> From: Peter Xu [mailto:peterx@xxxxxxxxxx] > Sent: Tuesday, November 5, 2019 4:26 AM > To: Liu, Yi L <yi.l.liu@xxxxxxxxx> > Subject: Re: [RFC v2 15/22] intel_iommu: bind/unbind guest page table to host > > On Thu, Oct 24, 2019 at 08:34:36AM -0400, Liu Yi L wrote: > > This patch captures the guest PASID table entry modifications and > > propagates the changes to host to setup nested translation. The > > guest page table is configured as 1st level page table (GVA->GPA) > > whose translation result would further go through host VT-d 2nd > > level page table(GPA->HPA) under nested translation mode. This is > > a key part of vSVA support. > > > > Cc: Kevin Tian <kevin.tian@xxxxxxxxx> > > Cc: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx> > > Cc: Peter Xu <peterx@xxxxxxxxxx> > > Cc: Yi Sun <yi.y.sun@xxxxxxxxxxxxxxx> > > Signed-off-by: Liu Yi L <yi.l.liu@xxxxxxxxx> > > --- > > hw/i386/intel_iommu.c | 81 > ++++++++++++++++++++++++++++++++++++++++++ > > hw/i386/intel_iommu_internal.h | 20 +++++++++++ > > 2 files changed, 101 insertions(+) > > > > diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c > > index d8827c9..793b0de 100644 > > --- a/hw/i386/intel_iommu.c > > +++ b/hw/i386/intel_iommu.c > > @@ -41,6 +41,7 @@ > > #include "migration/vmstate.h" > > #include "trace.h" > > #include "qemu/jhash.h" > > +#include <linux/iommu.h> > > > > /* context entry operations */ > > #define VTD_CE_GET_RID2PASID(ce) \ > > @@ -695,6 +696,16 @@ static inline uint16_t > vtd_pe_get_domain_id(VTDPASIDEntry *pe) > > return VTD_SM_PASID_ENTRY_DID((pe)->val[1]); > > } > > > > +static inline uint32_t vtd_pe_get_fl_aw(VTDPASIDEntry *pe) > > +{ > > + return 48 + ((pe->val[2] >> 2) & VTD_SM_PASID_ENTRY_FLPM) * 9; > > +} > > + > > +static inline dma_addr_t vtd_pe_get_flpt_base(VTDPASIDEntry *pe) > > +{ > > + return pe->val[2] & VTD_SM_PASID_ENTRY_FLPTPTR; > > +} > > + > > static inline bool vtd_pdire_present(VTDPASIDDirEntry *pdire) > > { > > return pdire->val & 1; > > @@ -1850,6 +1861,67 @@ static void > vtd_context_global_invalidate(IntelIOMMUState *s) > > vtd_iommu_replay_all(s); > > } > > > > +static void vtd_bind_guest_pasid(IntelIOMMUState *s, VTDBus *vtd_bus, > > + int devfn, int pasid, VTDPASIDEntry *pe, VTDPASIDOp op) > > +{ > > +#ifdef __linux__ > > + VTDIOMMUContext *vtd_ic; > > + IOMMUCTXEventData event_data; > > + IOMMUCTXPASIDBindData bind; > > + struct iommu_gpasid_bind_data *g_bind_data; > > + > > + vtd_ic = vtd_bus->dev_ic[devfn]; > > + if (!vtd_ic) { > > + return; > > + } > > + > > + g_bind_data = g_malloc0(sizeof(*g_bind_data)); > > + bind.flag = 0; > > + g_bind_data->flags = 0; > > + g_bind_data->vtd.flags = 0; > > + switch (op) { > > + case VTD_PASID_BIND: > > + case VTD_PASID_UPDATE: > > + g_bind_data->version = IOMMU_GPASID_BIND_VERSION_1; > > + g_bind_data->format = IOMMU_PASID_FORMAT_INTEL_VTD; > > + g_bind_data->gpgd = vtd_pe_get_flpt_base(pe); > > + g_bind_data->addr_width = vtd_pe_get_fl_aw(pe); > > + g_bind_data->hpasid = pasid; > > + g_bind_data->gpasid = pasid; > > + g_bind_data->flags |= IOMMU_SVA_GPASID_VAL; > > + g_bind_data->vtd.flags = > > + (VTD_SM_PASID_ENTRY_SRE_BIT(pe->val[2]) ? 1 : 0) > > + | (VTD_SM_PASID_ENTRY_EAFE_BIT(pe->val[2]) ? 1 : 0) > > + | (VTD_SM_PASID_ENTRY_PCD_BIT(pe->val[1]) ? 1 : 0) > > + | (VTD_SM_PASID_ENTRY_PWT_BIT(pe->val[1]) ? 1 : 0) > > + | (VTD_SM_PASID_ENTRY_EMTE_BIT(pe->val[1]) ? 1 : 0) > > + | (VTD_SM_PASID_ENTRY_CD_BIT(pe->val[1]) ? 1 : 0); > > + g_bind_data->vtd.pat = VTD_SM_PASID_ENTRY_PAT(pe->val[1]); > > + g_bind_data->vtd.emt = VTD_SM_PASID_ENTRY_EMT(pe->val[1]); > > + bind.flag |= IOMMU_CTX_BIND_PASID; > > + break; > > + > > + case VTD_PASID_UNBIND: > > + g_bind_data->gpgd = 0; > > + g_bind_data->addr_width = 0; > > + g_bind_data->hpasid = pasid; > > + bind.flag |= IOMMU_CTX_UNBIND_PASID; > > + break; > > + > > + default: > > + printf("Unknown VTDPASIDOp!!\n"); > > Please don't use printf().. Here assert() suits. Will correct it. Thanks. > > > + break; > > + } > > + if (bind.flag) { > > Will this be untrue? If not, assert() works too. yes, it is possible. If an unknown VTDPASIDOp, then no switch case will initiate bind.flag. > > + event_data.event = IOMMU_CTX_EVENT_PASID_BIND; > > + bind.data = g_bind_data; > > + event_data.data = &bind; > > + iommu_ctx_event_notify(&vtd_ic->iommu_context, &event_data); > > + } > > + g_free(g_bind_data); > > +#endif > > +} > > + > > /* Do a context-cache device-selective invalidation. > > * @func_mask: FM field after shifting > > */ > > @@ -2528,12 +2600,17 @@ static gboolean vtd_flush_pasid(gpointer key, > gpointer value, > > pc_entry->pasid_cache_gen = s->pasid_cache_gen; > > if (!vtd_pasid_entry_compare(&pe, &pc_entry->pasid_entry)) { > > pc_entry->pasid_entry = pe; > > + vtd_bind_guest_pasid(s, vtd_bus, devfn, > > + pasid, &pe, VTD_PASID_UPDATE); > > /* > > * TODO: when pasid-base-iotlb(piotlb) infrastructure is > > * ready, should invalidate QEMU piotlb togehter with this > > * change. > > */ > > } > > + } else { > > + vtd_bind_guest_pasid(s, vtd_bus, devfn, > > + pasid, NULL, VTD_PASID_UNBIND); > > Please see the reply in the other thread on vtd_flush_pasid(). I've > filled in where I feel like this UNBIND should exist, I feel like your > current code could miss some places where you should unbind but didn't. I've replied in that thread regards to your comments. May you reconsider it here. Hope, it suits what you thought. If still something missed, pls feel free to point out. Regards, Yi Liu
