diff options
author | bsiegert <bsiegert@pkgsrc.org> | 2019-11-16 22:10:06 +0000 |
---|---|---|
committer | bsiegert <bsiegert@pkgsrc.org> | 2019-11-16 22:10:06 +0000 |
commit | 387a32675a6bf89df35eacce8df408e602f05aa9 (patch) | |
tree | 4835057d968c59a27c6003ed5b5e6d63147920f6 | |
parent | 2a831eaa34ca5c2eadedbc11252818c68f581d0c (diff) | |
download | pkgsrc-387a32675a6bf89df35eacce8df408e602f05aa9.tar.gz |
Pullup ticket #6086 - requested by bouyer
sysutils/xenkernel411: security fix
Revisions pulled up:
- sysutils/xenkernel411/Makefile 1.9-1.10
- sysutils/xenkernel411/distinfo 1.6-1.7
- sysutils/xenkernel411/patches/patch-XSA298 1.1-1.2
- sysutils/xenkernel411/patches/patch-XSA299 1.1
- sysutils/xenkernel411/patches/patch-XSA302 1.1-1.2
- sysutils/xenkernel411/patches/patch-XSA304 1.1-1.2
- sysutils/xenkernel411/patches/patch-XSA305 1.1-1.2
---
Module Name: pkgsrc
Committed By: bouyer
Date: Wed Nov 13 13:36:11 UTC 2019
Modified Files:
pkgsrc/sysutils/xenkernel411: Makefile distinfo
Added Files:
pkgsrc/sysutils/xenkernel411/patches: patch-XSA298 patch-XSA302
patch-XSA304 patch-XSA305
Log Message:
Add patches for relevant Xen security advisory up to XSA305 (everything
up to XSA297 is already fixed upstream).
Bump PKGREVISION
---
Module Name: pkgsrc
Committed By: bouyer
Date: Wed Nov 13 15:00:06 UTC 2019
Modified Files:
pkgsrc/sysutils/xenkernel411: Makefile distinfo
pkgsrc/sysutils/xenkernel411/patches: patch-XSA298 patch-XSA302
patch-XSA304 patch-XSA305
Added Files:
pkgsrc/sysutils/xenkernel411/patches: patch-XSA299
Log Message:
Apply patch fixing XSA299.
Bump PKGREVISION
-rw-r--r-- | sysutils/xenkernel411/Makefile | 4 | ||||
-rw-r--r-- | sysutils/xenkernel411/distinfo | 7 | ||||
-rw-r--r-- | sysutils/xenkernel411/patches/patch-XSA298 | 89 | ||||
-rw-r--r-- | sysutils/xenkernel411/patches/patch-XSA299 | 2413 | ||||
-rw-r--r-- | sysutils/xenkernel411/patches/patch-XSA302 | 537 | ||||
-rw-r--r-- | sysutils/xenkernel411/patches/patch-XSA304 | 481 | ||||
-rw-r--r-- | sysutils/xenkernel411/patches/patch-XSA305 | 482 |
7 files changed, 4010 insertions, 3 deletions
diff --git a/sysutils/xenkernel411/Makefile b/sysutils/xenkernel411/Makefile index 5753bffb774..e1426bd25f6 100644 --- a/sysutils/xenkernel411/Makefile +++ b/sysutils/xenkernel411/Makefile @@ -1,7 +1,7 @@ -# $NetBSD: Makefile,v 1.8 2019/08/30 13:16:27 bouyer Exp $ +# $NetBSD: Makefile,v 1.8.2.1 2019/11/16 22:10:06 bsiegert Exp $ VERSION= 4.11.2 -#PKGREVISION= 0 +PKGREVISION= 2 DISTNAME= xen-${VERSION} PKGNAME= xenkernel411-${VERSION} CATEGORIES= sysutils diff --git a/sysutils/xenkernel411/distinfo b/sysutils/xenkernel411/distinfo index ccf14678aaa..2ebc521bc0f 100644 --- a/sysutils/xenkernel411/distinfo +++ b/sysutils/xenkernel411/distinfo @@ -1,10 +1,15 @@ -$NetBSD: distinfo,v 1.5 2019/08/30 13:16:27 bouyer Exp $ +$NetBSD: distinfo,v 1.5.2.1 2019/11/16 22:10:06 bsiegert Exp $ SHA1 (xen411/xen-4.11.2.tar.gz) = 82766db0eca7ce65962732af8a31bb5cce1eb7ce RMD160 (xen411/xen-4.11.2.tar.gz) = 6dcb1ac3e72381474912607b30b59fa55d87d38b SHA512 (xen411/xen-4.11.2.tar.gz) = 48d3d926d35eb56c79c06d0abc6e6be2564fadb43367cc7f46881c669a75016707672179c2cca1c4cfb14af2cefd46e2e7f99470cddf7df2886d8435a2de814e Size (xen411/xen-4.11.2.tar.gz) = 25164925 bytes SHA1 (patch-Config.mk) = 9372a09efd05c9fbdbc06f8121e411fcb7c7ba65 +SHA1 (patch-XSA298) = 63e0f96ce3b945b16b98b51b423bafec14cf2be6 +SHA1 (patch-XSA299) = beb7ba1a8f9e0adda161c0da725ff053e674067e +SHA1 (patch-XSA302) = 12fbb7dfea27f53c70c8115487a2e30595549c2b +SHA1 (patch-XSA304) = f2c22732227e11a3e77c630f0264a689eed53399 +SHA1 (patch-XSA305) = eb5e0096cbf501fcbd7a5c5f9d1f932b557636b6 SHA1 (patch-xen_Makefile) = 465388d80de414ca3bb84faefa0f52d817e423a6 SHA1 (patch-xen_Rules.mk) = c743dc63f51fc280d529a7d9e08650292c171dac SHA1 (patch-xen_arch_x86_Rules.mk) = 0bedfc53a128a87b6a249ae04fbdf6a053bfb70b diff --git a/sysutils/xenkernel411/patches/patch-XSA298 b/sysutils/xenkernel411/patches/patch-XSA298 new file mode 100644 index 00000000000..cb8ca6b856d --- /dev/null +++ b/sysutils/xenkernel411/patches/patch-XSA298 @@ -0,0 +1,89 @@ +$NetBSD: patch-XSA298,v 1.2.2.2 2019/11/16 22:10:07 bsiegert Exp $ + +From: Jan Beulich <jbeulich@suse.com> +Subject: x86/PV: check GDT/LDT limits during emulation + +Accesses beyond the LDT limit originating from emulation would trigger +the ASSERT() in pv_map_ldt_shadow_page(). On production builds such +accesses would cause an attempt to promote the touched page (offset from +the present LDT base address) to a segment descriptor one. If this +happens to succeed, guest user mode would be able to elevate its +privileges to that of the guest kernel. This is particularly easy when +there's no LDT at all, in which case the LDT base stored internally to +Xen is simply zero. + +Also adjust the ASSERT() that was triggering: It was off by one to +begin with, and for production builds we also better use +ASSERT_UNREACHABLE() instead with suitable recovery code afterwards. + +This is XSA-298. + +Reported-by: Andrew Cooper <andrew.cooper3@citrix.com> +Signed-off-by: Jan Beulich <jbeulich@suse.com> +Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com> + +--- xen/arch/x86/pv/emul-gate-op.c.orig ++++ xen/arch/x86/pv/emul-gate-op.c +@@ -51,7 +51,13 @@ static int read_gate_descriptor(unsigned + const struct desc_struct *pdesc = gdt_ldt_desc_ptr(gate_sel); + + if ( (gate_sel < 4) || +- ((gate_sel >= FIRST_RESERVED_GDT_BYTE) && !(gate_sel & 4)) || ++ /* ++ * We're interested in call gates only, which occupy a single ++ * seg_desc_t for 32-bit and a consecutive pair of them for 64-bit. ++ */ ++ ((gate_sel >> 3) + !is_pv_32bit_vcpu(v) >= ++ (gate_sel & 4 ? v->arch.pv_vcpu.ldt_ents ++ : v->arch.pv_vcpu.gdt_ents)) || + __get_user(desc, pdesc) ) + return 0; + +@@ -70,7 +76,7 @@ static int read_gate_descriptor(unsigned + if ( !is_pv_32bit_vcpu(v) ) + { + if ( (*ar & 0x1f00) != 0x0c00 || +- (gate_sel >= FIRST_RESERVED_GDT_BYTE - 8 && !(gate_sel & 4)) || ++ /* Limit check done above already. */ + __get_user(desc, pdesc + 1) || + (desc.b & 0x1f00) ) + return 0; +--- xen/arch/x86/pv/emulate.c.orig ++++ xen/arch/x86/pv/emulate.c +@@ -31,7 +31,14 @@ int pv_emul_read_descriptor(unsigned int + { + struct desc_struct desc; + +- if ( sel < 4) ++ if ( sel < 4 || ++ /* ++ * Don't apply the GDT limit here, as the selector may be a Xen ++ * provided one. __get_user() will fail (without taking further ++ * action) for ones falling in the gap between guest populated ++ * and Xen ones. ++ */ ++ ((sel & 4) && (sel >> 3) >= v->arch.pv_vcpu.ldt_ents) ) + desc.b = desc.a = 0; + else if ( __get_user(desc, gdt_ldt_desc_ptr(sel)) ) + return 0; +--- xen/arch/x86/pv/mm.c.orig ++++ xen/arch/x86/pv/mm.c +@@ -92,12 +92,16 @@ bool pv_map_ldt_shadow_page(unsigned int + BUG_ON(unlikely(in_irq())); + + /* +- * Hardware limit checking should guarantee this property. NB. This is ++ * Prior limit checking should guarantee this property. NB. This is + * safe as updates to the LDT can only be made by MMUEXT_SET_LDT to the + * current vcpu, and vcpu_reset() will block until this vcpu has been + * descheduled before continuing. + */ +- ASSERT((offset >> 3) <= curr->arch.pv_vcpu.ldt_ents); ++ if ( unlikely((offset >> 3) >= curr->arch.pv_vcpu.ldt_ents) ) ++ { ++ ASSERT_UNREACHABLE(); ++ return false; ++ } + + if ( is_pv_32bit_domain(currd) ) + linear = (uint32_t)linear; diff --git a/sysutils/xenkernel411/patches/patch-XSA299 b/sysutils/xenkernel411/patches/patch-XSA299 new file mode 100644 index 00000000000..ea92bb68aad --- /dev/null +++ b/sysutils/xenkernel411/patches/patch-XSA299 @@ -0,0 +1,2413 @@ +$NetBSD: patch-XSA299,v 1.1.2.2 2019/11/16 22:10:07 bsiegert Exp $ + +From 852df269d247e177d5f2e9b8f3a4301a6fdd76bd Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 01/11] x86/mm: L1TF checks don't leave a partial entry + +On detection of a potential L1TF issue, most validation code returns +-ERESTART to allow the switch to shadow mode to happen and cause the +original operation to be restarted. + +However, in the validation code, the return value -ERESTART has been +repurposed to indicate 1) the function has partially completed +something which needs to be undone, and 2) calling put_page_type() +should cleanly undo it. This causes problems in several places. + +For L1 tables, on receiving an -ERESTART return from alloc_l1_table(), +alloc_page_type() will set PGT_partial on the page. If for some +reason the original operation never restarts, then on domain +destruction, relinquish_memory() will call free_page_type() on the +page. + +Unfortunately, alloc_ and free_l1_table() aren't set up to deal with +PGT_partial. When returning a failure, alloc_l1_table() always +de-validates whatever it's validated so far, and free_l1_table() +always devalidates the whole page. This means that if +relinquish_memory() calls free_page_type() on an L1 that didn't +complete due to an L1TF, it will call put_page_from_l1e() on "page +entries" that have never been validated. + +For L2+ tables, setting rc to ERESTART causes the rest of the +alloc_lN_table() function to *think* that the entry in question will +have PGT_partial set. This will cause it to set partial_pte = 1. If +relinqush_memory() then calls free_page_type() on one of those pages, +then free_lN_table() will call put_page_from_lNe() on the entry when +it shouldn't. + +Rather than indicating -ERESTART, indicate -EINTR. This is the code +to indicate that nothing has changed from when you started the call +(which is effectively how alloc_l1_table() handles errors). + +mod_lN_entry() shouldn't have any of these types of problems, so leave +potential changes there for a clean-up patch later. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 8 ++++---- + 1 file changed, 4 insertions(+), 4 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index e6a4cb28f8..8ced185b49 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -1110,7 +1110,7 @@ get_page_from_l2e( + int rc; + + if ( !(l2e_get_flags(l2e) & _PAGE_PRESENT) ) +- return pv_l1tf_check_l2e(d, l2e) ? -ERESTART : 1; ++ return pv_l1tf_check_l2e(d, l2e) ? -EINTR : 1; + + if ( unlikely((l2e_get_flags(l2e) & L2_DISALLOW_MASK)) ) + { +@@ -1142,7 +1142,7 @@ get_page_from_l3e( + int rc; + + if ( !(l3e_get_flags(l3e) & _PAGE_PRESENT) ) +- return pv_l1tf_check_l3e(d, l3e) ? -ERESTART : 1; ++ return pv_l1tf_check_l3e(d, l3e) ? -EINTR : 1; + + if ( unlikely((l3e_get_flags(l3e) & l3_disallow_mask(d))) ) + { +@@ -1175,7 +1175,7 @@ get_page_from_l4e( + int rc; + + if ( !(l4e_get_flags(l4e) & _PAGE_PRESENT) ) +- return pv_l1tf_check_l4e(d, l4e) ? -ERESTART : 1; ++ return pv_l1tf_check_l4e(d, l4e) ? -EINTR : 1; + + if ( unlikely((l4e_get_flags(l4e) & L4_DISALLOW_MASK)) ) + { +@@ -1404,7 +1404,7 @@ static int alloc_l1_table(struct page_info *page) + { + if ( !(l1e_get_flags(pl1e[i]) & _PAGE_PRESENT) ) + { +- ret = pv_l1tf_check_l1e(d, pl1e[i]) ? -ERESTART : 0; ++ ret = pv_l1tf_check_l1e(d, pl1e[i]) ? -EINTR : 0; + if ( ret ) + goto out; + } +-- +2.23.0 + +From 6bdddd7980eac0cc883945d823986f24682ca47a Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 02/11] x86/mm: Don't re-set PGT_pinned on a partially + de-validated page + +When unpinning pagetables, if an operation is interrupted, +relinquish_memory() re-sets PGT_pinned so that the un-pin will +pickedup again when the hypercall restarts. + +This is appropriate when put_page_and_type_preemptible() returns +-EINTR, which indicates that the page is back in its initial state +(i.e., completely validated). However, for -ERESTART, this leads to a +state where a page has both PGT_pinned and PGT_partial set. + +This happens to work at the moment, although it's not really a +"canonical" state; but in subsequent patches, where we need to make a +distinction in handling between PGT_validated and PGT_partial pages, +this causes issues. + +Move to a "canonical" state by: +- Only re-setting PGT_pinned on -EINTR +- Re-dropping the refcount held by PGT_pinned on -ERESTART + +In the latter case, the PGT_partial bit will be cleared further down +with the rest of the other PGT_partial pages. + +While here, clean up some trainling whitespace. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/domain.c | 31 ++++++++++++++++++++++++++++--- + 1 file changed, 28 insertions(+), 3 deletions(-) + +diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c +index 29f892c04c..8fbecbb169 100644 +--- xen/arch/x86/domain.c.orig ++++ xen/arch/x86/domain.c +@@ -112,7 +112,7 @@ static void play_dead(void) + * this case, heap corruption or #PF can occur (when heap debugging is + * enabled). For example, even printk() can involve tasklet scheduling, + * which touches per-cpu vars. +- * ++ * + * Consider very carefully when adding code to *dead_idle. Most hypervisor + * subsystems are unsafe to call. + */ +@@ -1838,9 +1838,34 @@ static int relinquish_memory( + break; + case -ERESTART: + case -EINTR: ++ /* ++ * -EINTR means PGT_validated has been re-set; re-set ++ * PGT_pinned again so that it gets picked up next time ++ * around. ++ * ++ * -ERESTART, OTOH, means PGT_partial is set instead. Put ++ * it back on the list, but don't set PGT_pinned; the ++ * section below will finish off de-validation. But we do ++ * need to drop the general ref associated with ++ * PGT_pinned, since put_page_and_type_preemptible() ++ * didn't do it. ++ * ++ * NB we can do an ASSERT for PGT_validated, since we ++ * "own" the type ref; but theoretically, the PGT_partial ++ * could be cleared by someone else. ++ */ ++ if ( ret == -EINTR ) ++ { ++ ASSERT(page->u.inuse.type_info & PGT_validated); ++ set_bit(_PGT_pinned, &page->u.inuse.type_info); ++ } ++ else ++ put_page(page); ++ + ret = -ERESTART; ++ ++ /* Put the page back on the list and drop the ref we grabbed above */ + page_list_add(page, list); +- set_bit(_PGT_pinned, &page->u.inuse.type_info); + put_page(page); + goto out; + default: +@@ -2062,7 +2087,7 @@ void vcpu_kick(struct vcpu *v) + * pending flag. These values may fluctuate (after all, we hold no + * locks) but the key insight is that each change will cause + * evtchn_upcall_pending to be polled. +- * ++ * + * NB2. We save the running flag across the unblock to avoid a needless + * IPI for domains that we IPI'd to unblock. + */ +-- +2.23.0 + +From 7c0a37005f52d10903ce22851b52ae9b6f4f0ee2 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 03/11] x86/mm: Separate out partial_pte tristate into + individual flags + +At the moment, partial_pte is a tri-state that contains two distinct bits +of information: + +1. If zero, the pte at index [nr_validated_ptes] is un-validated. If + non-zero, the pte was last seen with PGT_partial set. + +2. If positive, the pte at index [nr_validated_ptes] does not hold a + general reference count. If negative, it does. + +To make future patches more clear, separate out this functionality +into two distinct, named bits: PTF_partial_set (for #1) and +PTF_partial_general_ref (for #2). + +Additionally, a number of functions which need this information also +take other flags to control behavior (such as `preemptible` and +`defer`). These are hard to read in the caller (since you only see +'true' or 'false'), and ugly when many are added together. In +preparation for adding yet another flag in a future patch, collapse +all of these into a single `flag` variable. + +NB that this does mean checking for what was previously the '-1' +condition a bit more ugly in the put_page_from_lNe functions (since +you have to check for both partial_set and general ref); but this +clause will go away in a future patch. + +Also note that the original comment had an off-by-one error: +partial_flags (like partial_pte before it) concerns +plNe[nr_validated_ptes], not plNe[nr_validated_ptes+1]. + +No functional change intended. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 164 +++++++++++++++++++++++---------------- + xen/include/asm-x86/mm.h | 41 ++++++---- + 2 files changed, 127 insertions(+), 78 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index 8ced185b49..1c4f54e328 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -610,20 +610,34 @@ static int alloc_segdesc_page(struct page_info *page) + static int _get_page_type(struct page_info *page, unsigned long type, + bool preemptible); + ++/* ++ * The following flags are used to specify behavior of various get and ++ * put commands. The first two are also stored in page->partial_flags ++ * to indicate the state of the page pointed to by ++ * page->pte[page->nr_validated_entries]. See the comment in mm.h for ++ * more information. ++ */ ++#define PTF_partial_set (1 << 0) ++#define PTF_partial_general_ref (1 << 1) ++#define PTF_preemptible (1 << 2) ++#define PTF_defer (1 << 3) ++ + static int get_page_and_type_from_mfn( + mfn_t mfn, unsigned long type, struct domain *d, +- int partial, int preemptible) ++ unsigned int flags) + { + struct page_info *page = mfn_to_page(mfn); + int rc; ++ bool preemptible = flags & PTF_preemptible, ++ partial_ref = flags & PTF_partial_general_ref; + +- if ( likely(partial >= 0) && ++ if ( likely(!partial_ref) && + unlikely(!get_page_from_mfn(mfn, d)) ) + return -EINVAL; + + rc = _get_page_type(page, type, preemptible); + +- if ( unlikely(rc) && partial >= 0 && ++ if ( unlikely(rc) && !partial_ref && + (!preemptible || page != current->arch.old_guest_table) ) + put_page(page); + +@@ -1104,7 +1118,7 @@ get_page_from_l1e( + define_get_linear_pagetable(l2); + static int + get_page_from_l2e( +- l2_pgentry_t l2e, unsigned long pfn, struct domain *d, int partial) ++ l2_pgentry_t l2e, unsigned long pfn, struct domain *d, unsigned int flags) + { + unsigned long mfn = l2e_get_pfn(l2e); + int rc; +@@ -1119,8 +1133,9 @@ get_page_from_l2e( + return -EINVAL; + } + +- rc = get_page_and_type_from_mfn(_mfn(mfn), PGT_l1_page_table, d, +- partial, false); ++ ASSERT(!(flags & PTF_preemptible)); ++ ++ rc = get_page_and_type_from_mfn(_mfn(mfn), PGT_l1_page_table, d, flags); + if ( unlikely(rc == -EINVAL) && get_l2_linear_pagetable(l2e, pfn, d) ) + rc = 0; + +@@ -1137,7 +1152,7 @@ get_page_from_l2e( + define_get_linear_pagetable(l3); + static int + get_page_from_l3e( +- l3_pgentry_t l3e, unsigned long pfn, struct domain *d, int partial) ++ l3_pgentry_t l3e, unsigned long pfn, struct domain *d, unsigned int flags) + { + int rc; + +@@ -1152,7 +1167,7 @@ get_page_from_l3e( + } + + rc = get_page_and_type_from_mfn( +- l3e_get_mfn(l3e), PGT_l2_page_table, d, partial, 1); ++ l3e_get_mfn(l3e), PGT_l2_page_table, d, flags | PTF_preemptible); + if ( unlikely(rc == -EINVAL) && + !is_pv_32bit_domain(d) && + get_l3_linear_pagetable(l3e, pfn, d) ) +@@ -1170,7 +1185,7 @@ get_page_from_l3e( + define_get_linear_pagetable(l4); + static int + get_page_from_l4e( +- l4_pgentry_t l4e, unsigned long pfn, struct domain *d, int partial) ++ l4_pgentry_t l4e, unsigned long pfn, struct domain *d, unsigned int flags) + { + int rc; + +@@ -1185,7 +1200,7 @@ get_page_from_l4e( + } + + rc = get_page_and_type_from_mfn( +- l4e_get_mfn(l4e), PGT_l3_page_table, d, partial, 1); ++ l4e_get_mfn(l4e), PGT_l3_page_table, d, flags | PTF_preemptible); + if ( unlikely(rc == -EINVAL) && get_l4_linear_pagetable(l4e, pfn, d) ) + rc = 0; + +@@ -1275,7 +1290,7 @@ void put_page_from_l1e(l1_pgentry_t l1e, struct domain *l1e_owner) + * Note also that this automatically deals correctly with linear p.t.'s. + */ + static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, +- int partial, bool defer) ++ unsigned int flags) + { + int rc = 0; + +@@ -1295,12 +1310,13 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + struct page_info *pg = l2e_get_page(l2e); + struct page_info *ptpg = mfn_to_page(_mfn(pfn)); + +- if ( unlikely(partial > 0) ) ++ if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == ++ PTF_partial_set ) + { +- ASSERT(!defer); ++ ASSERT(!(flags & PTF_defer)); + rc = _put_page_type(pg, true, ptpg); + } +- else if ( defer ) ++ else if ( flags & PTF_defer ) + { + current->arch.old_guest_ptpg = ptpg; + current->arch.old_guest_table = pg; +@@ -1317,7 +1333,7 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + } + + static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, +- int partial, bool defer) ++ unsigned int flags) + { + struct page_info *pg; + int rc; +@@ -1340,13 +1356,14 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + + pg = l3e_get_page(l3e); + +- if ( unlikely(partial > 0) ) ++ if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == ++ PTF_partial_set ) + { +- ASSERT(!defer); ++ ASSERT(!(flags & PTF_defer)); + return _put_page_type(pg, true, mfn_to_page(_mfn(pfn))); + } + +- if ( defer ) ++ if ( flags & PTF_defer ) + { + current->arch.old_guest_ptpg = mfn_to_page(_mfn(pfn)); + current->arch.old_guest_table = pg; +@@ -1361,7 +1378,7 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + } + + static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, +- int partial, bool defer) ++ unsigned int flags) + { + int rc = 1; + +@@ -1370,13 +1387,14 @@ static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, + { + struct page_info *pg = l4e_get_page(l4e); + +- if ( unlikely(partial > 0) ) ++ if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == ++ PTF_partial_set ) + { +- ASSERT(!defer); ++ ASSERT(!(flags & PTF_defer)); + return _put_page_type(pg, true, mfn_to_page(_mfn(pfn))); + } + +- if ( defer ) ++ if ( flags & PTF_defer ) + { + current->arch.old_guest_ptpg = mfn_to_page(_mfn(pfn)); + current->arch.old_guest_table = pg; +@@ -1483,12 +1501,13 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + unsigned long pfn = mfn_x(page_to_mfn(page)); + l2_pgentry_t *pl2e; + unsigned int i; +- int rc = 0, partial = page->partial_pte; ++ int rc = 0; ++ unsigned int partial_flags = page->partial_flags; + + pl2e = map_domain_page(_mfn(pfn)); + + for ( i = page->nr_validated_ptes; i < L2_PAGETABLE_ENTRIES; +- i++, partial = 0 ) ++ i++, partial_flags = 0 ) + { + if ( i > page->nr_validated_ptes && hypercall_preempt_check() ) + { +@@ -1498,18 +1517,19 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + } + + if ( !is_guest_l2_slot(d, type, i) || +- (rc = get_page_from_l2e(pl2e[i], pfn, d, partial)) > 0 ) ++ (rc = get_page_from_l2e(pl2e[i], pfn, d, partial_flags)) > 0 ) + continue; + + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_pte = partial ?: 1; ++ /* Set 'set', retain 'general ref' */ ++ page->partial_flags = partial_flags | PTF_partial_set; + } + else if ( rc == -EINTR && i ) + { + page->nr_validated_ptes = i; +- page->partial_pte = 0; ++ page->partial_flags = 0; + rc = -ERESTART; + } + else if ( rc < 0 && rc != -EINTR ) +@@ -1518,7 +1538,7 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + if ( i ) + { + page->nr_validated_ptes = i; +- page->partial_pte = 0; ++ page->partial_flags = 0; + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; + } +@@ -1542,7 +1562,8 @@ static int alloc_l3_table(struct page_info *page) + unsigned long pfn = mfn_x(page_to_mfn(page)); + l3_pgentry_t *pl3e; + unsigned int i; +- int rc = 0, partial = page->partial_pte; ++ int rc = 0; ++ unsigned int partial_flags = page->partial_flags; + + pl3e = map_domain_page(_mfn(pfn)); + +@@ -1557,7 +1578,7 @@ static int alloc_l3_table(struct page_info *page) + memset(pl3e + 4, 0, (L3_PAGETABLE_ENTRIES - 4) * sizeof(*pl3e)); + + for ( i = page->nr_validated_ptes; i < L3_PAGETABLE_ENTRIES; +- i++, partial = 0 ) ++ i++, partial_flags = 0 ) + { + if ( i > page->nr_validated_ptes && hypercall_preempt_check() ) + { +@@ -1574,20 +1595,22 @@ static int alloc_l3_table(struct page_info *page) + else + rc = get_page_and_type_from_mfn( + l3e_get_mfn(pl3e[i]), +- PGT_l2_page_table | PGT_pae_xen_l2, d, partial, 1); ++ PGT_l2_page_table | PGT_pae_xen_l2, d, ++ partial_flags | PTF_preemptible); + } +- else if ( (rc = get_page_from_l3e(pl3e[i], pfn, d, partial)) > 0 ) ++ else if ( (rc = get_page_from_l3e(pl3e[i], pfn, d, partial_flags)) > 0 ) + continue; + + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_pte = partial ?: 1; ++ /* Set 'set', leave 'general ref' set if this entry was set */ ++ page->partial_flags = partial_flags | PTF_partial_set; + } + else if ( rc == -EINTR && i ) + { + page->nr_validated_ptes = i; +- page->partial_pte = 0; ++ page->partial_flags = 0; + rc = -ERESTART; + } + if ( rc < 0 ) +@@ -1604,7 +1627,7 @@ static int alloc_l3_table(struct page_info *page) + if ( i ) + { + page->nr_validated_ptes = i; +- page->partial_pte = 0; ++ page->partial_flags = 0; + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; + } +@@ -1736,19 +1759,21 @@ static int alloc_l4_table(struct page_info *page) + unsigned long pfn = mfn_x(page_to_mfn(page)); + l4_pgentry_t *pl4e = map_domain_page(_mfn(pfn)); + unsigned int i; +- int rc = 0, partial = page->partial_pte; ++ int rc = 0; ++ unsigned int partial_flags = page->partial_flags; + + for ( i = page->nr_validated_ptes; i < L4_PAGETABLE_ENTRIES; +- i++, partial = 0 ) ++ i++, partial_flags = 0 ) + { + if ( !is_guest_l4_slot(d, i) || +- (rc = get_page_from_l4e(pl4e[i], pfn, d, partial)) > 0 ) ++ (rc = get_page_from_l4e(pl4e[i], pfn, d, partial_flags)) > 0 ) + continue; + + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_pte = partial ?: 1; ++ /* Set 'set', leave 'general ref' set if this entry was set */ ++ page->partial_flags = partial_flags | PTF_partial_set; + } + else if ( rc < 0 ) + { +@@ -1758,7 +1783,7 @@ static int alloc_l4_table(struct page_info *page) + if ( i ) + { + page->nr_validated_ptes = i; +- page->partial_pte = 0; ++ page->partial_flags = 0; + if ( rc == -EINTR ) + rc = -ERESTART; + else +@@ -1811,19 +1836,20 @@ static int free_l2_table(struct page_info *page) + struct domain *d = page_get_owner(page); + unsigned long pfn = mfn_x(page_to_mfn(page)); + l2_pgentry_t *pl2e; +- int rc = 0, partial = page->partial_pte; +- unsigned int i = page->nr_validated_ptes - !partial; ++ int rc = 0; ++ unsigned int partial_flags = page->partial_flags, ++ i = page->nr_validated_ptes - !(partial_flags & PTF_partial_set); + + pl2e = map_domain_page(_mfn(pfn)); + + for ( ; ; ) + { + if ( is_guest_l2_slot(d, page->u.inuse.type_info, i) ) +- rc = put_page_from_l2e(pl2e[i], pfn, partial, false); ++ rc = put_page_from_l2e(pl2e[i], pfn, partial_flags); + if ( rc < 0 ) + break; + +- partial = 0; ++ partial_flags = 0; + + if ( !i-- ) + break; +@@ -1845,12 +1871,14 @@ static int free_l2_table(struct page_info *page) + else if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_pte = partial ?: -1; ++ page->partial_flags = (partial_flags & PTF_partial_set) ? ++ partial_flags : ++ (PTF_partial_set | PTF_partial_general_ref); + } + else if ( rc == -EINTR && i < L2_PAGETABLE_ENTRIES - 1 ) + { + page->nr_validated_ptes = i + 1; +- page->partial_pte = 0; ++ page->partial_flags = 0; + rc = -ERESTART; + } + +@@ -1862,18 +1890,19 @@ static int free_l3_table(struct page_info *page) + struct domain *d = page_get_owner(page); + unsigned long pfn = mfn_x(page_to_mfn(page)); + l3_pgentry_t *pl3e; +- int rc = 0, partial = page->partial_pte; +- unsigned int i = page->nr_validated_ptes - !partial; ++ int rc = 0; ++ unsigned int partial_flags = page->partial_flags, ++ i = page->nr_validated_ptes - !(partial_flags & PTF_partial_set); + + pl3e = map_domain_page(_mfn(pfn)); + + for ( ; ; ) + { +- rc = put_page_from_l3e(pl3e[i], pfn, partial, 0); ++ rc = put_page_from_l3e(pl3e[i], pfn, partial_flags); + if ( rc < 0 ) + break; + +- partial = 0; ++ partial_flags = 0; + if ( rc == 0 ) + pl3e[i] = unadjust_guest_l3e(pl3e[i], d); + +@@ -1892,12 +1921,14 @@ static int free_l3_table(struct page_info *page) + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_pte = partial ?: -1; ++ page->partial_flags = (partial_flags & PTF_partial_set) ? ++ partial_flags : ++ (PTF_partial_set | PTF_partial_general_ref); + } + else if ( rc == -EINTR && i < L3_PAGETABLE_ENTRIES - 1 ) + { + page->nr_validated_ptes = i + 1; +- page->partial_pte = 0; ++ page->partial_flags = 0; + rc = -ERESTART; + } + return rc > 0 ? 0 : rc; +@@ -1908,26 +1939,29 @@ static int free_l4_table(struct page_info *page) + struct domain *d = page_get_owner(page); + unsigned long pfn = mfn_x(page_to_mfn(page)); + l4_pgentry_t *pl4e = map_domain_page(_mfn(pfn)); +- int rc = 0, partial = page->partial_pte; +- unsigned int i = page->nr_validated_ptes - !partial; ++ int rc = 0; ++ unsigned partial_flags = page->partial_flags, ++ i = page->nr_validated_ptes - !(partial_flags & PTF_partial_set); + + do { + if ( is_guest_l4_slot(d, i) ) +- rc = put_page_from_l4e(pl4e[i], pfn, partial, 0); ++ rc = put_page_from_l4e(pl4e[i], pfn, partial_flags); + if ( rc < 0 ) + break; +- partial = 0; ++ partial_flags = 0; + } while ( i-- ); + + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_pte = partial ?: -1; ++ page->partial_flags = (partial_flags & PTF_partial_set) ? ++ partial_flags : ++ (PTF_partial_set | PTF_partial_general_ref); + } + else if ( rc == -EINTR && i < L4_PAGETABLE_ENTRIES - 1 ) + { + page->nr_validated_ptes = i + 1; +- page->partial_pte = 0; ++ page->partial_flags = 0; + rc = -ERESTART; + } + +@@ -2203,7 +2237,7 @@ static int mod_l2_entry(l2_pgentry_t *pl2e, + return -EBUSY; + } + +- put_page_from_l2e(ol2e, pfn, 0, true); ++ put_page_from_l2e(ol2e, pfn, PTF_defer); + + return rc; + } +@@ -2271,7 +2305,7 @@ static int mod_l3_entry(l3_pgentry_t *pl3e, + if ( !create_pae_xen_mappings(d, pl3e) ) + BUG(); + +- put_page_from_l3e(ol3e, pfn, 0, 1); ++ put_page_from_l3e(ol3e, pfn, PTF_defer); + return rc; + } + +@@ -2334,7 +2368,7 @@ static int mod_l4_entry(l4_pgentry_t *pl4e, + return -EFAULT; + } + +- put_page_from_l4e(ol4e, pfn, 0, 1); ++ put_page_from_l4e(ol4e, pfn, PTF_defer); + return rc; + } + +@@ -2598,7 +2632,7 @@ int free_page_type(struct page_info *page, unsigned long type, + if ( !(type & PGT_partial) ) + { + page->nr_validated_ptes = 1U << PAGETABLE_ORDER; +- page->partial_pte = 0; ++ page->partial_flags = 0; + } + + switch ( type & PGT_type_mask ) +@@ -2889,7 +2923,7 @@ static int _get_page_type(struct page_info *page, unsigned long type, + if ( !(x & PGT_partial) ) + { + page->nr_validated_ptes = 0; +- page->partial_pte = 0; ++ page->partial_flags = 0; + } + page->linear_pt_count = 0; + rc = alloc_page_type(page, type, preemptible); +@@ -3064,7 +3098,7 @@ int new_guest_cr3(mfn_t mfn) + return 0; + } + +- rc = get_page_and_type_from_mfn(mfn, PGT_root_page_table, d, 0, 1); ++ rc = get_page_and_type_from_mfn(mfn, PGT_root_page_table, d, PTF_preemptible); + switch ( rc ) + { + case 0: +@@ -3452,7 +3486,7 @@ long do_mmuext_op( + if ( op.arg1.mfn != 0 ) + { + rc = get_page_and_type_from_mfn( +- _mfn(op.arg1.mfn), PGT_root_page_table, currd, 0, 1); ++ _mfn(op.arg1.mfn), PGT_root_page_table, currd, PTF_preemptible); + + if ( unlikely(rc) ) + { +diff --git a/xen/include/asm-x86/mm.h b/xen/include/asm-x86/mm.h +index 1ea173c555..46cba52941 100644 +--- xen/include/asm-x86/mm.h.orig ++++ xen/include/asm-x86/mm.h +@@ -228,19 +228,34 @@ struct page_info + * setting the flag must not drop that reference, whereas the instance + * clearing it will have to. + * +- * If @partial_pte is positive then PTE at @nr_validated_ptes+1 has +- * been partially validated. This implies that the general reference +- * to the page (acquired from get_page_from_lNe()) would be dropped +- * (again due to the apparent failure) and hence must be re-acquired +- * when resuming the validation, but must not be dropped when picking +- * up the page for invalidation. ++ * If partial_flags & PTF_partial_set is set, then the page at ++ * at @nr_validated_ptes had PGT_partial set as a result of an ++ * operation on the current page. (That page may or may not ++ * still have PGT_partial set.) + * +- * If @partial_pte is negative then PTE at @nr_validated_ptes+1 has +- * been partially invalidated. This is basically the opposite case of +- * above, i.e. the general reference to the page was not dropped in +- * put_page_from_lNe() (due to the apparent failure), and hence it +- * must be dropped when the put operation is resumed (and completes), +- * but it must not be acquired if picking up the page for validation. ++ * If PTF_partial_general_ref is set, then the PTE at ++ * @nr_validated_ptef holds a general reference count for the ++ * page. ++ * ++ * This happens: ++ * - During de-validation, if de-validation of the page was ++ * interrupted ++ * - During validation, if an invalid entry is encountered and ++ * validation is preemptible ++ * - During validation, if PTF_partial_general_ref was set on ++ * this entry to begin with (perhaps because we're picking ++ * up from a partial de-validation). ++ * ++ * When resuming validation, if PTF_partial_general_ref is clear, ++ * then a general reference must be re-acquired; if it is set, no ++ * reference should be acquired. ++ * ++ * When resuming de-validation, if PTF_partial_general_ref is ++ * clear, no reference should be dropped; if it is set, a ++ * reference should be dropped. ++ * ++ * NB that PTF_partial_set and PTF_partial_general_ref are ++ * defined in mm.c, the only place where they are used. + * + * The 3rd field, @linear_pt_count, indicates + * - by a positive value, how many same-level page table entries a page +@@ -251,7 +266,7 @@ struct page_info + struct { + u16 nr_validated_ptes:PAGETABLE_ORDER + 1; + u16 :16 - PAGETABLE_ORDER - 1 - 2; +- s16 partial_pte:2; ++ u16 partial_flags:2; + s16 linear_pt_count; + }; + +-- +2.23.0 + +From 20b8a6702c6839bafd252789396b443d4b5c5474 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 04/11] x86/mm: Use flags for _put_page_type rather than a + boolean + +This is in mainly in preparation for _put_page_type taking the +partial_flags value in the future. It also makes it easier to read in +the caller (since you see a flag name rather than `true` or `false`). + +No functional change intended. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 25 +++++++++++++------------ + 1 file changed, 13 insertions(+), 12 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index 1c4f54e328..e2fba15d86 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -1207,7 +1207,7 @@ get_page_from_l4e( + return rc; + } + +-static int _put_page_type(struct page_info *page, bool preemptible, ++static int _put_page_type(struct page_info *page, unsigned int flags, + struct page_info *ptpg); + + void put_page_from_l1e(l1_pgentry_t l1e, struct domain *l1e_owner) +@@ -1314,7 +1314,7 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + PTF_partial_set ) + { + ASSERT(!(flags & PTF_defer)); +- rc = _put_page_type(pg, true, ptpg); ++ rc = _put_page_type(pg, PTF_preemptible, ptpg); + } + else if ( flags & PTF_defer ) + { +@@ -1323,7 +1323,7 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + } + else + { +- rc = _put_page_type(pg, true, ptpg); ++ rc = _put_page_type(pg, PTF_preemptible, ptpg); + if ( likely(!rc) ) + put_page(pg); + } +@@ -1360,7 +1360,7 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + PTF_partial_set ) + { + ASSERT(!(flags & PTF_defer)); +- return _put_page_type(pg, true, mfn_to_page(_mfn(pfn))); ++ return _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); + } + + if ( flags & PTF_defer ) +@@ -1370,7 +1370,7 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + return 0; + } + +- rc = _put_page_type(pg, true, mfn_to_page(_mfn(pfn))); ++ rc = _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); + if ( likely(!rc) ) + put_page(pg); + +@@ -1391,7 +1391,7 @@ static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, + PTF_partial_set ) + { + ASSERT(!(flags & PTF_defer)); +- return _put_page_type(pg, true, mfn_to_page(_mfn(pfn))); ++ return _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); + } + + if ( flags & PTF_defer ) +@@ -1401,7 +1401,7 @@ static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, + return 0; + } + +- rc = _put_page_type(pg, true, mfn_to_page(_mfn(pfn))); ++ rc = _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); + if ( likely(!rc) ) + put_page(pg); + } +@@ -2701,10 +2701,11 @@ static int _put_final_page_type(struct page_info *page, unsigned long type, + } + + +-static int _put_page_type(struct page_info *page, bool preemptible, ++static int _put_page_type(struct page_info *page, unsigned int flags, + struct page_info *ptpg) + { + unsigned long nx, x, y = page->u.inuse.type_info; ++ bool preemptible = flags & PTF_preemptible; + + ASSERT(current_locked_page_ne_check(page)); + +@@ -2911,7 +2912,7 @@ static int _get_page_type(struct page_info *page, unsigned long type, + + if ( unlikely(iommu_ret) ) + { +- _put_page_type(page, false, NULL); ++ _put_page_type(page, 0, NULL); + rc = iommu_ret; + goto out; + } +@@ -2938,7 +2939,7 @@ static int _get_page_type(struct page_info *page, unsigned long type, + + void put_page_type(struct page_info *page) + { +- int rc = _put_page_type(page, false, NULL); ++ int rc = _put_page_type(page, 0, NULL); + ASSERT(rc == 0); + (void)rc; + } +@@ -2955,7 +2956,7 @@ int get_page_type(struct page_info *page, unsigned long type) + + int put_page_type_preemptible(struct page_info *page) + { +- return _put_page_type(page, true, NULL); ++ return _put_page_type(page, PTF_preemptible, NULL); + } + + int get_page_type_preemptible(struct page_info *page, unsigned long type) +@@ -2972,7 +2973,7 @@ int put_old_guest_table(struct vcpu *v) + if ( !v->arch.old_guest_table ) + return 0; + +- switch ( rc = _put_page_type(v->arch.old_guest_table, true, ++ switch ( rc = _put_page_type(v->arch.old_guest_table, PTF_preemptible, + v->arch.old_guest_ptpg) ) + { + case -EINTR: +-- +2.23.0 + +From 7b3f9f9a797459902bebba962e31be5cbfe7b515 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 05/11] x86/mm: Rework get_page_and_type_from_mfn conditional + +Make it easier to read by declaring the conditions in which we will +retain the ref, rather than the conditions under which we release it. + +The only way (page == current->arch.old_guest_table) can be true is if +preemptible is true; so remove this from the query itself, and add an +ASSERT() to that effect on the opposite path. + +No functional change intended. + +NB that alloc_lN_table() mishandle the "linear pt failure" situation +described in the comment; this will be addressed in a future patch. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 39 +++++++++++++++++++++++++++++++++++++-- + 1 file changed, 37 insertions(+), 2 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index e2fba15d86..eaf7b14245 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -637,8 +637,43 @@ static int get_page_and_type_from_mfn( + + rc = _get_page_type(page, type, preemptible); + +- if ( unlikely(rc) && !partial_ref && +- (!preemptible || page != current->arch.old_guest_table) ) ++ /* ++ * Retain the refcount if: ++ * - page is fully validated (rc == 0) ++ * - page is not validated (rc < 0) but: ++ * - We came in with a reference (partial_ref) ++ * - page is partially validated but there's been an error ++ * (page == current->arch.old_guest_table) ++ * ++ * The partial_ref-on-error clause is worth an explanation. There ++ * are two scenarios where partial_ref might be true coming in: ++ * - mfn has been partially demoted as type `type`; i.e. has ++ * PGT_partial set ++ * - mfn has been partially demoted as L(type+1) (i.e., a linear ++ * page; e.g. we're being called from get_page_from_l2e with ++ * type == PGT_l1_table, but the mfn is PGT_l2_table) ++ * ++ * If there's an error, in the first case, _get_page_type will ++ * either return -ERESTART, in which case we want to retain the ++ * ref (as the caller will consider it retained), or -EINVAL, in ++ * which case old_guest_table will be set; in both cases, we need ++ * to retain the ref. ++ * ++ * In the second case, if there's an error, _get_page_type() can ++ * *only* return -EINVAL, and *never* set old_guest_table. In ++ * that case we also want to retain the reference, to allow the ++ * page to continue to be torn down (i.e., PGT_partial cleared) ++ * safely. ++ * ++ * Also note that we shouldn't be able to leave with the reference ++ * count retained unless we succeeded, or the operation was ++ * preemptible. ++ */ ++ if ( likely(!rc) || partial_ref ) ++ /* nothing */; ++ else if ( page == current->arch.old_guest_table ) ++ ASSERT(preemptible); ++ else + put_page(page); + + return rc; +-- +2.23.0 + +From d28893777be56ef51562ed32502377974f738fd3 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 06/11] x86/mm: Have alloc_l[23]_table clear partial_flags when + preempting + +In order to allow recursive pagetable promotions and demotions to be +interrupted, Xen must keep track of the state of the sub-pages +promoted or demoted. This is stored in two elements in the page +struct: nr_entries_validated and partial_flags. + +The rule is that entries [0, nr_entries_validated) should always be +validated and hold a general reference count. If partial_flags is +zero, then [nr_entries_validated] is not validated and no reference +count is held. If PTF_partial_set is set, then [nr_entries_validated] +is partially validated. + +At the moment, a distinction is made between promotion and demotion +with regard to whether the entry itself "holds" a general reference +count: when entry promotion is interrupted (i.e., returns -ERESTART), +the entry is not considered to hold a reference; when entry demotion +is interrupted, the entry is still considered to hold a general +reference. + +PTF_partial_general_ref is used to distinguish between these cases. +If clear, it's a partial promotion => no general reference count held +by the entry; if set, it's partial demotion, so a general reference +count held. Because promotions and demotions can be interleaved, this +value is passed to get_page_and_type_from_mfn and put_page_from_l*e, +to be able to properly handle reference counts. + +Unfortunately, when alloc_l[23]_table check hypercall_preempt_check() +and return -ERESTART, they set nr_entries_validated, but don't clear +partial_flags. + +If we were picking up from a previously-interrupted promotion, that +means that PTF_partial_set would be set even though +[nr_entries_validated] was not partially validated. This means that +if the page in this state were de-validated, put_page_type() would +erroneously be called on that entry. + +Perhaps worse, if we were racing with a de-validation, then we might +leave both PTF_partial_set and PTF_partial_general_ref; and when +de-validation picked up again, both the type and the general ref would +be erroneously dropped from [nr_entries_validated]. + +In a sense, the real issue here is code duplication. Rather than +duplicate the interruption code, set rc to -EINTR and fall through to +the code which already handles that case correctly. + +Given the logic at this point, it should be impossible for +partial_flags to be non-zero; add an ASSERT() to catch any changes. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 18 ++++-------------- + 1 file changed, 4 insertions(+), 14 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index eaf7b14245..053465cb7c 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -1545,13 +1545,8 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + i++, partial_flags = 0 ) + { + if ( i > page->nr_validated_ptes && hypercall_preempt_check() ) +- { +- page->nr_validated_ptes = i; +- rc = -ERESTART; +- break; +- } +- +- if ( !is_guest_l2_slot(d, type, i) || ++ rc = -EINTR; ++ else if ( !is_guest_l2_slot(d, type, i) || + (rc = get_page_from_l2e(pl2e[i], pfn, d, partial_flags)) > 0 ) + continue; + +@@ -1616,13 +1611,8 @@ static int alloc_l3_table(struct page_info *page) + i++, partial_flags = 0 ) + { + if ( i > page->nr_validated_ptes && hypercall_preempt_check() ) +- { +- page->nr_validated_ptes = i; +- rc = -ERESTART; +- break; +- } +- +- if ( is_pv_32bit_domain(d) && (i == 3) ) ++ rc = -EINTR; ++ else if ( is_pv_32bit_domain(d) && (i == 3) ) + { + if ( !(l3e_get_flags(pl3e[i]) & _PAGE_PRESENT) || + (l3e_get_flags(pl3e[i]) & l3_disallow_mask(d)) ) +-- +2.23.0 + +From f608a53c25806a7a4318cbe225bc5f5bbf154d69 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 07/11] x86/mm: Always retain a general ref on partial + +In order to allow recursive pagetable promotions and demotions to be +interrupted, Xen must keep track of the state of the sub-pages +promoted or demoted. This is stored in two elements in the page struct: +nr_entries_validated and partial_flags. + +The rule is that entries [0, nr_entries_validated) should always be +validated and hold a general reference count. If partial_flags is +zero, then [nr_entries_validated] is not validated and no reference +count is held. If PTF_partial_set is set, then [nr_entries_validated] +is partially validated. + +At the moment, a distinction is made between promotion and demotion +with regard to whether the entry itself "holds" a general reference +count: when entry promotion is interrupted (i.e., returns -ERESTART), +the entry is not considered to hold a reference; when entry demotion +is interrupted, the entry is still considered to hold a general +reference. + +PTF_partial_general_ref is used to distinguish between these cases. +If clear, it's a partial promotion => no general reference count held +by the entry; if set, it's partial demotion, so a general reference +count held. Because promotions and demotions can be interleaved, this +value is passed to get_page_and_type_from_mfn and put_page_from_l*e, +to be able to properly handle reference counts. + +Unfortunately, because a refcount is not held, it is possible to +engineer a situation where PFT_partial_set is set but the page in +question has been assigned to another domain. A sketch is provided in +the appendix. + +Fix this by having the parent page table entry hold a general +reference count whenever PFT_partial_set is set. (For clarity of +change, keep two separate flags. These will be collapsed in a +subsequent changeset.) + +This has two basic implications. On the put_page_from_lNe() side, +this mean that the (partial_set && !partial_ref) case can never happen, +and no longer needs to be special-cased. + +Secondly, because both flags are set together, there's no need to carry over +existing bits from partial_pte. + +(NB there is still another issue with calling _put_page_type() on a +page which had PGT_partial set; that will be handled in a subsequent +patch.) + +On the get_page_and_type_from_mfn() side, we need to distinguish +between callers which hold a reference on partial (i.e., +alloc_lN_table()), and those which do not (new_cr3, PIN_LN_TABLE, and +so on): pass a flag if the type should be retained on interruption. + +NB that since l1 promotion can't be preempted, that get_page_from_l2e +can't return -ERESTART. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +----- +* Appendix: Engineering PTF_partial_set while a page belongs to a + foreign domain + +Suppose A is a page which can be promoted to an l3, and B is a page +which can be promoted to an l2, and A[x] points to B. B has +PGC_allocated set but no other general references. + +V1: PIN_L3 A. + A is validated, B is validated. + A.type_count = 1 | PGT_validated | PGT_pinned + B.type_count = 1 | PGT_validated + B.count = 2 | PGC_allocated (A[x] holds a general ref) + +V1: UNPIN A. + A begins de-validation. + Arrange to be interrupted when i < x + V1->old_guest_table = A + V1->old_guest_table_ref_held = false + A.type_count = 1 | PGT_partial + A.nr_validated_entries = i < x + B.type_count = 0 + B.count = 1 | PGC_allocated + +V2: MOD_L4_ENTRY to point some l4e to A. + Picks up re-validation of A. + Arrange to be interrupted halfway through B's validation + B.type_count = 1 | PGT_partial + B.count = 2 | PGC_allocated (PGT_partial holds a general ref) + A.type_count = 1 | PGT_partial + A.nr_validated_entries = x + A.partial_pte = PTF_partial_set + +V3: MOD_L3_ENTRY to point some other l3e (not in A) to B. + Validates B. + B.type_count = 1 | PGT_validated + B.count = 2 | PGC_allocated ("other l3e" holds a general ref) + +V3: MOD_L3_ENTRY to clear l3e pointing to B. + Devalidates B. + B.type_count = 0 + B.count = 1 | PGC_allocated + +V3: decrease_reservation(B) + Clears PGC_allocated + B.count = 0 => B is freed + +B gets assigned to a different domain + +V1: Restarts UNPIN of A + put_old_guest_table(A) + ... + free_l3_table(A) + +Now since A.partial_flags has PTF_partial_set, free_l3_table() will +call put_page_from_l3e() on A[x], which points to B, while B is owned +by another domain. + +If A[x] held a general refcount for B on partial validation, as it does +for partial de-validation, then B would still have a reference count of +1 after PGC_allocated was freed; so B wouldn't be freed until after +put_page_from_l3e() had happend on A[x]. +--- + xen/arch/x86/mm.c | 84 +++++++++++++++++++++++----------------- + xen/include/asm-x86/mm.h | 15 ++++--- + 2 files changed, 58 insertions(+), 41 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index 053465cb7c..68a9e74002 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -617,10 +617,11 @@ static int _get_page_type(struct page_info *page, unsigned long type, + * page->pte[page->nr_validated_entries]. See the comment in mm.h for + * more information. + */ +-#define PTF_partial_set (1 << 0) +-#define PTF_partial_general_ref (1 << 1) +-#define PTF_preemptible (1 << 2) +-#define PTF_defer (1 << 3) ++#define PTF_partial_set (1 << 0) ++#define PTF_partial_general_ref (1 << 1) ++#define PTF_preemptible (1 << 2) ++#define PTF_defer (1 << 3) ++#define PTF_retain_ref_on_restart (1 << 4) + + static int get_page_and_type_from_mfn( + mfn_t mfn, unsigned long type, struct domain *d, +@@ -629,7 +630,11 @@ static int get_page_and_type_from_mfn( + struct page_info *page = mfn_to_page(mfn); + int rc; + bool preemptible = flags & PTF_preemptible, +- partial_ref = flags & PTF_partial_general_ref; ++ partial_ref = flags & PTF_partial_general_ref, ++ partial_set = flags & PTF_partial_set, ++ retain_ref = flags & PTF_retain_ref_on_restart; ++ ++ ASSERT(partial_ref == partial_set); + + if ( likely(!partial_ref) && + unlikely(!get_page_from_mfn(mfn, d)) ) +@@ -642,13 +647,15 @@ static int get_page_and_type_from_mfn( + * - page is fully validated (rc == 0) + * - page is not validated (rc < 0) but: + * - We came in with a reference (partial_ref) ++ * - page is partially validated (rc == -ERESTART), and the ++ * caller has asked the ref to be retained in that case + * - page is partially validated but there's been an error + * (page == current->arch.old_guest_table) + * + * The partial_ref-on-error clause is worth an explanation. There + * are two scenarios where partial_ref might be true coming in: +- * - mfn has been partially demoted as type `type`; i.e. has +- * PGT_partial set ++ * - mfn has been partially promoted / demoted as type `type`; ++ * i.e. has PGT_partial set + * - mfn has been partially demoted as L(type+1) (i.e., a linear + * page; e.g. we're being called from get_page_from_l2e with + * type == PGT_l1_table, but the mfn is PGT_l2_table) +@@ -671,7 +678,8 @@ static int get_page_and_type_from_mfn( + */ + if ( likely(!rc) || partial_ref ) + /* nothing */; +- else if ( page == current->arch.old_guest_table ) ++ else if ( page == current->arch.old_guest_table || ++ (retain_ref && rc == -ERESTART) ) + ASSERT(preemptible); + else + put_page(page); +@@ -1348,8 +1356,8 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == + PTF_partial_set ) + { +- ASSERT(!(flags & PTF_defer)); +- rc = _put_page_type(pg, PTF_preemptible, ptpg); ++ /* partial_set should always imply partial_ref */ ++ BUG(); + } + else if ( flags & PTF_defer ) + { +@@ -1394,8 +1402,8 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == + PTF_partial_set ) + { +- ASSERT(!(flags & PTF_defer)); +- return _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); ++ /* partial_set should always imply partial_ref */ ++ BUG(); + } + + if ( flags & PTF_defer ) +@@ -1425,8 +1433,8 @@ static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, + if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == + PTF_partial_set ) + { +- ASSERT(!(flags & PTF_defer)); +- return _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); ++ /* partial_set should always imply partial_ref */ ++ BUG(); + } + + if ( flags & PTF_defer ) +@@ -1550,13 +1558,22 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + (rc = get_page_from_l2e(pl2e[i], pfn, d, partial_flags)) > 0 ) + continue; + +- if ( rc == -ERESTART ) +- { +- page->nr_validated_ptes = i; +- /* Set 'set', retain 'general ref' */ +- page->partial_flags = partial_flags | PTF_partial_set; +- } +- else if ( rc == -EINTR && i ) ++ /* ++ * It shouldn't be possible for get_page_from_l2e to return ++ * -ERESTART, since we never call this with PTF_preemptible. ++ * (alloc_l1_table may return -EINTR on an L1TF-vulnerable ++ * entry.) ++ * ++ * NB that while on a "clean" promotion, we can never get ++ * PGT_partial. It is possible to arrange for an l2e to ++ * contain a partially-devalidated l2; but in that case, both ++ * of the following functions will fail anyway (the first ++ * because the page in question is not an l1; the second ++ * because the page is not fully validated). ++ */ ++ ASSERT(rc != -ERESTART); ++ ++ if ( rc == -EINTR && i ) + { + page->nr_validated_ptes = i; + page->partial_flags = 0; +@@ -1565,6 +1582,7 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + else if ( rc < 0 && rc != -EINTR ) + { + gdprintk(XENLOG_WARNING, "Failure in alloc_l2_table: slot %#x\n", i); ++ ASSERT(current->arch.old_guest_table == NULL); + if ( i ) + { + page->nr_validated_ptes = i; +@@ -1621,16 +1639,17 @@ static int alloc_l3_table(struct page_info *page) + rc = get_page_and_type_from_mfn( + l3e_get_mfn(pl3e[i]), + PGT_l2_page_table | PGT_pae_xen_l2, d, +- partial_flags | PTF_preemptible); ++ partial_flags | PTF_preemptible | PTF_retain_ref_on_restart); + } +- else if ( (rc = get_page_from_l3e(pl3e[i], pfn, d, partial_flags)) > 0 ) ++ else if ( (rc = get_page_from_l3e(pl3e[i], pfn, d, ++ partial_flags | PTF_retain_ref_on_restart)) > 0 ) + continue; + + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; + /* Set 'set', leave 'general ref' set if this entry was set */ +- page->partial_flags = partial_flags | PTF_partial_set; ++ page->partial_flags = PTF_partial_set | PTF_partial_general_ref; + } + else if ( rc == -EINTR && i ) + { +@@ -1791,14 +1810,15 @@ static int alloc_l4_table(struct page_info *page) + i++, partial_flags = 0 ) + { + if ( !is_guest_l4_slot(d, i) || +- (rc = get_page_from_l4e(pl4e[i], pfn, d, partial_flags)) > 0 ) ++ (rc = get_page_from_l4e(pl4e[i], pfn, d, ++ partial_flags | PTF_retain_ref_on_restart)) > 0 ) + continue; + + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; + /* Set 'set', leave 'general ref' set if this entry was set */ +- page->partial_flags = partial_flags | PTF_partial_set; ++ page->partial_flags = PTF_partial_set | PTF_partial_general_ref; + } + else if ( rc < 0 ) + { +@@ -1896,9 +1916,7 @@ static int free_l2_table(struct page_info *page) + else if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_flags = (partial_flags & PTF_partial_set) ? +- partial_flags : +- (PTF_partial_set | PTF_partial_general_ref); ++ page->partial_flags = PTF_partial_set | PTF_partial_general_ref; + } + else if ( rc == -EINTR && i < L2_PAGETABLE_ENTRIES - 1 ) + { +@@ -1946,9 +1964,7 @@ static int free_l3_table(struct page_info *page) + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_flags = (partial_flags & PTF_partial_set) ? +- partial_flags : +- (PTF_partial_set | PTF_partial_general_ref); ++ page->partial_flags = PTF_partial_set | PTF_partial_general_ref; + } + else if ( rc == -EINTR && i < L3_PAGETABLE_ENTRIES - 1 ) + { +@@ -1979,9 +1995,7 @@ static int free_l4_table(struct page_info *page) + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_flags = (partial_flags & PTF_partial_set) ? +- partial_flags : +- (PTF_partial_set | PTF_partial_general_ref); ++ page->partial_flags = PTF_partial_set | PTF_partial_general_ref; + } + else if ( rc == -EINTR && i < L4_PAGETABLE_ENTRIES - 1 ) + { +diff --git a/xen/include/asm-x86/mm.h b/xen/include/asm-x86/mm.h +index 46cba52941..dc9cb869dd 100644 +--- xen/include/asm-x86/mm.h.orig ++++ xen/include/asm-x86/mm.h +@@ -238,22 +238,25 @@ struct page_info + * page. + * + * This happens: +- * - During de-validation, if de-validation of the page was ++ * - During validation or de-validation, if the operation was + * interrupted + * - During validation, if an invalid entry is encountered and + * validation is preemptible + * - During validation, if PTF_partial_general_ref was set on +- * this entry to begin with (perhaps because we're picking +- * up from a partial de-validation). ++ * this entry to begin with (perhaps because it picked up a ++ * previous operation) + * +- * When resuming validation, if PTF_partial_general_ref is clear, +- * then a general reference must be re-acquired; if it is set, no +- * reference should be acquired. ++ * When resuming validation, if PTF_partial_general_ref is ++ * clear, then a general reference must be re-acquired; if it ++ * is set, no reference should be acquired. + * + * When resuming de-validation, if PTF_partial_general_ref is + * clear, no reference should be dropped; if it is set, a + * reference should be dropped. + * ++ * NB at the moment, PTF_partial_set should be set if and only if ++ * PTF_partial_general_ref is set. ++ * + * NB that PTF_partial_set and PTF_partial_general_ref are + * defined in mm.c, the only place where they are used. + * +-- +2.23.0 + +From 6811df7fb7a1d4bb5a75fec9cf41519b5c86c605 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 08/11] x86/mm: Collapse PTF_partial_set and + PTF_partial_general_ref into one + +...now that they are equivalent. No functional change intended. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 50 +++++++++++----------------------------- + xen/include/asm-x86/mm.h | 29 +++++++++++------------ + 2 files changed, 26 insertions(+), 53 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index 68a9e74002..4970b19aff 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -612,13 +612,12 @@ static int _get_page_type(struct page_info *page, unsigned long type, + + /* + * The following flags are used to specify behavior of various get and +- * put commands. The first two are also stored in page->partial_flags +- * to indicate the state of the page pointed to by ++ * put commands. The first is also stored in page->partial_flags to ++ * indicate the state of the page pointed to by + * page->pte[page->nr_validated_entries]. See the comment in mm.h for + * more information. + */ + #define PTF_partial_set (1 << 0) +-#define PTF_partial_general_ref (1 << 1) + #define PTF_preemptible (1 << 2) + #define PTF_defer (1 << 3) + #define PTF_retain_ref_on_restart (1 << 4) +@@ -630,13 +629,10 @@ static int get_page_and_type_from_mfn( + struct page_info *page = mfn_to_page(mfn); + int rc; + bool preemptible = flags & PTF_preemptible, +- partial_ref = flags & PTF_partial_general_ref, + partial_set = flags & PTF_partial_set, + retain_ref = flags & PTF_retain_ref_on_restart; + +- ASSERT(partial_ref == partial_set); +- +- if ( likely(!partial_ref) && ++ if ( likely(!partial_set) && + unlikely(!get_page_from_mfn(mfn, d)) ) + return -EINVAL; + +@@ -646,14 +642,14 @@ static int get_page_and_type_from_mfn( + * Retain the refcount if: + * - page is fully validated (rc == 0) + * - page is not validated (rc < 0) but: +- * - We came in with a reference (partial_ref) ++ * - We came in with a reference (partial_set) + * - page is partially validated (rc == -ERESTART), and the + * caller has asked the ref to be retained in that case + * - page is partially validated but there's been an error + * (page == current->arch.old_guest_table) + * +- * The partial_ref-on-error clause is worth an explanation. There +- * are two scenarios where partial_ref might be true coming in: ++ * The partial_set-on-error clause is worth an explanation. There ++ * are two scenarios where partial_set might be true coming in: + * - mfn has been partially promoted / demoted as type `type`; + * i.e. has PGT_partial set + * - mfn has been partially demoted as L(type+1) (i.e., a linear +@@ -676,7 +672,7 @@ static int get_page_and_type_from_mfn( + * count retained unless we succeeded, or the operation was + * preemptible. + */ +- if ( likely(!rc) || partial_ref ) ++ if ( likely(!rc) || partial_set ) + /* nothing */; + else if ( page == current->arch.old_guest_table || + (retain_ref && rc == -ERESTART) ) +@@ -1353,13 +1349,7 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + struct page_info *pg = l2e_get_page(l2e); + struct page_info *ptpg = mfn_to_page(_mfn(pfn)); + +- if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == +- PTF_partial_set ) +- { +- /* partial_set should always imply partial_ref */ +- BUG(); +- } +- else if ( flags & PTF_defer ) ++ if ( flags & PTF_defer ) + { + current->arch.old_guest_ptpg = ptpg; + current->arch.old_guest_table = pg; +@@ -1399,13 +1389,6 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + + pg = l3e_get_page(l3e); + +- if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == +- PTF_partial_set ) +- { +- /* partial_set should always imply partial_ref */ +- BUG(); +- } +- + if ( flags & PTF_defer ) + { + current->arch.old_guest_ptpg = mfn_to_page(_mfn(pfn)); +@@ -1430,13 +1413,6 @@ static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, + { + struct page_info *pg = l4e_get_page(l4e); + +- if ( (flags & (PTF_partial_set | PTF_partial_general_ref)) == +- PTF_partial_set ) +- { +- /* partial_set should always imply partial_ref */ +- BUG(); +- } +- + if ( flags & PTF_defer ) + { + current->arch.old_guest_ptpg = mfn_to_page(_mfn(pfn)); +@@ -1649,7 +1625,7 @@ static int alloc_l3_table(struct page_info *page) + { + page->nr_validated_ptes = i; + /* Set 'set', leave 'general ref' set if this entry was set */ +- page->partial_flags = PTF_partial_set | PTF_partial_general_ref; ++ page->partial_flags = PTF_partial_set; + } + else if ( rc == -EINTR && i ) + { +@@ -1818,7 +1794,7 @@ static int alloc_l4_table(struct page_info *page) + { + page->nr_validated_ptes = i; + /* Set 'set', leave 'general ref' set if this entry was set */ +- page->partial_flags = PTF_partial_set | PTF_partial_general_ref; ++ page->partial_flags = PTF_partial_set; + } + else if ( rc < 0 ) + { +@@ -1916,7 +1892,7 @@ static int free_l2_table(struct page_info *page) + else if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_flags = PTF_partial_set | PTF_partial_general_ref; ++ page->partial_flags = PTF_partial_set; + } + else if ( rc == -EINTR && i < L2_PAGETABLE_ENTRIES - 1 ) + { +@@ -1964,7 +1940,7 @@ static int free_l3_table(struct page_info *page) + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_flags = PTF_partial_set | PTF_partial_general_ref; ++ page->partial_flags = PTF_partial_set; + } + else if ( rc == -EINTR && i < L3_PAGETABLE_ENTRIES - 1 ) + { +@@ -1995,7 +1971,7 @@ static int free_l4_table(struct page_info *page) + if ( rc == -ERESTART ) + { + page->nr_validated_ptes = i; +- page->partial_flags = PTF_partial_set | PTF_partial_general_ref; ++ page->partial_flags = PTF_partial_set; + } + else if ( rc == -EINTR && i < L4_PAGETABLE_ENTRIES - 1 ) + { +diff --git a/xen/include/asm-x86/mm.h b/xen/include/asm-x86/mm.h +index dc9cb869dd..c6ba9e4d73 100644 +--- xen/include/asm-x86/mm.h.orig ++++ xen/include/asm-x86/mm.h +@@ -233,7 +233,7 @@ struct page_info + * operation on the current page. (That page may or may not + * still have PGT_partial set.) + * +- * If PTF_partial_general_ref is set, then the PTE at ++ * Additionally, if PTF_partial_set is set, then the PTE at + * @nr_validated_ptef holds a general reference count for the + * page. + * +@@ -242,23 +242,20 @@ struct page_info + * interrupted + * - During validation, if an invalid entry is encountered and + * validation is preemptible +- * - During validation, if PTF_partial_general_ref was set on +- * this entry to begin with (perhaps because it picked up a ++ * - During validation, if PTF_partial_set was set on this ++ * entry to begin with (perhaps because it picked up a + * previous operation) + * +- * When resuming validation, if PTF_partial_general_ref is +- * clear, then a general reference must be re-acquired; if it +- * is set, no reference should be acquired. ++ * When resuming validation, if PTF_partial_set is clear, then ++ * a general reference must be re-acquired; if it is set, no ++ * reference should be acquired. + * +- * When resuming de-validation, if PTF_partial_general_ref is +- * clear, no reference should be dropped; if it is set, a +- * reference should be dropped. ++ * When resuming de-validation, if PTF_partial_set is clear, ++ * no reference should be dropped; if it is set, a reference ++ * should be dropped. + * +- * NB at the moment, PTF_partial_set should be set if and only if +- * PTF_partial_general_ref is set. +- * +- * NB that PTF_partial_set and PTF_partial_general_ref are +- * defined in mm.c, the only place where they are used. ++ * NB that PTF_partial_set is defined in mm.c, the only place ++ * where it is used. + * + * The 3rd field, @linear_pt_count, indicates + * - by a positive value, how many same-level page table entries a page +@@ -268,8 +265,8 @@ struct page_info + */ + struct { + u16 nr_validated_ptes:PAGETABLE_ORDER + 1; +- u16 :16 - PAGETABLE_ORDER - 1 - 2; +- u16 partial_flags:2; ++ u16 :16 - PAGETABLE_ORDER - 1 - 1; ++ u16 partial_flags:1; + s16 linear_pt_count; + }; + +-- +2.23.0 + +From a6098b8920b02149220641cb13358e9012b5fc4d Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 09/11] x86/mm: Properly handle linear pagetable promotion + failures + +In order to allow recursive pagetable promotions and demotions to be +interrupted, Xen must keep track of the state of the sub-pages +promoted or demoted. This is stored in two elements in the page +struct: nr_entries_validated and partial_flags. + +The rule is that entries [0, nr_entries_validated) should always be +validated and hold a general reference count. If partial_flags is +zero, then [nr_entries_validated] is not validated and no reference +count is held. If PTF_partial_set is set, then [nr_entries_validated] +is partially validated, and a general reference count is held. + +Unfortunately, in cases where an entry began with PTF_partial_set set, +and get_page_from_lNe() returns -EINVAL, the PTF_partial_set bit is +erroneously dropped. (This scenario can be engineered mainly by the +use of interleaving of promoting and demoting a page which has "linear +pagetable" entries; see the appendix for a sketch.) This means that +we will "leak" a general reference count on the page in question, +preventing the page from being freed. + +Fix this by setting page->partial_flags to the partial_flags local +variable. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +----- +Appendix + +Suppose A and B can both be promoted to L2 pages, and A[x] points to B. + +V1: PIN_L2 B. + B.type_count = 1 | PGT_validated + B.count = 2 | PGC_allocated + +V1: MOD_L3_ENTRY pointing something to A. + In the process of validating A[x], grab an extra type / ref on B: + B.type_count = 2 | PGT_validated + B.count = 3 | PGC_allocated + A.type_count = 1 | PGT_validated + A.count = 2 | PGC_allocated + +V1: UNPIN B. + B.type_count = 1 | PGT_validate + B.count = 2 | PGC_allocated + +V1: MOD_L3_ENTRY removing the reference to A. + De-validate A, down to A[x], which points to B. + Drop the final type on B. Arrange to be interrupted. + B.type_count = 1 | PGT_partial + B.count = 2 | PGC_allocated + A.type_count = 1 | PGT_partial + A.nr_validated_entries = x + A.partial_pte = -1 + +V2: MOD_L3_ENTRY adds a reference to A. + +At this point, get_page_from_l2e(A[x]) tries +get_page_and_type_from_mfn(), which fails because it's the wrong type; +and get_l2_linear_pagetable() also fails, because B isn't validated as +an l2 anymore. +--- + xen/arch/x86/mm.c | 6 +++--- + 1 file changed, 3 insertions(+), 3 deletions(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index 4970b19aff..cfb7538403 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -1562,7 +1562,7 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + if ( i ) + { + page->nr_validated_ptes = i; +- page->partial_flags = 0; ++ page->partial_flags = partial_flags; + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; + } +@@ -1647,7 +1647,7 @@ static int alloc_l3_table(struct page_info *page) + if ( i ) + { + page->nr_validated_ptes = i; +- page->partial_flags = 0; ++ page->partial_flags = partial_flags; + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; + } +@@ -1804,7 +1804,7 @@ static int alloc_l4_table(struct page_info *page) + if ( i ) + { + page->nr_validated_ptes = i; +- page->partial_flags = 0; ++ page->partial_flags = partial_flags; + if ( rc == -EINTR ) + rc = -ERESTART; + else +-- +2.23.0 + +From eabd77b59f4006128501d6e15f9e620dfb349420 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:49 +0100 +Subject: [PATCH 10/11] x86/mm: Fix nested de-validation on error + +If an invalid entry is discovered when validating a page-table tree, +the entire tree which has so far been validated must be de-validated. +Since this may take a long time, alloc_l[2-4]_table() set current +vcpu's old_guest_table immediately; put_old_guest_table() will make +sure that put_page_type() will be called to finish off the +de-validation before any other MMU operations can happen on the vcpu. + +The invariant for partial pages should be: + +* Entries [0, nr_validated_ptes) should be completely validated; + put_page_type() will de-validate these. + +* If [nr_validated_ptes] is partially validated, partial_flags should + set PTF_partiaL_set. put_page_type() will be called on this page to + finish off devalidation, and the appropriate refcount adjustments + will be done. + +alloc_l[2-3]_table() indicates partial validation to its callers by +setting current->old_guest_table. + +Unfortunately, this is mishandled. + +Take the case where validating lNe[x] returns an error. + +First, alloc_l3_table() doesn't check old_guest_table at all; as a +result, partial_flags is not set when it should be. nr_validated_ptes +is set to x; and since PFT_partial_set clear, de-validation resumes at +nr_validated_ptes-1. This means that the l2 page at pl3e[x] will not +have put_page_type() called on it when de-validating the rest of the +l3: it will be stuck in the PGT_partial state until the domain is +destroyed, or until it is re-used as an l2. (Any other page type will +fail.) + +Worse, alloc_l4_table(), rather than setting PTF_partial_set as it +should, sets nr_validated_ptes to x+1. When de-validating, since +partial is 0, this will correctly resume calling put_page_type at [x]; +but, if the put_page_type() is never called, but instead +get_page_type() is called, validation will pick up at [x+1], +neglecting to validate [x]. If the rest of the validation succeeds, +the l4 will be validated even though [x] is invalid. + +Fix this in both cases by setting PTF_partial_set if old_guest_table +is set. + +While here, add some safety catches: +- old_guest_table must point to the page contained in + [nr_validated_ptes]. +- alloc_l1_page shouldn't set old_guest_table + +If we experience one of these situations in production builds, it's +safer to avoid calling put_page_type for the pages in question. If +they have PGT_partial set, they will be cleaned up on domain +destruction; if not, we have no idea whether a type count is safe to +drop. Retaining an extra type ref that should have been dropped may +trigger a BUG() on the free_domain_page() path, but dropping a type +count that shouldn't be dropped may cause a privilege escalation. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +--- + xen/arch/x86/mm.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++- + 1 file changed, 54 insertions(+), 1 deletion(-) + +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index cfb7538403..aa03cb8b40 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -1561,6 +1561,20 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + ASSERT(current->arch.old_guest_table == NULL); + if ( i ) + { ++ /* ++ * alloc_l1_table() doesn't set old_guest_table; it does ++ * its own tear-down immediately on failure. If it ++ * did we'd need to check it and set partial_flags as we ++ * do in alloc_l[34]_table(). ++ * ++ * Note on the use of ASSERT: if it's non-null and ++ * hasn't been cleaned up yet, it should have ++ * PGT_partial set; and so the type will be cleaned up ++ * on domain destruction. Unfortunately, we would ++ * leak the general ref held by old_guest_table; but ++ * leaking a page is less bad than a host crash. ++ */ ++ ASSERT(current->arch.old_guest_table == NULL); + page->nr_validated_ptes = i; + page->partial_flags = partial_flags; + current->arch.old_guest_ptpg = NULL; +@@ -1588,6 +1602,7 @@ static int alloc_l3_table(struct page_info *page) + unsigned int i; + int rc = 0; + unsigned int partial_flags = page->partial_flags; ++ l3_pgentry_t l3e = l3e_empty(); + + pl3e = map_domain_page(_mfn(pfn)); + +@@ -1634,7 +1649,11 @@ static int alloc_l3_table(struct page_info *page) + rc = -ERESTART; + } + if ( rc < 0 ) ++ { ++ /* XSA-299 Backport: Copy l3e for checking */ ++ l3e = pl3e[i]; + break; ++ } + + pl3e[i] = adjust_guest_l3e(pl3e[i], d); + } +@@ -1648,6 +1667,24 @@ static int alloc_l3_table(struct page_info *page) + { + page->nr_validated_ptes = i; + page->partial_flags = partial_flags; ++ if ( current->arch.old_guest_table ) ++ { ++ /* ++ * We've experienced a validation failure. If ++ * old_guest_table is set, "transfer" the general ++ * reference count to pl3e[nr_validated_ptes] by ++ * setting PTF_partial_set. ++ * ++ * As a precaution, check that old_guest_table is the ++ * page pointed to by pl3e[nr_validated_ptes]. If ++ * not, it's safer to leak a type ref on production ++ * builds. ++ */ ++ if ( current->arch.old_guest_table == l3e_get_page(l3e) ) ++ page->partial_flags = PTF_partial_set; ++ else ++ ASSERT_UNREACHABLE(); ++ } + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; + } +@@ -1810,7 +1847,23 @@ static int alloc_l4_table(struct page_info *page) + else + { + if ( current->arch.old_guest_table ) +- page->nr_validated_ptes++; ++ { ++ /* ++ * We've experienced a validation failure. If ++ * old_guest_table is set, "transfer" the general ++ * reference count to pl3e[nr_validated_ptes] by ++ * setting PTF_partial_set. ++ * ++ * As a precaution, check that old_guest_table is the ++ * page pointed to by pl4e[nr_validated_ptes]. If ++ * not, it's safer to leak a type ref on production ++ * builds. ++ */ ++ if ( current->arch.old_guest_table == l4e_get_page(pl4e[i]) ) ++ page->partial_flags = PTF_partial_set; ++ else ++ ASSERT_UNREACHABLE(); ++ } + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; + } +-- +2.23.0 + +From f0086e3ac65c8bcabb84c1c29ab00b0c8a187555 Mon Sep 17 00:00:00 2001 +From: George Dunlap <george.dunlap@citrix.com> +Date: Thu, 10 Oct 2019 17:57:50 +0100 +Subject: [PATCH 11/11] x86/mm: Don't drop a type ref unless you held a ref to + begin with + +Validation and de-validation of pagetable trees may take arbitrarily +large amounts of time, and so must be preemptible. This is indicated +by setting the PGT_partial bit in the type_info, and setting +nr_validated_entries and partial_flags appropriately. Specifically, +if the entry at [nr_validated_entries] is partially validated, +partial_flags should have the PGT_partial_set bit set, and the entry +should hold a general reference count. During de-validation, +put_page_type() is called on partially validated entries. + +Unfortunately, there are a number of issues with the current algorithm. + +First, doing a "normal" put_page_type() is not safe when no type ref +is held: there is nothing to stop another vcpu from coming along and +picking up validation again: at which point the put_page_type may drop +the only page ref on an in-use page. Some examples are listed in the +appendix. + +The core issue is that put_page_type() is being called both to clean +up PGT_partial, and to drop a type count; and has no way of knowing +which is which; and so if in between, PGT_partial is cleared, +put_page_type() will drop the type ref erroneously. + +What is needed is to distinguish between two states: +- Dropping a type ref which is held +- Cleaning up a page which has been partially de/validated + +Fix this by telling put_page_type() which of the two activities you +intend. + +When cleaning up a partial de/validation, take no action unless you +find a page partially validated. + +If put_page_type() is called without PTF_partial_set, and finds the +page in a PGT_partial state anyway, then there's certainly been a +misaccounting somewhere, and carrying on would almost certainly cause +a security issue, so crash the host instead. + +In put_page_from_lNe, pass partial_flags on to _put_page_type(). + +old_guest_table may be set either with a fully validated page (when +using the "deferred put" pattern), or with a partially validated page +(when a normal "de-validation" is interrupted, or when a validation +fails part-way through due to invalid entries). Add a flag, +old_guest_table_partial, to indicate which of these it is, and use +that to pass the appropriate flag to _put_page_type(). + +While here, delete stray trailing whitespace. + +This is part of XSA-299. + +Reported-by: George Dunlap <george.dunlap@citrix.com> +Signed-off-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +----- +Appendix: + +Suppose page A, when interpreted as an l3 pagetable, contains all +valid entries; and suppose A[x] points to page B, which when +interpreted as an l2 pagetable, contains all valid entries. + +P1: PIN_L3_TABLE + A -> PGT_l3_table | 1 | valid + B -> PGT_l2_table | 1 | valid + +P1: UNPIN_TABLE + > Arrange to interrupt after B has been de-validated + B: + type_info -> PGT_l2_table | 0 + A: + type_info -> PGT_l3_table | 1 | partial + nr_validated_enties -> (less than x) + +P2: mod_l4_entry to point to A + > Arrange for this to be interrupted while B is being validated + B: + type_info -> PGT_l2_table | 1 | partial + (nr_validated_entires &c set as appropriate) + A: + type_info -> PGT_l3_table | 1 | partial + nr_validated_entries -> x + partial_pte = 1 + +P3: mod_l3_entry some other unrelated l3 to point to B: + B: + type_info -> PGT_l2_table | 1 + +P1: Restart UNPIN_TABLE + +At this point, since A.nr_validate_entries == x and A.partial_pte != +0, free_l3_table() will call put_page_from_l3e() on pl3e[x], dropping +its type count to 0 while it's still being pointed to by some other l3 + +A similar issue arises with old_guest_table. Consider the following +scenario: + +Suppose A is a page which, when interpreted as an l2, has valid entries +until entry x, which is invalid. + +V1: PIN_L2_TABLE(A) + <Validate until we try to validate [x], get -EINVAL> + A -> PGT_l2_table | 1 | PGT_partial + V1 -> old_guest_table = A + <delayed> + +V2: PIN_L2_TABLE(A) + <Pick up where V1 left off, try to re-validate [x], get -EINVAL> + A -> PGT_l2_table | 1 | PGT_partial + V2 -> old_guest_table = A + <restart> + put_old_guest_table() + _put_page_type(A) + A -> PGT_l2_table | 0 + +V1: <restart> + put_old_guest_table() + _put_page_type(A) # UNDERFLOW + +Indeed, it is possible to engineer for old_guest_table for every vcpu +a guest has to point to the same page. +--- + xen/arch/x86/domain.c | 6 +++ + xen/arch/x86/mm.c | 99 +++++++++++++++++++++++++++++++----- + xen/include/asm-x86/domain.h | 4 +- + 3 files changed, 95 insertions(+), 14 deletions(-) + +diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c +index 8fbecbb169..c880568dd4 100644 +--- xen/arch/x86/domain.c.orig ++++ xen/arch/x86/domain.c +@@ -1074,9 +1074,15 @@ int arch_set_info_guest( + rc = -ERESTART; + /* Fallthrough */ + case -ERESTART: ++ /* ++ * NB that we're putting the kernel-mode table ++ * here, which we've already successfully ++ * validated above; hence partial = false; ++ */ + v->arch.old_guest_ptpg = NULL; + v->arch.old_guest_table = + pagetable_get_page(v->arch.guest_table); ++ v->arch.old_guest_table_partial = false; + v->arch.guest_table = pagetable_null(); + break; + default: +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index aa03cb8b40..c701c7ef14 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -1353,10 +1353,11 @@ static int put_page_from_l2e(l2_pgentry_t l2e, unsigned long pfn, + { + current->arch.old_guest_ptpg = ptpg; + current->arch.old_guest_table = pg; ++ current->arch.old_guest_table_partial = false; + } + else + { +- rc = _put_page_type(pg, PTF_preemptible, ptpg); ++ rc = _put_page_type(pg, flags | PTF_preemptible, ptpg); + if ( likely(!rc) ) + put_page(pg); + } +@@ -1379,6 +1380,7 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + unsigned long mfn = l3e_get_pfn(l3e); + int writeable = l3e_get_flags(l3e) & _PAGE_RW; + ++ ASSERT(!(flags & PTF_partial_set)); + ASSERT(!(mfn & ((1UL << (L3_PAGETABLE_SHIFT - PAGE_SHIFT)) - 1))); + do { + put_data_page(mfn_to_page(_mfn(mfn)), writeable); +@@ -1391,12 +1393,14 @@ static int put_page_from_l3e(l3_pgentry_t l3e, unsigned long pfn, + + if ( flags & PTF_defer ) + { ++ ASSERT(!(flags & PTF_partial_set)); + current->arch.old_guest_ptpg = mfn_to_page(_mfn(pfn)); + current->arch.old_guest_table = pg; ++ current->arch.old_guest_table_partial = false; + return 0; + } + +- rc = _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); ++ rc = _put_page_type(pg, flags | PTF_preemptible, mfn_to_page(_mfn(pfn))); + if ( likely(!rc) ) + put_page(pg); + +@@ -1415,12 +1419,15 @@ static int put_page_from_l4e(l4_pgentry_t l4e, unsigned long pfn, + + if ( flags & PTF_defer ) + { ++ ASSERT(!(flags & PTF_partial_set)); + current->arch.old_guest_ptpg = mfn_to_page(_mfn(pfn)); + current->arch.old_guest_table = pg; ++ current->arch.old_guest_table_partial = false; + return 0; + } + +- rc = _put_page_type(pg, PTF_preemptible, mfn_to_page(_mfn(pfn))); ++ rc = _put_page_type(pg, flags | PTF_preemptible, ++ mfn_to_page(_mfn(pfn))); + if ( likely(!rc) ) + put_page(pg); + } +@@ -1525,6 +1532,14 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + + pl2e = map_domain_page(_mfn(pfn)); + ++ /* ++ * NB that alloc_l2_table will never set partial_pte on an l2; but ++ * free_l2_table might if a linear_pagetable entry is interrupted ++ * partway through de-validation. In that circumstance, ++ * get_page_from_l2e() will always return -EINVAL; and we must ++ * retain the type ref by doing the normal partial_flags tracking. ++ */ ++ + for ( i = page->nr_validated_ptes; i < L2_PAGETABLE_ENTRIES; + i++, partial_flags = 0 ) + { +@@ -1579,6 +1594,7 @@ static int alloc_l2_table(struct page_info *page, unsigned long type) + page->partial_flags = partial_flags; + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; ++ current->arch.old_guest_table_partial = true; + } + } + if ( rc < 0 ) +@@ -1681,12 +1697,16 @@ static int alloc_l3_table(struct page_info *page) + * builds. + */ + if ( current->arch.old_guest_table == l3e_get_page(l3e) ) ++ { ++ ASSERT(current->arch.old_guest_table_partial); + page->partial_flags = PTF_partial_set; ++ } + else + ASSERT_UNREACHABLE(); + } + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; ++ current->arch.old_guest_table_partial = true; + } + while ( i-- > 0 ) + pl3e[i] = unadjust_guest_l3e(pl3e[i], d); +@@ -1860,12 +1880,16 @@ static int alloc_l4_table(struct page_info *page) + * builds. + */ + if ( current->arch.old_guest_table == l4e_get_page(pl4e[i]) ) ++ { ++ ASSERT(current->arch.old_guest_table_partial); + page->partial_flags = PTF_partial_set; ++ } + else + ASSERT_UNREACHABLE(); + } + current->arch.old_guest_ptpg = NULL; + current->arch.old_guest_table = page; ++ current->arch.old_guest_table_partial = true; + } + } + } +@@ -2782,6 +2806,28 @@ static int _put_page_type(struct page_info *page, unsigned int flags, + x = y; + nx = x - 1; + ++ /* ++ * Is this expected to do a full reference drop, or only ++ * cleanup partial validation / devalidation? ++ * ++ * If the former, the caller must hold a "full" type ref; ++ * which means the page must be validated. If the page is ++ * *not* fully validated, continuing would almost certainly ++ * open up a security hole. An exception to this is during ++ * domain destruction, where PGT_validated can be dropped ++ * without dropping a type ref. ++ * ++ * If the latter, do nothing unless type PGT_partial is set. ++ * If it is set, the type count must be 1. ++ */ ++ if ( !(flags & PTF_partial_set) ) ++ BUG_ON((x & PGT_partial) || ++ !((x & PGT_validated) || page_get_owner(page)->is_dying)); ++ else if ( !(x & PGT_partial) ) ++ return 0; ++ else ++ BUG_ON((x & PGT_count_mask) != 1); ++ + ASSERT((x & PGT_count_mask) != 0); + + switch ( nx & (PGT_locked | PGT_count_mask) ) +@@ -3041,17 +3087,34 @@ int put_old_guest_table(struct vcpu *v) + if ( !v->arch.old_guest_table ) + return 0; + +- switch ( rc = _put_page_type(v->arch.old_guest_table, PTF_preemptible, +- v->arch.old_guest_ptpg) ) ++ rc = _put_page_type(v->arch.old_guest_table, ++ PTF_preemptible | ++ ( v->arch.old_guest_table_partial ? ++ PTF_partial_set : 0 ), ++ v->arch.old_guest_ptpg); ++ ++ if ( rc == -ERESTART || rc == -EINTR ) + { +- case -EINTR: +- case -ERESTART: ++ v->arch.old_guest_table_partial = (rc == -ERESTART); + return -ERESTART; +- case 0: +- put_page(v->arch.old_guest_table); + } + ++ /* ++ * It shouldn't be possible for _put_page_type() to return ++ * anything else at the moment; but if it does happen in ++ * production, leaking the type ref is probably the best thing to ++ * do. Either way, drop the general ref held by old_guest_table. ++ */ ++ ASSERT(rc == 0); ++ ++ put_page(v->arch.old_guest_table); + v->arch.old_guest_table = NULL; ++ v->arch.old_guest_ptpg = NULL; ++ /* ++ * Safest default if someone sets old_guest_table without ++ * explicitly setting old_guest_table_partial. ++ */ ++ v->arch.old_guest_table_partial = true; + + return rc; + } +@@ -3201,11 +3264,11 @@ int new_guest_cr3(mfn_t mfn) + switch ( rc = put_page_and_type_preemptible(page) ) + { + case -EINTR: +- rc = -ERESTART; +- /* fallthrough */ + case -ERESTART: + curr->arch.old_guest_ptpg = NULL; + curr->arch.old_guest_table = page; ++ curr->arch.old_guest_table_partial = (rc == -ERESTART); ++ rc = -ERESTART; + break; + default: + BUG_ON(rc); +@@ -3479,6 +3542,7 @@ long do_mmuext_op( + { + curr->arch.old_guest_ptpg = NULL; + curr->arch.old_guest_table = page; ++ curr->arch.old_guest_table_partial = false; + } + } + } +@@ -3513,6 +3577,11 @@ long do_mmuext_op( + case -ERESTART: + curr->arch.old_guest_ptpg = NULL; + curr->arch.old_guest_table = page; ++ /* ++ * EINTR means we still hold the type ref; ERESTART ++ * means PGT_partial holds the type ref ++ */ ++ curr->arch.old_guest_table_partial = (rc == -ERESTART); + rc = 0; + break; + default: +@@ -3581,11 +3650,15 @@ long do_mmuext_op( + switch ( rc = put_page_and_type_preemptible(page) ) + { + case -EINTR: +- rc = -ERESTART; +- /* fallthrough */ + case -ERESTART: + curr->arch.old_guest_ptpg = NULL; + curr->arch.old_guest_table = page; ++ /* ++ * EINTR means we still hold the type ref; ++ * ERESTART means PGT_partial holds the ref ++ */ ++ curr->arch.old_guest_table_partial = (rc == -ERESTART); ++ rc = -ERESTART; + break; + default: + BUG_ON(rc); +diff --git a/xen/include/asm-x86/domain.h b/xen/include/asm-x86/domain.h +index 1ac5a96c08..360c38bd83 100644 +--- xen/include/asm-x86/domain.h.orig ++++ xen/include/asm-x86/domain.h +@@ -309,7 +309,7 @@ struct arch_domain + + struct paging_domain paging; + struct p2m_domain *p2m; +- /* To enforce lock ordering in the pod code wrt the ++ /* To enforce lock ordering in the pod code wrt the + * page_alloc lock */ + int page_alloc_unlock_level; + +@@ -542,6 +542,8 @@ struct arch_vcpu + struct page_info *old_guest_table; /* partially destructed pagetable */ + struct page_info *old_guest_ptpg; /* containing page table of the */ + /* former, if any */ ++ bool old_guest_table_partial; /* Are we dropping a type ref, or just ++ * finishing up a partial de-validation? */ + /* guest_table holds a ref to the page, and also a type-count unless + * shadow refcounts are in use */ + pagetable_t shadow_table[4]; /* (MFN) shadow(s) of guest */ +-- +2.23.0 + diff --git a/sysutils/xenkernel411/patches/patch-XSA302 b/sysutils/xenkernel411/patches/patch-XSA302 new file mode 100644 index 00000000000..36929eb5a1f --- /dev/null +++ b/sysutils/xenkernel411/patches/patch-XSA302 @@ -0,0 +1,537 @@ +$NetBSD: patch-XSA302,v 1.2.2.2 2019/11/16 22:10:07 bsiegert Exp $ + +From bbca29f88d9ad9c7e91125a3b5d5f13a23e5801f Mon Sep 17 00:00:00 2001 +From: Jan Beulich <jbeulich@suse.com> +Date: Wed, 2 Oct 2019 13:36:59 +0200 +Subject: [PATCH 1/2] IOMMU: add missing HVM check +MIME-Version: 1.0 +Content-Type: text/plain; charset=UTF-8 +Content-Transfer-Encoding: 8bit + +Fix an unguarded d->arch.hvm access in assign_device(). + +Signed-off-by: Jan Beulich <jbeulich@suse.com> +Reviewed-by: Roger Pau Monné <roger.pau@citrix.com> +Acked-by: Andrew Cooper <andrew.cooper3@citrix.com> + +(cherry picked from commit 41fd1009cd7416b73d745a77c24b4e8d1a296fe6) +Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> +--- + xen/drivers/passthrough/pci.c | 3 ++- + 1 file changed, 2 insertions(+), 1 deletion(-) + +diff --git a/xen/drivers/passthrough/pci.c b/xen/drivers/passthrough/pci.c +index f51cae7f4e..037aba7c94 100644 +--- xen/drivers/passthrough/pci.c.orig ++++ xen/drivers/passthrough/pci.c +@@ -1416,7 +1416,8 @@ static int assign_device(struct domain *d, u16 seg, u8 bus, u8 devfn, u32 flag) + /* Prevent device assign if mem paging or mem sharing have been + * enabled for this domain */ + if ( unlikely(!need_iommu(d) && +- (d->arch.hvm_domain.mem_sharing_enabled || ++ ((is_hvm_domain(d) && ++ d->arch.hvm_domain.mem_sharing_enabled) || + vm_event_check_ring(d->vm_event_paging) || + p2m_get_hostp2m(d)->global_logdirty)) ) + return -EXDEV; +-- +2.11.0 + +From ec99857f59f7f06236f11ca8b0b2303e5e745cc4 Mon Sep 17 00:00:00 2001 +From: Paul Durrant <paul.durrant@citrix.com> +Date: Mon, 14 Oct 2019 17:52:59 +0100 +Subject: [PATCH 2/2] passthrough: quarantine PCI devices + +When a PCI device is assigned to an untrusted domain, it is possible for +that domain to program the device to DMA to an arbitrary address. The +IOMMU is used to protect the host from malicious DMA by making sure that +the device addresses can only target memory assigned to the guest. However, +when the guest domain is torn down the device is assigned back to dom0, +thus allowing any in-flight DMA to potentially target critical host data. + +This patch introduces a 'quarantine' for PCI devices using dom_io. When +the toolstack makes a device assignable (by binding it to pciback), it +will now also assign it to DOMID_IO and the device will only be assigned +back to dom0 when the device is made unassignable again. Whilst device is +assignable it will only ever transfer between dom_io and guest domains. +dom_io is actually only used as a sentinel domain for quarantining purposes; +it is not configured with any IOMMU mappings. Assignment to dom_io simply +means that the device's initiator (requestor) identifier is not present in +the IOMMU's device table and thus any DMA transactions issued will be +terminated with a fault condition. + +In addition, a fix to assignment handling is made for VT-d. Failure +during the assignment step should not lead to a device still being +associated with its prior owner. Hand the device to DomIO temporarily, +until the assignment step has completed successfully. Remove the PI +hooks from the source domain then earlier as well. + +Failure of the recovery reassign_device_ownership() may not go silent: +There e.g. may still be left over RMRR mappings in the domain assignment +to which has failed, and hence we can't allow that domain to continue +executing. + +NOTE: This patch also includes one printk() cleanup; the + "XEN_DOMCTL_assign_device: " tag is dropped in iommu_do_pci_domctl(), + since similar printk()-s elsewhere also don't log such a tag. + +This is XSA-302. + +Signed-off-by: Paul Durrant <paul.durrant@citrix.com> +Signed-off-by: Jan Beulich <jbeulich@suse.com> +Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> +--- + tools/libxl/libxl_pci.c | 25 +++++++++++- + xen/arch/x86/mm.c | 2 + + xen/common/domctl.c | 14 ++++++- + xen/drivers/passthrough/amd/pci_amd_iommu.c | 10 ++++- + xen/drivers/passthrough/iommu.c | 9 +++++ + xen/drivers/passthrough/pci.c | 59 ++++++++++++++++++++++------- + xen/drivers/passthrough/vtd/iommu.c | 40 ++++++++++++++++--- + xen/include/xen/pci.h | 3 ++ + 8 files changed, 138 insertions(+), 24 deletions(-) + +diff --git a/tools/libxl/libxl_pci.c b/tools/libxl/libxl_pci.c +index 4755a0c93c..81890a91ac 100644 +--- tools/libxl/libxl_pci.c.orig ++++ tools/libxl/libxl_pci.c +@@ -754,6 +754,7 @@ static int libxl__device_pci_assignable_add(libxl__gc *gc, + libxl_device_pci *pcidev, + int rebind) + { ++ libxl_ctx *ctx = libxl__gc_owner(gc); + unsigned dom, bus, dev, func; + char *spath, *driver_path = NULL; + int rc; +@@ -779,7 +780,7 @@ static int libxl__device_pci_assignable_add(libxl__gc *gc, + } + if ( rc ) { + LOG(WARN, PCI_BDF" already assigned to pciback", dom, bus, dev, func); +- return 0; ++ goto quarantine; + } + + /* Check to see if there's already a driver that we need to unbind from */ +@@ -810,6 +811,19 @@ static int libxl__device_pci_assignable_add(libxl__gc *gc, + return ERROR_FAIL; + } + ++quarantine: ++ /* ++ * DOMID_IO is just a sentinel domain, without any actual mappings, ++ * so always pass XEN_DOMCTL_DEV_RDM_RELAXED to avoid assignment being ++ * unnecessarily denied. ++ */ ++ rc = xc_assign_device(ctx->xch, DOMID_IO, pcidev_encode_bdf(pcidev), ++ XEN_DOMCTL_DEV_RDM_RELAXED); ++ if ( rc < 0 ) { ++ LOG(ERROR, "failed to quarantine "PCI_BDF, dom, bus, dev, func); ++ return ERROR_FAIL; ++ } ++ + return 0; + } + +@@ -817,9 +831,18 @@ static int libxl__device_pci_assignable_remove(libxl__gc *gc, + libxl_device_pci *pcidev, + int rebind) + { ++ libxl_ctx *ctx = libxl__gc_owner(gc); + int rc; + char *driver_path; + ++ /* De-quarantine */ ++ rc = xc_deassign_device(ctx->xch, DOMID_IO, pcidev_encode_bdf(pcidev)); ++ if ( rc < 0 ) { ++ LOG(ERROR, "failed to de-quarantine "PCI_BDF, pcidev->domain, pcidev->bus, ++ pcidev->dev, pcidev->func); ++ return ERROR_FAIL; ++ } ++ + /* Unbind from pciback */ + if ( (rc=pciback_dev_is_assigned(gc, pcidev)) < 0 ) { + return ERROR_FAIL; +diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c +index e6a4cb28f8..c1ab57f9a5 100644 +--- xen/arch/x86/mm.c.orig ++++ xen/arch/x86/mm.c +@@ -295,9 +295,11 @@ void __init arch_init_memory(void) + * Initialise our DOMID_IO domain. + * This domain owns I/O pages that are within the range of the page_info + * array. Mappings occur at the priv of the caller. ++ * Quarantined PCI devices will be associated with this domain. + */ + dom_io = domain_create(DOMID_IO, NULL); + BUG_ON(IS_ERR(dom_io)); ++ INIT_LIST_HEAD(&dom_io->arch.pdev_list); + + /* + * Initialise our COW domain. +diff --git a/xen/common/domctl.c b/xen/common/domctl.c +index 9b7bc083ee..741d774cd1 100644 +--- xen/common/domctl.c.orig ++++ xen/common/domctl.c +@@ -392,6 +392,16 @@ long do_domctl(XEN_GUEST_HANDLE_PARAM(xen_domctl_t) u_domctl) + + switch ( op->cmd ) + { ++ case XEN_DOMCTL_assign_device: ++ case XEN_DOMCTL_deassign_device: ++ if ( op->domain == DOMID_IO ) ++ { ++ d = dom_io; ++ break; ++ } ++ else if ( op->domain == DOMID_INVALID ) ++ return -ESRCH; ++ /* fall through */ + case XEN_DOMCTL_test_assign_device: + if ( op->domain == DOMID_INVALID ) + { +@@ -413,7 +423,7 @@ long do_domctl(XEN_GUEST_HANDLE_PARAM(xen_domctl_t) u_domctl) + + if ( !domctl_lock_acquire() ) + { +- if ( d ) ++ if ( d && d != dom_io ) + rcu_unlock_domain(d); + return hypercall_create_continuation( + __HYPERVISOR_domctl, "h", u_domctl); +@@ -1148,7 +1158,7 @@ long do_domctl(XEN_GUEST_HANDLE_PARAM(xen_domctl_t) u_domctl) + domctl_lock_release(); + + domctl_out_unlock_domonly: +- if ( d ) ++ if ( d && d != dom_io ) + rcu_unlock_domain(d); + + if ( copyback && __copy_to_guest(u_domctl, op, 1) ) +diff --git a/xen/drivers/passthrough/amd/pci_amd_iommu.c b/xen/drivers/passthrough/amd/pci_amd_iommu.c +index 12d2695b89..ec8baae717 100644 +--- xen/drivers/passthrough/amd/pci_amd_iommu.c.orig ++++ xen/drivers/passthrough/amd/pci_amd_iommu.c +@@ -118,6 +118,10 @@ static void amd_iommu_setup_domain_device( + u8 bus = pdev->bus; + const struct domain_iommu *hd = dom_iommu(domain); + ++ /* dom_io is used as a sentinel for quarantined devices */ ++ if ( domain == dom_io ) ++ return; ++ + BUG_ON( !hd->arch.root_table || !hd->arch.paging_mode || + !iommu->dev_table.buffer ); + +@@ -305,6 +309,10 @@ void amd_iommu_disable_domain_device(struct domain *domain, + int req_id; + u8 bus = pdev->bus; + ++ /* dom_io is used as a sentinel for quarantined devices */ ++ if ( domain == dom_io ) ++ return; ++ + BUG_ON ( iommu->dev_table.buffer == NULL ); + req_id = get_dma_requestor_id(iommu->seg, PCI_BDF2(bus, devfn)); + dte = iommu->dev_table.buffer + (req_id * IOMMU_DEV_TABLE_ENTRY_SIZE); +@@ -391,7 +399,7 @@ static int amd_iommu_assign_device(struct domain *d, u8 devfn, + ivrs_mappings[req_id].read_permission); + } + +- return reassign_device(hardware_domain, d, devfn, pdev); ++ return reassign_device(pdev->domain, d, devfn, pdev); + } + + static void deallocate_next_page_table(struct page_info *pg, int level) +diff --git a/xen/drivers/passthrough/iommu.c b/xen/drivers/passthrough/iommu.c +index 04b0be37d3..8027d96f1c 100644 +--- xen/drivers/passthrough/iommu.c.orig ++++ xen/drivers/passthrough/iommu.c +@@ -219,6 +219,9 @@ void iommu_teardown(struct domain *d) + { + const struct domain_iommu *hd = dom_iommu(d); + ++ if ( d == dom_io ) ++ return; ++ + d->need_iommu = 0; + hd->platform_ops->teardown(d); + tasklet_schedule(&iommu_pt_cleanup_tasklet); +@@ -229,6 +232,9 @@ int iommu_construct(struct domain *d) + if ( need_iommu(d) > 0 ) + return 0; + ++ if ( d == dom_io ) ++ return 0; ++ + if ( !iommu_use_hap_pt(d) ) + { + int rc; +@@ -404,6 +410,9 @@ int __init iommu_setup(void) + printk("I/O virtualisation %sabled\n", iommu_enabled ? "en" : "dis"); + if ( iommu_enabled ) + { ++ if ( iommu_domain_init(dom_io) ) ++ panic("Could not set up quarantine\n"); ++ + printk(" - Dom0 mode: %s\n", + iommu_passthrough ? "Passthrough" : + iommu_dom0_strict ? "Strict" : "Relaxed"); +diff --git a/xen/drivers/passthrough/pci.c b/xen/drivers/passthrough/pci.c +index 037aba7c94..fb010a547b 100644 +--- xen/drivers/passthrough/pci.c.orig ++++ xen/drivers/passthrough/pci.c +@@ -1389,19 +1389,29 @@ static int iommu_remove_device(struct pci_dev *pdev) + return hd->platform_ops->remove_device(pdev->devfn, pci_to_dev(pdev)); + } + +-/* +- * If the device isn't owned by the hardware domain, it means it already +- * has been assigned to other domain, or it doesn't exist. +- */ + static int device_assigned(u16 seg, u8 bus, u8 devfn) + { + struct pci_dev *pdev; ++ int rc = 0; + + pcidevs_lock(); +- pdev = pci_get_pdev_by_domain(hardware_domain, seg, bus, devfn); ++ ++ pdev = pci_get_pdev(seg, bus, devfn); ++ ++ if ( !pdev ) ++ rc = -ENODEV; ++ /* ++ * If the device exists and it is not owned by either the hardware ++ * domain or dom_io then it must be assigned to a guest, or be ++ * hidden (owned by dom_xen). ++ */ ++ else if ( pdev->domain != hardware_domain && ++ pdev->domain != dom_io ) ++ rc = -EBUSY; ++ + pcidevs_unlock(); + +- return pdev ? 0 : -EBUSY; ++ return rc; + } + + static int assign_device(struct domain *d, u16 seg, u8 bus, u8 devfn, u32 flag) +@@ -1415,7 +1425,8 @@ static int assign_device(struct domain *d, u16 seg, u8 bus, u8 devfn, u32 flag) + + /* Prevent device assign if mem paging or mem sharing have been + * enabled for this domain */ +- if ( unlikely(!need_iommu(d) && ++ if ( d != dom_io && ++ unlikely(!need_iommu(d) && + ((is_hvm_domain(d) && + d->arch.hvm_domain.mem_sharing_enabled) || + vm_event_check_ring(d->vm_event_paging) || +@@ -1432,12 +1443,20 @@ static int assign_device(struct domain *d, u16 seg, u8 bus, u8 devfn, u32 flag) + return rc; + } + +- pdev = pci_get_pdev_by_domain(hardware_domain, seg, bus, devfn); ++ pdev = pci_get_pdev(seg, bus, devfn); ++ ++ rc = -ENODEV; + if ( !pdev ) +- { +- rc = pci_get_pdev(seg, bus, devfn) ? -EBUSY : -ENODEV; + goto done; +- } ++ ++ rc = 0; ++ if ( d == pdev->domain ) ++ goto done; ++ ++ rc = -EBUSY; ++ if ( pdev->domain != hardware_domain && ++ pdev->domain != dom_io ) ++ goto done; + + if ( pdev->msix ) + msixtbl_init(d); +@@ -1460,6 +1479,10 @@ static int assign_device(struct domain *d, u16 seg, u8 bus, u8 devfn, u32 flag) + } + + done: ++ /* The device is assigned to dom_io so mark it as quarantined */ ++ if ( !rc && d == dom_io ) ++ pdev->quarantine = true; ++ + if ( !has_arch_pdevs(d) && need_iommu(d) ) + iommu_teardown(d); + pcidevs_unlock(); +@@ -1472,6 +1495,7 @@ int deassign_device(struct domain *d, u16 seg, u8 bus, u8 devfn) + { + const struct domain_iommu *hd = dom_iommu(d); + struct pci_dev *pdev = NULL; ++ struct domain *target; + int ret = 0; + + if ( !iommu_enabled || !hd->platform_ops ) +@@ -1482,12 +1506,16 @@ int deassign_device(struct domain *d, u16 seg, u8 bus, u8 devfn) + if ( !pdev ) + return -ENODEV; + ++ /* De-assignment from dom_io should de-quarantine the device */ ++ target = (pdev->quarantine && pdev->domain != dom_io) ? ++ dom_io : hardware_domain; ++ + while ( pdev->phantom_stride ) + { + devfn += pdev->phantom_stride; + if ( PCI_SLOT(devfn) != PCI_SLOT(pdev->devfn) ) + break; +- ret = hd->platform_ops->reassign_device(d, hardware_domain, devfn, ++ ret = hd->platform_ops->reassign_device(d, target, devfn, + pci_to_dev(pdev)); + if ( !ret ) + continue; +@@ -1498,7 +1526,7 @@ int deassign_device(struct domain *d, u16 seg, u8 bus, u8 devfn) + } + + devfn = pdev->devfn; +- ret = hd->platform_ops->reassign_device(d, hardware_domain, devfn, ++ ret = hd->platform_ops->reassign_device(d, target, devfn, + pci_to_dev(pdev)); + if ( ret ) + { +@@ -1508,6 +1536,9 @@ int deassign_device(struct domain *d, u16 seg, u8 bus, u8 devfn) + return ret; + } + ++ if ( pdev->domain == hardware_domain ) ++ pdev->quarantine = false; ++ + pdev->fault.count = 0; + + if ( !has_arch_pdevs(d) && need_iommu(d) ) +@@ -1686,7 +1717,7 @@ int iommu_do_pci_domctl( + ret = hypercall_create_continuation(__HYPERVISOR_domctl, + "h", u_domctl); + else if ( ret ) +- printk(XENLOG_G_ERR "XEN_DOMCTL_assign_device: " ++ printk(XENLOG_G_ERR + "assign %04x:%02x:%02x.%u to dom%d failed (%d)\n", + seg, bus, PCI_SLOT(devfn), PCI_FUNC(devfn), + d->domain_id, ret); +diff --git a/xen/drivers/passthrough/vtd/iommu.c b/xen/drivers/passthrough/vtd/iommu.c +index 4c719d4ee7..19f7d13013 100644 +--- xen/drivers/passthrough/vtd/iommu.c.orig ++++ xen/drivers/passthrough/vtd/iommu.c +@@ -1338,6 +1338,10 @@ int domain_context_mapping_one( + int agaw, rc, ret; + bool_t flush_dev_iotlb; + ++ /* dom_io is used as a sentinel for quarantined devices */ ++ if ( domain == dom_io ) ++ return 0; ++ + ASSERT(pcidevs_locked()); + spin_lock(&iommu->lock); + maddr = bus_to_context_maddr(iommu, bus); +@@ -1573,6 +1577,10 @@ int domain_context_unmap_one( + int iommu_domid, rc, ret; + bool_t flush_dev_iotlb; + ++ /* dom_io is used as a sentinel for quarantined devices */ ++ if ( domain == dom_io ) ++ return 0; ++ + ASSERT(pcidevs_locked()); + spin_lock(&iommu->lock); + +@@ -1705,6 +1713,10 @@ static int domain_context_unmap(struct domain *domain, u8 devfn, + goto out; + } + ++ /* dom_io is used as a sentinel for quarantined devices */ ++ if ( domain == dom_io ) ++ goto out; ++ + /* + * if no other devices under the same iommu owned by this domain, + * clear iommu in iommu_bitmap and clear domain_id in domid_bitmp +@@ -2389,6 +2401,15 @@ static int reassign_device_ownership( + if ( ret ) + return ret; + ++ if ( devfn == pdev->devfn ) ++ { ++ list_move(&pdev->domain_list, &dom_io->arch.pdev_list); ++ pdev->domain = dom_io; ++ } ++ ++ if ( !has_arch_pdevs(source) ) ++ vmx_pi_hooks_deassign(source); ++ + if ( !has_arch_pdevs(target) ) + vmx_pi_hooks_assign(target); + +@@ -2407,15 +2428,13 @@ static int reassign_device_ownership( + pdev->domain = target; + } + +- if ( !has_arch_pdevs(source) ) +- vmx_pi_hooks_deassign(source); +- + return ret; + } + + static int intel_iommu_assign_device( + struct domain *d, u8 devfn, struct pci_dev *pdev, u32 flag) + { ++ struct domain *s = pdev->domain; + struct acpi_rmrr_unit *rmrr; + int ret = 0, i; + u16 bdf, seg; +@@ -2458,8 +2477,8 @@ static int intel_iommu_assign_device( + } + } + +- ret = reassign_device_ownership(hardware_domain, d, devfn, pdev); +- if ( ret ) ++ ret = reassign_device_ownership(s, d, devfn, pdev); ++ if ( ret || d == dom_io ) + return ret; + + /* Setup rmrr identity mapping */ +@@ -2472,11 +2491,20 @@ static int intel_iommu_assign_device( + ret = rmrr_identity_mapping(d, 1, rmrr, flag); + if ( ret ) + { +- reassign_device_ownership(d, hardware_domain, devfn, pdev); ++ int rc; ++ ++ rc = reassign_device_ownership(d, s, devfn, pdev); + printk(XENLOG_G_ERR VTDPREFIX + " cannot map reserved region (%"PRIx64",%"PRIx64"] for Dom%d (%d)\n", + rmrr->base_address, rmrr->end_address, + d->domain_id, ret); ++ if ( rc ) ++ { ++ printk(XENLOG_ERR VTDPREFIX ++ " failed to reclaim %04x:%02x:%02x.%u from %pd (%d)\n", ++ seg, bus, PCI_SLOT(devfn), PCI_FUNC(devfn), d, rc); ++ domain_crash(d); ++ } + break; + } + } +diff --git a/xen/include/xen/pci.h b/xen/include/xen/pci.h +index 4cfa774615..066364bdef 100644 +--- xen/include/xen/pci.h.orig ++++ xen/include/xen/pci.h +@@ -88,6 +88,9 @@ struct pci_dev { + + nodeid_t node; /* NUMA node */ + ++ /* Device to be quarantined, don't automatically re-assign to dom0 */ ++ bool quarantine; ++ + enum pdev_type { + DEV_TYPE_PCI_UNKNOWN, + DEV_TYPE_PCIe_ENDPOINT, +-- +2.11.0 + diff --git a/sysutils/xenkernel411/patches/patch-XSA304 b/sysutils/xenkernel411/patches/patch-XSA304 new file mode 100644 index 00000000000..83b52f7bd31 --- /dev/null +++ b/sysutils/xenkernel411/patches/patch-XSA304 @@ -0,0 +1,481 @@ +$NetBSD: patch-XSA304,v 1.2.2.2 2019/11/16 22:10:07 bsiegert Exp $ + +From: Andrew Cooper <andrew.cooper3@citrix.com> +Subject: x86/vtd: Hide superpage support for SandyBridge IOMMUs + +Something causes SandyBridge IOMMUs to choke when sharing EPT pagetables, and +an EPT superpage gets shattered. The root cause is still under investigation, +but the end result is unusable in combination with CVE-2018-12207 protections. + +This is part of XSA-304 / CVE-2018-12207 + +Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> + +diff --git a/xen/drivers/passthrough/vtd/extern.h b/xen/drivers/passthrough/vtd/extern.h +index fb7edfaef9..d698b1d50a 100644 +--- xen/drivers/passthrough/vtd/extern.h.orig ++++ xen/drivers/passthrough/vtd/extern.h +@@ -96,6 +96,8 @@ void vtd_ops_postamble_quirk(struct iommu* iommu); + int __must_check me_wifi_quirk(struct domain *domain, + u8 bus, u8 devfn, int map); + void pci_vtd_quirk(const struct pci_dev *); ++void quirk_iommu_caps(struct iommu *iommu); ++ + bool_t platform_supports_intremap(void); + bool_t platform_supports_x2apic(void); + +diff --git a/xen/drivers/passthrough/vtd/iommu.c b/xen/drivers/passthrough/vtd/iommu.c +index f242e30caf..8712d3b4dc 100644 +--- xen/drivers/passthrough/vtd/iommu.c.orig ++++ xen/drivers/passthrough/vtd/iommu.c +@@ -1211,6 +1211,8 @@ int __init iommu_alloc(struct acpi_drhd_unit *drhd) + if ( !(iommu->cap + 1) || !(iommu->ecap + 1) ) + return -ENODEV; + ++ quirk_iommu_caps(iommu); ++ + if ( cap_fault_reg_offset(iommu->cap) + + cap_num_fault_regs(iommu->cap) * PRIMARY_FAULT_REG_LEN >= PAGE_SIZE || + ecap_iotlb_offset(iommu->ecap) >= PAGE_SIZE ) +diff --git a/xen/drivers/passthrough/vtd/quirks.c b/xen/drivers/passthrough/vtd/quirks.c +index d6db862678..b02688e316 100644 +--- xen/drivers/passthrough/vtd/quirks.c.orig ++++ xen/drivers/passthrough/vtd/quirks.c +@@ -540,3 +540,28 @@ void pci_vtd_quirk(const struct pci_dev *pdev) + break; + } + } ++ ++void __init quirk_iommu_caps(struct iommu *iommu) ++{ ++ /* ++ * IOMMU Quirks: ++ * ++ * SandyBridge IOMMUs claim support for 2M and 1G superpages, but don't ++ * implement superpages internally. ++ * ++ * There are issues changing the walk length under in-flight DMA, which ++ * has manifested as incompatibility between EPT/IOMMU sharing and the ++ * workaround for CVE-2018-12207 / XSA-304. Hide the superpages ++ * capabilities in the IOMMU, which will prevent Xen from sharing the EPT ++ * and IOMMU pagetables. ++ * ++ * Detection of SandyBridge unfortunately has to be done by processor ++ * model because the client parts don't expose their IOMMUs as PCI devices ++ * we could match with a Device ID. ++ */ ++ if ( boot_cpu_data.x86_vendor == X86_VENDOR_INTEL && ++ boot_cpu_data.x86 == 6 && ++ (boot_cpu_data.x86_model == 0x2a || ++ boot_cpu_data.x86_model == 0x2d) ) ++ iommu->cap &= ~(0xful << 34); ++} +From: Andrew Cooper <andrew.cooper3@citrix.com> +Subject: x86/vtx: Disable executable EPT superpages to work around + CVE-2018-12207 + +CVE-2018-12207 covers a set of errata on various Intel processors, whereby a +machine check exception can be generated in a corner case when an executable +mapping changes size or cacheability without TLB invalidation. HVM guest +kernels can trigger this to DoS the host. + +To mitigate, in affected hardware, all EPT superpages are marked NX. When an +instruction fetch violation is observed against the superpage, the superpage +is shattered to 4k and has execute permissions restored. This prevents the +guest kernel from being able to create the necessary preconditions in the iTLB +to exploit the vulnerability. + +This does come with a workload-dependent performance overhead, caused by +increased TLB pressure. Performance can be restored, if guest kernels are +trusted not to mount an attack, by specifying ept=exec-sp on the command line. + +This is part of XSA-304 / CVE-2018-12207 + +Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> +Acked-by: George Dunlap <george.dunlap@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> + +diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown +index c63a07d29b..684671cb7b 100644 +--- docs/misc/xen-command-line.markdown.orig ++++ docs/misc/xen-command-line.markdown +@@ -828,7 +828,7 @@ effect the inverse meaning. + >> set as UC. + + ### ept (Intel) +-> `= List of ( {no-}pml | {no-}ad )` ++> `= List of [ {no-}pml, {no-}ad, {no-}exec-sp ]` + + Controls EPT related features. + +@@ -851,6 +851,16 @@ Controls EPT related features. + + >> Have hardware keep accessed/dirty (A/D) bits updated. + ++* The `exec-sp` boolean controls whether EPT superpages with execute ++ permissions are permitted. In general this is good for performance. ++ ++ However, on processors vulnerable CVE-2018-12207, HVM guest kernels can ++ use executable superpages to crash the host. By default, executable ++ superpages are disabled on affected hardware. ++ ++ If HVM guest kernels are trusted not to mount a DoS against the system, ++ this option can enabled to regain performance. ++ + ### extra\_guest\_irqs + > `= [<domU number>][,<dom0 number>]` + +diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c +index f4a6a37149..1924434960 100644 +--- xen/arch/x86/hvm/hvm.c.orig ++++ xen/arch/x86/hvm/hvm.c +@@ -1706,6 +1706,7 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla, + struct p2m_domain *p2m, *hostp2m; + int rc, fall_through = 0, paged = 0; + int sharing_enomem = 0; ++ unsigned int page_order = 0; + vm_event_request_t *req_ptr = NULL; + bool_t ap2m_active, sync = 0; + +@@ -1774,7 +1775,7 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla, + hostp2m = p2m_get_hostp2m(currd); + mfn = get_gfn_type_access(hostp2m, gfn, &p2mt, &p2ma, + P2M_ALLOC | (npfec.write_access ? P2M_UNSHARE : 0), +- NULL); ++ &page_order); + + if ( ap2m_active ) + { +@@ -1786,7 +1787,7 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla, + goto out; + } + +- mfn = get_gfn_type_access(p2m, gfn, &p2mt, &p2ma, 0, NULL); ++ mfn = get_gfn_type_access(p2m, gfn, &p2mt, &p2ma, 0, &page_order); + } + else + p2m = hostp2m; +@@ -1828,6 +1829,24 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla, + break; + } + ++ /* ++ * Workaround for XSA-304 / CVE-2018-12207. If we take an execution ++ * fault against a non-executable superpage, shatter it to regain ++ * execute permissions. ++ */ ++ if ( page_order > 0 && npfec.insn_fetch && npfec.present && !violation ) ++ { ++ int res = p2m_set_entry(p2m, _gfn(gfn), mfn, PAGE_ORDER_4K, ++ p2mt, p2ma); ++ ++ if ( res ) ++ printk(XENLOG_ERR "Failed to shatter gfn %"PRI_gfn": %d\n", ++ gfn, res); ++ ++ rc = !res; ++ goto out_put_gfn; ++ } ++ + if ( violation ) + { + /* Should #VE be emulated for this fault? */ +diff --git a/xen/arch/x86/hvm/vmx/vmcs.c b/xen/arch/x86/hvm/vmx/vmcs.c +index 493986e84a..8821a3b536 100644 +--- xen/arch/x86/hvm/vmx/vmcs.c.orig ++++ xen/arch/x86/hvm/vmx/vmcs.c +@@ -67,6 +67,7 @@ integer_param("ple_window", ple_window); + + static bool_t __read_mostly opt_pml_enabled = 1; + static s8 __read_mostly opt_ept_ad = -1; ++int8_t __read_mostly opt_ept_exec_sp = -1; + + /* + * The 'ept' parameter controls functionalities that depend on, or impact the +@@ -94,6 +95,8 @@ static int __init parse_ept_param(const char *s) + opt_pml_enabled = val; + else if ( !cmdline_strcmp(s, "ad") ) + opt_ept_ad = val; ++ else if ( !cmdline_strcmp(s, "exec-sp") ) ++ opt_ept_exec_sp = val; + else + rc = -EINVAL; + +diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c +index 840dc2b44d..a568d62643 100644 +--- xen/arch/x86/hvm/vmx/vmx.c.orig ++++ xen/arch/x86/hvm/vmx/vmx.c +@@ -2415,6 +2415,102 @@ static void pi_notification_interrupt(struct cpu_user_regs *regs) + static void __init lbr_tsx_fixup_check(void); + static void __init bdw_erratum_bdf14_fixup_check(void); + ++/* ++ * Calculate whether the CPU is vulnerable to Instruction Fetch page ++ * size-change MCEs. ++ */ ++static bool __init has_if_pschange_mc(void) ++{ ++ uint64_t caps = 0; ++ ++ /* ++ * If we are virtualised, there is nothing we can do. Our EPT tables are ++ * shadowed by our hypervisor, and not walked by hardware. ++ */ ++ if ( cpu_has_hypervisor ) ++ return false; ++ ++ if ( boot_cpu_has(X86_FEATURE_ARCH_CAPS) ) ++ rdmsrl(MSR_ARCH_CAPABILITIES, caps); ++ ++ if ( caps & ARCH_CAPS_IF_PSCHANGE_MC_NO ) ++ return false; ++ ++ /* ++ * IF_PSCHANGE_MC is only known to affect Intel Family 6 processors at ++ * this time. ++ */ ++ if ( boot_cpu_data.x86_vendor != X86_VENDOR_INTEL || ++ boot_cpu_data.x86 != 6 ) ++ return false; ++ ++ switch ( boot_cpu_data.x86_model ) ++ { ++ /* ++ * Core processors since at least Nehalem are vulnerable. ++ */ ++ case 0x1f: /* Auburndale / Havendale */ ++ case 0x1e: /* Nehalem */ ++ case 0x1a: /* Nehalem EP */ ++ case 0x2e: /* Nehalem EX */ ++ case 0x25: /* Westmere */ ++ case 0x2c: /* Westmere EP */ ++ case 0x2f: /* Westmere EX */ ++ case 0x2a: /* SandyBridge */ ++ case 0x2d: /* SandyBridge EP/EX */ ++ case 0x3a: /* IvyBridge */ ++ case 0x3e: /* IvyBridge EP/EX */ ++ case 0x3c: /* Haswell */ ++ case 0x3f: /* Haswell EX/EP */ ++ case 0x45: /* Haswell D */ ++ case 0x46: /* Haswell H */ ++ case 0x3d: /* Broadwell */ ++ case 0x47: /* Broadwell H */ ++ case 0x4f: /* Broadwell EP/EX */ ++ case 0x56: /* Broadwell D */ ++ case 0x4e: /* Skylake M */ ++ case 0x5e: /* Skylake D */ ++ case 0x55: /* Skylake-X / Cascade Lake */ ++ case 0x8e: /* Kaby / Coffee / Whiskey Lake M */ ++ case 0x9e: /* Kaby / Coffee / Whiskey Lake D */ ++ return true; ++ ++ /* ++ * Atom processors are not vulnerable. ++ */ ++ case 0x1c: /* Pineview */ ++ case 0x26: /* Lincroft */ ++ case 0x27: /* Penwell */ ++ case 0x35: /* Cloverview */ ++ case 0x36: /* Cedarview */ ++ case 0x37: /* Baytrail / Valleyview (Silvermont) */ ++ case 0x4d: /* Avaton / Rangely (Silvermont) */ ++ case 0x4c: /* Cherrytrail / Brasswell */ ++ case 0x4a: /* Merrifield */ ++ case 0x5a: /* Moorefield */ ++ case 0x5c: /* Goldmont */ ++ case 0x5d: /* SoFIA 3G Granite/ES2.1 */ ++ case 0x65: /* SoFIA LTE AOSP */ ++ case 0x5f: /* Denverton */ ++ case 0x6e: /* Cougar Mountain */ ++ case 0x75: /* Lightning Mountain */ ++ case 0x7a: /* Gemini Lake */ ++ case 0x86: /* Jacobsville */ ++ ++ /* ++ * Knights processors are not vulnerable. ++ */ ++ case 0x57: /* Knights Landing */ ++ case 0x85: /* Knights Mill */ ++ return false; ++ ++ default: ++ printk("Unrecognised CPU model %#x - assuming vulnerable to IF_PSCHANGE_MC\n", ++ boot_cpu_data.x86_model); ++ return true; ++ } ++} ++ + const struct hvm_function_table * __init start_vmx(void) + { + set_in_cr4(X86_CR4_VMXE); +@@ -2435,6 +2531,17 @@ const struct hvm_function_table * __init start_vmx(void) + */ + if ( cpu_has_vmx_ept && (cpu_has_vmx_pat || opt_force_ept) ) + { ++ bool cpu_has_bug_pschange_mc = has_if_pschange_mc(); ++ ++ if ( opt_ept_exec_sp == -1 ) ++ { ++ /* Default to non-executable superpages on vulnerable hardware. */ ++ opt_ept_exec_sp = !cpu_has_bug_pschange_mc; ++ ++ if ( cpu_has_bug_pschange_mc ) ++ printk("VMX: Disabling executable EPT superpages due to CVE-2018-12207\n"); ++ } ++ + vmx_function_table.hap_supported = 1; + vmx_function_table.altp2m_supported = 1; + +diff --git a/xen/arch/x86/mm/p2m-ept.c b/xen/arch/x86/mm/p2m-ept.c +index ce46201d45..93e08f89a2 100644 +--- xen/arch/x86/mm/p2m-ept.c.orig ++++ xen/arch/x86/mm/p2m-ept.c +@@ -215,6 +215,12 @@ static void ept_p2m_type_to_flags(struct p2m_domain *p2m, ept_entry_t *entry, + break; + } + ++ /* ++ * Don't create executable superpages if we need to shatter them to ++ * protect against CVE-2018-12207. ++ */ ++ if ( !opt_ept_exec_sp && is_epte_superpage(entry) ) ++ entry->x = 0; + } + + #define GUEST_TABLE_MAP_FAILED 0 +diff --git a/xen/include/asm-x86/hvm/vmx/vmx.h b/xen/include/asm-x86/hvm/vmx/vmx.h +index 89619e4afd..20eb7f6082 100644 +--- xen/include/asm-x86/hvm/vmx/vmx.h.orig ++++ xen/include/asm-x86/hvm/vmx/vmx.h +@@ -28,6 +28,8 @@ + #include <asm/hvm/trace.h> + #include <asm/hvm/vmx/vmcs.h> + ++extern int8_t opt_ept_exec_sp; ++ + typedef union { + struct { + u64 r : 1, /* bit 0 - Read permission */ +diff --git a/xen/include/asm-x86/msr-index.h b/xen/include/asm-x86/msr-index.h +index b8151d2d9f..89ae3e03f1 100644 +--- xen/include/asm-x86/msr-index.h.orig ++++ xen/include/asm-x86/msr-index.h +@@ -54,6 +54,7 @@ + #define ARCH_CAPS_SKIP_L1DFL (_AC(1, ULL) << 3) + #define ARCH_CAPS_SSB_NO (_AC(1, ULL) << 4) + #define ARCH_CAPS_MDS_NO (_AC(1, ULL) << 5) ++#define ARCH_CAPS_IF_PSCHANGE_MC_NO (_AC(1, ULL) << 6) + + #define MSR_FLUSH_CMD 0x0000010b + #define FLUSH_CMD_L1D (_AC(1, ULL) << 0) +From: Andrew Cooper <andrew.cooper3@citrix.com> +Subject: x86/vtx: Allow runtime modification of the exec-sp setting + +See patch for details. + +Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> +Reviewed-by: George Dunlap <george.dunlap@citrix.com> + +diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown +index 684671cb7b..33ed1ffc40 100644 +--- docs/misc/xen-command-line.markdown.orig ++++ docs/misc/xen-command-line.markdown +@@ -861,6 +861,21 @@ Controls EPT related features. + If HVM guest kernels are trusted not to mount a DoS against the system, + this option can enabled to regain performance. + ++ This boolean may be modified at runtime using `xl set-parameters ++ ept=[no-]exec-sp` to switch between fast and secure. ++ ++ * When switching from secure to fast, preexisting HVM domains will run ++ at their current performance until they are rebooted; new domains will ++ run without any overhead. ++ ++ * When switching from fast to secure, all HVM domains will immediately ++ suffer a performance penalty. ++ ++ **Warning: No guarantee is made that this runtime option will be retained ++ indefinitely, or that it will retain this exact behaviour. It is ++ intended as an emergency option for people who first chose fast, then ++ change their minds to secure, and wish not to reboot.** ++ + ### extra\_guest\_irqs + > `= [<domU number>][,<dom0 number>]` + +diff --git a/xen/arch/x86/hvm/vmx/vmcs.c b/xen/arch/x86/hvm/vmx/vmcs.c +index 8821a3b536..15376e25ba 100644 +--- xen/arch/x86/hvm/vmx/vmcs.c.orig ++++ xen/arch/x86/hvm/vmx/vmcs.c +@@ -107,6 +107,41 @@ static int __init parse_ept_param(const char *s) + } + custom_param("ept", parse_ept_param); + ++static int parse_ept_param_runtime(const char *s) ++{ ++ int val; ++ ++ if ( !cpu_has_vmx_ept || !hvm_funcs.hap_supported || ++ !(hvm_funcs.hap_capabilities & ++ (HVM_HAP_SUPERPAGE_2MB | HVM_HAP_SUPERPAGE_1GB)) ) ++ { ++ printk("VMX: EPT not available, or not in use - ignoring\n"); ++ return 0; ++ } ++ ++ if ( (val = parse_boolean("exec-sp", s, NULL)) < 0 ) ++ return -EINVAL; ++ ++ if ( val != opt_ept_exec_sp ) ++ { ++ struct domain *d; ++ ++ opt_ept_exec_sp = val; ++ ++ rcu_read_lock(&domlist_read_lock); ++ for_each_domain ( d ) ++ if ( paging_mode_hap(d) ) ++ p2m_change_entry_type_global(d, p2m_ram_rw, p2m_ram_rw); ++ rcu_read_unlock(&domlist_read_lock); ++ } ++ ++ printk("VMX: EPT executable superpages %sabled\n", ++ val ? "en" : "dis"); ++ ++ return 0; ++} ++custom_runtime_only_param("ept", parse_ept_param_runtime); ++ + /* Dynamic (run-time adjusted) execution control flags. */ + u32 vmx_pin_based_exec_control __read_mostly; + u32 vmx_cpu_based_exec_control __read_mostly; +diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c +index 2b62bc61dd..97c417fc3e 100644 +--- xen/arch/x86/mm/p2m.c.orig ++++ xen/arch/x86/mm/p2m.c +@@ -257,17 +257,22 @@ int p2m_is_logdirty_range(struct p2m_domain *p2m, unsigned long start, + return 0; + } + ++/* ++ * May be called with ot = nt = p2m_ram_rw for its side effect of ++ * recalculating all PTEs in the p2m. ++ */ + void p2m_change_entry_type_global(struct domain *d, + p2m_type_t ot, p2m_type_t nt) + { + struct p2m_domain *p2m = p2m_get_hostp2m(d); + +- ASSERT(ot != nt); + ASSERT(p2m_is_changeable(ot) && p2m_is_changeable(nt)); + + p2m_lock(p2m); + p2m->change_entry_type_global(p2m, ot, nt); +- p2m->global_logdirty = (nt == p2m_ram_logdirty); ++ /* Don't allow 'recalculate' operations to change the logdirty state. */ ++ if ( ot != nt ) ++ p2m->global_logdirty = (nt == p2m_ram_logdirty); + p2m_unlock(p2m); + } + diff --git a/sysutils/xenkernel411/patches/patch-XSA305 b/sysutils/xenkernel411/patches/patch-XSA305 new file mode 100644 index 00000000000..5827a342213 --- /dev/null +++ b/sysutils/xenkernel411/patches/patch-XSA305 @@ -0,0 +1,482 @@ +$NetBSD: patch-XSA305,v 1.2.2.2 2019/11/16 22:10:07 bsiegert Exp $ + +From: Andrew Cooper <andrew.cooper3@citrix.com> +Subject: x86/tsx: Introduce tsx= to use MSR_TSX_CTRL when available + +To protect against the TSX Async Abort speculative vulnerability, Intel have +released new microcode for affected parts which introduce the MSR_TSX_CTRL +control, which allows TSX to be turned off. This will be architectural on +future parts. + +Introduce tsx= to provide a global on/off for TSX, including its enumeration +via CPUID. Provide stub virtualisation of this MSR, as it is not exposed to +guests at the moment. + +VMs may have booted before microcode is loaded, or before hosts have rebooted, +and they still want to migrate freely. A VM which booted seeing TSX can +migrate safely to hosts with TSX disabled - TSX will start unconditionally +aborting, but still behave in a manner compatible with the ABI. + +The guest-visible behaviour is equivalent to late loading the microcode and +setting the RTM_DISABLE bit in the course of live patching. + +This is part of XSA-305 / CVE-2019-11135 + +Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> + +diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown +index 684671cb7b..b86d26399a 100644 +--- docs/misc/xen-command-line.markdown.orig ++++ docs/misc/xen-command-line.markdown +@@ -1948,6 +1948,20 @@ pages) must also be specified via the tbuf\_size parameter. + ### tsc (x86) + > `= unstable | skewed | stable:socket` + ++### tsx ++ = <bool> ++ ++ Applicability: x86 ++ Default: true ++ ++Controls for the use of Transactional Synchronization eXtensions. ++ ++On Intel parts released in Q3 2019 (with updated microcode), and future parts, ++a control has been introduced which allows TSX to be turned off. ++ ++On systems with the ability to turn TSX off, this boolean offers system wide ++control of whether TSX is enabled or disabled. ++ + ### ucode (x86) + > `= [<integer> | scan]` + +diff --git a/xen/arch/x86/Makefile b/xen/arch/x86/Makefile +index da1e4827f4..4c82d9f710 100644 +--- xen/arch/x86/Makefile.orig ++++ xen/arch/x86/Makefile +@@ -65,6 +65,7 @@ obj-y += sysctl.o + obj-y += time.o + obj-y += trace.o + obj-y += traps.o ++obj-y += tsx.o + obj-y += usercopy.o + obj-y += x86_emulate.o + obj-$(CONFIG_TBOOT) += tboot.o +diff --git a/xen/arch/x86/cpuid.c b/xen/arch/x86/cpuid.c +index 5e11970701..04aefa555d 100644 +--- xen/arch/x86/cpuid.c.orig ++++ xen/arch/x86/cpuid.c +@@ -622,6 +622,20 @@ void recalculate_cpuid_policy(struct domain *d) + if ( cpu_has_itsc && (d->disable_migrate || d->arch.vtsc) ) + __set_bit(X86_FEATURE_ITSC, max_fs); + ++ /* ++ * On hardware with MSR_TSX_CTRL, the admin may have elected to disable ++ * TSX and hide the feature bits. Migrating-in VMs may have been booted ++ * pre-mitigation when the TSX features were visbile. ++ * ++ * This situation is compatible (albeit with a perf hit to any TSX code in ++ * the guest), so allow the feature bits to remain set. ++ */ ++ if ( cpu_has_tsx_ctrl ) ++ { ++ __set_bit(X86_FEATURE_HLE, max_fs); ++ __set_bit(X86_FEATURE_RTM, max_fs); ++ } ++ + /* Clamp the toolstacks choices to reality. */ + for ( i = 0; i < ARRAY_SIZE(fs); i++ ) + fs[i] &= max_fs[i]; +diff --git a/xen/arch/x86/msr.c b/xen/arch/x86/msr.c +index ebc0665615..35d99a98a1 100644 +--- xen/arch/x86/msr.c.orig ++++ xen/arch/x86/msr.c +@@ -153,6 +153,7 @@ int guest_rdmsr(const struct vcpu *v, uint32_t msr, uint64_t *val) + case MSR_FLUSH_CMD: + /* Write-only */ + case MSR_TSX_FORCE_ABORT: ++ case MSR_TSX_CTRL: + /* Not offered to guests. */ + goto gp_fault; + +@@ -233,6 +234,7 @@ int guest_wrmsr(struct vcpu *v, uint32_t msr, uint64_t val) + case MSR_ARCH_CAPABILITIES: + /* Read-only */ + case MSR_TSX_FORCE_ABORT: ++ case MSR_TSX_CTRL: + /* Not offered to guests. */ + goto gp_fault; + +diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c +index 657160549f..dc13ad6c36 100644 +--- xen/arch/x86/setup.c.orig ++++ xen/arch/x86/setup.c +@@ -1551,6 +1551,8 @@ void __init noreturn __start_xen(unsigned long mbi_p) + + early_microcode_init(); + ++ tsx_init(); /* Needs microcode. May change HLE/RTM feature bits. */ ++ + identify_cpu(&boot_cpu_data); + + set_in_cr4(X86_CR4_OSFXSR | X86_CR4_OSXMMEXCPT); +diff --git a/xen/arch/x86/smpboot.c b/xen/arch/x86/smpboot.c +index fd52a10cf9..bdc118d88b 100644 +--- xen/arch/x86/smpboot.c.orig ++++ xen/arch/x86/smpboot.c +@@ -376,6 +376,8 @@ void start_secondary(void *unused) + if ( boot_cpu_has(X86_FEATURE_IBRSB) ) + wrmsrl(MSR_SPEC_CTRL, default_xen_spec_ctrl); + ++ tsx_init(); /* Needs microcode. May change HLE/RTM feature bits. */ ++ + if ( xen_guest ) + hypervisor_ap_setup(); + +diff --git a/xen/arch/x86/tsx.c b/xen/arch/x86/tsx.c +new file mode 100644 +index 0000000000..a8ec2ccc69 +--- /dev/null ++++ xen/arch/x86/tsx.c +@@ -0,0 +1,74 @@ ++#include <xen/init.h> ++#include <asm/msr.h> ++ ++/* ++ * Valid values: ++ * 1 => Explicit tsx=1 ++ * 0 => Explicit tsx=0 ++ * -1 => Default, implicit tsx=1 ++ * ++ * This is arranged such that the bottom bit encodes whether TSX is actually ++ * disabled, while identifying various explicit (>=0) and implicit (<0) ++ * conditions. ++ */ ++int8_t __read_mostly opt_tsx = -1; ++int8_t __read_mostly cpu_has_tsx_ctrl = -1; ++ ++static int __init parse_tsx(const char *s) ++{ ++ int rc = 0, val = parse_bool(s, NULL); ++ ++ if ( val >= 0 ) ++ opt_tsx = val; ++ else ++ rc = -EINVAL; ++ ++ return rc; ++} ++custom_param("tsx", parse_tsx); ++ ++void tsx_init(void) ++{ ++ /* ++ * This function is first called between microcode being loaded, and CPUID ++ * being scanned generally. Calculate from raw data whether MSR_TSX_CTRL ++ * is available. ++ */ ++ if ( unlikely(cpu_has_tsx_ctrl < 0) ) ++ { ++ uint64_t caps = 0; ++ ++ if ( boot_cpu_data.cpuid_level >= 7 && ++ (cpuid_count_edx(7, 0) & cpufeat_mask(X86_FEATURE_ARCH_CAPS)) ) ++ rdmsrl(MSR_ARCH_CAPABILITIES, caps); ++ ++ cpu_has_tsx_ctrl = !!(caps & ARCH_CAPS_TSX_CTRL); ++ } ++ ++ if ( cpu_has_tsx_ctrl ) ++ { ++ uint64_t val; ++ ++ rdmsrl(MSR_TSX_CTRL, val); ++ ++ val &= ~(TSX_CTRL_RTM_DISABLE | TSX_CTRL_CPUID_CLEAR); ++ /* Check bottom bit only. Higher bits are various sentinals. */ ++ if ( !(opt_tsx & 1) ) ++ val |= TSX_CTRL_RTM_DISABLE | TSX_CTRL_CPUID_CLEAR; ++ ++ wrmsrl(MSR_TSX_CTRL, val); ++ } ++ else if ( opt_tsx >= 0 ) ++ printk_once(XENLOG_WARNING ++ "MSR_TSX_CTRL not available - Ignoring tsx= setting\n"); ++} ++ ++/* ++ * Local variables: ++ * mode: C ++ * c-file-style: "BSD" ++ * c-basic-offset: 4 ++ * tab-width: 4 ++ * indent-tabs-mode: nil ++ * End: ++ */ +diff --git a/xen/include/asm-x86/msr-index.h b/xen/include/asm-x86/msr-index.h +index 89ae3e03f1..5ee7a37c12 100644 +--- xen/include/asm-x86/msr-index.h.orig ++++ xen/include/asm-x86/msr-index.h +@@ -55,6 +55,7 @@ + #define ARCH_CAPS_SSB_NO (_AC(1, ULL) << 4) + #define ARCH_CAPS_MDS_NO (_AC(1, ULL) << 5) + #define ARCH_CAPS_IF_PSCHANGE_MC_NO (_AC(1, ULL) << 6) ++#define ARCH_CAPS_TSX_CTRL (_AC(1, ULL) << 7) + + #define MSR_FLUSH_CMD 0x0000010b + #define FLUSH_CMD_L1D (_AC(1, ULL) << 0) +@@ -62,6 +63,10 @@ + #define MSR_TSX_FORCE_ABORT 0x0000010f + #define TSX_FORCE_ABORT_RTM (_AC(1, ULL) << 0) + ++#define MSR_TSX_CTRL 0x00000122 ++#define TSX_CTRL_RTM_DISABLE (_AC(1, ULL) << 0) ++#define TSX_CTRL_CPUID_CLEAR (_AC(1, ULL) << 1) ++ + /* Intel MSRs. Some also available on other CPUs */ + #define MSR_IA32_PERFCTR0 0x000000c1 + #define MSR_IA32_A_PERFCTR0 0x000004c1 +diff --git a/xen/include/asm-x86/processor.h b/xen/include/asm-x86/processor.h +index 20d1ecb332..66224f23b9 100644 +--- xen/include/asm-x86/processor.h.orig ++++ xen/include/asm-x86/processor.h +@@ -258,6 +258,16 @@ static always_inline unsigned int cpuid_count_ebx( + return ebx; + } + ++static always_inline unsigned int cpuid_count_edx( ++ unsigned int leaf, unsigned int subleaf) ++{ ++ unsigned int edx, tmp; ++ ++ cpuid_count(leaf, subleaf, &tmp, &tmp, &tmp, &edx); ++ ++ return edx; ++} ++ + static always_inline void cpuid_count_leaf(uint32_t leaf, uint32_t subleaf, + struct cpuid_leaf *data) + { +@@ -610,6 +620,9 @@ static inline uint8_t get_cpu_family(uint32_t raw, uint8_t *model, + return fam; + } + ++extern int8_t opt_tsx, cpu_has_tsx_ctrl; ++void tsx_init(void); ++ + #endif /* !__ASSEMBLY__ */ + + #endif /* __ASM_X86_PROCESSOR_H */ +diff --git a/xen/include/xen/lib.h b/xen/include/xen/lib.h +index 750f809968..be223a6950 100644 +--- xen/include/xen/lib.h.orig ++++ xen/include/xen/lib.h +@@ -116,6 +116,16 @@ extern int printk_ratelimit(void); + #define gprintk(lvl, fmt, args...) \ + printk(XENLOG_GUEST lvl "%pv " fmt, current, ## args) + ++#define printk_once(fmt, args...) \ ++({ \ ++ static bool __read_mostly once_; \ ++ if ( unlikely(!once_) ) \ ++ { \ ++ once_ = true; \ ++ printk(fmt, ## args); \ ++ } \ ++}) ++ + #ifdef NDEBUG + + static inline void +From: Andrew Cooper <andrew.cooper3@citrix.com> +Subject: x86/spec-ctrl: Mitigate the TSX Asynchronous Abort sidechannel + +See patch documentation and comments. + +This is part of XSA-305 / CVE-2019-11135 + +Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> +Reviewed-by: Jan Beulich <jbeulich@suse.com> + +diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown +index b86d26399a..31635a473a 100644 +--- docs/misc/xen-command-line.markdown.orig ++++ docs/misc/xen-command-line.markdown +@@ -1841,7 +1841,7 @@ extreme care.** + An overall boolean value, `spec-ctrl=no`, can be specified to turn off all + mitigations, including pieces of infrastructure used to virtualise certain + mitigation features for guests. This also includes settings which `xpti`, +-`smt`, `pv-l1tf` control, unless the respective option(s) have been ++`smt`, `pv-l1tf`, `tsx` control, unless the respective option(s) have been + specified earlier on the command line. + + Alternatively, a slightly more restricted `spec-ctrl=no-xen` can be used to +@@ -1952,7 +1952,7 @@ pages) must also be specified via the tbuf\_size parameter. + = <bool> + + Applicability: x86 +- Default: true ++ Default: false on parts vulnerable to TAA, true otherwise + + Controls for the use of Transactional Synchronization eXtensions. + +@@ -1962,6 +1962,19 @@ a control has been introduced which allows TSX to be turned off. + On systems with the ability to turn TSX off, this boolean offers system wide + control of whether TSX is enabled or disabled. + ++On parts vulnerable to CVE-2019-11135 / TSX Asynchronous Abort, the following ++logic applies: ++ ++ * An explicit `tsx=` choice is honoured, even if it is `true` and would ++ result in a vulnerable system. ++ ++ * When no explicit `tsx=` choice is given, parts vulnerable to TAA will be ++ mitigated by disabling TSX, as this is the lowest overhead option. ++ ++ * If the use of TSX is important, the more expensive TAA mitigations can be ++ opted in to with `smt=0 spec-ctrl=md-clear`, at which point TSX will remain ++ active by default. ++ + ### ucode (x86) + > `= [<integer> | scan]` + +diff --git a/xen/arch/x86/spec_ctrl.c b/xen/arch/x86/spec_ctrl.c +index 2fe16b423d..ab196b156d 100644 +--- xen/arch/x86/spec_ctrl.c.orig ++++ xen/arch/x86/spec_ctrl.c +@@ -152,6 +152,9 @@ static int __init parse_spec_ctrl(const char *s) + if ( opt_pv_l1tf_domu < 0 ) + opt_pv_l1tf_domu = 0; + ++ if ( opt_tsx == -1 ) ++ opt_tsx = -3; ++ + disable_common: + opt_rsb_pv = false; + opt_rsb_hvm = false; +@@ -362,7 +365,7 @@ static void __init print_details(enum ind_thunk thunk, uint64_t caps) + printk("Speculative mitigation facilities:\n"); + + /* Hardware features which pertain to speculative mitigations. */ +- printk(" Hardware features:%s%s%s%s%s%s%s%s%s%s%s%s\n", ++ printk(" Hardware features:%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", + (_7d0 & cpufeat_mask(X86_FEATURE_IBRSB)) ? " IBRS/IBPB" : "", + (_7d0 & cpufeat_mask(X86_FEATURE_STIBP)) ? " STIBP" : "", + (_7d0 & cpufeat_mask(X86_FEATURE_L1D_FLUSH)) ? " L1D_FLUSH" : "", +@@ -374,7 +377,9 @@ static void __init print_details(enum ind_thunk thunk, uint64_t caps) + (caps & ARCH_CAPS_RSBA) ? " RSBA" : "", + (caps & ARCH_CAPS_SKIP_L1DFL) ? " SKIP_L1DFL": "", + (caps & ARCH_CAPS_SSB_NO) ? " SSB_NO" : "", +- (caps & ARCH_CAPS_MDS_NO) ? " MDS_NO" : ""); ++ (caps & ARCH_CAPS_MDS_NO) ? " MDS_NO" : "", ++ (caps & ARCH_CAPS_TSX_CTRL) ? " TSX_CTRL" : "", ++ (caps & ARCH_CAPS_TAA_NO) ? " TAA_NO" : ""); + + /* Compiled-in support which pertains to mitigations. */ + if ( IS_ENABLED(CONFIG_INDIRECT_THUNK) || IS_ENABLED(CONFIG_SHADOW_PAGING) ) +@@ -388,7 +393,7 @@ static void __init print_details(enum ind_thunk thunk, uint64_t caps) + "\n"); + + /* Settings for Xen's protection, irrespective of guests. */ +- printk(" Xen settings: BTI-Thunk %s, SPEC_CTRL: %s%s, Other:%s%s%s\n", ++ printk(" Xen settings: BTI-Thunk %s, SPEC_CTRL: %s%s%s, Other:%s%s%s\n", + thunk == THUNK_NONE ? "N/A" : + thunk == THUNK_RETPOLINE ? "RETPOLINE" : + thunk == THUNK_LFENCE ? "LFENCE" : +@@ -397,6 +402,8 @@ static void __init print_details(enum ind_thunk thunk, uint64_t caps) + (default_xen_spec_ctrl & SPEC_CTRL_IBRS) ? "IBRS+" : "IBRS-", + !boot_cpu_has(X86_FEATURE_SSBD) ? "" : + (default_xen_spec_ctrl & SPEC_CTRL_SSBD) ? " SSBD+" : " SSBD-", ++ !(caps & ARCH_CAPS_TSX_CTRL) ? "" : ++ (opt_tsx & 1) ? " TSX+" : " TSX-", + opt_ibpb ? " IBPB" : "", + opt_l1d_flush ? " L1D_FLUSH" : "", + opt_md_clear_pv || opt_md_clear_hvm ? " VERW" : ""); +@@ -911,6 +918,7 @@ void __init init_speculation_mitigations(void) + { + enum ind_thunk thunk = THUNK_DEFAULT; + bool use_spec_ctrl = false, ibrs = false, hw_smt_enabled; ++ bool cpu_has_bug_taa; + uint64_t caps = 0; + + if ( boot_cpu_has(X86_FEATURE_ARCH_CAPS) ) +@@ -1140,6 +1148,53 @@ void __init init_speculation_mitigations(void) + "enabled. Mitigations will not be fully effective. Please\n" + "choose an explicit smt=<bool> setting. See XSA-297.\n"); + ++ /* ++ * Vulnerability to TAA is a little complicated to quantify. ++ * ++ * In the pipeline, it is just another way to get speculative access to ++ * stale load port, store buffer or fill buffer data, and therefore can be ++ * considered a superset of MDS (on TSX-capable parts). On parts which ++ * predate MDS_NO, the existing VERW flushing will mitigate this ++ * sidechannel as well. ++ * ++ * On parts which contain MDS_NO, the lack of VERW flushing means that an ++ * attacker can still use TSX to target microarchitectural buffers to leak ++ * secrets. Therefore, we consider TAA to be the set of TSX-capable parts ++ * which have MDS_NO but lack TAA_NO. ++ * ++ * Note: cpu_has_rtm (== hle) could already be hidden by `tsx=0` on the ++ * cmdline. MSR_TSX_CTRL will only appear on TSX-capable parts, so ++ * we check both to spot TSX in a microcode/cmdline independent way. ++ */ ++ cpu_has_bug_taa = ++ (cpu_has_rtm || (caps & ARCH_CAPS_TSX_CTRL)) && ++ (caps & (ARCH_CAPS_MDS_NO | ARCH_CAPS_TAA_NO)) == ARCH_CAPS_MDS_NO; ++ ++ /* ++ * On TAA-affected hardware, disabling TSX is the preferred mitigation, vs ++ * the MDS mitigation of disabling HT and using VERW flushing. ++ * ++ * On CPUs which advertise MDS_NO, VERW has no flushing side effect until ++ * the TSX_CTRL microcode is loaded, despite the MD_CLEAR CPUID bit being ++ * advertised, and there isn't a MD_CLEAR_2 flag to use... ++ * ++ * If we're on affected hardware, able to do something about it (which ++ * implies that VERW now works), no explicit TSX choice and traditional ++ * MDS mitigations (no-SMT, VERW) not obviosuly in use (someone might ++ * plausibly value TSX higher than Hyperthreading...), disable TSX to ++ * mitigate TAA. ++ */ ++ if ( opt_tsx == -1 && cpu_has_bug_taa && (caps & ARCH_CAPS_TSX_CTRL) && ++ ((hw_smt_enabled && opt_smt) || ++ !boot_cpu_has(X86_FEATURE_SC_VERW_IDLE)) ) ++ { ++ setup_clear_cpu_cap(X86_FEATURE_HLE); ++ setup_clear_cpu_cap(X86_FEATURE_RTM); ++ ++ opt_tsx = 0; ++ tsx_init(); ++ } ++ + print_details(thunk, caps); + + /* +diff --git a/xen/arch/x86/tsx.c b/xen/arch/x86/tsx.c +index a8ec2ccc69..2d202a0d4e 100644 +--- xen/arch/x86/tsx.c.orig ++++ xen/arch/x86/tsx.c +@@ -5,7 +5,8 @@ + * Valid values: + * 1 => Explicit tsx=1 + * 0 => Explicit tsx=0 +- * -1 => Default, implicit tsx=1 ++ * -1 => Default, implicit tsx=1, may change to 0 to mitigate TAA ++ * -3 => Implicit tsx=1 (feed-through from spec-ctrl=0) + * + * This is arranged such that the bottom bit encodes whether TSX is actually + * disabled, while identifying various explicit (>=0) and implicit (<0) +diff --git a/xen/include/asm-x86/msr-index.h b/xen/include/asm-x86/msr-index.h +index 5ee7a37c12..1761a01f1f 100644 +--- xen/include/asm-x86/msr-index.h.orig ++++ xen/include/asm-x86/msr-index.h +@@ -56,6 +56,7 @@ + #define ARCH_CAPS_MDS_NO (_AC(1, ULL) << 5) + #define ARCH_CAPS_IF_PSCHANGE_MC_NO (_AC(1, ULL) << 6) + #define ARCH_CAPS_TSX_CTRL (_AC(1, ULL) << 7) ++#define ARCH_CAPS_TAA_NO (_AC(1, ULL) << 8) + + #define MSR_FLUSH_CMD 0x0000010b + #define FLUSH_CMD_L1D (_AC(1, ULL) << 0) |