From: Yi Liu <yi.l.liu@intel.com>
To: Zhenzhong Duan <zhenzhong.duan@intel.com>, <qemu-devel@nongnu.org>
Cc: <alex.williamson@redhat.com>, <clg@redhat.com>,
<eric.auger@redhat.com>, <mst@redhat.com>, <jasowang@redhat.com>,
<peterx@redhat.com>, <ddutile@redhat.com>, <jgg@nvidia.com>,
<nicolinc@nvidia.com>, <skolothumtho@nvidia.com>,
<joao.m.martins@oracle.com>, <clement.mathieu--drif@eviden.com>,
<kevin.tian@intel.com>, <chao.p.peng@intel.com>
Subject: Re: [PATCH v6 12/22] intel_iommu: Handle PASID cache invalidation
Date: Sun, 12 Oct 2025 22:58:18 +0800 [thread overview]
Message-ID: <42881757-3e8d-437e-80e6-aa2d409523f6@intel.com> (raw)
In-Reply-To: <20250918085803.796942-13-zhenzhong.duan@intel.com>
On 2025/9/18 16:57, Zhenzhong Duan wrote:
> This adds PASID cache sync for RID_PASID, non-RID_PASID isn't supported.
>
> Adds an new entry VTDPASIDCacheEntry in VTDAddressSpace to cache the pasid
> entry and track PASID usage and future PASID tagged DMA address translation
> support in vIOMMU.
>
> When guest triggers pasid cache invalidation, QEMU will capture it and
> update or invalidate pasid cache.
>
> vIOMMU emulator could figure out the reason by fetching latest guest pasid
> entry in memory and compare it with cached PASID entry if it's valid.
>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> hw/i386/intel_iommu_internal.h | 19 +++-
> include/hw/i386/intel_iommu.h | 6 ++
> hw/i386/intel_iommu.c | 157 ++++++++++++++++++++++++++++++---
> hw/i386/trace-events | 3 +
> 4 files changed, 173 insertions(+), 12 deletions(-)
>
> diff --git a/hw/i386/intel_iommu_internal.h b/hw/i386/intel_iommu_internal.h
> index 9cdc8d5dbb..d400bcee21 100644
> --- a/hw/i386/intel_iommu_internal.h
> +++ b/hw/i386/intel_iommu_internal.h
> @@ -316,6 +316,7 @@ typedef enum VTDFaultReason {
> * request while disabled */
> VTD_FR_IR_SID_ERR = 0x26, /* Invalid Source-ID */
>
> + VTD_FR_RTADDR_INV_TTM = 0x31, /* Invalid TTM in RTADDR */
> /* PASID directory entry access failure */
> VTD_FR_PASID_DIR_ACCESS_ERR = 0x50,
> /* The Present(P) field of pasid directory entry is 0 */
> @@ -493,6 +494,15 @@ typedef union VTDInvDesc VTDInvDesc;
> #define VTD_INV_DESC_PIOTLB_RSVD_VAL0 0xfff000000000f1c0ULL
> #define VTD_INV_DESC_PIOTLB_RSVD_VAL1 0xf80ULL
>
> +/* PASID-cache Invalidate Descriptor (pc_inv_dsc) fields */
> +#define VTD_INV_DESC_PASIDC_G(x) extract64((x)->val[0], 4, 2)
> +#define VTD_INV_DESC_PASIDC_G_DSI 0
> +#define VTD_INV_DESC_PASIDC_G_PASID_SI 1
> +#define VTD_INV_DESC_PASIDC_G_GLOBAL 3
> +#define VTD_INV_DESC_PASIDC_DID(x) extract64((x)->val[0], 16, 16)
> +#define VTD_INV_DESC_PASIDC_PASID(x) extract64((x)->val[0], 32, 20)
> +#define VTD_INV_DESC_PASIDC_RSVD_VAL0 0xfff000000000f1c0ULL
> +
> /* Information about page-selective IOTLB invalidate */
> struct VTDIOTLBPageInvInfo {
> uint16_t domain_id;
> @@ -552,6 +562,13 @@ 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
>
> +typedef struct VTDPASIDCacheInfo {
> + uint8_t type;
> + uint16_t did;
> + uint32_t pasid;
> + bool reset;
> +} VTDPASIDCacheInfo;
> +
> /* PASID Table Related Definitions */
> #define VTD_PASID_DIR_BASE_ADDR_MASK (~0xfffULL)
> #define VTD_PASID_TABLE_BASE_ADDR_MASK (~0xfffULL)
> @@ -573,7 +590,7 @@ typedef struct VTDRootEntry VTDRootEntry;
> #define VTD_SM_PASID_ENTRY_PT (4ULL << 6)
>
> #define VTD_SM_PASID_ENTRY_AW 7ULL /* Adjusted guest-address-width */
> -#define VTD_SM_PASID_ENTRY_DID(val) ((val) & VTD_DOMAIN_ID_MASK)
> +#define VTD_SM_PASID_ENTRY_DID(x) extract64((x)->val[1], 0, 16)
>
> #define VTD_SM_PASID_ENTRY_FSPM 3ULL
> #define VTD_SM_PASID_ENTRY_FSPTPTR (~0xfffULL)
> diff --git a/include/hw/i386/intel_iommu.h b/include/hw/i386/intel_iommu.h
> index 3351892da0..ff01e5c82d 100644
> --- a/include/hw/i386/intel_iommu.h
> +++ b/include/hw/i386/intel_iommu.h
> @@ -95,6 +95,11 @@ struct VTDPASIDEntry {
> uint64_t val[8];
> };
>
> +typedef struct VTDPASIDCacheEntry {
> + struct VTDPASIDEntry pasid_entry;
> + bool valid;
> +} VTDPASIDCacheEntry;
> +
> struct VTDAddressSpace {
> PCIBus *bus;
> uint8_t devfn;
> @@ -107,6 +112,7 @@ struct VTDAddressSpace {
> MemoryRegion iommu_ir_fault; /* Interrupt region for catching fault */
> IntelIOMMUState *iommu_state;
> VTDContextCacheEntry context_cache_entry;
> + VTDPASIDCacheEntry pasid_cache_entry;
> QLIST_ENTRY(VTDAddressSpace) next;
> /* Superset of notifier flags that this address space has */
> IOMMUNotifierFlag notifier_flags;
> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
> index d37d47115a..24061f6dc6 100644
> --- a/hw/i386/intel_iommu.c
> +++ b/hw/i386/intel_iommu.c
> @@ -1614,7 +1614,7 @@ static uint16_t vtd_get_domain_id(IntelIOMMUState *s,
>
> if (s->root_scalable) {
> vtd_ce_get_pasid_entry(s, ce, &pe, pasid);
> - return VTD_SM_PASID_ENTRY_DID(pe.val[1]);
> + return VTD_SM_PASID_ENTRY_DID(&pe);
> }
>
> return VTD_CONTEXT_ENTRY_DID(ce->hi);
> @@ -3074,6 +3074,144 @@ static bool vtd_process_piotlb_desc(IntelIOMMUState *s,
> return true;
> }
>
> +static inline int vtd_dev_get_pe_from_pasid(VTDAddressSpace *vtd_as,
> + VTDPASIDEntry *pe)
> +{
> + IntelIOMMUState *s = vtd_as->iommu_state;
> + VTDContextEntry ce;
> + int ret;
> +
> + if (!s->root_scalable) {
> + return -VTD_FR_RTADDR_INV_TTM;
> + }
> +
> + ret = vtd_dev_to_context_entry(s, pci_bus_num(vtd_as->bus), vtd_as->devfn,
> + &ce);
> + if (ret) {
> + return ret;
> + }
> +
> + return vtd_ce_get_pasid_entry(s, &ce, pe, vtd_as->pasid);
> +}
> +
> +/*
> + * For each IOMMUFD backed device, update or invalidate pasid cache based on
> + * the value in memory.
> + */
> +static void vtd_pasid_cache_sync_locked(gpointer key, gpointer value,
> + gpointer user_data)
> +{
> + VTDPASIDCacheInfo *pc_info = user_data;
> + VTDAddressSpace *vtd_as = value;
> + VTDPASIDCacheEntry *pc_entry = &vtd_as->pasid_cache_entry;
> + VTDPASIDEntry pe;
> + uint16_t did;
> +
> + /* Ignore emulated device or legacy VFIO backed device */
> + if (!vtd_find_hiod_iommufd(vtd_as)) {
> + return;
> + }
> +
> + /* non-RID_PASID isn't supported yet */
> + assert(vtd_as->pasid == PCI_NO_PASID);
> +
> + if (vtd_dev_get_pe_from_pasid(vtd_as, &pe)) {
> + /*
> + * No valid pasid entry in guest memory. e.g. pasid entry was modified
> + * to be either all-zero or non-present. Either case means existing
> + * pasid cache should be invalidated.
> + */
> + pc_entry->valid = false;
> + return;
> + }
> +
> + /*
> + * VTD_INV_DESC_PASIDC_G_DSI and VTD_INV_DESC_PASIDC_G_PASID_SI require
> + * DID check. If DID doesn't match the value in cache or memory, then
> + * it's not a pasid entry we want to invalidate.
I think comparing DID applies to the case in which pc_entry->valid is
true. If pc_entry->valid is false, this means no cached pc_entry yet. If
pe in guest memory is valid, the pc_entry should be updated/set hence
the bind_pasid operation (added in later patch) would be conducted.
> + */
> + switch (pc_info->type) {
> + case VTD_INV_DESC_PASIDC_G_PASID_SI:
> + case VTD_INV_DESC_PASIDC_G_DSI:
> + if (pc_entry->valid) {
> + did = VTD_SM_PASID_ENTRY_DID(&pc_entry->pasid_entry);
> + if (pc_info->did == did) {
> + break;
> + }
> + }
> + did = VTD_SM_PASID_ENTRY_DID(&pe);
> + if (pc_info->did == did) {
> + break;
> + }
> + return;
> + }
> +
> + pc_entry->pasid_entry = pe;
> + pc_entry->valid = true;
> +}
> +
> +static void vtd_pasid_cache_sync(IntelIOMMUState *s, VTDPASIDCacheInfo *pc_info)
> +{
> + if (!s->fsts || !s->root_scalable || !s->dmar_enabled) {
> + return;
> + }
> +
> + vtd_iommu_lock(s);
> + g_hash_table_foreach(s->vtd_address_spaces, vtd_pasid_cache_sync_locked,
> + pc_info);
> + vtd_iommu_unlock(s);
> +}
> +
> +static bool vtd_process_pasid_desc(IntelIOMMUState *s,
> + VTDInvDesc *inv_desc)
> +{
> + uint16_t did;
> + uint32_t pasid;
> + VTDPASIDCacheInfo pc_info = {};
> + uint64_t mask[4] = {VTD_INV_DESC_PASIDC_RSVD_VAL0, VTD_INV_DESC_ALL_ONE,
> + VTD_INV_DESC_ALL_ONE, VTD_INV_DESC_ALL_ONE};
> +
> + if (!vtd_inv_desc_reserved_check(s, inv_desc, mask, true,
> + __func__, "pasid cache inv")) {
> + return false;
> + }
> +
> + did = VTD_INV_DESC_PASIDC_DID(inv_desc);
> + pasid = VTD_INV_DESC_PASIDC_PASID(inv_desc);
> + pc_info.type = VTD_INV_DESC_PASIDC_G(inv_desc);
> +
> + switch (pc_info.type) {
> + case VTD_INV_DESC_PASIDC_G_DSI:
> + trace_vtd_inv_desc_pasid_cache_dsi(did);
> + pc_info.did = did;
> + break;
> +
> + case VTD_INV_DESC_PASIDC_G_PASID_SI:
> + /* PASID selective implies a DID selective */
> + trace_vtd_inv_desc_pasid_cache_psi(did, pasid);
> + /* Currently non-RID_PASID invalidation requests are ignored */
I'm a bit doubting if this is safe given the ATS path (for emulated
device) is merged. ATS path supports non-RID_PASID if emulated device
has PASID cap. The lucky thing is that the ATS path does not have
pasid level cache. So skipping invalidation for non-RID_PASID is not
harmful so far. Just a note to other reviewers although I didn't see a
problem here.
> + if (pasid != RID_PASID) {
> + return true;
> + }
> + pc_info.did = did;
> + pc_info.pasid = pasid;
> + break;
> +
Regards,
Yi Liu
next prev parent reply other threads:[~2025-10-12 14:52 UTC|newest]
Thread overview: 62+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-18 8:57 [PATCH v6 00/22] intel_iommu: Enable first stage translation for passthrough device Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 01/22] intel_iommu: Rename vtd_ce_get_rid2pasid_entry to vtd_ce_get_pasid_entry Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 02/22] intel_iommu: Delete RPS capability related supporting code Zhenzhong Duan
2025-09-30 13:49 ` Eric Auger
2025-10-09 10:10 ` Duan, Zhenzhong
2025-10-12 12:30 ` Yi Liu
2025-09-18 8:57 ` [PATCH v6 03/22] intel_iommu: Update terminology to match VTD spec Zhenzhong Duan
2025-09-30 7:45 ` Eric Auger
2025-10-12 12:30 ` Yi Liu
2025-10-13 6:20 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 04/22] hw/pci: Export pci_device_get_iommu_bus_devfn() and return bool Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 05/22] hw/pci: Introduce pci_device_get_viommu_flags() Zhenzhong Duan
2025-09-23 18:47 ` Nicolin Chen
2025-09-24 7:05 ` Duan, Zhenzhong
2025-09-24 8:21 ` Nicolin Chen
2025-09-26 2:54 ` Duan, Zhenzhong
2025-09-30 13:55 ` Eric Auger
2025-10-12 12:26 ` Yi Liu
2025-10-13 6:24 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 06/22] intel_iommu: Implement get_viommu_flags() callback Zhenzhong Duan
2025-10-12 12:28 ` Yi Liu
2025-10-13 6:26 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 07/22] intel_iommu: Introduce a new structure VTDHostIOMMUDevice Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 08/22] vfio/iommufd: Force creating nesting parent HWPT Zhenzhong Duan
2025-09-30 14:19 ` Eric Auger
2025-10-12 12:33 ` Yi Liu
2025-09-18 8:57 ` [PATCH v6 09/22] intel_iommu: Stick to system MR for IOMMUFD backed host device when x-fls=on Zhenzhong Duan
2025-09-30 15:04 ` Eric Auger
2025-10-09 10:10 ` Duan, Zhenzhong
2025-10-12 12:51 ` Yi Liu
2025-09-18 8:57 ` [PATCH v6 10/22] intel_iommu: Check for compatibility with IOMMUFD backed device when x-flts=on Zhenzhong Duan
2025-10-12 12:55 ` Yi Liu
2025-10-13 6:48 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 11/22] intel_iommu: Fail passthrough device under PCI bridge if x-flts=on Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 12/22] intel_iommu: Handle PASID cache invalidation Zhenzhong Duan
2025-10-12 14:58 ` Yi Liu [this message]
2025-10-13 7:37 ` Duan, Zhenzhong
2025-10-13 12:53 ` Yi Liu
2025-10-14 6:25 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 13/22] intel_iommu: Reset pasid cache when system level reset Zhenzhong Duan
2025-10-13 10:25 ` Yi Liu
2025-10-14 5:56 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 14/22] intel_iommu: Add some macros and inline functions Zhenzhong Duan
2025-10-13 10:25 ` Yi Liu
2025-10-14 6:15 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 15/22] intel_iommu: Bind/unbind guest page table to host Zhenzhong Duan
2025-10-20 13:49 ` Yi Liu
2025-10-21 8:26 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 16/22] intel_iommu: Propagate PASID-based iotlb invalidation " Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 17/22] intel_iommu: Replay all pasid bindings when either SRTP or TE bit is changed Zhenzhong Duan
2025-09-18 8:57 ` [PATCH v6 18/22] iommufd: Introduce a helper function to extract vendor capabilities Zhenzhong Duan
2025-09-23 19:45 ` Nicolin Chen
2025-09-24 8:05 ` Duan, Zhenzhong
2025-09-24 8:27 ` Nicolin Chen
2025-09-26 2:54 ` Duan, Zhenzhong
2025-09-18 8:57 ` [PATCH v6 19/22] vfio: Add a new element bypass_ro in VFIOContainerBase Zhenzhong Duan
2025-09-26 12:25 ` Cédric Le Goater
2025-09-18 8:57 ` [PATCH v6 20/22] Workaround for ERRATA_772415_SPR17 Zhenzhong Duan
2025-09-18 8:58 ` [PATCH v6 21/22] intel_iommu: Enable host device when x-flts=on in scalable mode Zhenzhong Duan
2025-09-18 8:58 ` [PATCH v6 22/22] docs/devel: Add IOMMUFD nesting documentation Zhenzhong Duan
2025-09-18 10:00 ` Cédric Le Goater
2025-09-19 2:17 ` Duan, Zhenzhong
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=42881757-3e8d-437e-80e6-aa2d409523f6@intel.com \
--to=yi.l.liu@intel.com \
--cc=alex.williamson@redhat.com \
--cc=chao.p.peng@intel.com \
--cc=clement.mathieu--drif@eviden.com \
--cc=clg@redhat.com \
--cc=ddutile@redhat.com \
--cc=eric.auger@redhat.com \
--cc=jasowang@redhat.com \
--cc=jgg@nvidia.com \
--cc=joao.m.martins@oracle.com \
--cc=kevin.tian@intel.com \
--cc=mst@redhat.com \
--cc=nicolinc@nvidia.com \
--cc=peterx@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=skolothumtho@nvidia.com \
--cc=zhenzhong.duan@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.