diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-05-18 17:40:19 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-05-18 17:40:19 +0000 |
commit | 9f0fc191371843c4fc000a226b0a26b6c059aacd (patch) | |
tree | 35f8be3ef04506ac891ad001e8c41e535ae8d01d /drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c | |
parent | Releasing progress-linux version 6.6.15-2~progress7.99u1. (diff) | |
download | linux-9f0fc191371843c4fc000a226b0a26b6c059aacd.tar.xz linux-9f0fc191371843c4fc000a226b0a26b6c059aacd.zip |
Merging upstream version 6.7.7.
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c')
-rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c | 220 |
1 files changed, 96 insertions, 124 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c index e3b76fd28d..c9c653cfc7 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c @@ -42,9 +42,11 @@ #include "nbio_v4_3.h" #include "gfxhub_v3_0.h" #include "gfxhub_v3_0_3.h" +#include "gfxhub_v11_5_0.h" #include "mmhub_v3_0.h" #include "mmhub_v3_0_1.h" #include "mmhub_v3_0_2.h" +#include "mmhub_v3_3.h" #include "athub_v3_0.h" @@ -71,7 +73,8 @@ gmc_v11_0_vm_fault_interrupt_state(struct amdgpu_device *adev, * fini/suspend, so the overall state doesn't * change over the course of suspend/resume. */ - if (!adev->in_s0ix) + if (!adev->in_s0ix && (adev->in_runpm || adev->in_suspend || + amdgpu_in_reset(adev))) amdgpu_gmc_set_vm_fault_masks(adev, AMDGPU_GFXHUB(0), false); break; case AMDGPU_IRQ_STATE_ENABLE: @@ -117,6 +120,9 @@ static int gmc_v11_0_process_interrupt(struct amdgpu_device *adev, status = RREG32(hub->vm_l2_pro_fault_status); WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); + + amdgpu_vm_update_fault_cache(adev, entry->pasid, addr, status, + entry->vmid_src ? AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0)); } if (printk_ratelimit()) { @@ -184,27 +190,50 @@ static bool gmc_v11_0_get_vmid_pasid_mapping_info( return !!(*p_pasid); } -/* - * GART - * VMID 0 is the physical GPU addresses as used by the kernel. - * VMIDs 1-15 are used for userspace clients and are handled - * by the amdgpu vm/hsa code. +/** + * gmc_v11_0_flush_gpu_tlb - gart tlb flush callback + * + * @adev: amdgpu_device pointer + * @vmid: vm instance to flush + * @vmhub: which hub to flush + * @flush_type: the flush type + * + * Flush the TLB for the requested page table. */ - -static void gmc_v11_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, - unsigned int vmhub, uint32_t flush_type) +static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, + uint32_t vmhub, uint32_t flush_type) { bool use_semaphore = gmc_v11_0_use_invalidate_semaphore(adev, vmhub); struct amdgpu_vmhub *hub = &adev->vmhub[vmhub]; u32 inv_req = hub->vmhub_funcs->get_invalidate_req(vmid, flush_type); - u32 tmp; /* Use register 17 for GART */ const unsigned int eng = 17; + unsigned char hub_ip; + u32 sem, req, ack; unsigned int i; - unsigned char hub_ip = 0; + u32 tmp; + + if ((vmhub == AMDGPU_GFXHUB(0)) && !adev->gfx.is_poweron) + return; + + sem = hub->vm_inv_eng0_sem + hub->eng_distance * eng; + req = hub->vm_inv_eng0_req + hub->eng_distance * eng; + ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng; + + /* flush hdp cache */ + adev->hdp.funcs->flush_hdp(adev, NULL); + + /* For SRIOV run time, driver shouldn't access the register through MMIO + * Directly use kiq to do the vm invalidation instead + */ + if ((adev->gfx.kiq[0].ring.sched.ready || adev->mes.ring.sched.ready) && + (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) { + amdgpu_virt_kiq_reg_write_reg_wait(adev, req, ack, inv_req, + 1 << vmid, GET_INST(GC, 0)); + return; + } - hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ? - GC_HWIP : MMHUB_HWIP; + hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ? GC_HWIP : MMHUB_HWIP; spin_lock(&adev->gmc.invalidate_lock); /* @@ -218,8 +247,7 @@ static void gmc_v11_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, if (use_semaphore) { for (i = 0; i < adev->usec_timeout; i++) { /* a read return value of 1 means semaphore acuqire */ - tmp = RREG32_RLC_NO_KIQ(hub->vm_inv_eng0_sem + - hub->eng_distance * eng, hub_ip); + tmp = RREG32_RLC_NO_KIQ(sem, hub_ip); if (tmp & 0x1) break; udelay(1); @@ -229,12 +257,11 @@ static void gmc_v11_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, DRM_ERROR("Timeout waiting for sem acquire in VM flush!\n"); } - WREG32_RLC_NO_KIQ(hub->vm_inv_eng0_req + hub->eng_distance * eng, inv_req, hub_ip); + WREG32_RLC_NO_KIQ(req, inv_req, hub_ip); /* Wait for ACK with a delay.*/ for (i = 0; i < adev->usec_timeout; i++) { - tmp = RREG32_RLC_NO_KIQ(hub->vm_inv_eng0_ack + - hub->eng_distance * eng, hub_ip); + tmp = RREG32_RLC_NO_KIQ(ack, hub_ip); tmp &= 1 << vmid; if (tmp) break; @@ -244,12 +271,7 @@ static void gmc_v11_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, /* TODO: It needs to continue working on debugging with semaphore for GFXHUB as well. */ if (use_semaphore) - /* - * add semaphore release after invalidation, - * write with 0 means semaphore release - */ - WREG32_RLC_NO_KIQ(hub->vm_inv_eng0_sem + - hub->eng_distance * eng, 0, hub_ip); + WREG32_RLC_NO_KIQ(sem, 0, hub_ip); /* Issue additional private vm invalidation to MMHUB */ if ((vmhub != AMDGPU_GFXHUB(0)) && @@ -266,50 +288,8 @@ static void gmc_v11_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, spin_unlock(&adev->gmc.invalidate_lock); - if (i < adev->usec_timeout) - return; - - DRM_ERROR("Timeout waiting for VM flush ACK!\n"); -} - -/** - * gmc_v11_0_flush_gpu_tlb - gart tlb flush callback - * - * @adev: amdgpu_device pointer - * @vmid: vm instance to flush - * @vmhub: which hub to flush - * @flush_type: the flush type - * - * Flush the TLB for the requested page table. - */ -static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, - uint32_t vmhub, uint32_t flush_type) -{ - if ((vmhub == AMDGPU_GFXHUB(0)) && !adev->gfx.is_poweron) - return; - - /* flush hdp cache */ - adev->hdp.funcs->flush_hdp(adev, NULL); - - /* For SRIOV run time, driver shouldn't access the register through MMIO - * Directly use kiq to do the vm invalidation instead - */ - if ((adev->gfx.kiq[0].ring.sched.ready || adev->mes.ring.sched.ready) && - (amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) { - struct amdgpu_vmhub *hub = &adev->vmhub[vmhub]; - const unsigned int eng = 17; - u32 inv_req = hub->vmhub_funcs->get_invalidate_req(vmid, flush_type); - u32 req = hub->vm_inv_eng0_req + hub->eng_distance * eng; - u32 ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng; - - amdgpu_virt_kiq_reg_write_reg_wait(adev, req, ack, inv_req, - 1 << vmid); - return; - } - - mutex_lock(&adev->mman.gtt_window_lock); - gmc_v11_0_flush_vm_hub(adev, vmid, vmhub, 0); - mutex_unlock(&adev->mman.gtt_window_lock); + if (i >= adev->usec_timeout) + dev_err(adev->dev, "Timeout waiting for VM flush ACK!\n"); } /** @@ -323,59 +303,31 @@ static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, * * Flush the TLB for the requested pasid. */ -static int gmc_v11_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev, - uint16_t pasid, uint32_t flush_type, - bool all_hub, uint32_t inst) +static void gmc_v11_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev, + uint16_t pasid, uint32_t flush_type, + bool all_hub, uint32_t inst) { + uint16_t queried; int vmid, i; - signed long r; - uint32_t seq; - uint16_t queried_pasid; - bool ret; - struct amdgpu_ring *ring = &adev->gfx.kiq[0].ring; - struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; - - if (amdgpu_emu_mode == 0 && ring->sched.ready) { - spin_lock(&adev->gfx.kiq[0].ring_lock); - /* 2 dwords flush + 8 dwords fence */ - amdgpu_ring_alloc(ring, kiq->pmf->invalidate_tlbs_size + 8); - kiq->pmf->kiq_invalidate_tlbs(ring, - pasid, flush_type, all_hub); - r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT); - if (r) { - amdgpu_ring_undo(ring); - spin_unlock(&adev->gfx.kiq[0].ring_lock); - return -ETIME; - } - - amdgpu_ring_commit(ring); - spin_unlock(&adev->gfx.kiq[0].ring_lock); - r = amdgpu_fence_wait_polling(ring, seq, adev->usec_timeout); - if (r < 1) { - dev_err(adev->dev, "wait for kiq fence error: %ld.\n", r); - return -ETIME; - } - - return 0; - } for (vmid = 1; vmid < 16; vmid++) { - - ret = gmc_v11_0_get_vmid_pasid_mapping_info(adev, vmid, - &queried_pasid); - if (ret && queried_pasid == pasid) { - if (all_hub) { - for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) - gmc_v11_0_flush_gpu_tlb(adev, vmid, - i, flush_type); - } else { - gmc_v11_0_flush_gpu_tlb(adev, vmid, - AMDGPU_GFXHUB(0), flush_type); - } + bool valid; + + valid = gmc_v11_0_get_vmid_pasid_mapping_info(adev, vmid, + &queried); + if (!valid || queried != pasid) + continue; + + if (all_hub) { + for_each_set_bit(i, adev->vmhubs_mask, + AMDGPU_MAX_VMHUBS) + gmc_v11_0_flush_gpu_tlb(adev, vmid, i, + flush_type); + } else { + gmc_v11_0_flush_gpu_tlb(adev, vmid, AMDGPU_GFXHUB(0), + flush_type); } } - - return 0; } static uint64_t gmc_v11_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring, @@ -498,8 +450,7 @@ static void gmc_v11_0_get_vm_pde(struct amdgpu_device *adev, int level, uint64_t *addr, uint64_t *flags) { if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM)) - *addr = adev->vm_manager.vram_base_offset + *addr - - adev->gmc.vram_start; + *addr = amdgpu_gmc_vram_mc2pa(adev, *addr); BUG_ON(*addr & 0xFFFF00000000003FULL); if (!adev->gmc.translate_further) @@ -542,6 +493,7 @@ static void gmc_v11_0_get_vm_pte(struct amdgpu_device *adev, } if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT | + AMDGPU_GEM_CREATE_EXT_COHERENT | AMDGPU_GEM_CREATE_UNCACHED)) *flags = (*flags & ~AMDGPU_PTE_MTYPE_NV10_MASK) | AMDGPU_PTE_MTYPE_NV10(MTYPE_UC); @@ -587,7 +539,7 @@ static void gmc_v11_0_set_gmc_funcs(struct amdgpu_device *adev) static void gmc_v11_0_set_umc_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[UMC_HWIP][0]) { + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { case IP_VERSION(8, 10, 0): adev->umc.channel_inst_num = UMC_V8_10_CHANNEL_INSTANCE_NUM; adev->umc.umc_inst_num = UMC_V8_10_UMC_INSTANCE_NUM; @@ -610,13 +562,16 @@ static void gmc_v11_0_set_umc_funcs(struct amdgpu_device *adev) static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(3, 0, 1): adev->mmhub.funcs = &mmhub_v3_0_1_funcs; break; case IP_VERSION(3, 0, 2): adev->mmhub.funcs = &mmhub_v3_0_2_funcs; break; + case IP_VERSION(3, 3, 0): + adev->mmhub.funcs = &mmhub_v3_3_funcs; + break; default: adev->mmhub.funcs = &mmhub_v3_0_funcs; break; @@ -625,10 +580,13 @@ static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev) static void gmc_v11_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 3): adev->gfxhub.funcs = &gfxhub_v3_0_3_funcs; break; + case IP_VERSION(11, 5, 0): + adev->gfxhub.funcs = &gfxhub_v11_5_0_funcs; + break; default: adev->gfxhub.funcs = &gfxhub_v3_0_funcs; break; @@ -679,9 +637,13 @@ static void gmc_v11_0_vram_gtt_location(struct amdgpu_device *adev, base = adev->mmhub.funcs->get_fb_location(adev); + amdgpu_gmc_set_agp_default(adev, mc); amdgpu_gmc_vram_location(adev, &adev->gmc, base); - amdgpu_gmc_gart_location(adev, mc); - amdgpu_gmc_agp_location(adev, mc); + amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_HIGH); + if (!amdgpu_sriov_vf(adev) && + (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(11, 5, 0)) && + (amdgpu_agp == 1)) + amdgpu_gmc_agp_location(adev, mc); /* base offset of vram pages */ if (amdgpu_sriov_vf(adev)) @@ -766,6 +728,8 @@ static int gmc_v11_0_sw_init(void *handle) adev->mmhub.funcs->init(adev); + adev->gfxhub.funcs->init(adev); + spin_lock_init(&adev->gmc.invalidate_lock); r = amdgpu_atomfirmware_get_vram_info(adev, @@ -775,12 +739,13 @@ static int gmc_v11_0_sw_init(void *handle) adev->gmc.vram_type = vram_type; adev->gmc.vram_vendor = vram_vendor; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 3): case IP_VERSION(11, 0, 4): + case IP_VERSION(11, 5, 0): set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask); set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask); /* @@ -935,8 +900,10 @@ static int gmc_v11_0_gart_enable(struct amdgpu_device *adev) static int gmc_v11_0_hw_init(void *handle) { - int r; struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int r; + + adev->gmc.flush_pasid_uses_kiq = !amdgpu_emu_mode; /* The sequence of these two function calls matters.*/ gmc_v11_0_init_golden_registers(adev); @@ -974,6 +941,11 @@ static int gmc_v11_0_hw_fini(void *handle) } amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0); + + if (adev->gmc.ecc_irq.funcs && + amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC)) + amdgpu_irq_put(adev, &adev->gmc.ecc_irq, 0); + gmc_v11_0_gart_disable(adev); return 0; |