From: Peter Xu <peterx@redhat.com>
To: "Liu, Yi L" <yi.l.liu@intel.com>
Cc: "qemu-devel@nongnu.org" <qemu-devel@nongnu.org>,
"alex.williamson@redhat.com" <alex.williamson@redhat.com>,
"eric.auger@redhat.com" <eric.auger@redhat.com>,
"pbonzini@redhat.com" <pbonzini@redhat.com>,
"mst@redhat.com" <mst@redhat.com>,
"david@gibson.dropbear.id.au" <david@gibson.dropbear.id.au>,
"Tian, Kevin" <kevin.tian@intel.com>,
"Tian, Jun J" <jun.j.tian@intel.com>,
"Sun, Yi Y" <yi.y.sun@intel.com>,
"kvm@vger.kernel.org" <kvm@vger.kernel.org>,
"Wu, Hao" <hao.wu@intel.com>,
"jean-philippe@linaro.org" <jean-philippe@linaro.org>,
Jacob Pan <jacob.jun.pan@linux.intel.com>,
Yi Sun <yi.y.sun@linux.intel.com>,
Richard Henderson <rth@twiddle.net>
Subject: Re: [PATCH v1 14/22] intel_iommu: bind/unbind guest page table to host
Date: Wed, 25 Mar 2020 10:56:46 -0400 [thread overview]
Message-ID: <20200325145646.GB354390@xz-x1> (raw)
In-Reply-To: <A2975661238FB949B60364EF0F2C25743A2021CD@SHSMSX104.ccr.corp.intel.com>
On Wed, Mar 25, 2020 at 12:42:58PM +0000, Liu, Yi L wrote:
> > From: Peter Xu
> > Sent: Wednesday, March 25, 2020 1:47 AM
> > To: Liu, Yi L <yi.l.liu@intel.com>
> > Subject: Re: [PATCH v1 14/22] intel_iommu: bind/unbind guest page table to host
> >
> > On Sun, Mar 22, 2020 at 05:36:11AM -0700, Liu Yi L wrote:
> > > This patch captures the guest PASID table entry modifications and
> > > propagates the changes to host to setup dual stage DMA 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 the key
> > > part of vSVA support, and also a key to support IOVA over 1st- level
> > > page table for Intel VT-d in virtualization environment.
> > >
> > > Cc: Kevin Tian <kevin.tian@intel.com>
> > > Cc: Jacob Pan <jacob.jun.pan@linux.intel.com>
> > > Cc: Peter Xu <peterx@redhat.com>
> > > Cc: Yi Sun <yi.y.sun@linux.intel.com>
> > > Cc: Paolo Bonzini <pbonzini@redhat.com>
> > > Cc: Richard Henderson <rth@twiddle.net>
> > > Signed-off-by: Liu Yi L <yi.l.liu@intel.com>
> > > ---
> > > hw/i386/intel_iommu.c | 98
> > +++++++++++++++++++++++++++++++++++++++---
> > > hw/i386/intel_iommu_internal.h | 25 +++++++++++
> > > 2 files changed, 118 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c index
> > > c985cae..0423c83 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;
> > > @@ -1856,6 +1867,81 @@ static void
> > vtd_context_global_invalidate(IntelIOMMUState *s)
> > > vtd_iommu_replay_all(s);
> > > }
> > >
> > > +/**
> > > + * Caller should hold iommu_lock.
> > > + */
> > > +static int vtd_bind_guest_pasid(IntelIOMMUState *s, VTDBus *vtd_bus,
> > > + int devfn, int pasid, VTDPASIDEntry *pe,
> > > + VTDPASIDOp op) {
> > > + VTDHostIOMMUContext *vtd_dev_icx;
> > > + HostIOMMUContext *host_icx;
> > > + DualIOMMUStage1BindData *bind_data;
> > > + struct iommu_gpasid_bind_data *g_bind_data;
> > > + int ret = -1;
> > > +
> > > + vtd_dev_icx = vtd_bus->dev_icx[devfn];
> > > + if (!vtd_dev_icx) {
> > > + return -EINVAL;
> > > + }
> > > +
> > > + host_icx = vtd_dev_icx->host_icx;
> > > + if (!host_icx) {
> > > + return -EINVAL;
> > > + }
> > > +
> > > + if (!(host_icx->stage1_formats
> > > + & IOMMU_PASID_FORMAT_INTEL_VTD)) {
> > > + error_report_once("IOMMU Stage 1 format is not
> > > + compatible!\n");
> >
> > Shouldn't we fail with this?
>
> oh, yes. no need to go further though host should also fail it.
>
> > > + }
> > > +
> > > + bind_data = g_malloc0(sizeof(*bind_data));
> > > + bind_data->pasid = pasid;
> > > + g_bind_data = &bind_data->bind_data.gpasid_bind;
> > > +
> > > + g_bind_data->flags = 0;
> > > + g_bind_data->vtd.flags = 0;
> > > + switch (op) {
> > > + case VTD_PASID_BIND:
> > > + case VTD_PASID_UPDATE:
> >
> > Is VTD_PASID_UPDATE used anywhere?
> >
> > But since it's called "UPDATE"... I really want to confirm with you that the bind() to
> > the kernel will handle the UPDATE case, right? I mean, we need to unbind first if
> > there is an existing pgtable pointer.
>
> I guess you mean host kernel. right? Actually, it's fine. host kernel
> only needs to fill in the latest pgtable pointer and permission configs
> to the pasid entry and then issue a cache invalidation. No need to do
> unbind firstly since kernel always needs to flush cache after modifying
> a pasid entry (includes valid->valid).
Great.
>
> >
> > If the answer is yes, then I think we're good, but we really need to comment it
> > somewhere about the fact.
> >
> > > + g_bind_data->version = IOMMU_UAPI_VERSION;
> > > + 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]);
> > > + ret = host_iommu_ctx_bind_stage1_pgtbl(host_icx, bind_data);
> > > + break;
> > > + case VTD_PASID_UNBIND:
> > > + g_bind_data->version = IOMMU_UAPI_VERSION;
> > > + g_bind_data->format = IOMMU_PASID_FORMAT_INTEL_VTD;
> > > + g_bind_data->gpgd = 0;
> > > + g_bind_data->addr_width = 0;
> > > + g_bind_data->hpasid = pasid;
> > > + g_bind_data->gpasid = pasid;
> > > + g_bind_data->flags |= IOMMU_SVA_GPASID_VAL;
> > > + ret = host_iommu_ctx_unbind_stage1_pgtbl(host_icx, bind_data);
> > > + break;
> > > + default:
> > > + error_report_once("Unknown VTDPASIDOp!!!\n");
> > > + break;
> > > + }
> > > +
> > > + g_free(bind_data);
> > > +
> > > + return ret;
> > > +}
> > > +
> > > /* Do a context-cache device-selective invalidation.
> > > * @func_mask: FM field after shifting
> > > */
> > > @@ -2481,10 +2567,10 @@ static inline void
> > > vtd_fill_in_pe_in_cache(IntelIOMMUState *s,
> > >
> > > pc_entry->pasid_entry = *pe;
> > > pc_entry->pasid_cache_gen = s->pasid_cache_gen;
> > > - /*
> > > - * TODO:
> > > - * - send pasid bind to host for passthru devices
> > > - */
> > > + vtd_bind_guest_pasid(s, vtd_pasid_as->vtd_bus,
> > > + vtd_pasid_as->devfn,
> > > + vtd_pasid_as->pasid,
> > > + pe, VTD_PASID_BIND);
> > > }
> > >
> > > /**
> > > @@ -2574,11 +2660,13 @@ static gboolean vtd_flush_pasid(gpointer key,
> > gpointer value,
> > > * - when pasid-base-iotlb(piotlb) infrastructure is ready,
> > > * should invalidate QEMU piotlb togehter with this change.
> > > */
> > > +
> > > return false;
> > > remove:
> > > + vtd_bind_guest_pasid(s, vtd_bus, devfn,
> > > + pasid, NULL, VTD_PASID_UNBIND);
> > > /*
> > > * TODO:
> > > - * - send pasid bind to host for passthru devices
> > > * - when pasid-base-iotlb(piotlb) infrastructure is ready,
> > > * should invalidate QEMU piotlb togehter with this change.
> > > */
> > > diff --git a/hw/i386/intel_iommu_internal.h
> > > b/hw/i386/intel_iommu_internal.h index 01fd95c..4451acf 100644
> > > --- a/hw/i386/intel_iommu_internal.h
> > > +++ b/hw/i386/intel_iommu_internal.h
> > > @@ -516,6 +516,20 @@ typedef struct VTDRootEntry VTDRootEntry;
> > > #define VTD_SM_CONTEXT_ENTRY_RSVD_VAL0(aw) (0x1e0ULL |
> > ~VTD_HAW_MASK(aw))
> > > #define VTD_SM_CONTEXT_ENTRY_RSVD_VAL1 0xffffffffffe00000ULL
> > >
> > > +enum VTD_DUAL_STAGE_UAPI {
> > > + UAPI_BIND_GPASID,
> > > + UAPI_NUM
> > > +};
> > > +typedef enum VTD_DUAL_STAGE_UAPI VTD_DUAL_STAGE_UAPI;
> > > +
> > > +enum VTDPASIDOp {
> > > + VTD_PASID_BIND,
> > > + VTD_PASID_UNBIND,
> > > + VTD_PASID_UPDATE,
> >
> > Same here (whether to drop?).
> >
> If above reply doesn't make sense, may drop it.
Your reply makes perfect sense, but still, could we drop it because
it's not used? :)
I suggest drop UPDATE, then either:
- comment at VTD_PASID_BIND that when binding exists, we'll update
the entry so the caller does not need to call unbind, or,
- rename BIND to BIND_UPDATE to show that
What do you think?
--
Peter Xu
next prev parent reply other threads:[~2020-03-25 14:56 UTC|newest]
Thread overview: 80+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-03-22 12:35 [PATCH v1 00/22] intel_iommu: expose Shared Virtual Addressing to VMs Liu Yi L
2020-03-22 12:35 ` [PATCH v1 01/22] scripts/update-linux-headers: Import iommu.h Liu Yi L
2020-03-22 12:35 ` [PATCH v1 02/22] header file update VFIO/IOMMU vSVA APIs Liu Yi L
2020-03-29 16:32 ` Auger Eric
2020-03-30 7:06 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 03/22] vfio: check VFIO_TYPE1_NESTING_IOMMU support Liu Yi L
2020-03-22 12:36 ` [PATCH v1 04/22] hw/iommu: introduce HostIOMMUContext Liu Yi L
2020-03-23 20:58 ` Peter Xu
2020-03-24 10:00 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 05/22] hw/pci: modify pci_setup_iommu() to set PCIIOMMUOps Liu Yi L
2020-03-22 12:36 ` [PATCH v1 06/22] hw/pci: introduce pci_device_set/unset_iommu_context() Liu Yi L
2020-03-23 21:15 ` Peter Xu
2020-03-24 10:02 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 07/22] intel_iommu: add set/unset_iommu_context callback Liu Yi L
2020-03-23 21:29 ` Peter Xu
2020-03-24 11:15 ` Liu, Yi L
2020-03-24 15:24 ` Peter Xu
2020-03-25 9:37 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 08/22] vfio: init HostIOMMUContext per-container Liu Yi L
[not found] ` <20200323213943.GR127076@xz-x1>
2020-03-24 13:03 ` Liu, Yi L
2020-03-24 14:45 ` Peter Xu
2020-03-25 9:30 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 09/22] vfio/common: check PASID alloc/free availability Liu Yi L
2020-03-23 22:06 ` Peter Xu
2020-03-24 11:18 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 10/22] intel_iommu: add virtual command capability support Liu Yi L
2020-03-22 12:36 ` [PATCH v1 11/22] intel_iommu: process PASID cache invalidation Liu Yi L
2020-03-22 12:36 ` [PATCH v1 12/22] intel_iommu: add PASID cache management infrastructure Liu Yi L
2020-03-24 17:32 ` Peter Xu
2020-03-25 12:20 ` Liu, Yi L
2020-03-25 14:52 ` Peter Xu
2020-03-26 6:15 ` Liu, Yi L
2020-03-26 13:57 ` Liu, Yi L
2020-03-26 15:53 ` Peter Xu
2020-03-27 1:33 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 13/22] vfio: add bind stage-1 page table support Liu Yi L
2020-03-24 17:41 ` Peter Xu
2020-03-25 9:49 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 14/22] intel_iommu: bind/unbind guest page table to host Liu Yi L
2020-03-24 17:46 ` Peter Xu
2020-03-25 12:42 ` Liu, Yi L
2020-03-25 14:56 ` Peter Xu [this message]
2020-03-26 3:04 ` Liu, Yi L
2020-03-25 12:47 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 15/22] intel_iommu: replay guest pasid bindings " Liu Yi L
2020-03-24 18:00 ` Peter Xu
2020-03-25 13:14 ` Liu, Yi L
2020-03-25 15:06 ` Peter Xu
2020-03-26 3:17 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 16/22] intel_iommu: replay pasid binds after context cache invalidation Liu Yi L
2020-03-24 18:07 ` Peter Xu
2020-03-25 13:18 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 17/22] intel_iommu: do not pass down pasid bind for PASID #0 Liu Yi L
2020-03-24 18:13 ` Peter Xu
2020-03-25 10:42 ` Liu, Yi L
2020-03-25 15:12 ` Peter Xu
2020-03-26 2:42 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 18/22] vfio: add support for flush iommu stage-1 cache Liu Yi L
2020-03-24 18:19 ` Peter Xu
2020-03-25 10:40 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 19/22] intel_iommu: process PASID-based iotlb invalidation Liu Yi L
2020-03-24 18:26 ` Peter Xu
2020-03-25 13:36 ` Liu, Yi L
2020-03-25 15:15 ` Peter Xu
2020-03-29 11:17 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 20/22] intel_iommu: propagate PASID-based iotlb invalidation to host Liu Yi L
2020-03-24 18:34 ` Peter Xu
2020-03-25 13:21 ` Liu, Yi L
2020-03-26 5:41 ` Liu, Yi L
2020-03-26 13:02 ` Peter Xu
2020-03-26 13:22 ` Peter Xu
2020-03-26 13:33 ` Liu, Yi L
2020-03-26 13:23 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 21/22] intel_iommu: process PASID-based Device-TLB invalidation Liu Yi L
2020-03-24 18:36 ` Peter Xu
2020-03-25 9:19 ` Liu, Yi L
2020-03-22 12:36 ` [PATCH v1 22/22] intel_iommu: modify x-scalable-mode to be string option Liu Yi L
2020-03-24 18:39 ` Peter Xu
2020-03-25 13:22 ` Liu, Yi L
2020-03-22 13:25 ` [PATCH v1 00/22] intel_iommu: expose Shared Virtual Addressing to VMs no-reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200325145646.GB354390@xz-x1 \
--to=peterx@redhat.com \
--cc=alex.williamson@redhat.com \
--cc=david@gibson.dropbear.id.au \
--cc=eric.auger@redhat.com \
--cc=hao.wu@intel.com \
--cc=jacob.jun.pan@linux.intel.com \
--cc=jean-philippe@linaro.org \
--cc=jun.j.tian@intel.com \
--cc=kevin.tian@intel.com \
--cc=kvm@vger.kernel.org \
--cc=mst@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=rth@twiddle.net \
--cc=yi.l.liu@intel.com \
--cc=yi.y.sun@intel.com \
--cc=yi.y.sun@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).