond_exec(ring, true);
> + return r;
> }
>
> -static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring)
> +static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool
> +resume)
> {
> + struct amdgpu_device *adev = ring->adev;
> struct v9_de_ib_state de_payload = {0};
> - uint64_t csa_addr, gds_addr;
> + uint64_t offset, gds_addr, de_payload_gpu_addr;
> + void *de_payload_cpu_addr;
> int cnt;
>
> - csa_addr = amdgpu_csa_vaddr(ring->adev);
> - gds_addr = csa_addr + 4096;
> + if (ring->is_mes_queue) {
> + offset = offsetof(struct amdgpu_mes_ctx_meta_data,
> + gfx[0].gfx_meta_data) +
> + offsetof(struct v9_gfx_meta_data, de_payload);
> + de_payload_gpu_addr =
> + amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
> + de_payload_cpu_addr =
> + amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset);
> +
> + offset = offsetof(struct amdgpu_mes_ctx_meta_data,
> + gfx[0].gds_backup) +
> + offsetof(struct v9_gfx_meta_data, de_payload);
> + gds_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
> + } else {
> + offset = offsetof(struct v9_gfx_meta_data, de_payload);
> + de_payload_gpu_addr = amdgpu_csa_vaddr(ring->adev) + offset;
> + de_payload_cpu_addr = adev->virt.csa_cpu_addr + offset;
> +
> + gds_addr = ALIGN(amdgpu_csa_vaddr(ring->adev) +
> + AMDGPU_CSA_SIZE - adev->gds.gds_size,
> + PAGE_SIZE);
> + }
> +
> de_payload.gds_backup_addrlo = lower_32_bits(gds_addr);
> de_payload.gds_backup_addrhi = upper_32_bits(gds_addr);
>
> @@ -5656,9 +5766,15 @@ static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring)
> WRITE_DATA_DST_SEL(8) |
> WR_CONFIRM) |
> WRITE_DATA_CACHE_POLICY(0));
> - amdgpu_ring_write(ring, lower_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, de_payload)));
> - amdgpu_ring_write(ring, upper_32_bits(csa_addr + offsetof(struct v9_gfx_meta_data, de_payload)));
> - amdgpu_ring_write_multiple(ring, (void *)&de_payload, sizeof(de_payload) >> 2);
> + amdgpu_ring_write(ring, lower_32_bits(de_payload_gpu_addr));
> + amdgpu_ring_write(ring, upper_32_bits(de_payload_gpu_addr));
> +
> + if (resume)
> + amdgpu_ring_write_multiple(ring, de_payload_cpu_addr,
> + sizeof(de_payload) >> 2);
> + else
> + amdgpu_ring_write_multiple(ring, (void *)&de_payload,
> + sizeof(de_payload) >> 2);
> }
>
> static void gfx_v9_0_ring_emit_frame_cntl(struct amdgpu_ring *ring,
> bool start, @@ -5674,8 +5790,10 @@ static void gfx_v9_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
> {
> uint32_t dw2 = 0;
>
> - if (amdgpu_sriov_vf(ring->adev))
> - gfx_v9_0_ring_emit_ce_meta(ring);
> + if (amdgpu_sriov_vf(ring->adev) || amdgpu_mcbp)
> + gfx_v9_0_ring_emit_ce_meta(ring,
> + (!amdgpu_sriov_vf(ring->adev) &&
> + flags &
> + AMDGPU_IB_PREEMPTED) ? true : false);
>
> dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
> if (flags & AMDGPU_HAVE_CTX_SWITCH) { @@ -7024,6 +7142,7 @@
> static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = {
> .emit_cntxcntl = gfx_v9_ring_emit_cntxcntl,
> .init_cond_exec = gfx_v9_0_ring_emit_init_cond_exec,
> .patch_cond_exec = gfx_v9_0_ring_emit_patch_cond_exec,
> + .preempt_ib = gfx_v9_0_ring_preempt_ib,
> .emit_frame_cntl = gfx_v9_0_ring_emit_frame_cntl,
> .emit_wreg = gfx_v9_0_ring_emit_wreg,
> .emit_reg_wait = gfx_v9_0_ring_emit_reg_wait, diff --git
> a/drivers/gpu/drm/amd/amdgpu/soc15d.h
> b/drivers/gpu/drm/amd/amdgpu/soc15d.h
> index 799925d22fc8..614e9f8467fb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/soc15d.h
> +++ b/drivers/gpu/drm/amd/amdgpu/soc15d.h
> @@ -162,6 +162,7 @@
> * 2 - Bypass
> */
> #define INDIRECT_BUFFER_PRE_ENB(x) ((x) << 21)
> +#define INDIRECT_BUFFER_PRE_RESUME(x) ((x) << 30)
> #define PACKET3_COPY_DATA 0x40
> #define PACKET3_PFP_SYNC_ME 0x42
> #define PACKET3_COND_WRITE 0x45
> @@ -184,6 +185,7 @@
> #define EOP_TC_ACTION_EN (1 << 17) /* L2 */
> #define EOP_TC_NC_ACTION_EN (1 << 19)
> #define EOP_TC_MD_ACTION_EN (1 << 21) /* L2 metadata */
> +#define EOP_EXEC (1 << 28) /* For Trailing Fence */
>
> #define DATA_SEL(x) ((x) << 29)
> /* 0 - discard
