From cb5fae143d79d255251921066dbf8eae16383639 Mon Sep 17 00:00:00 2001 From: Huang Rui Date: Thu, 8 Aug 2019 20:05:15 +0800 Subject: drm/amdgpu: job is secure iff CS is secure (v5) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Mark a job as secure, if and only if the command submission flag has the secure flag set. v2: fix the null job pointer while in vmid 0 submission. v3: Context --> Command submission. v4: filling cs parser with cs->in.flags v5: move the job secure flag setting out of amdgpu_cs_submit() Signed-off-by: Huang Rui Signed-off-by: Luben Tuikov Reviewed-by: Alex Deucher Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 3eee5c7d83e0..99de770a8e9f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -232,6 +232,8 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs if (ret) goto free_all_kdata; + p->job->secure = cs->in.flags & AMDGPU_CS_FLAGS_SECURE; + if (p->ctx->vram_lost_counter != p->job->vram_lost_counter) { ret = -ECANCELED; goto free_all_kdata; -- cgit From 0bb5d5b03f78aeb5f87d47877eb15532875c64da Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Wed, 22 Apr 2020 17:56:56 -0400 Subject: drm/amdgpu: Move to a per-IB secure flag (TMZ) Move from a per-CS secure flag (TMZ) to a per-IB secure flag. Signed-off-by: Luben Tuikov Reviewed-by: Huang Rui Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 -- drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 23 ++++++++++++++++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_job.h | 3 --- drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 9 ++++----- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 23 +++++++---------------- drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c | 3 +-- drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c | 3 +-- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 3 +-- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 20 ++++++-------------- include/uapi/drm/amdgpu_drm.h | 7 ++++--- 10 files changed, 44 insertions(+), 52 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 99de770a8e9f..3eee5c7d83e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -232,8 +232,6 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs if (ret) goto free_all_kdata; - p->job->secure = cs->in.flags & AMDGPU_CS_FLAGS_SECURE; - if (p->ctx->vram_lost_counter != p->job->vram_lost_counter) { ret = -ECANCELED; goto free_all_kdata; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index 045951d2b46c..cba22039df6c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -133,6 +133,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, uint64_t fence_ctx; uint32_t status = 0, alloc_size; unsigned fence_flags = 0; + bool secure; unsigned i; int r = 0; @@ -214,9 +215,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, if (job && ring->funcs->emit_cntxcntl) { status |= job->preamble_status; status |= job->preemption_status; - amdgpu_ring_emit_cntxcntl(ring, status, job->secure); + amdgpu_ring_emit_cntxcntl(ring, status); } + secure = false; for (i = 0; i < num_ibs; ++i) { ib = &ibs[i]; @@ -228,12 +230,27 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, !amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */ continue; + /* If this IB is TMZ, add frame TMZ start packet, + * else, turn off TMZ. + */ + if (ib->flags & AMDGPU_IB_FLAGS_SECURE && ring->funcs->emit_tmz) { + if (!secure) { + secure = true; + amdgpu_ring_emit_tmz(ring, true); + } + } else if (secure) { + secure = false; + amdgpu_ring_emit_tmz(ring, false); + } + amdgpu_ring_emit_ib(ring, job, ib, status); status &= ~AMDGPU_HAVE_CTX_SWITCH; } - if (ring->funcs->emit_tmz) - amdgpu_ring_emit_tmz(ring, false, job ? job->secure : false); + if (secure) { + secure = false; + amdgpu_ring_emit_tmz(ring, false); + } #ifdef CONFIG_X86_64 if (!(adev->flags & AMD_IS_APU)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h index 7f5ccee476a4..81caac9b958a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h @@ -62,9 +62,6 @@ struct amdgpu_job { /* user fence handling */ uint64_t uf_addr; uint64_t uf_sequence; - - /* the job is due to a secure command submission */ - bool secure; }; int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 5956eff2d784..7d39064f9361 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -168,8 +168,7 @@ struct amdgpu_ring_funcs { void (*begin_use)(struct amdgpu_ring *ring); void (*end_use)(struct amdgpu_ring *ring); void (*emit_switch_buffer) (struct amdgpu_ring *ring); - void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags, - bool trusted); + void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags); void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg, uint32_t reg_val_offs); void (*emit_wreg)(struct amdgpu_ring *ring, uint32_t reg, uint32_t val); @@ -178,7 +177,7 @@ struct amdgpu_ring_funcs { void (*emit_reg_write_reg_wait)(struct amdgpu_ring *ring, uint32_t reg0, uint32_t reg1, uint32_t ref, uint32_t mask); - void (*emit_tmz)(struct amdgpu_ring *ring, bool start, bool trusted); + void (*emit_tmz)(struct amdgpu_ring *ring, bool start); /* Try to soft recover the ring to make the fence signal */ void (*soft_recovery)(struct amdgpu_ring *ring, unsigned vmid); int (*preempt_ib)(struct amdgpu_ring *ring); @@ -252,12 +251,12 @@ struct amdgpu_ring { #define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as)) #define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r)) #define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r)) -#define amdgpu_ring_emit_cntxcntl(r, d, s) (r)->funcs->emit_cntxcntl((r), (d), (s)) +#define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d)) #define amdgpu_ring_emit_rreg(r, d, o) (r)->funcs->emit_rreg((r), (d), (o)) #define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v)) #define amdgpu_ring_emit_reg_wait(r, d, v, m) (r)->funcs->emit_reg_wait((r), (d), (v), (m)) #define amdgpu_ring_emit_reg_write_reg_wait(r, d0, d1, v, m) (r)->funcs->emit_reg_write_reg_wait((r), (d0), (d1), (v), (m)) -#define amdgpu_ring_emit_tmz(r, b, s) (r)->funcs->emit_tmz((r), (b), (s)) +#define amdgpu_ring_emit_tmz(r, b) (r)->funcs->emit_tmz((r), (b)) #define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib))) #define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r)) #define amdgpu_ring_patch_cond_exec(r,o) (r)->funcs->patch_cond_exec((r),(o)) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 473c1c145332..404c6d470515 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -3037,8 +3037,7 @@ static int gfx_v10_0_rlc_backdoor_autoload_enable(struct amdgpu_device *adev); static int gfx_v10_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev); static void gfx_v10_0_ring_emit_ce_meta(struct amdgpu_ring *ring, bool resume); static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume); -static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start, - bool trusted); +static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start); static void gfx10_kiq_set_resources(struct amdgpu_ring *kiq_ring, uint64_t queue_mask) { @@ -7436,8 +7435,7 @@ static void gfx_v10_0_ring_emit_sb(struct amdgpu_ring *ring) } static void gfx_v10_0_ring_emit_cntxcntl(struct amdgpu_ring *ring, - uint32_t flags, - bool trusted) + uint32_t flags) { uint32_t dw2 = 0; @@ -7445,8 +7443,6 @@ static void gfx_v10_0_ring_emit_cntxcntl(struct amdgpu_ring *ring, gfx_v10_0_ring_emit_ce_meta(ring, (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false); - gfx_v10_0_ring_emit_tmz(ring, true, trusted); - dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */ if (flags & AMDGPU_HAVE_CTX_SWITCH) { /* set load_global_config & load_global_uconfig */ @@ -7603,17 +7599,12 @@ static void gfx_v10_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume) sizeof(de_payload) >> 2); } -static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start, - bool trusted) +static void gfx_v10_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start) { - amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0)); - /* - * cmd = 0: frame begin - * cmd = 1: frame end - */ - amdgpu_ring_write(ring, - ((amdgpu_is_tmz(ring->adev) && trusted) ? FRAME_TMZ : 0) - | FRAME_CMD(start ? 0 : 1)); + if (amdgpu_is_tmz(ring->adev)) { + amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0)); + amdgpu_ring_write(ring, FRAME_TMZ | FRAME_CMD(start ? 0 : 1)); + } } static void gfx_v10_0_ring_emit_rreg(struct amdgpu_ring *ring, uint32_t reg, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index 283b7fc10f98..aa1e1be852dd 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -2969,8 +2969,7 @@ static uint64_t gfx_v6_0_get_gpu_clock_counter(struct amdgpu_device *adev) return clock; } -static void gfx_v6_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags, - bool trusted) +static void gfx_v6_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags) { if (flags & AMDGPU_HAVE_CTX_SWITCH) gfx_v6_0_ring_emit_vgt_flush(ring); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index f26e91354ba8..e5a88cad44cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -2320,8 +2320,7 @@ static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring, amdgpu_ring_write(ring, control); } -static void gfx_v7_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags, - bool trusted) +static void gfx_v7_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags) { uint32_t dw2 = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index d1312d829252..2fcf6865abba 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -6329,8 +6329,7 @@ static void gfx_v8_ring_emit_sb(struct amdgpu_ring *ring) amdgpu_ring_write(ring, 0); } -static void gfx_v8_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags, - bool trusted) +static void gfx_v8_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags) { uint32_t dw2 = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index bae5dd6ea348..4e042e974983 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -5442,29 +5442,21 @@ static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring) amdgpu_ring_write_multiple(ring, (void *)&de_payload, sizeof(de_payload) >> 2); } -static void gfx_v9_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start, - bool trusted) +static void gfx_v9_0_ring_emit_tmz(struct amdgpu_ring *ring, bool start) { - amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0)); - /* - * cmd = 0: frame begin - * cmd = 1: frame end - */ - amdgpu_ring_write(ring, - ((amdgpu_is_tmz(ring->adev) && trusted) ? FRAME_TMZ : 0) - | FRAME_CMD(start ? 0 : 1)); + if (amdgpu_is_tmz(ring->adev)) { + amdgpu_ring_write(ring, PACKET3(PACKET3_FRAME_CONTROL, 0)); + amdgpu_ring_write(ring, FRAME_TMZ | FRAME_CMD(start ? 0 : 1)); + } } -static void gfx_v9_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags, - bool trusted) +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); - gfx_v9_0_ring_emit_tmz(ring, true, trusted); - dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */ if (flags & AMDGPU_HAVE_CTX_SWITCH) { /* set load_global_config & load_global_uconfig */ diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index bea72eb8c147..e01b673f0449 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -558,9 +558,6 @@ struct drm_amdgpu_cs_chunk { __u64 chunk_data; }; -/* Flag the command submission as secure */ -#define AMDGPU_CS_FLAGS_SECURE (1 << 0) - struct drm_amdgpu_cs_in { /** Rendering context id */ __u32 ctx_id; @@ -601,6 +598,10 @@ union drm_amdgpu_cs { */ #define AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID (1 << 4) +/* Flag the IB as secure (TMZ) + */ +#define AMDGPU_IB_FLAGS_SECURE (1 << 5) + struct drm_amdgpu_cs_chunk_ib { __u32 _pad; /** AMDGPU_IB_FLAG_* */ -- cgit From 9ecefb19c3a6626c27ea7ee72d431f22462e1d54 Mon Sep 17 00:00:00 2001 From: Christian König Date: Wed, 1 Apr 2020 11:18:21 +0200 Subject: drm/amdgpu: cleanup IB pool handling a bit MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Fix the coding style, move and rename the definitions to better match what they are supposed to be doing. Signed-off-by: Christian König Reviewed-by: Felix Kuehling Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 11 +---- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 3 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c | 65 +++++++++++++++-------------- drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 13 ++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_test.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 10 +++-- drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c | 5 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c | 11 +++-- drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c | 3 +- 10 files changed, 71 insertions(+), 56 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 589d8783fa21..99e5f474505d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -204,8 +204,6 @@ extern int amdgpu_cik_support; #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS 3000 #define AMDGPU_MAX_USEC_TIMEOUT 100000 /* 100 ms */ #define AMDGPU_FENCE_JIFFIES_TIMEOUT (HZ / 2) -/* AMDGPU_IB_POOL_SIZE must be a power of 2 */ -#define AMDGPU_IB_POOL_SIZE 16 #define AMDGPU_DEBUGFS_MAX_COMPONENTS 32 #define AMDGPUFB_CONN_LIMIT 4 #define AMDGPU_BIOS_NUM_SCRATCH 16 @@ -402,13 +400,6 @@ struct amdgpu_sa_bo { int amdgpu_fence_slab_init(void); void amdgpu_fence_slab_fini(void); -enum amdgpu_ib_pool_type { - AMDGPU_IB_POOL_NORMAL = 0, - AMDGPU_IB_POOL_VM, - AMDGPU_IB_POOL_DIRECT, - - AMDGPU_IB_POOL_MAX -}; /* * IRQS. */ @@ -866,7 +857,7 @@ struct amdgpu_device { unsigned num_rings; struct amdgpu_ring *rings[AMDGPU_MAX_RINGS]; bool ib_pool_ready; - struct amdgpu_sa_manager ring_tmp_bo[AMDGPU_IB_POOL_MAX]; + struct amdgpu_sa_manager ib_pools[AMDGPU_IB_POOL_MAX]; struct amdgpu_sched gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX]; /* interrupts */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 3eee5c7d83e0..7653f62b1b2d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -924,7 +924,8 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev, ring = to_amdgpu_ring(entity->rq->sched); r = amdgpu_ib_get(adev, vm, ring->funcs->parse_cs ? - chunk_ib->ib_bytes : 0, AMDGPU_IB_POOL_NORMAL, ib); + chunk_ib->ib_bytes : 0, + AMDGPU_IB_POOL_DELAYED, ib); if (r) { DRM_ERROR("Failed to get ib !\n"); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index aebbbb573884..c24366aacf3a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -61,14 +61,13 @@ * Returns 0 on success, error on failure. */ int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm, - unsigned size, - enum amdgpu_ib_pool_type pool_type, - struct amdgpu_ib *ib) + unsigned size, enum amdgpu_ib_pool_type pool_type, + struct amdgpu_ib *ib) { int r; if (size) { - r = amdgpu_sa_bo_new(&adev->ring_tmp_bo[pool_type], + r = amdgpu_sa_bo_new(&adev->ib_pools[pool_type], &ib->sa_bo, size, 256); if (r) { dev_err(adev->dev, "failed to get a new IB (%d)\n", r); @@ -305,30 +304,32 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, */ int amdgpu_ib_pool_init(struct amdgpu_device *adev) { - int r, i; unsigned size; + int r, i; - if (adev->ib_pool_ready) { + if (adev->ib_pool_ready) return 0; - } + for (i = 0; i < AMDGPU_IB_POOL_MAX; i++) { if (i == AMDGPU_IB_POOL_DIRECT) size = PAGE_SIZE * 2; else - size = AMDGPU_IB_POOL_SIZE*64*1024; - r = amdgpu_sa_bo_manager_init(adev, &adev->ring_tmp_bo[i], - size, - AMDGPU_GPU_PAGE_SIZE, - AMDGPU_GEM_DOMAIN_GTT); - if (r) { - for (i--; i >= 0; i--) - amdgpu_sa_bo_manager_fini(adev, &adev->ring_tmp_bo[i]); - return r; - } + size = AMDGPU_IB_POOL_SIZE; + + r = amdgpu_sa_bo_manager_init(adev, &adev->ib_pools[i], + size, AMDGPU_GPU_PAGE_SIZE, + AMDGPU_GEM_DOMAIN_GTT); + if (r) + goto error; } adev->ib_pool_ready = true; return 0; + +error: + while (i--) + amdgpu_sa_bo_manager_fini(adev, &adev->ib_pools[i]); + return r; } /** @@ -343,11 +344,12 @@ void amdgpu_ib_pool_fini(struct amdgpu_device *adev) { int i; - if (adev->ib_pool_ready) { - for (i = 0; i < AMDGPU_IB_POOL_MAX; i++) - amdgpu_sa_bo_manager_fini(adev, &adev->ring_tmp_bo[i]); - adev->ib_pool_ready = false; - } + if (!adev->ib_pool_ready) + return; + + for (i = 0; i < AMDGPU_IB_POOL_MAX; i++) + amdgpu_sa_bo_manager_fini(adev, &adev->ib_pools[i]); + adev->ib_pool_ready = false; } /** @@ -362,9 +364,9 @@ void amdgpu_ib_pool_fini(struct amdgpu_device *adev) */ int amdgpu_ib_ring_tests(struct amdgpu_device *adev) { - unsigned i; - int r, ret = 0; long tmo_gfx, tmo_mm; + int r, ret = 0; + unsigned i; tmo_mm = tmo_gfx = AMDGPU_IB_TEST_TIMEOUT; if (amdgpu_sriov_vf(adev)) { @@ -442,15 +444,16 @@ static int amdgpu_debugfs_sa_info(struct seq_file *m, void *data) struct drm_device *dev = node->minor->dev; struct amdgpu_device *adev = dev->dev_private; - seq_printf(m, "-------------------- NORMAL -------------------- \n"); - amdgpu_sa_bo_dump_debug_info(&adev->ring_tmp_bo[AMDGPU_IB_POOL_NORMAL], m); - seq_printf(m, "---------------------- VM ---------------------- \n"); - amdgpu_sa_bo_dump_debug_info(&adev->ring_tmp_bo[AMDGPU_IB_POOL_VM], m); - seq_printf(m, "-------------------- DIRECT--------------------- \n"); - amdgpu_sa_bo_dump_debug_info(&adev->ring_tmp_bo[AMDGPU_IB_POOL_DIRECT], m); + seq_printf(m, "--------------------- DELAYED --------------------- \n"); + amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_DELAYED], + m); + seq_printf(m, "-------------------- IMMEDIATE -------------------- \n"); + amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_IMMEDIATE], + m); + seq_printf(m, "--------------------- DIRECT ---------------------- \n"); + amdgpu_sa_bo_dump_debug_info(&adev->ib_pools[AMDGPU_IB_POOL_DIRECT], m); return 0; - } static const struct drm_info_list amdgpu_debugfs_sa_list[] = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 7390261095b7..107e80063553 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -50,6 +50,8 @@ #define to_amdgpu_ring(s) container_of((s), struct amdgpu_ring, sched) +#define AMDGPU_IB_POOL_SIZE (1024 * 1024) + enum amdgpu_ring_type { AMDGPU_RING_TYPE_GFX = AMDGPU_HW_IP_GFX, AMDGPU_RING_TYPE_COMPUTE = AMDGPU_HW_IP_COMPUTE, @@ -63,6 +65,17 @@ enum amdgpu_ring_type { AMDGPU_RING_TYPE_KIQ }; +enum amdgpu_ib_pool_type { + /* Normal submissions to the top of the pipeline. */ + AMDGPU_IB_POOL_DELAYED, + /* Immediate submissions to the bottom of the pipeline. */ + AMDGPU_IB_POOL_IMMEDIATE, + /* Direct submission to the ring buffer during init and reset. */ + AMDGPU_IB_POOL_DIRECT, + + AMDGPU_IB_POOL_MAX +}; + struct amdgpu_device; struct amdgpu_ring; struct amdgpu_ib; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c index 476f1f89aaad..2f4d5ca9894f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c @@ -44,7 +44,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev) /* Number of tests = * (Total GTT - IB pool - writeback page - ring buffers) / test size */ - n = adev->gmc.gart_size - AMDGPU_IB_POOL_SIZE*64*1024; + n = adev->gmc.gart_size - AMDGPU_IB_POOL_SIZE; for (i = 0; i < AMDGPU_MAX_RINGS; ++i) if (adev->rings[i]) n -= adev->rings[i]->ring_size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 1296499f0f54..ea0199a8f9c9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -333,7 +333,7 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, num_bytes = num_pages * 8; r = amdgpu_job_alloc_with_ib(adev, num_dw * 4 + num_bytes, - AMDGPU_IB_POOL_NORMAL, &job); + AMDGPU_IB_POOL_DELAYED, &job); if (r) return r; @@ -2122,6 +2122,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, struct dma_fence **fence, bool direct_submit, bool vm_needs_flush, bool tmz) { + enum amdgpu_ib_pool_type pool = direct_submit ? AMDGPU_IB_POOL_DIRECT : + AMDGPU_IB_POOL_DELAYED; struct amdgpu_device *adev = ring->adev; struct amdgpu_job *job; @@ -2139,8 +2141,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, num_loops = DIV_ROUND_UP(byte_count, max_bytes); num_dw = ALIGN(num_loops * adev->mman.buffer_funcs->copy_num_dw, 8); - r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, - direct_submit ? AMDGPU_IB_POOL_DIRECT : AMDGPU_IB_POOL_NORMAL, &job); + r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, pool, &job); if (r) return r; @@ -2229,7 +2230,8 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo, /* for IB padding */ num_dw += 64; - r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, AMDGPU_IB_POOL_NORMAL, &job); + r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, AMDGPU_IB_POOL_DELAYED, + &job); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index 550282d9c1fc..5100ebe8858d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -1056,8 +1056,8 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo, goto err; } - r = amdgpu_job_alloc_with_ib(adev, 64, - direct ? AMDGPU_IB_POOL_DIRECT : AMDGPU_IB_POOL_NORMAL, &job); + r = amdgpu_job_alloc_with_ib(adev, 64, direct ? AMDGPU_IB_POOL_DIRECT : + AMDGPU_IB_POOL_DELAYED, &job); if (r) goto err; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index d090455282e5..ecaa2d7483b2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -447,7 +447,7 @@ static int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle, int i, r; r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, - AMDGPU_IB_POOL_DIRECT, &job); + AMDGPU_IB_POOL_DIRECT, &job); if (r) return r; @@ -526,7 +526,8 @@ static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, int i, r; r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, - direct ? AMDGPU_IB_POOL_DIRECT : AMDGPU_IB_POOL_NORMAL, &job); + direct ? AMDGPU_IB_POOL_DIRECT : + AMDGPU_IB_POOL_DELAYED, &job); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c index fbd451f3559a..b96c8d9a1946 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c @@ -61,11 +61,12 @@ static int amdgpu_vm_sdma_prepare(struct amdgpu_vm_update_params *p, struct dma_resv *resv, enum amdgpu_sync_mode sync_mode) { + enum amdgpu_ib_pool_type pool = p->direct ? AMDGPU_IB_POOL_IMMEDIATE : + AMDGPU_IB_POOL_DELAYED; unsigned int ndw = AMDGPU_VM_SDMA_MIN_NUM_DW; int r; - r = amdgpu_job_alloc_with_ib(p->adev, ndw * 4, - p->direct ? AMDGPU_IB_POOL_VM : AMDGPU_IB_POOL_NORMAL, &p->job); + r = amdgpu_job_alloc_with_ib(p->adev, ndw * 4, pool, &p->job); if (r) return r; @@ -199,6 +200,8 @@ static int amdgpu_vm_sdma_update(struct amdgpu_vm_update_params *p, uint64_t addr, unsigned count, uint32_t incr, uint64_t flags) { + enum amdgpu_ib_pool_type pool = p->direct ? AMDGPU_IB_POOL_IMMEDIATE : + AMDGPU_IB_POOL_DELAYED; unsigned int i, ndw, nptes; uint64_t *pte; int r; @@ -224,8 +227,8 @@ static int amdgpu_vm_sdma_update(struct amdgpu_vm_update_params *p, ndw = max(ndw, AMDGPU_VM_SDMA_MIN_NUM_DW); ndw = min(ndw, AMDGPU_VM_SDMA_MAX_NUM_DW); - r = amdgpu_job_alloc_with_ib(p->adev, ndw * 4, - p->direct ? AMDGPU_IB_POOL_VM : AMDGPU_IB_POOL_NORMAL, &p->job); + r = amdgpu_job_alloc_with_ib(p->adev, ndw * 4, pool, + &p->job); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index eff25c72c6c6..edaa50d850a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -372,7 +372,8 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, * translation. Avoid this by doing the invalidation from the SDMA * itself. */ - r = amdgpu_job_alloc_with_ib(adev, 16 * 4, AMDGPU_IB_POOL_VM, &job); + r = amdgpu_job_alloc_with_ib(adev, 16 * 4, AMDGPU_IB_POOL_IMMEDIATE, + &job); if (r) goto error_alloc; -- cgit