From patchwork Sat Feb 22 08:07:16 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Yi Liu X-Patchwork-Id: 1242413 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=nongnu.org (client-ip=209.51.188.17; helo=lists.gnu.org; envelope-from=qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=intel.com Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 48Pgyh40Ytz9sNg for ; Sat, 22 Feb 2020 19:11:40 +1100 (AEDT) Received: from localhost ([::1]:40028 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1j5Psw-00014P-Fm for incoming@patchwork.ozlabs.org; Sat, 22 Feb 2020 03:11:38 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]:57532) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1j5Pjm-0002Bf-Jt for qemu-devel@nongnu.org; Sat, 22 Feb 2020 03:02:13 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1j5Pji-00081x-UM for qemu-devel@nongnu.org; Sat, 22 Feb 2020 03:02:10 -0500 Received: from mga17.intel.com ([192.55.52.151]:49242) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1j5Pji-0007zy-LK for qemu-devel@nongnu.org; Sat, 22 Feb 2020 03:02:06 -0500 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga006.jf.intel.com ([10.7.209.51]) by fmsmga107.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 22 Feb 2020 00:01:58 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.70,471,1574150400"; d="scan'208";a="240547695" Received: from jacob-builder.jf.intel.com ([10.7.199.155]) by orsmga006.jf.intel.com with ESMTP; 22 Feb 2020 00:01:57 -0800 From: Liu Yi L To: qemu-devel@nongnu.org, alex.williamson@redhat.com, peterx@redhat.com Subject: [RFC v3.1 15/22] intel_iommu: replay guest pasid bindings to host Date: Sat, 22 Feb 2020 00:07:16 -0800 Message-Id: <1582358843-51931-16-git-send-email-yi.l.liu@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1582358843-51931-1-git-send-email-yi.l.liu@intel.com> References: <1582358843-51931-1-git-send-email-yi.l.liu@intel.com> X-detected-operating-system: by eggs.gnu.org: Genre and OS details not recognized. X-Received-From: 192.55.52.151 X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: kevin.tian@intel.com, yi.l.liu@intel.com, Yi Sun , Eduardo Habkost , kvm@vger.kernel.org, mst@redhat.com, jun.j.tian@intel.com, eric.auger@redhat.com, yi.y.sun@intel.com, Jacob Pan , pbonzini@redhat.com, Richard Henderson , david@gibson.dropbear.id.au Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: "Qemu-devel" This patch adds guest pasid bindings replay for domain selective pasid cache invalidation(dsi) and global pasid cache invalidation by walking guest pasid table. Reason: Guest OS may flush the pasid cache with a larger granularity. e.g. guest does a svm_bind() but flush the pasid cache with global or domain selective pasid cache invalidation instead of pasid selective(psi) pasid cache invalidation. Regards to such case, it works in host. Per spec, a global or domain selective pasid cache invalidation should be able to cover what a pasid selective invalidation does. The only concern is performance deduction since dsi and global cache invalidation will flush more than psi. To align with native, vIOMMU needs emulator needs to do replay for the two invalidation granularity to reflect the latest pasid bindings in guest pasid table. Cc: Kevin Tian Cc: Jacob Pan Cc: Peter Xu Cc: Yi Sun Cc: Paolo Bonzini Cc: Richard Henderson Cc: Eduardo Habkost Signed-off-by: Liu Yi L --- hw/i386/intel_iommu.c | 183 ++++++++++++++++++++++++++++++++++++++--- hw/i386/intel_iommu_internal.h | 1 + 2 files changed, 173 insertions(+), 11 deletions(-) diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c index 8bd27b1..e7c9677 100644 --- a/hw/i386/intel_iommu.c +++ b/hw/i386/intel_iommu.c @@ -68,6 +68,8 @@ static void vtd_address_space_refresh_all(IntelIOMMUState *s); static void vtd_address_space_unmap(VTDAddressSpace *as, IOMMUNotifier *n); static void vtd_pasid_cache_reset(IntelIOMMUState *s); +static int vtd_update_pe_cache_for_dev(IntelIOMMUState *s, + VTDBus *vtd_bus, int devfn, int pasid, VTDPASIDEntry *pe); static void vtd_panic_require_caching_mode(void) { @@ -2631,6 +2633,127 @@ remove: return true; } +/** + * Constant information used during pasid table walk + @vtd_bus, @devfn: device info + * @flags: indicates if it is domain selective walk + * @did: domain ID of the pasid table walk + */ +typedef struct { + VTDBus *vtd_bus; + uint16_t devfn; +#define VTD_PASID_TABLE_DID_SEL_WALK (1ULL << 0); + uint32_t flags; + uint16_t did; +} vtd_pasid_table_walk_info; + +static bool vtd_sm_pasid_table_walk_one(IntelIOMMUState *s, + dma_addr_t pt_base, + int start, + int end, + vtd_pasid_table_walk_info *info) +{ + VTDPASIDEntry pe; + int pasid = start; + int pasid_next; + + while (pasid < end) { + pasid_next = pasid + 1; + + if (!vtd_get_pe_in_pasid_leaf_table(s, pasid, pt_base, &pe) + && vtd_pe_present(&pe)) { + if (vtd_update_pe_cache_for_dev(s, info->vtd_bus, + info->devfn, pasid, &pe)) { + error_report_once("%s, bus: %d, devfn: %d, pasid: %d", + __func__, + pci_bus_num(info->vtd_bus->bus), + info->devfn, pasid); + return false; + } + } + pasid = pasid_next; + } + return true; +} + +/* + * Currently, VT-d scalable mode pasid table is a two level table, + * this function aims to loop a range of PASIDs in a given pasid + * table to identify the pasid config in guest. + */ +static void vtd_sm_pasid_table_walk(IntelIOMMUState *s, + dma_addr_t pdt_base, + int start, + int end, + vtd_pasid_table_walk_info *info) +{ + VTDPASIDDirEntry pdire; + int pasid = start; + int pasid_next; + dma_addr_t pt_base; + + while (pasid < end) { + pasid_next = pasid + VTD_PASID_TBL_ENTRY_NUM; + if (!vtd_get_pdire_from_pdir_table(pdt_base, pasid, &pdire) + && vtd_pdire_present(&pdire)) { + pt_base = pdire.val & VTD_PASID_TABLE_BASE_ADDR_MASK; + if (!vtd_sm_pasid_table_walk_one(s, + pt_base, pasid, pasid_next, info)) { + break; + } + } + pasid = pasid_next; + } +} + +/** + * This function replay the guest pasid bindings to hots by + * walking the guest PASID table. This ensures host will have + * latest guest pasid bindings. + */ +static void vtd_replay_guest_pasid_bindings(IntelIOMMUState *s, + uint16_t *did, + bool is_dsi) +{ + VTDContextEntry ce; + VTDBus *vtd_bus; + int bus_n, devfn; + vtd_pasid_table_walk_info info; + + if (is_dsi) { + info.flags = VTD_PASID_TABLE_DID_SEL_WALK; + info.did = *did; + } + + /* + * In this replay, only needs to care about the devices which + * has iommu_context created. For the one not have iommu_context, + * it is not necessary to replay the bindings since their cache + * could be re-created in the next DMA address transaltion. + */ + for (bus_n = 0; bus_n < PCI_BUS_MAX; bus_n++) { + vtd_bus = vtd_find_as_from_bus_num(s, bus_n); + if (!vtd_bus) { + continue; + } + for (devfn = 0; devfn < PCI_DEVFN_MAX; devfn++) { + PCIDevice *dev; + + dev = vtd_bus->bus->devices[devfn]; + if (pci_device_host_iommu_context(dev) && + !vtd_dev_to_context_entry(s, bus_n, devfn, &ce)) { + info.vtd_bus = vtd_bus; + info.devfn = devfn; + vtd_sm_pasid_table_walk(s, + VTD_CE_GET_PASID_DIR_TABLE(&ce), + 0, + VTD_MAX_HPASID, + &info); + } + } + } +} + static int vtd_pasid_cache_dsi(IntelIOMMUState *s, uint16_t domain_id) { VTDPASIDCacheInfo pc_info; @@ -2649,13 +2772,14 @@ static int vtd_pasid_cache_dsi(IntelIOMMUState *s, uint16_t domain_id) vtd_iommu_unlock(s); /* - * TODO: Domain selective PASID cache invalidation - * flushes all the pasid caches within a domain. To - * be safe, after invalidating the pasid caches, emulator - * needs to replay the pasid bindings by walking guest - * pasid dir and pasid table. e.g. When the guest setup a - * new PASID entry then send a PASID DSI. + * Domain selective PASID cache invalidation flushes + * all the pasid caches within a domain. To be safe, + * after invalidating the pasid caches, emulator needs + * to replay the pasid bindings by walking guest pasid + * dir and pasid table. e.g. When the guest setup a new + * PASID entry then send a PASID DSI. */ + vtd_replay_guest_pasid_bindings(s, &domain_id, true); return 0; } @@ -2723,6 +2847,42 @@ static inline void vtd_fill_in_pe_cache( } /** + * This function updates the pasid entry cached in &vtd_pasid_as. + */ +static int vtd_update_pe_cache_for_dev(IntelIOMMUState *s, + VTDBus *vtd_bus, + int devfn, int pasid, + VTDPASIDEntry *pe) +{ + VTDPASIDAddressSpace *vtd_pasid_as; + VTDPASIDCacheEntry *pc_entry; + int ret; + + vtd_iommu_lock(s); + vtd_pasid_as = vtd_add_find_pasid_as(s, vtd_bus, + devfn, pasid); + if (!vtd_pasid_as) { + error_report_once("%s, fatal error happened!\n", __func__); + ret = -1; + goto out; + } + + pc_entry = &vtd_pasid_as->pasid_cache_entry; + if (pc_entry->pasid_cache_gen == s->pasid_cache_gen && + vtd_pasid_entry_compare(pe, &pc_entry->pasid_entry)) { + /* No need to go further as cached pasid entry is latest */ + ret = 0; + goto out; + } + + vtd_fill_in_pe_cache(vtd_pasid_as, pe); + ret = 0; +out: + vtd_iommu_unlock(s); + return ret; +} + +/** * Caller of this function should hold iommu_lock */ static void vtd_new_pasid_bind_for_dev(IntelIOMMUState *s, VTDBus *vtd_bus, @@ -2869,12 +3029,13 @@ static int vtd_pasid_cache_gsi(IntelIOMMUState *s) vtd_iommu_unlock(s); /* - * TODO: Global PASID cache invalidation may be - * flushes all the pasid caches. To be safe, after - * invalidating the pasid caches, emulator needs - * to replay the pasid bindings by walking guest - * pasid dir and pasid table. + * Global PASID cache invalidation flushes all + * the pasid caches. To be safe, after invalidating + * the pasid caches, emulator needs to replay the + * pasid bindings by walking guest pasid dir and + * pasid table. */ + vtd_replay_guest_pasid_bindings(s, NULL, false); return 0; } diff --git a/hw/i386/intel_iommu_internal.h b/hw/i386/intel_iommu_internal.h index 9ee5856..46cec5c 100644 --- a/hw/i386/intel_iommu_internal.h +++ b/hw/i386/intel_iommu_internal.h @@ -554,6 +554,7 @@ typedef struct VTDPASIDCacheInfo VTDPASIDCacheInfo; #define VTD_PASID_TABLE_BITS_MASK (0x3fULL) #define VTD_PASID_TABLE_INDEX(pasid) ((pasid) & VTD_PASID_TABLE_BITS_MASK) #define VTD_PASID_ENTRY_FPD (1ULL << 1) /* Fault Processing Disable */ +#define VTD_PASID_TBL_ENTRY_NUM (1ULL << 6) /* PASID Granular Translation Type Mask */ #define VTD_PASID_ENTRY_P 1ULL