/*- * Copyright (c) 2011 NetApp, Inc. * Copyright (c) 2018 Intel Corporation * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include "vpci_priv.h" /** * @pre vdev != NULL */ static inline struct msix_table_entry *get_msix_table_entry(const struct pci_vdev *vdev, uint32_t index) { void *hva = hpa2hva(vdev->msix.mmio_hpa + vdev->msix.table_offset); return ((struct msix_table_entry *)hva + index); } /** * @brief Writing MSI-X Capability Structure * * @pre vdev != NULL * @pre vdev->pdev != NULL */ void write_pt_vmsix_cap_reg(struct pci_vdev *vdev, uint32_t offset, uint32_t bytes, uint32_t val) { uint32_t msgctrl; if (write_vmsix_cap_reg(vdev, offset, bytes, val)) { msgctrl = pci_vdev_read_vcfg(vdev, vdev->msix.capoff + PCIR_MSIX_CTRL, 2U); /* If MSI Enable is being set, make sure INTxDIS bit is set */ if ((msgctrl & PCIM_MSIXCTRL_MSIX_ENABLE) != 0U) { enable_disable_pci_intx(vdev->pdev->bdf, false); } pci_pdev_write_cfg(vdev->pdev->bdf, vdev->msix.capoff + PCIR_MSIX_CTRL, 2U, msgctrl); } } /** * @pre vdev != NULL */ static void mask_one_msix_vector(const struct pci_vdev *vdev, uint32_t index) { uint32_t vector_control; struct msix_table_entry *pentry = get_msix_table_entry(vdev, index); stac(); vector_control = pentry->vector_control | PCIM_MSIX_VCTRL_MASK; mmio_write32(vector_control, (void *)&(pentry->vector_control)); clac(); } /** * @pre vdev != NULL * @pre vdev->vpci != NULL * @pre vdev->pdev != NULL */ static void remap_one_vmsix_entry(const struct pci_vdev *vdev, uint32_t index) { const struct msix_table_entry *ventry; struct msix_table_entry *pentry; struct msi_info info = {}; int32_t ret; mask_one_msix_vector(vdev, index); ventry = &vdev->msix.table_entries[index]; if ((ventry->vector_control & PCIM_MSIX_VCTRL_MASK) == 0U) { info.addr.full = vdev->msix.table_entries[index].addr; info.data.full = vdev->msix.table_entries[index].data; ret = ptirq_prepare_msix_remap(vpci2vm(vdev->vpci), vdev->bdf.value, vdev->pdev->bdf.value, (uint16_t)index, &info, INVALID_IRTE_ID); if (ret == 0) { /* Write the table entry to the physical structure */ pentry = get_msix_table_entry(vdev, index); /* * PCI 3.0 Spec allows writing to Message Address and Message Upper Address * fields with a single QWORD write, but some hardware can accept 32 bits * write only */ stac(); mmio_write32((uint32_t)(info.addr.full), (void *)&(pentry->addr)); mmio_write32((uint32_t)(info.addr.full >> 32U), (void *)((char *)&(pentry->addr) + 4U)); mmio_write32(info.data.full, (void *)&(pentry->data)); mmio_write32(vdev->msix.table_entries[index].vector_control, (void *)&(pentry->vector_control)); clac(); } } } /** * @pre io_req != NULL * @pre priv_data != NULL */ static int32_t pt_vmsix_handle_table_mmio_access(struct io_request *io_req, void *priv_data) { struct acrn_mmio_request *mmio = &io_req->reqs.mmio_request; struct pci_vdev *vdev; uint32_t index; int32_t ret = 0; vdev = (struct pci_vdev *)priv_data; if (vdev->user == vdev) { index = rw_vmsix_table(vdev, io_req); if ((mmio->direction == ACRN_IOREQ_DIR_WRITE) && (index < vdev->msix.table_count)) { if (vdev->msix.is_vmsix_on_msi) { remap_one_vmsix_entry_on_msi(vdev, index); } else { remap_one_vmsix_entry(vdev, index); } } } else { ret = -EFAULT; } return ret; } /* * @pre vdev != NULL * @pre vdev->vpci != NULL */ static void vdev_pt_unmap_msix(struct pci_vdev *vdev) { uint32_t i; uint64_t addr_hi, addr_lo; struct pci_msix *msix = &vdev->msix; /* Mask all table entries */ for (i = 0U; i < msix->table_count; i++) { msix->table_entries[i].vector_control = PCIM_MSIX_VCTRL_MASK; msix->table_entries[i].addr = 0U; msix->table_entries[i].data = 0U; } if (msix->mmio_gpa != 0UL) { addr_lo = msix->mmio_gpa + msix->table_offset; addr_hi = addr_lo + (msix->table_count * MSIX_TABLE_ENTRY_SIZE); addr_lo = round_page_down(addr_lo); addr_hi = round_page_up(addr_hi); unregister_mmio_emulation_handler(vpci2vm(vdev->vpci), addr_lo, addr_hi); msix->mmio_gpa = 0UL; } } /* * @pre vdev != NULL * @pre vdev->vpci != NULL */ void vdev_pt_map_msix(struct pci_vdev *vdev, bool hold_lock) { struct pci_vbar *vbar; uint64_t addr_hi, addr_lo; struct pci_msix *msix = &vdev->msix; vbar = &vdev->vbars[msix->table_bar]; if (vbar->base_gpa != 0UL) { struct acrn_vm *vm = vpci2vm(vdev->vpci); addr_lo = vbar->base_gpa + msix->table_offset; addr_hi = addr_lo + (msix->table_count * MSIX_TABLE_ENTRY_SIZE); addr_lo = round_page_down(addr_lo); addr_hi = round_page_up(addr_hi); register_mmio_emulation_handler(vm, pt_vmsix_handle_table_mmio_access, addr_lo, addr_hi, vdev, hold_lock); ept_del_mr(vm, (uint64_t *)vm->arch_vm.nworld_eptp, addr_lo, addr_hi - addr_lo); msix->mmio_gpa = vbar->base_gpa; } } /** * @pre vdev != NULL * @pre vdev->vpci != NULL */ static void vdev_pt_unmap_mem_vbar(struct pci_vdev *vdev, uint32_t idx) { struct pci_vbar *vbar = &vdev->vbars[idx]; if (vbar->base_gpa != 0UL) { struct acrn_vm *vm = vpci2vm(vdev->vpci); ept_del_mr(vm, (uint64_t *)(vm->arch_vm.nworld_eptp), vbar->base_gpa, /* GPA (old vbar) */ vbar->size); } if ((has_msix_cap(vdev) && (idx == vdev->msix.table_bar))) { vdev_pt_unmap_msix(vdev); } } /** * @pre vdev != NULL * @pre vdev->vpci != NULL */ static void vdev_pt_map_mem_vbar(struct pci_vdev *vdev, uint32_t idx) { struct pci_vbar *vbar = &vdev->vbars[idx]; if (vbar->base_gpa != 0UL) { struct acrn_vm *vm = vpci2vm(vdev->vpci); ept_add_mr(vm, (uint64_t *)(vm->arch_vm.nworld_eptp), vbar->base_hpa, /* HPA (pbar) */ vbar->base_gpa, /* GPA (new vbar) */ vbar->size, EPT_WR | EPT_RD | EPT_UNCACHED); } if (has_msix_cap(vdev) && (idx == vdev->msix.table_bar)) { vdev_pt_map_msix(vdev, true); } } /** * @brief Allow IO bar access * @pre vdev != NULL * @pre vdev->vpci != NULL */ static void vdev_pt_allow_io_vbar(struct pci_vdev *vdev, uint32_t idx) { struct acrn_vm *vm = vpci2vm(vdev->vpci); /* For SOS, all port IO access is allowed by default, so skip SOS here */ if (!is_sos_vm(vm)) { struct pci_vbar *vbar = &vdev->vbars[idx]; if (vbar->base_gpa != 0UL) { allow_guest_pio_access(vm, (uint16_t)vbar->base_gpa, (uint32_t)(vbar->size)); } } } /** * @brief Deny IO bar access * @pre vdev != NULL * @pre vdev->vpci != NULL */ static void vdev_pt_deny_io_vbar(struct pci_vdev *vdev, uint32_t idx) { struct acrn_vm *vm = vpci2vm(vdev->vpci); /* For SOS, all port IO access is allowed by default, so skip SOS here */ if (!is_sos_vm(vm)) { struct pci_vbar *vbar = &vdev->vbars[idx]; if (vbar->base_gpa != 0UL) { deny_guest_pio_access(vm, (uint16_t)(vbar->base_gpa), (uint32_t)(vbar->size)); } } } /** * @pre vdev != NULL */ void vdev_pt_write_vbar(struct pci_vdev *vdev, uint32_t idx, uint32_t val) { struct pci_vbar *vbar = &vdev->vbars[idx]; if (is_pci_io_bar(vbar)) { vpci_update_one_vbar(vdev, idx, val, vdev_pt_allow_io_vbar, vdev_pt_deny_io_vbar); } else if (is_pci_mem_bar(vbar)) { vpci_update_one_vbar(vdev, idx, val, vdev_pt_map_mem_vbar, vdev_pt_unmap_mem_vbar); } } /** * PCI base address register (bar) virtualization: * * Virtualize the PCI bars (up to 6 bars at byte offset 0x10~0x24 for type 0 PCI device, * 2 bars at byte offset 0x10-0x14 for type 1 PCI device) of the PCI configuration space * header. * * pbar: bar for the physical PCI device (pci_pdev), the value of pbar (hpa) is assigned * by platform firmware during boot. It is assumed a valid hpa is always assigned to a * mmio pbar, hypervisor shall not change the value of a pbar. * * vbar: for each pci_pdev, it has a virtual PCI device (pci_vdev) counterpart. pci_vdev * virtualizes all the bars (called vbars). a vbar can be initialized by hypervisor by * assigning a gpa to it; if vbar has a value of 0 (unassigned), guest may assign * and program a gpa to it. The guest only sees the vbars, it will not see and can * never change the pbars. * * Hypervisor traps guest changes to the mmio vbar (gpa) to establish ept mapping * between vbar(gpa) and pbar(hpa). pbar should always align on 4K boundary. * * @param vdev Pointer to a vdev structure * @param is_sriov_bar When the first parameter vdev is a SRIOV PF vdev, the function * init_bars is used to initialize normal PCIe BARs of PF vdev if the * parameter is_sriov_bar is false, the function init_bars is used to * initialize SRIOV VF BARs of PF vdev if parameter is_sriov_bar is true * Otherwise, the parameter is_sriov_bar should be false if the first * parameter vdev is not SRIOV PF vdev * * @pre vdev != NULL * @pre vdev->vpci != NULL * @pre vdev->pdev != NULL * * @return None */ static void init_bars(struct pci_vdev *vdev, bool is_sriov_bar) { uint32_t idx, bar_cnt; struct pci_vbar *vbar; uint32_t size32, offset, lo, hi = 0U; union pci_bdf pbdf; uint64_t mask; if (is_sriov_bar) { bar_cnt = PCI_BAR_COUNT; } else { vdev->nr_bars = vdev->pdev->nr_bars; bar_cnt = vdev->nr_bars; } pbdf.value = vdev->pdev->bdf.value; for (idx = 0U; idx < bar_cnt; idx++) { if (is_sriov_bar) { vbar = &vdev->sriov.vbars[idx]; offset = sriov_bar_offset(vdev, idx); } else { vbar = &vdev->vbars[idx]; offset = pci_bar_offset(idx); } lo = pci_pdev_read_cfg(pbdf, offset, 4U); vbar->bar_type.bits = lo; if (is_pci_reserved_bar(vbar)) { continue; } mask = (is_pci_io_bar(vbar)) ? PCI_BASE_ADDRESS_IO_MASK : PCI_BASE_ADDRESS_MEM_MASK; vbar->base_hpa = (uint64_t)lo & mask; if (is_pci_mem64lo_bar(vbar)) { hi = pci_pdev_read_cfg(pbdf, offset + 4U, 4U); vbar->base_hpa |= ((uint64_t)hi << 32U); } if (vbar->base_hpa != 0UL) { pci_pdev_write_cfg(pbdf, offset, 4U, ~0U); size32 = pci_pdev_read_cfg(pbdf, offset, 4U); pci_pdev_write_cfg(pbdf, offset, 4U, lo); vbar->mask = size32 & mask; vbar->bar_type.bits &= (~mask); vbar->size = (uint64_t)size32 & mask; if (is_prelaunched_vm(vpci2vm(vdev->vpci))) { lo = (uint32_t)vdev->pci_dev_config->vbar_base[idx]; } if (is_pci_mem64lo_bar(vbar)) { idx++; if (is_sriov_bar) { offset = sriov_bar_offset(vdev, idx); } else { offset = pci_bar_offset(idx); } pci_pdev_write_cfg(pbdf, offset, 4U, ~0U); size32 = pci_pdev_read_cfg(pbdf, offset, 4U); pci_pdev_write_cfg(pbdf, offset, 4U, hi); vbar->size |= ((uint64_t)size32 << 32U); vbar->size = vbar->size & ~(vbar->size - 1UL); vbar->size = round_page_up(vbar->size); if (is_sriov_bar) { vbar = &vdev->sriov.vbars[idx]; } else { vbar = &vdev->vbars[idx]; } vbar->mask = size32; vbar->is_mem64hi = true; if (is_prelaunched_vm(vpci2vm(vdev->vpci))) { hi = (uint32_t)(vdev->pci_dev_config->vbar_base[idx - 1U] >> 32U); } /* if it is parsing SRIOV VF BARs, no need to write vdev bars */ if (!is_sriov_bar) { pci_vdev_write_vbar(vdev, idx - 1U, lo); pci_vdev_write_vbar(vdev, idx, hi); } } else { vbar->size = vbar->size & ~(vbar->size - 1UL); if (is_pci_mem32_bar(vbar)) { vbar->size = round_page_up(vbar->size); } /* if it is parsing SRIOV VF BARs, no need to write vdev bar */ if (!is_sriov_bar) { pci_vdev_write_vbar(vdev, idx, lo); } } } } /* Initialize MSIx mmio hpa and size after BARs initialization */ if (has_msix_cap(vdev) && (!is_sriov_bar)) { vdev->msix.mmio_hpa = vdev->vbars[vdev->msix.table_bar].base_hpa; vdev->msix.mmio_size = vdev->vbars[vdev->msix.table_bar].size; } } /** * @pre vdev != NULL * @pre vdev->pdev != NULL */ void init_vmsix_pt(struct pci_vdev *vdev) { struct pci_pdev *pdev = vdev->pdev; vdev->msix.capoff = pdev->msix.capoff; vdev->msix.caplen = pdev->msix.caplen; vdev->msix.table_bar = pdev->msix.table_bar; vdev->msix.table_offset = pdev->msix.table_offset; vdev->msix.table_count = pdev->msix.table_count; if (has_msix_cap(vdev)) { (void)memcpy_s((void *)&vdev->cfgdata.data_8[pdev->msix.capoff], pdev->msix.caplen, (void *)&pdev->msix.cap[0U], pdev->msix.caplen); } } /** * @pre vdev != NULL * @pre vdev->vpci != NULL */ void deinit_vmsix_pt(struct pci_vdev *vdev) { if (has_msix_cap(vdev)) { if (vdev->msix.table_count != 0U) { ptirq_remove_msix_remapping(vpci2vm(vdev->vpci), vdev->pdev->bdf.value, vdev->msix.table_count); (void)memset((void *)&vdev->msix.table_entries, 0U, sizeof(vdev->msix.table_entries)); vdev->msix.is_vmsix_on_msi_programmed = false; } } } void vdev_pt_hide_sriov_cap(struct pci_vdev *vdev) { uint32_t pre_pos = vdev->pdev->sriov.pre_pos; uint32_t pre_hdr, hdr, vhdr; pre_hdr = pci_pdev_read_cfg(vdev->pdev->bdf, pre_pos, 4U); hdr = pci_pdev_read_cfg(vdev->pdev->bdf, vdev->pdev->sriov.capoff, 4U); vhdr = pre_hdr & 0xfffffU; vhdr |= hdr & 0xfff00000U; pci_vdev_write_vcfg(vdev, pre_pos, 4U, vhdr); vdev->pdev->sriov.hide_sriov = true; pr_acrnlog("Hide sriov cap for %02x:%02x.%x", vdev->pdev->bdf.bits.b, vdev->pdev->bdf.bits.d, vdev->pdev->bdf.bits.f); } /* TODO: * The OpRegion is not 4KB aligned, while under some platforms, * it will take up to 16KB. In this case, OpRegion overlay 5 pages. * So set GPU_OPREGION_SIZE to 0x5000U(20KB) here. * * The solution that pass-thru OpRegion has potential security issue. * Will take the copy + emulation solution to expose host OpRegion to guest later. */ void passthru_gpu_opregion(struct pci_vdev *vdev) { uint32_t gpu_opregion_hpa, gpu_opregion_gpa, gpu_asls_phys; gpu_opregion_gpa = GPU_OPREGION_GPA; gpu_asls_phys = pci_pdev_read_cfg(vdev->pdev->bdf, PCIR_ASLS_CTL, 4U); gpu_opregion_hpa = gpu_asls_phys & PCIM_ASLS_OPREGION_MASK; ept_add_mr(vpci2vm(vdev->vpci), vpci2vm(vdev->vpci)->arch_vm.nworld_eptp, gpu_opregion_hpa, gpu_opregion_gpa, GPU_OPREGION_SIZE, EPT_RD | EPT_UNCACHED); pci_vdev_write_vcfg(vdev, PCIR_ASLS_CTL, 4U, gpu_opregion_gpa | (gpu_asls_phys & ~PCIM_ASLS_OPREGION_MASK)); } /* * @brief Initialize a specified passthrough vdev structure. * * The function init_vdev_pt is used to initialize a vdev structure. If a vdev structure supports * SRIOV capability that the vdev represents a SRIOV physical function(PF) virtual device, then * function init_vdev_pt can initialize PF vdev SRIOV capability if parameter is_pf_vdev is true. * * @param vdev pointer to vdev data structure * @param is_pf_vdev indicate the first parameter vdev is the data structure of a PF, which contains * the SR-IOV capability * * @pre vdev != NULL * @pre vdev->vpci != NULL * @pre vdev->pdev != NULL * * @return None */ void init_vdev_pt(struct pci_vdev *vdev, bool is_pf_vdev) { uint16_t pci_command; uint32_t offset; for (offset = 0U; offset < PCI_CFG_HEADER_LENGTH; offset += 4U) { pci_vdev_write_vcfg(vdev, offset, 4U, pci_pdev_read_cfg(vdev->pdev->bdf, offset, 4U)); } /* Initialize the vdev BARs except SRIOV VF, VF BARs are initialized directly from create_vf function */ if (vdev->phyfun == NULL) { init_bars(vdev, is_pf_vdev); init_vmsix_on_msi(vdev); if (is_prelaunched_vm(vpci2vm(vdev->vpci)) && (!is_pf_vdev)) { pci_command = (uint16_t)pci_pdev_read_cfg(vdev->pdev->bdf, PCIR_COMMAND, 2U); /* Disable INTX */ pci_command |= 0x400U; pci_pdev_write_cfg(vdev->pdev->bdf, PCIR_COMMAND, 2U, pci_command); if (vdev->pdev->bdf.value == CONFIG_GPU_SBDF) { passthru_gpu_opregion(vdev); } } } else { if (vdev->phyfun->vpci != vdev->vpci) { /* VF is assigned to a UOS */ uint32_t vid, did; vdev->nr_bars = PCI_BAR_COUNT; /* SRIOV VF Vendor ID and Device ID initialization */ vid = pci_pdev_read_cfg(vdev->phyfun->bdf, PCIR_VENDOR, 2U); did = pci_pdev_read_cfg(vdev->phyfun->bdf, (vdev->phyfun->sriov.capoff + PCIR_SRIOV_VF_DEV_ID), 2U); pci_vdev_write_vcfg(vdev, PCIR_VENDOR, 2U, vid); pci_vdev_write_vcfg(vdev, PCIR_DEVICE, 2U, did); } else { /* VF is unassinged */ uint32_t bar_idx; for (bar_idx = 0U; bar_idx < vdev->nr_bars; bar_idx++) { vdev_pt_map_mem_vbar(vdev, bar_idx); } } } if (!is_sos_vm(vpci2vm(vdev->vpci)) && (has_sriov_cap(vdev))) { vdev_pt_hide_sriov_cap(vdev); } } /* * @brief Destruct a specified passthrough vdev structure. * * The function deinit_vdev_pt is the destructor corresponding to the function init_vdev_pt. * * @param vdev pointer to vdev data structure * * @pre vdev != NULL * * @return None */ void deinit_vdev_pt(struct pci_vdev *vdev) { /* Check if the vdev is an unassigned SR-IOV VF device */ if ((vdev->phyfun != NULL) && (vdev->phyfun->vpci == vdev->vpci)) { uint32_t bar_idx; /* Delete VF MMIO from EPT table since the VF physical device has gone */ for (bar_idx = 0U; bar_idx < vdev->nr_bars; bar_idx++) { vdev_pt_unmap_mem_vbar(vdev, bar_idx); } } }