* [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM
@ 2014-09-17 20:51 Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 01/21] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
` (21 more replies)
0 siblings, 22 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
The ARM virtualization extension provides 2-stage paging, a similar mechanisms
to Intel's EPT, which can be used to trace the memory accesses performed by
the guest systems. This series moves the mem_access and mem_event codebase
into Xen common, performs some code cleanup and architecture specific division
of components, then sets up the necessary infrastructure in the ARM code
to deliver the event on R/W/X traps. Finally, we turn on the compilation of
mem_access and mem_event on ARM and perform the necessary changes to the tools side.
This version of the series has been fully tested and is functional on both an
Arndale board and on Intel hardware.
This PATCH series is also available at:
https://github.com/tklengyel/xen/tree/arm_memaccess7
Julien Grall (1):
xen/arm: Implement domain_get_maximum_gpfn
Tamas K Lengyel (20):
xen: Relocate mem_access and mem_event into common.
xen: Relocate struct npfec definition into common
xen: Relocate mem_event_op domctl and access_op memop into common.
xen: Relocate set_access_required domctl into common
xen: Relocate p2m_access_t into common and swap the order
xen: Relocate p2m_mem_access_resume to mem_access common
xen: Relocate p2m_mem_access_check into common and refactor it.
x86/p2m: Typo fix for spelling ambiguous
x86/p2m: Fix conversion macro of p2m_access to XENMEM_access
xen/mem_event: Clean out superfluous white-spaces
xen/mem_event: Relax error condition on debug builds
xen/mem_event: Abstract architecture specific sanity checks
xen/mem_access: Abstract architecture specific sanity check
xen/arm: p2m changes for mem_access support
xen/arm: Add p2m_set_permission and p2m_shatter_page helpers.
xen/arm: Data abort exception (R/W) mem_events.
xen/arm: Instruction prefetch abort (X) mem_event handling
xen/arm: Enable the compilation of mem_access and mem_event on ARM.
tools/libxc: Allocate magic page for mem access on ARM
tools/tests: Enable xen-access on ARM
MAINTAINERS | 6 +
config/arm32.mk | 1 +
config/arm64.mk | 1 +
config/x86_32.mk | 2 +
config/x86_64.mk | 2 +
tools/libxc/xc_dom_arm.c | 6 +-
tools/tests/xen-access/Makefile | 9 +-
tools/tests/xen-access/xen-access.c | 79 ++--
xen/Rules.mk | 1 +
xen/arch/arm/mm.c | 2 +-
xen/arch/arm/p2m.c | 419 +++++++++++++++++----
xen/arch/arm/traps.c | 74 +++-
xen/arch/x86/domctl.c | 24 +-
xen/arch/x86/hvm/hvm.c | 122 +-----
xen/arch/x86/mm/Makefile | 2 -
xen/arch/x86/mm/hap/nested_ept.c | 6 +-
xen/arch/x86/mm/hap/nested_hap.c | 8 +-
xen/arch/x86/mm/mem_access.c | 133 -------
xen/arch/x86/mm/mem_event.c | 705 -----------------------------------
xen/arch/x86/mm/mem_paging.c | 2 +-
xen/arch/x86/mm/mem_sharing.c | 6 +-
xen/arch/x86/mm/mm-locks.h | 299 ---------------
xen/arch/x86/mm/p2m-ept.c | 9 +-
xen/arch/x86/mm/p2m-pod.c | 10 +-
xen/arch/x86/mm/p2m-pt.c | 12 +-
xen/arch/x86/mm/p2m.c | 136 +------
xen/arch/x86/mm/paging.c | 4 +-
xen/arch/x86/x86_64/compat/mm.c | 8 +-
xen/arch/x86/x86_64/mm.c | 8 +-
xen/common/Makefile | 2 +
xen/common/compat/memory.c | 5 +
xen/common/domain.c | 1 +
xen/common/domctl.c | 22 ++
xen/common/mem_access.c | 298 +++++++++++++++
xen/common/mem_event.c | 723 ++++++++++++++++++++++++++++++++++++
xen/common/memory.c | 72 +++-
xen/include/asm-arm/mm.h | 12 +-
xen/include/asm-arm/p2m.h | 112 ++++--
xen/include/asm-arm/processor.h | 70 +++-
xen/include/asm-x86/config.h | 6 +
xen/include/asm-x86/hvm/hvm.h | 8 +-
xen/include/asm-x86/mem_access.h | 39 --
xen/include/asm-x86/mem_event.h | 82 ----
xen/include/asm-x86/mm-locks.h | 299 +++++++++++++++
xen/include/asm-x86/mm.h | 23 --
xen/include/asm-x86/p2m.h | 56 ++-
xen/include/asm-x86/page.h | 4 -
xen/include/xen/mem_access.h | 74 ++++
xen/include/xen/mem_event.h | 143 +++++++
xen/include/xen/mm.h | 31 ++
xen/include/xen/p2m-common.h | 29 ++
xen/include/xsm/dummy.h | 26 +-
xen/include/xsm/xsm.h | 29 +-
xen/xsm/dummy.c | 7 +-
xen/xsm/flask/hooks.c | 33 +-
55 files changed, 2495 insertions(+), 1807 deletions(-)
delete mode 100644 xen/arch/x86/mm/mem_access.c
delete mode 100644 xen/arch/x86/mm/mem_event.c
delete mode 100644 xen/arch/x86/mm/mm-locks.h
create mode 100644 xen/common/mem_access.c
create mode 100644 xen/common/mem_event.c
delete mode 100644 xen/include/asm-x86/mem_access.h
delete mode 100644 xen/include/asm-x86/mem_event.h
create mode 100644 xen/include/asm-x86/mm-locks.h
create mode 100644 xen/include/xen/mem_access.h
create mode 100644 xen/include/xen/mem_event.h
--
2.1.0
^ permalink raw reply [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 01/21] xen: Relocate mem_access and mem_event into common.
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 02/21] xen: Relocate struct npfec definition " Tamas K Lengyel
` (20 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
In preparation to add support for ARM LPAE mem_event, relocate mem_access,
mem_event and auxiliary functions into common Xen code.
This patch makes no functional changes to the X86 side, for ARM mem_event
and mem_access functions are just defined as placeholder stubs, and are
actually enabled later in the series.
Edits that are only header path adjustments:
xen/arch/x86/domctl.c
xen/arch/x86/mm/hap/nested_ept.c
xen/arch/x86/mm/hap/nested_hap.c
xen/arch/x86/mm/mem_paging.c
xen/arch/x86/mm/mem_sharing.c
xen/arch/x86/mm/p2m-pod.c
xen/arch/x86/mm/p2m-pt.c
xen/arch/x86/mm/p2m.c
xen/arch/x86/x86_64/compat/mm.c
xen/arch/x86/x86_64/mm.c
Makefile adjustments for new/removed code:
xen/common/Makefile
xen/arch/x86/mm/Makefile
Relocated prepare_ring_for_helper and destroy_ring_for_helper functions:
xen/include/xen/mm.h
xen/common/memory.c
xen/include/asm-x86/hvm/hvm.h
xen/arch/x86/hvm/hvm.c
Code movement of mem_event and mem_access:
xen/arch/x86/mm/mem_access.c -> xen/common/mem_access.c
xen/arch/x86/mm/mem_event.c -> xen/common/mem_event.c
xen/include/asm-x86/mem_access.h -> xen/include/xen/mem_access.h
xen/include/asm-x86/mem_event.h -> xen/include/xen/mem_event.h
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Tim Deegan <tim@xen.org>
---
v5: Make <xen/mem-event.h> include <xen/sched.h> by default.
Style fix with grouping of #includes.
v4: Make <xen/mem_access.h> include <public/memory.h> by default.
v3: Replace asm/domain.h with xen/sched.h in mem_event.c to better
accomodate for the new code location.
Replace #ifdef CONFIG_X86 wrappers with HAS_MEM_ACCESS flags.
v2: Update MAINTAINERS.
More descriptive commit message to aid in the review process.
---
MAINTAINERS | 6 +
xen/Rules.mk | 1 +
xen/arch/x86/Rules.mk | 1 +
xen/arch/x86/domctl.c | 2 +-
xen/arch/x86/hvm/hvm.c | 63 +---
xen/arch/x86/mm/Makefile | 2 -
xen/arch/x86/mm/hap/nested_ept.c | 6 +-
xen/arch/x86/mm/hap/nested_hap.c | 6 +-
xen/arch/x86/mm/mem_access.c | 133 --------
xen/arch/x86/mm/mem_event.c | 705 ---------------------------------------
xen/arch/x86/mm/mem_paging.c | 2 +-
xen/arch/x86/mm/mem_sharing.c | 4 +-
xen/arch/x86/mm/p2m-pod.c | 8 +-
xen/arch/x86/mm/p2m-pt.c | 10 +-
xen/arch/x86/mm/p2m.c | 8 +-
xen/arch/x86/x86_64/compat/mm.c | 4 +-
xen/arch/x86/x86_64/mm.c | 4 +-
xen/common/Makefile | 2 +
xen/common/domain.c | 1 +
xen/common/mem_access.c | 133 ++++++++
xen/common/mem_event.c | 705 +++++++++++++++++++++++++++++++++++++++
xen/common/memory.c | 63 ++++
xen/include/asm-arm/mm.h | 1 -
xen/include/asm-x86/config.h | 3 +
xen/include/asm-x86/hvm/hvm.h | 6 -
xen/include/asm-x86/mem_access.h | 39 ---
xen/include/asm-x86/mem_event.h | 82 -----
xen/include/asm-x86/mm.h | 2 -
xen/include/xen/mem_access.h | 60 ++++
xen/include/xen/mem_event.h | 143 ++++++++
xen/include/xen/mm.h | 6 +
31 files changed, 1154 insertions(+), 1057 deletions(-)
delete mode 100644 xen/arch/x86/mm/mem_access.c
delete mode 100644 xen/arch/x86/mm/mem_event.c
create mode 100644 xen/common/mem_access.c
create mode 100644 xen/common/mem_event.c
delete mode 100644 xen/include/asm-x86/mem_access.h
delete mode 100644 xen/include/asm-x86/mem_event.h
create mode 100644 xen/include/xen/mem_access.h
create mode 100644 xen/include/xen/mem_event.h
diff --git a/MAINTAINERS b/MAINTAINERS
index 266e47b..f659180 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -337,6 +337,12 @@ F: xen/arch/x86/mm/mem_sharing.c
F: xen/arch/x86/mm/mem_paging.c
F: tools/memshr
+MEMORY EVENT AND ACCESS
+M: Tim Deegan <tim@xen.org>
+S: Supported
+F: xen/common/mem_event.c
+F: xen/common/mem_access.c
+
XENTRACE
M: George Dunlap <george.dunlap@eu.citrix.com>
S: Supported
diff --git a/xen/Rules.mk b/xen/Rules.mk
index b49f3c8..dc15b09 100644
--- a/xen/Rules.mk
+++ b/xen/Rules.mk
@@ -57,6 +57,7 @@ CFLAGS-$(HAS_ACPI) += -DHAS_ACPI
CFLAGS-$(HAS_GDBSX) += -DHAS_GDBSX
CFLAGS-$(HAS_PASSTHROUGH) += -DHAS_PASSTHROUGH
CFLAGS-$(HAS_DEVICE_TREE) += -DHAS_DEVICE_TREE
+CFLAGS-$(HAS_MEM_ACCESS) += -DHAS_MEM_ACCESS
CFLAGS-$(HAS_PCI) += -DHAS_PCI
CFLAGS-$(HAS_IOPORTS) += -DHAS_IOPORTS
CFLAGS-$(frame_pointer) += -fno-omit-frame-pointer -DCONFIG_FRAME_POINTER
diff --git a/xen/arch/x86/Rules.mk b/xen/arch/x86/Rules.mk
index 576985e..bd4e342 100644
--- a/xen/arch/x86/Rules.mk
+++ b/xen/arch/x86/Rules.mk
@@ -12,6 +12,7 @@ HAS_NS16550 := y
HAS_EHCI := y
HAS_KEXEC := y
HAS_GDBSX := y
+HAS_MEM_ACCESS := y
xenoprof := y
#
diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index 7a5de43..26a3ea1 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -30,7 +30,7 @@
#include <xen/hypercall.h> /* for arch_do_domctl */
#include <xsm/xsm.h>
#include <xen/iommu.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
#include <public/mem_event.h>
#include <asm/mem_sharing.h>
#include <asm/xstate.h>
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 8d905d3..7649d36 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -35,6 +35,9 @@
#include <xen/paging.h>
#include <xen/cpu.h>
#include <xen/wait.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.h>
+#include <xen/rangeset.h>
#include <asm/shadow.h>
#include <asm/hap.h>
#include <asm/current.h>
@@ -63,10 +66,7 @@
#include <public/hvm/ioreq.h>
#include <public/version.h>
#include <public/memory.h>
-#include <asm/mem_event.h>
-#include <asm/mem_access.h>
#include <public/mem_event.h>
-#include <xen/rangeset.h>
#include <public/arch-x86/cpuid.h>
bool_t __read_mostly hvm_enabled;
@@ -484,19 +484,6 @@ static void hvm_free_ioreq_gmfn(struct domain *d, unsigned long gmfn)
clear_bit(i, &d->arch.hvm_domain.ioreq_gmfn.mask);
}
-void destroy_ring_for_helper(
- void **_va, struct page_info *page)
-{
- void *va = *_va;
-
- if ( va != NULL )
- {
- unmap_domain_page_global(va);
- put_page_and_type(page);
- *_va = NULL;
- }
-}
-
static void hvm_unmap_ioreq_page(struct hvm_ioreq_server *s, bool_t buf)
{
struct hvm_ioreq_page *iorp = buf ? &s->bufioreq : &s->ioreq;
@@ -504,50 +491,6 @@ static void hvm_unmap_ioreq_page(struct hvm_ioreq_server *s, bool_t buf)
destroy_ring_for_helper(&iorp->va, iorp->page);
}
-int prepare_ring_for_helper(
- struct domain *d, unsigned long gmfn, struct page_info **_page,
- void **_va)
-{
- struct page_info *page;
- p2m_type_t p2mt;
- void *va;
-
- page = get_page_from_gfn(d, gmfn, &p2mt, P2M_UNSHARE);
- if ( p2m_is_paging(p2mt) )
- {
- if ( page )
- put_page(page);
- p2m_mem_paging_populate(d, gmfn);
- return -ENOENT;
- }
- if ( p2m_is_shared(p2mt) )
- {
- if ( page )
- put_page(page);
- return -ENOENT;
- }
- if ( !page )
- return -EINVAL;
-
- if ( !get_page_type(page, PGT_writable_page) )
- {
- put_page(page);
- return -EINVAL;
- }
-
- va = __map_domain_page_global(page);
- if ( va == NULL )
- {
- put_page_and_type(page);
- return -ENOMEM;
- }
-
- *_va = va;
- *_page = page;
-
- return 0;
-}
-
static int hvm_map_ioreq_page(
struct hvm_ioreq_server *s, bool_t buf, unsigned long gmfn)
{
diff --git a/xen/arch/x86/mm/Makefile b/xen/arch/x86/mm/Makefile
index 73dcdf4..ed4b1f8 100644
--- a/xen/arch/x86/mm/Makefile
+++ b/xen/arch/x86/mm/Makefile
@@ -6,10 +6,8 @@ obj-y += p2m.o p2m-pt.o p2m-ept.o p2m-pod.o
obj-y += guest_walk_2.o
obj-y += guest_walk_3.o
obj-$(x86_64) += guest_walk_4.o
-obj-$(x86_64) += mem_event.o
obj-$(x86_64) += mem_paging.o
obj-$(x86_64) += mem_sharing.o
-obj-$(x86_64) += mem_access.o
guest_walk_%.o: guest_walk.c Makefile
$(CC) $(CFLAGS) -DGUEST_PAGING_LEVELS=$* -c $< -o $@
diff --git a/xen/arch/x86/mm/hap/nested_ept.c b/xen/arch/x86/mm/hap/nested_ept.c
index 0d044bc..cbbc4e9 100644
--- a/xen/arch/x86/mm/hap/nested_ept.c
+++ b/xen/arch/x86/mm/hap/nested_ept.c
@@ -17,14 +17,14 @@
* this program; if not, write to the Free Software Foundation, Inc., 59 Temple
* Place - Suite 330, Boston, MA 02111-1307 USA.
*/
+#include <xen/mem_event.h>
+#include <xen/event.h>
+#include <public/mem_event.h>
#include <asm/domain.h>
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
-#include <asm/mem_event.h>
-#include <public/mem_event.h>
#include <asm/mem_sharing.h>
-#include <xen/event.h>
#include <asm/hap.h>
#include <asm/hvm/support.h>
diff --git a/xen/arch/x86/mm/hap/nested_hap.c b/xen/arch/x86/mm/hap/nested_hap.c
index 137a87c..a4bb835 100644
--- a/xen/arch/x86/mm/hap/nested_hap.c
+++ b/xen/arch/x86/mm/hap/nested_hap.c
@@ -19,14 +19,14 @@
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
+#include <xen/mem_event.h>
+#include <xen/event.h>
+#include <public/mem_event.h>
#include <asm/domain.h>
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
-#include <asm/mem_event.h>
-#include <public/mem_event.h>
#include <asm/mem_sharing.h>
-#include <xen/event.h>
#include <asm/hap.h>
#include <asm/hvm/support.h>
diff --git a/xen/arch/x86/mm/mem_access.c b/xen/arch/x86/mm/mem_access.c
deleted file mode 100644
index e8465a5..0000000
--- a/xen/arch/x86/mm/mem_access.c
+++ /dev/null
@@ -1,133 +0,0 @@
-/******************************************************************************
- * arch/x86/mm/mem_access.c
- *
- * Memory access support.
- *
- * Copyright (c) 2011 Virtuata, Inc.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- */
-
-
-#include <xen/sched.h>
-#include <xen/guest_access.h>
-#include <xen/hypercall.h>
-#include <asm/p2m.h>
-#include <asm/mem_event.h>
-#include <xsm/xsm.h>
-
-
-int mem_access_memop(unsigned long cmd,
- XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg)
-{
- long rc;
- xen_mem_access_op_t mao;
- struct domain *d;
-
- if ( copy_from_guest(&mao, arg, 1) )
- return -EFAULT;
-
- rc = rcu_lock_live_remote_domain_by_id(mao.domid, &d);
- if ( rc )
- return rc;
-
- rc = -EINVAL;
- if ( !is_hvm_domain(d) )
- goto out;
-
- rc = xsm_mem_event_op(XSM_DM_PRIV, d, XENMEM_access_op);
- if ( rc )
- goto out;
-
- rc = -ENODEV;
- if ( unlikely(!d->mem_event->access.ring_page) )
- goto out;
-
- switch ( mao.op )
- {
- case XENMEM_access_op_resume:
- p2m_mem_access_resume(d);
- rc = 0;
- break;
-
- case XENMEM_access_op_set_access:
- {
- unsigned long start_iter = cmd & ~MEMOP_CMD_MASK;
-
- rc = -EINVAL;
- if ( (mao.pfn != ~0ull) &&
- (mao.nr < start_iter ||
- ((mao.pfn + mao.nr - 1) < mao.pfn) ||
- ((mao.pfn + mao.nr - 1) > domain_get_maximum_gpfn(d))) )
- break;
-
- rc = p2m_set_mem_access(d, mao.pfn, mao.nr, start_iter,
- MEMOP_CMD_MASK, mao.access);
- if ( rc > 0 )
- {
- ASSERT(!(rc & MEMOP_CMD_MASK));
- rc = hypercall_create_continuation(__HYPERVISOR_memory_op, "lh",
- XENMEM_access_op | rc, arg);
- }
- break;
- }
-
- case XENMEM_access_op_get_access:
- {
- xenmem_access_t access;
-
- rc = -EINVAL;
- if ( (mao.pfn > domain_get_maximum_gpfn(d)) && mao.pfn != ~0ull )
- break;
-
- rc = p2m_get_mem_access(d, mao.pfn, &access);
- if ( rc != 0 )
- break;
-
- mao.access = access;
- rc = __copy_field_to_guest(arg, &mao, access) ? -EFAULT : 0;
-
- break;
- }
-
- default:
- rc = -ENOSYS;
- break;
- }
-
- out:
- rcu_unlock_domain(d);
- return rc;
-}
-
-int mem_access_send_req(struct domain *d, mem_event_request_t *req)
-{
- int rc = mem_event_claim_slot(d, &d->mem_event->access);
- if ( rc < 0 )
- return rc;
-
- mem_event_put_request(d, &d->mem_event->access, req);
-
- return 0;
-}
-
-/*
- * Local variables:
- * mode: C
- * c-file-style: "BSD"
- * c-basic-offset: 4
- * indent-tabs-mode: nil
- * End:
- */
diff --git a/xen/arch/x86/mm/mem_event.c b/xen/arch/x86/mm/mem_event.c
deleted file mode 100644
index ba7e71e..0000000
--- a/xen/arch/x86/mm/mem_event.c
+++ /dev/null
@@ -1,705 +0,0 @@
-/******************************************************************************
- * arch/x86/mm/mem_event.c
- *
- * Memory event support.
- *
- * Copyright (c) 2009 Citrix Systems, Inc. (Patrick Colp)
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- */
-
-
-#include <asm/domain.h>
-#include <xen/event.h>
-#include <xen/wait.h>
-#include <asm/p2m.h>
-#include <asm/mem_event.h>
-#include <asm/mem_paging.h>
-#include <asm/mem_access.h>
-#include <asm/mem_sharing.h>
-#include <xsm/xsm.h>
-
-/* for public/io/ring.h macros */
-#define xen_mb() mb()
-#define xen_rmb() rmb()
-#define xen_wmb() wmb()
-
-#define mem_event_ring_lock_init(_med) spin_lock_init(&(_med)->ring_lock)
-#define mem_event_ring_lock(_med) spin_lock(&(_med)->ring_lock)
-#define mem_event_ring_unlock(_med) spin_unlock(&(_med)->ring_lock)
-
-static int mem_event_enable(
- struct domain *d,
- xen_domctl_mem_event_op_t *mec,
- struct mem_event_domain *med,
- int pause_flag,
- int param,
- xen_event_channel_notification_t notification_fn)
-{
- int rc;
- unsigned long ring_gfn = d->arch.hvm_domain.params[param];
-
- /* Only one helper at a time. If the helper crashed,
- * the ring is in an undefined state and so is the guest.
- */
- if ( med->ring_page )
- return -EBUSY;
-
- /* The parameter defaults to zero, and it should be
- * set to something */
- if ( ring_gfn == 0 )
- return -ENOSYS;
-
- mem_event_ring_lock_init(med);
- mem_event_ring_lock(med);
-
- rc = prepare_ring_for_helper(d, ring_gfn, &med->ring_pg_struct,
- &med->ring_page);
- if ( rc < 0 )
- goto err;
-
- /* Set the number of currently blocked vCPUs to 0. */
- med->blocked = 0;
-
- /* Allocate event channel */
- rc = alloc_unbound_xen_event_channel(d->vcpu[0],
- current->domain->domain_id,
- notification_fn);
- if ( rc < 0 )
- goto err;
-
- med->xen_port = mec->port = rc;
-
- /* Prepare ring buffer */
- FRONT_RING_INIT(&med->front_ring,
- (mem_event_sring_t *)med->ring_page,
- PAGE_SIZE);
-
- /* Save the pause flag for this particular ring. */
- med->pause_flag = pause_flag;
-
- /* Initialize the last-chance wait queue. */
- init_waitqueue_head(&med->wq);
-
- mem_event_ring_unlock(med);
- return 0;
-
- err:
- destroy_ring_for_helper(&med->ring_page,
- med->ring_pg_struct);
- mem_event_ring_unlock(med);
-
- return rc;
-}
-
-static unsigned int mem_event_ring_available(struct mem_event_domain *med)
-{
- int avail_req = RING_FREE_REQUESTS(&med->front_ring);
- avail_req -= med->target_producers;
- avail_req -= med->foreign_producers;
-
- BUG_ON(avail_req < 0);
-
- return avail_req;
-}
-
-/*
- * mem_event_wake_blocked() will wakeup vcpus waiting for room in the
- * ring. These vCPUs were paused on their way out after placing an event,
- * but need to be resumed where the ring is capable of processing at least
- * one event from them.
- */
-static void mem_event_wake_blocked(struct domain *d, struct mem_event_domain *med)
-{
- struct vcpu *v;
- int online = d->max_vcpus;
- unsigned int avail_req = mem_event_ring_available(med);
-
- if ( avail_req == 0 || med->blocked == 0 )
- return;
-
- /*
- * We ensure that we only have vCPUs online if there are enough free slots
- * for their memory events to be processed. This will ensure that no
- * memory events are lost (due to the fact that certain types of events
- * cannot be replayed, we need to ensure that there is space in the ring
- * for when they are hit).
- * See comment below in mem_event_put_request().
- */
- for_each_vcpu ( d, v )
- if ( test_bit(med->pause_flag, &v->pause_flags) )
- online--;
-
- ASSERT(online == (d->max_vcpus - med->blocked));
-
- /* We remember which vcpu last woke up to avoid scanning always linearly
- * from zero and starving higher-numbered vcpus under high load */
- if ( d->vcpu )
- {
- int i, j, k;
-
- for (i = med->last_vcpu_wake_up + 1, j = 0; j < d->max_vcpus; i++, j++)
- {
- k = i % d->max_vcpus;
- v = d->vcpu[k];
- if ( !v )
- continue;
-
- if ( !(med->blocked) || online >= avail_req )
- break;
-
- if ( test_and_clear_bit(med->pause_flag, &v->pause_flags) )
- {
- vcpu_unpause(v);
- online++;
- med->blocked--;
- med->last_vcpu_wake_up = k;
- }
- }
- }
-}
-
-/*
- * In the event that a vCPU attempted to place an event in the ring and
- * was unable to do so, it is queued on a wait queue. These are woken as
- * needed, and take precedence over the blocked vCPUs.
- */
-static void mem_event_wake_queued(struct domain *d, struct mem_event_domain *med)
-{
- unsigned int avail_req = mem_event_ring_available(med);
-
- if ( avail_req > 0 )
- wake_up_nr(&med->wq, avail_req);
-}
-
-/*
- * mem_event_wake() will wakeup all vcpus waiting for the ring to
- * become available. If we have queued vCPUs, they get top priority. We
- * are guaranteed that they will go through code paths that will eventually
- * call mem_event_wake() again, ensuring that any blocked vCPUs will get
- * unpaused once all the queued vCPUs have made it through.
- */
-void mem_event_wake(struct domain *d, struct mem_event_domain *med)
-{
- if (!list_empty(&med->wq.list))
- mem_event_wake_queued(d, med);
- else
- mem_event_wake_blocked(d, med);
-}
-
-static int mem_event_disable(struct domain *d, struct mem_event_domain *med)
-{
- if ( med->ring_page )
- {
- struct vcpu *v;
-
- mem_event_ring_lock(med);
-
- if ( !list_empty(&med->wq.list) )
- {
- mem_event_ring_unlock(med);
- return -EBUSY;
- }
-
- /* Free domU's event channel and leave the other one unbound */
- free_xen_event_channel(d->vcpu[0], med->xen_port);
-
- /* Unblock all vCPUs */
- for_each_vcpu ( d, v )
- {
- if ( test_and_clear_bit(med->pause_flag, &v->pause_flags) )
- {
- vcpu_unpause(v);
- med->blocked--;
- }
- }
-
- destroy_ring_for_helper(&med->ring_page,
- med->ring_pg_struct);
- mem_event_ring_unlock(med);
- }
-
- return 0;
-}
-
-static inline void mem_event_release_slot(struct domain *d,
- struct mem_event_domain *med)
-{
- /* Update the accounting */
- if ( current->domain == d )
- med->target_producers--;
- else
- med->foreign_producers--;
-
- /* Kick any waiters */
- mem_event_wake(d, med);
-}
-
-/*
- * mem_event_mark_and_pause() tags vcpu and put it to sleep.
- * The vcpu will resume execution in mem_event_wake_waiters().
- */
-void mem_event_mark_and_pause(struct vcpu *v, struct mem_event_domain *med)
-{
- if ( !test_and_set_bit(med->pause_flag, &v->pause_flags) )
- {
- vcpu_pause_nosync(v);
- med->blocked++;
- }
-}
-
-/*
- * This must be preceded by a call to claim_slot(), and is guaranteed to
- * succeed. As a side-effect however, the vCPU may be paused if the ring is
- * overly full and its continued execution would cause stalling and excessive
- * waiting. The vCPU will be automatically unpaused when the ring clears.
- */
-void mem_event_put_request(struct domain *d,
- struct mem_event_domain *med,
- mem_event_request_t *req)
-{
- mem_event_front_ring_t *front_ring;
- int free_req;
- unsigned int avail_req;
- RING_IDX req_prod;
-
- if ( current->domain != d )
- {
- req->flags |= MEM_EVENT_FLAG_FOREIGN;
- ASSERT( !(req->flags & MEM_EVENT_FLAG_VCPU_PAUSED) );
- }
-
- mem_event_ring_lock(med);
-
- /* Due to the reservations, this step must succeed. */
- front_ring = &med->front_ring;
- free_req = RING_FREE_REQUESTS(front_ring);
- ASSERT(free_req > 0);
-
- /* Copy request */
- req_prod = front_ring->req_prod_pvt;
- memcpy(RING_GET_REQUEST(front_ring, req_prod), req, sizeof(*req));
- req_prod++;
-
- /* Update ring */
- front_ring->req_prod_pvt = req_prod;
- RING_PUSH_REQUESTS(front_ring);
-
- /* We've actually *used* our reservation, so release the slot. */
- mem_event_release_slot(d, med);
-
- /* Give this vCPU a black eye if necessary, on the way out.
- * See the comments above wake_blocked() for more information
- * on how this mechanism works to avoid waiting. */
- avail_req = mem_event_ring_available(med);
- if( current->domain == d && avail_req < d->max_vcpus )
- mem_event_mark_and_pause(current, med);
-
- mem_event_ring_unlock(med);
-
- notify_via_xen_event_channel(d, med->xen_port);
-}
-
-int mem_event_get_response(struct domain *d, struct mem_event_domain *med, mem_event_response_t *rsp)
-{
- mem_event_front_ring_t *front_ring;
- RING_IDX rsp_cons;
-
- mem_event_ring_lock(med);
-
- front_ring = &med->front_ring;
- rsp_cons = front_ring->rsp_cons;
-
- if ( !RING_HAS_UNCONSUMED_RESPONSES(front_ring) )
- {
- mem_event_ring_unlock(med);
- return 0;
- }
-
- /* Copy response */
- memcpy(rsp, RING_GET_RESPONSE(front_ring, rsp_cons), sizeof(*rsp));
- rsp_cons++;
-
- /* Update ring */
- front_ring->rsp_cons = rsp_cons;
- front_ring->sring->rsp_event = rsp_cons + 1;
-
- /* Kick any waiters -- since we've just consumed an event,
- * there may be additional space available in the ring. */
- mem_event_wake(d, med);
-
- mem_event_ring_unlock(med);
-
- return 1;
-}
-
-void mem_event_cancel_slot(struct domain *d, struct mem_event_domain *med)
-{
- mem_event_ring_lock(med);
- mem_event_release_slot(d, med);
- mem_event_ring_unlock(med);
-}
-
-static int mem_event_grab_slot(struct mem_event_domain *med, int foreign)
-{
- unsigned int avail_req;
-
- if ( !med->ring_page )
- return -ENOSYS;
-
- mem_event_ring_lock(med);
-
- avail_req = mem_event_ring_available(med);
- if ( avail_req == 0 )
- {
- mem_event_ring_unlock(med);
- return -EBUSY;
- }
-
- if ( !foreign )
- med->target_producers++;
- else
- med->foreign_producers++;
-
- mem_event_ring_unlock(med);
-
- return 0;
-}
-
-/* Simple try_grab wrapper for use in the wait_event() macro. */
-static int mem_event_wait_try_grab(struct mem_event_domain *med, int *rc)
-{
- *rc = mem_event_grab_slot(med, 0);
- return *rc;
-}
-
-/* Call mem_event_grab_slot() until the ring doesn't exist, or is available. */
-static int mem_event_wait_slot(struct mem_event_domain *med)
-{
- int rc = -EBUSY;
- wait_event(med->wq, mem_event_wait_try_grab(med, &rc) != -EBUSY);
- return rc;
-}
-
-bool_t mem_event_check_ring(struct mem_event_domain *med)
-{
- return (med->ring_page != NULL);
-}
-
-/*
- * Determines whether or not the current vCPU belongs to the target domain,
- * and calls the appropriate wait function. If it is a guest vCPU, then we
- * use mem_event_wait_slot() to reserve a slot. As long as there is a ring,
- * this function will always return 0 for a guest. For a non-guest, we check
- * for space and return -EBUSY if the ring is not available.
- *
- * Return codes: -ENOSYS: the ring is not yet configured
- * -EBUSY: the ring is busy
- * 0: a spot has been reserved
- *
- */
-int __mem_event_claim_slot(struct domain *d, struct mem_event_domain *med,
- bool_t allow_sleep)
-{
- if ( (current->domain == d) && allow_sleep )
- return mem_event_wait_slot(med);
- else
- return mem_event_grab_slot(med, (current->domain != d));
-}
-
-/* Registered with Xen-bound event channel for incoming notifications. */
-static void mem_paging_notification(struct vcpu *v, unsigned int port)
-{
- if ( likely(v->domain->mem_event->paging.ring_page != NULL) )
- p2m_mem_paging_resume(v->domain);
-}
-
-/* Registered with Xen-bound event channel for incoming notifications. */
-static void mem_access_notification(struct vcpu *v, unsigned int port)
-{
- if ( likely(v->domain->mem_event->access.ring_page != NULL) )
- p2m_mem_access_resume(v->domain);
-}
-
-/* Registered with Xen-bound event channel for incoming notifications. */
-static void mem_sharing_notification(struct vcpu *v, unsigned int port)
-{
- if ( likely(v->domain->mem_event->share.ring_page != NULL) )
- mem_sharing_sharing_resume(v->domain);
-}
-
-int do_mem_event_op(int op, uint32_t domain, void *arg)
-{
- int ret;
- struct domain *d;
-
- ret = rcu_lock_live_remote_domain_by_id(domain, &d);
- if ( ret )
- return ret;
-
- ret = xsm_mem_event_op(XSM_DM_PRIV, d, op);
- if ( ret )
- goto out;
-
- switch (op)
- {
- case XENMEM_paging_op:
- ret = mem_paging_memop(d, (xen_mem_event_op_t *) arg);
- break;
- case XENMEM_sharing_op:
- ret = mem_sharing_memop(d, (xen_mem_sharing_op_t *) arg);
- break;
- default:
- ret = -ENOSYS;
- }
-
- out:
- rcu_unlock_domain(d);
- return ret;
-}
-
-/* Clean up on domain destruction */
-void mem_event_cleanup(struct domain *d)
-{
- if ( d->mem_event->paging.ring_page ) {
- /* Destroying the wait queue head means waking up all
- * queued vcpus. This will drain the list, allowing
- * the disable routine to complete. It will also drop
- * all domain refs the wait-queued vcpus are holding.
- * Finally, because this code path involves previously
- * pausing the domain (domain_kill), unpausing the
- * vcpus causes no harm. */
- destroy_waitqueue_head(&d->mem_event->paging.wq);
- (void)mem_event_disable(d, &d->mem_event->paging);
- }
- if ( d->mem_event->access.ring_page ) {
- destroy_waitqueue_head(&d->mem_event->access.wq);
- (void)mem_event_disable(d, &d->mem_event->access);
- }
- if ( d->mem_event->share.ring_page ) {
- destroy_waitqueue_head(&d->mem_event->share.wq);
- (void)mem_event_disable(d, &d->mem_event->share);
- }
-}
-
-int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
- XEN_GUEST_HANDLE_PARAM(void) u_domctl)
-{
- int rc;
-
- rc = xsm_mem_event_control(XSM_PRIV, d, mec->mode, mec->op);
- if ( rc )
- return rc;
-
- if ( unlikely(d == current->domain) )
- {
- gdprintk(XENLOG_INFO, "Tried to do a memory event op on itself.\n");
- return -EINVAL;
- }
-
- if ( unlikely(d->is_dying) )
- {
- gdprintk(XENLOG_INFO, "Ignoring memory event op on dying domain %u\n",
- d->domain_id);
- return 0;
- }
-
- if ( unlikely(d->vcpu == NULL) || unlikely(d->vcpu[0] == NULL) )
- {
- gdprintk(XENLOG_INFO,
- "Memory event op on a domain (%u) with no vcpus\n",
- d->domain_id);
- return -EINVAL;
- }
-
- rc = -ENOSYS;
-
- switch ( mec->mode )
- {
- case XEN_DOMCTL_MEM_EVENT_OP_PAGING:
- {
- struct mem_event_domain *med = &d->mem_event->paging;
- rc = -EINVAL;
-
- switch( mec->op )
- {
- case XEN_DOMCTL_MEM_EVENT_OP_PAGING_ENABLE:
- {
- struct p2m_domain *p2m = p2m_get_hostp2m(d);
-
- rc = -EOPNOTSUPP;
- /* pvh fixme: p2m_is_foreign types need addressing */
- if ( is_pvh_vcpu(current) || is_pvh_domain(hardware_domain) )
- break;
-
- rc = -ENODEV;
- /* Only HAP is supported */
- if ( !hap_enabled(d) )
- break;
-
- /* No paging if iommu is used */
- rc = -EMLINK;
- if ( unlikely(need_iommu(d)) )
- break;
-
- rc = -EXDEV;
- /* Disallow paging in a PoD guest */
- if ( p2m->pod.entry_count )
- break;
-
- rc = mem_event_enable(d, mec, med, _VPF_mem_paging,
- HVM_PARAM_PAGING_RING_PFN,
- mem_paging_notification);
- }
- break;
-
- case XEN_DOMCTL_MEM_EVENT_OP_PAGING_DISABLE:
- {
- if ( med->ring_page )
- rc = mem_event_disable(d, med);
- }
- break;
-
- default:
- rc = -ENOSYS;
- break;
- }
- }
- break;
-
- case XEN_DOMCTL_MEM_EVENT_OP_ACCESS:
- {
- struct mem_event_domain *med = &d->mem_event->access;
- rc = -EINVAL;
-
- switch( mec->op )
- {
- case XEN_DOMCTL_MEM_EVENT_OP_ACCESS_ENABLE:
- {
- rc = -ENODEV;
- /* Only HAP is supported */
- if ( !hap_enabled(d) )
- break;
-
- /* Currently only EPT is supported */
- if ( !cpu_has_vmx )
- break;
-
- rc = mem_event_enable(d, mec, med, _VPF_mem_access,
- HVM_PARAM_ACCESS_RING_PFN,
- mem_access_notification);
- }
- break;
-
- case XEN_DOMCTL_MEM_EVENT_OP_ACCESS_DISABLE:
- {
- if ( med->ring_page )
- rc = mem_event_disable(d, med);
- }
- break;
-
- default:
- rc = -ENOSYS;
- break;
- }
- }
- break;
-
- case XEN_DOMCTL_MEM_EVENT_OP_SHARING:
- {
- struct mem_event_domain *med = &d->mem_event->share;
- rc = -EINVAL;
-
- switch( mec->op )
- {
- case XEN_DOMCTL_MEM_EVENT_OP_SHARING_ENABLE:
- {
- rc = -EOPNOTSUPP;
- /* pvh fixme: p2m_is_foreign types need addressing */
- if ( is_pvh_vcpu(current) || is_pvh_domain(hardware_domain) )
- break;
-
- rc = -ENODEV;
- /* Only HAP is supported */
- if ( !hap_enabled(d) )
- break;
-
- rc = mem_event_enable(d, mec, med, _VPF_mem_sharing,
- HVM_PARAM_SHARING_RING_PFN,
- mem_sharing_notification);
- }
- break;
-
- case XEN_DOMCTL_MEM_EVENT_OP_SHARING_DISABLE:
- {
- if ( med->ring_page )
- rc = mem_event_disable(d, med);
- }
- break;
-
- default:
- rc = -ENOSYS;
- break;
- }
- }
- break;
-
- default:
- rc = -ENOSYS;
- }
-
- return rc;
-}
-
-void mem_event_vcpu_pause(struct vcpu *v)
-{
- ASSERT(v == current);
-
- atomic_inc(&v->mem_event_pause_count);
- vcpu_pause_nosync(v);
-}
-
-void mem_event_vcpu_unpause(struct vcpu *v)
-{
- int old, new, prev = v->mem_event_pause_count.counter;
-
- /* All unpause requests as a result of toolstack responses. Prevent
- * underflow of the vcpu pause count. */
- do
- {
- old = prev;
- new = old - 1;
-
- if ( new < 0 )
- {
- printk(XENLOG_G_WARNING
- "%pv mem_event: Too many unpause attempts\n", v);
- return;
- }
-
- prev = cmpxchg(&v->mem_event_pause_count.counter, old, new);
- } while ( prev != old );
-
- vcpu_unpause(v);
-}
-
-/*
- * Local variables:
- * mode: C
- * c-file-style: "BSD"
- * c-basic-offset: 4
- * indent-tabs-mode: nil
- * End:
- */
diff --git a/xen/arch/x86/mm/mem_paging.c b/xen/arch/x86/mm/mem_paging.c
index 235776d..65f6a3d 100644
--- a/xen/arch/x86/mm/mem_paging.c
+++ b/xen/arch/x86/mm/mem_paging.c
@@ -22,7 +22,7 @@
#include <asm/p2m.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
int mem_paging_memop(struct domain *d, xen_mem_event_op_t *mec)
diff --git a/xen/arch/x86/mm/mem_sharing.c b/xen/arch/x86/mm/mem_sharing.c
index 79188b9..7c0fc7d 100644
--- a/xen/arch/x86/mm/mem_sharing.c
+++ b/xen/arch/x86/mm/mem_sharing.c
@@ -27,12 +27,12 @@
#include <xen/mm.h>
#include <xen/grant_table.h>
#include <xen/sched.h>
+#include <xen/rcupdate.h>
+#include <xen/mem_event.h>
#include <asm/page.h>
#include <asm/string.h>
#include <asm/p2m.h>
-#include <asm/mem_event.h>
#include <asm/atomic.h>
-#include <xen/rcupdate.h>
#include <asm/event.h>
#include <xsm/xsm.h>
diff --git a/xen/arch/x86/mm/p2m-pod.c b/xen/arch/x86/mm/p2m-pod.c
index bd4c7c8..43f507c 100644
--- a/xen/arch/x86/mm/p2m-pod.c
+++ b/xen/arch/x86/mm/p2m-pod.c
@@ -20,16 +20,16 @@
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
+#include <xen/iommu.h>
+#include <xen/mem_event.h>
+#include <xen/event.h>
+#include <public/mem_event.h>
#include <asm/domain.h>
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
#include <asm/hvm/vmx/vmx.h> /* ept_p2m_init() */
-#include <xen/iommu.h>
-#include <asm/mem_event.h>
-#include <public/mem_event.h>
#include <asm/mem_sharing.h>
-#include <xen/event.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/svm/amd-iommu-proto.h>
diff --git a/xen/arch/x86/mm/p2m-pt.c b/xen/arch/x86/mm/p2m-pt.c
index 085ab6f..e48b63a 100644
--- a/xen/arch/x86/mm/p2m-pt.c
+++ b/xen/arch/x86/mm/p2m-pt.c
@@ -25,16 +25,16 @@
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
+#include <xen/iommu.h>
+#include <xen/mem_event.h>
+#include <xen/event.h>
+#include <xen/trace.h>
+#include <public/mem_event.h>
#include <asm/domain.h>
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
-#include <xen/iommu.h>
-#include <asm/mem_event.h>
-#include <public/mem_event.h>
#include <asm/mem_sharing.h>
-#include <xen/event.h>
-#include <xen/trace.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/svm/amd-iommu-proto.h>
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index 32776c3..a9f120a 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -24,16 +24,16 @@
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
+#include <xen/iommu.h>
+#include <xen/mem_event.h>
+#include <xen/event.h>
+#include <public/mem_event.h>
#include <asm/domain.h>
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
#include <asm/hvm/vmx/vmx.h> /* ept_p2m_init() */
-#include <xen/iommu.h>
-#include <asm/mem_event.h>
-#include <public/mem_event.h>
#include <asm/mem_sharing.h>
-#include <xen/event.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/svm/amd-iommu-proto.h>
#include <xsm/xsm.h>
diff --git a/xen/arch/x86/x86_64/compat/mm.c b/xen/arch/x86/x86_64/compat/mm.c
index 69c6195..c079702 100644
--- a/xen/arch/x86/x86_64/compat/mm.c
+++ b/xen/arch/x86/x86_64/compat/mm.c
@@ -1,10 +1,10 @@
#include <xen/event.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.h>
#include <xen/multicall.h>
#include <compat/memory.h>
#include <compat/xen.h>
-#include <asm/mem_event.h>
#include <asm/mem_sharing.h>
-#include <asm/mem_access.h>
int compat_set_gdt(XEN_GUEST_HANDLE_PARAM(uint) frame_list, unsigned int entries)
{
diff --git a/xen/arch/x86/x86_64/mm.c b/xen/arch/x86/x86_64/mm.c
index 4937f9a..0da6ddc 100644
--- a/xen/arch/x86/x86_64/mm.c
+++ b/xen/arch/x86/x86_64/mm.c
@@ -26,6 +26,8 @@
#include <xen/nodemask.h>
#include <xen/guest_access.h>
#include <xen/hypercall.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.h>
#include <asm/current.h>
#include <asm/asm_defns.h>
#include <asm/page.h>
@@ -35,9 +37,7 @@
#include <asm/msr.h>
#include <asm/setup.h>
#include <asm/numa.h>
-#include <asm/mem_event.h>
#include <asm/mem_sharing.h>
-#include <asm/mem_access.h>
#include <public/memory.h>
/* Parameters for PFN/MADDR compression. */
diff --git a/xen/common/Makefile b/xen/common/Makefile
index 3683ae3..b9f3387 100644
--- a/xen/common/Makefile
+++ b/xen/common/Makefile
@@ -51,6 +51,8 @@ obj-y += tmem_xen.o
obj-y += radix-tree.o
obj-y += rbtree.o
obj-y += lzo.o
+obj-$(HAS_MEM_ACCESS) += mem_access.o
+obj-$(HAS_MEM_ACCESS) += mem_event.o
obj-bin-$(CONFIG_X86) += $(foreach n,decompress bunzip2 unxz unlzma unlzo unlz4 earlycpio,$(n).init.o)
diff --git a/xen/common/domain.c b/xen/common/domain.c
index 62514b0..134bed6 100644
--- a/xen/common/domain.c
+++ b/xen/common/domain.c
@@ -15,6 +15,7 @@
#include <xen/domain.h>
#include <xen/mm.h>
#include <xen/event.h>
+#include <xen/mem_event.h>
#include <xen/time.h>
#include <xen/console.h>
#include <xen/softirq.h>
diff --git a/xen/common/mem_access.c b/xen/common/mem_access.c
new file mode 100644
index 0000000..9a8c1a9
--- /dev/null
+++ b/xen/common/mem_access.c
@@ -0,0 +1,133 @@
+/******************************************************************************
+ * mem_access.c
+ *
+ * Memory access support.
+ *
+ * Copyright (c) 2011 Virtuata, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+
+#include <xen/sched.h>
+#include <xen/guest_access.h>
+#include <xen/hypercall.h>
+#include <xen/mem_event.h>
+#include <public/memory.h>
+#include <asm/p2m.h>
+#include <xsm/xsm.h>
+
+int mem_access_memop(unsigned long cmd,
+ XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg)
+{
+ long rc;
+ xen_mem_access_op_t mao;
+ struct domain *d;
+
+ if ( copy_from_guest(&mao, arg, 1) )
+ return -EFAULT;
+
+ rc = rcu_lock_live_remote_domain_by_id(mao.domid, &d);
+ if ( rc )
+ return rc;
+
+ rc = -EINVAL;
+ if ( !is_hvm_domain(d) )
+ goto out;
+
+ rc = xsm_mem_event_op(XSM_DM_PRIV, d, XENMEM_access_op);
+ if ( rc )
+ goto out;
+
+ rc = -ENODEV;
+ if ( unlikely(!d->mem_event->access.ring_page) )
+ goto out;
+
+ switch ( mao.op )
+ {
+ case XENMEM_access_op_resume:
+ p2m_mem_access_resume(d);
+ rc = 0;
+ break;
+
+ case XENMEM_access_op_set_access:
+ {
+ unsigned long start_iter = cmd & ~MEMOP_CMD_MASK;
+
+ rc = -EINVAL;
+ if ( (mao.pfn != ~0ull) &&
+ (mao.nr < start_iter ||
+ ((mao.pfn + mao.nr - 1) < mao.pfn) ||
+ ((mao.pfn + mao.nr - 1) > domain_get_maximum_gpfn(d))) )
+ break;
+
+ rc = p2m_set_mem_access(d, mao.pfn, mao.nr, start_iter,
+ MEMOP_CMD_MASK, mao.access);
+ if ( rc > 0 )
+ {
+ ASSERT(!(rc & MEMOP_CMD_MASK));
+ rc = hypercall_create_continuation(__HYPERVISOR_memory_op, "lh",
+ XENMEM_access_op | rc, arg);
+ }
+ break;
+ }
+
+ case XENMEM_access_op_get_access:
+ {
+ xenmem_access_t access;
+
+ rc = -EINVAL;
+ if ( (mao.pfn > domain_get_maximum_gpfn(d)) && mao.pfn != ~0ull )
+ break;
+
+ rc = p2m_get_mem_access(d, mao.pfn, &access);
+ if ( rc != 0 )
+ break;
+
+ mao.access = access;
+ rc = __copy_field_to_guest(arg, &mao, access) ? -EFAULT : 0;
+
+ break;
+ }
+
+ default:
+ rc = -ENOSYS;
+ break;
+ }
+
+ out:
+ rcu_unlock_domain(d);
+ return rc;
+}
+
+int mem_access_send_req(struct domain *d, mem_event_request_t *req)
+{
+ int rc = mem_event_claim_slot(d, &d->mem_event->access);
+ if ( rc < 0 )
+ return rc;
+
+ mem_event_put_request(d, &d->mem_event->access, req);
+
+ return 0;
+}
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
new file mode 100644
index 0000000..336b638
--- /dev/null
+++ b/xen/common/mem_event.c
@@ -0,0 +1,705 @@
+/******************************************************************************
+ * mem_event.c
+ *
+ * Memory event support.
+ *
+ * Copyright (c) 2009 Citrix Systems, Inc. (Patrick Colp)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+
+#include <xen/sched.h>
+#include <xen/event.h>
+#include <xen/wait.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.h>
+#include <asm/p2m.h>
+#include <asm/mem_paging.h>
+#include <asm/mem_sharing.h>
+#include <xsm/xsm.h>
+
+/* for public/io/ring.h macros */
+#define xen_mb() mb()
+#define xen_rmb() rmb()
+#define xen_wmb() wmb()
+
+#define mem_event_ring_lock_init(_med) spin_lock_init(&(_med)->ring_lock)
+#define mem_event_ring_lock(_med) spin_lock(&(_med)->ring_lock)
+#define mem_event_ring_unlock(_med) spin_unlock(&(_med)->ring_lock)
+
+static int mem_event_enable(
+ struct domain *d,
+ xen_domctl_mem_event_op_t *mec,
+ struct mem_event_domain *med,
+ int pause_flag,
+ int param,
+ xen_event_channel_notification_t notification_fn)
+{
+ int rc;
+ unsigned long ring_gfn = d->arch.hvm_domain.params[param];
+
+ /* Only one helper at a time. If the helper crashed,
+ * the ring is in an undefined state and so is the guest.
+ */
+ if ( med->ring_page )
+ return -EBUSY;
+
+ /* The parameter defaults to zero, and it should be
+ * set to something */
+ if ( ring_gfn == 0 )
+ return -ENOSYS;
+
+ mem_event_ring_lock_init(med);
+ mem_event_ring_lock(med);
+
+ rc = prepare_ring_for_helper(d, ring_gfn, &med->ring_pg_struct,
+ &med->ring_page);
+ if ( rc < 0 )
+ goto err;
+
+ /* Set the number of currently blocked vCPUs to 0. */
+ med->blocked = 0;
+
+ /* Allocate event channel */
+ rc = alloc_unbound_xen_event_channel(d->vcpu[0],
+ current->domain->domain_id,
+ notification_fn);
+ if ( rc < 0 )
+ goto err;
+
+ med->xen_port = mec->port = rc;
+
+ /* Prepare ring buffer */
+ FRONT_RING_INIT(&med->front_ring,
+ (mem_event_sring_t *)med->ring_page,
+ PAGE_SIZE);
+
+ /* Save the pause flag for this particular ring. */
+ med->pause_flag = pause_flag;
+
+ /* Initialize the last-chance wait queue. */
+ init_waitqueue_head(&med->wq);
+
+ mem_event_ring_unlock(med);
+ return 0;
+
+ err:
+ destroy_ring_for_helper(&med->ring_page,
+ med->ring_pg_struct);
+ mem_event_ring_unlock(med);
+
+ return rc;
+}
+
+static unsigned int mem_event_ring_available(struct mem_event_domain *med)
+{
+ int avail_req = RING_FREE_REQUESTS(&med->front_ring);
+ avail_req -= med->target_producers;
+ avail_req -= med->foreign_producers;
+
+ BUG_ON(avail_req < 0);
+
+ return avail_req;
+}
+
+/*
+ * mem_event_wake_blocked() will wakeup vcpus waiting for room in the
+ * ring. These vCPUs were paused on their way out after placing an event,
+ * but need to be resumed where the ring is capable of processing at least
+ * one event from them.
+ */
+static void mem_event_wake_blocked(struct domain *d, struct mem_event_domain *med)
+{
+ struct vcpu *v;
+ int online = d->max_vcpus;
+ unsigned int avail_req = mem_event_ring_available(med);
+
+ if ( avail_req == 0 || med->blocked == 0 )
+ return;
+
+ /*
+ * We ensure that we only have vCPUs online if there are enough free slots
+ * for their memory events to be processed. This will ensure that no
+ * memory events are lost (due to the fact that certain types of events
+ * cannot be replayed, we need to ensure that there is space in the ring
+ * for when they are hit).
+ * See comment below in mem_event_put_request().
+ */
+ for_each_vcpu ( d, v )
+ if ( test_bit(med->pause_flag, &v->pause_flags) )
+ online--;
+
+ ASSERT(online == (d->max_vcpus - med->blocked));
+
+ /* We remember which vcpu last woke up to avoid scanning always linearly
+ * from zero and starving higher-numbered vcpus under high load */
+ if ( d->vcpu )
+ {
+ int i, j, k;
+
+ for (i = med->last_vcpu_wake_up + 1, j = 0; j < d->max_vcpus; i++, j++)
+ {
+ k = i % d->max_vcpus;
+ v = d->vcpu[k];
+ if ( !v )
+ continue;
+
+ if ( !(med->blocked) || online >= avail_req )
+ break;
+
+ if ( test_and_clear_bit(med->pause_flag, &v->pause_flags) )
+ {
+ vcpu_unpause(v);
+ online++;
+ med->blocked--;
+ med->last_vcpu_wake_up = k;
+ }
+ }
+ }
+}
+
+/*
+ * In the event that a vCPU attempted to place an event in the ring and
+ * was unable to do so, it is queued on a wait queue. These are woken as
+ * needed, and take precedence over the blocked vCPUs.
+ */
+static void mem_event_wake_queued(struct domain *d, struct mem_event_domain *med)
+{
+ unsigned int avail_req = mem_event_ring_available(med);
+
+ if ( avail_req > 0 )
+ wake_up_nr(&med->wq, avail_req);
+}
+
+/*
+ * mem_event_wake() will wakeup all vcpus waiting for the ring to
+ * become available. If we have queued vCPUs, they get top priority. We
+ * are guaranteed that they will go through code paths that will eventually
+ * call mem_event_wake() again, ensuring that any blocked vCPUs will get
+ * unpaused once all the queued vCPUs have made it through.
+ */
+void mem_event_wake(struct domain *d, struct mem_event_domain *med)
+{
+ if (!list_empty(&med->wq.list))
+ mem_event_wake_queued(d, med);
+ else
+ mem_event_wake_blocked(d, med);
+}
+
+static int mem_event_disable(struct domain *d, struct mem_event_domain *med)
+{
+ if ( med->ring_page )
+ {
+ struct vcpu *v;
+
+ mem_event_ring_lock(med);
+
+ if ( !list_empty(&med->wq.list) )
+ {
+ mem_event_ring_unlock(med);
+ return -EBUSY;
+ }
+
+ /* Free domU's event channel and leave the other one unbound */
+ free_xen_event_channel(d->vcpu[0], med->xen_port);
+
+ /* Unblock all vCPUs */
+ for_each_vcpu ( d, v )
+ {
+ if ( test_and_clear_bit(med->pause_flag, &v->pause_flags) )
+ {
+ vcpu_unpause(v);
+ med->blocked--;
+ }
+ }
+
+ destroy_ring_for_helper(&med->ring_page,
+ med->ring_pg_struct);
+ mem_event_ring_unlock(med);
+ }
+
+ return 0;
+}
+
+static inline void mem_event_release_slot(struct domain *d,
+ struct mem_event_domain *med)
+{
+ /* Update the accounting */
+ if ( current->domain == d )
+ med->target_producers--;
+ else
+ med->foreign_producers--;
+
+ /* Kick any waiters */
+ mem_event_wake(d, med);
+}
+
+/*
+ * mem_event_mark_and_pause() tags vcpu and put it to sleep.
+ * The vcpu will resume execution in mem_event_wake_waiters().
+ */
+void mem_event_mark_and_pause(struct vcpu *v, struct mem_event_domain *med)
+{
+ if ( !test_and_set_bit(med->pause_flag, &v->pause_flags) )
+ {
+ vcpu_pause_nosync(v);
+ med->blocked++;
+ }
+}
+
+/*
+ * This must be preceded by a call to claim_slot(), and is guaranteed to
+ * succeed. As a side-effect however, the vCPU may be paused if the ring is
+ * overly full and its continued execution would cause stalling and excessive
+ * waiting. The vCPU will be automatically unpaused when the ring clears.
+ */
+void mem_event_put_request(struct domain *d,
+ struct mem_event_domain *med,
+ mem_event_request_t *req)
+{
+ mem_event_front_ring_t *front_ring;
+ int free_req;
+ unsigned int avail_req;
+ RING_IDX req_prod;
+
+ if ( current->domain != d )
+ {
+ req->flags |= MEM_EVENT_FLAG_FOREIGN;
+ ASSERT( !(req->flags & MEM_EVENT_FLAG_VCPU_PAUSED) );
+ }
+
+ mem_event_ring_lock(med);
+
+ /* Due to the reservations, this step must succeed. */
+ front_ring = &med->front_ring;
+ free_req = RING_FREE_REQUESTS(front_ring);
+ ASSERT(free_req > 0);
+
+ /* Copy request */
+ req_prod = front_ring->req_prod_pvt;
+ memcpy(RING_GET_REQUEST(front_ring, req_prod), req, sizeof(*req));
+ req_prod++;
+
+ /* Update ring */
+ front_ring->req_prod_pvt = req_prod;
+ RING_PUSH_REQUESTS(front_ring);
+
+ /* We've actually *used* our reservation, so release the slot. */
+ mem_event_release_slot(d, med);
+
+ /* Give this vCPU a black eye if necessary, on the way out.
+ * See the comments above wake_blocked() for more information
+ * on how this mechanism works to avoid waiting. */
+ avail_req = mem_event_ring_available(med);
+ if( current->domain == d && avail_req < d->max_vcpus )
+ mem_event_mark_and_pause(current, med);
+
+ mem_event_ring_unlock(med);
+
+ notify_via_xen_event_channel(d, med->xen_port);
+}
+
+int mem_event_get_response(struct domain *d, struct mem_event_domain *med, mem_event_response_t *rsp)
+{
+ mem_event_front_ring_t *front_ring;
+ RING_IDX rsp_cons;
+
+ mem_event_ring_lock(med);
+
+ front_ring = &med->front_ring;
+ rsp_cons = front_ring->rsp_cons;
+
+ if ( !RING_HAS_UNCONSUMED_RESPONSES(front_ring) )
+ {
+ mem_event_ring_unlock(med);
+ return 0;
+ }
+
+ /* Copy response */
+ memcpy(rsp, RING_GET_RESPONSE(front_ring, rsp_cons), sizeof(*rsp));
+ rsp_cons++;
+
+ /* Update ring */
+ front_ring->rsp_cons = rsp_cons;
+ front_ring->sring->rsp_event = rsp_cons + 1;
+
+ /* Kick any waiters -- since we've just consumed an event,
+ * there may be additional space available in the ring. */
+ mem_event_wake(d, med);
+
+ mem_event_ring_unlock(med);
+
+ return 1;
+}
+
+void mem_event_cancel_slot(struct domain *d, struct mem_event_domain *med)
+{
+ mem_event_ring_lock(med);
+ mem_event_release_slot(d, med);
+ mem_event_ring_unlock(med);
+}
+
+static int mem_event_grab_slot(struct mem_event_domain *med, int foreign)
+{
+ unsigned int avail_req;
+
+ if ( !med->ring_page )
+ return -ENOSYS;
+
+ mem_event_ring_lock(med);
+
+ avail_req = mem_event_ring_available(med);
+ if ( avail_req == 0 )
+ {
+ mem_event_ring_unlock(med);
+ return -EBUSY;
+ }
+
+ if ( !foreign )
+ med->target_producers++;
+ else
+ med->foreign_producers++;
+
+ mem_event_ring_unlock(med);
+
+ return 0;
+}
+
+/* Simple try_grab wrapper for use in the wait_event() macro. */
+static int mem_event_wait_try_grab(struct mem_event_domain *med, int *rc)
+{
+ *rc = mem_event_grab_slot(med, 0);
+ return *rc;
+}
+
+/* Call mem_event_grab_slot() until the ring doesn't exist, or is available. */
+static int mem_event_wait_slot(struct mem_event_domain *med)
+{
+ int rc = -EBUSY;
+ wait_event(med->wq, mem_event_wait_try_grab(med, &rc) != -EBUSY);
+ return rc;
+}
+
+bool_t mem_event_check_ring(struct mem_event_domain *med)
+{
+ return (med->ring_page != NULL);
+}
+
+/*
+ * Determines whether or not the current vCPU belongs to the target domain,
+ * and calls the appropriate wait function. If it is a guest vCPU, then we
+ * use mem_event_wait_slot() to reserve a slot. As long as there is a ring,
+ * this function will always return 0 for a guest. For a non-guest, we check
+ * for space and return -EBUSY if the ring is not available.
+ *
+ * Return codes: -ENOSYS: the ring is not yet configured
+ * -EBUSY: the ring is busy
+ * 0: a spot has been reserved
+ *
+ */
+int __mem_event_claim_slot(struct domain *d, struct mem_event_domain *med,
+ bool_t allow_sleep)
+{
+ if ( (current->domain == d) && allow_sleep )
+ return mem_event_wait_slot(med);
+ else
+ return mem_event_grab_slot(med, (current->domain != d));
+}
+
+/* Registered with Xen-bound event channel for incoming notifications. */
+static void mem_paging_notification(struct vcpu *v, unsigned int port)
+{
+ if ( likely(v->domain->mem_event->paging.ring_page != NULL) )
+ p2m_mem_paging_resume(v->domain);
+}
+
+/* Registered with Xen-bound event channel for incoming notifications. */
+static void mem_access_notification(struct vcpu *v, unsigned int port)
+{
+ if ( likely(v->domain->mem_event->access.ring_page != NULL) )
+ p2m_mem_access_resume(v->domain);
+}
+
+/* Registered with Xen-bound event channel for incoming notifications. */
+static void mem_sharing_notification(struct vcpu *v, unsigned int port)
+{
+ if ( likely(v->domain->mem_event->share.ring_page != NULL) )
+ mem_sharing_sharing_resume(v->domain);
+}
+
+int do_mem_event_op(int op, uint32_t domain, void *arg)
+{
+ int ret;
+ struct domain *d;
+
+ ret = rcu_lock_live_remote_domain_by_id(domain, &d);
+ if ( ret )
+ return ret;
+
+ ret = xsm_mem_event_op(XSM_DM_PRIV, d, op);
+ if ( ret )
+ goto out;
+
+ switch (op)
+ {
+ case XENMEM_paging_op:
+ ret = mem_paging_memop(d, (xen_mem_event_op_t *) arg);
+ break;
+ case XENMEM_sharing_op:
+ ret = mem_sharing_memop(d, (xen_mem_sharing_op_t *) arg);
+ break;
+ default:
+ ret = -ENOSYS;
+ }
+
+ out:
+ rcu_unlock_domain(d);
+ return ret;
+}
+
+/* Clean up on domain destruction */
+void mem_event_cleanup(struct domain *d)
+{
+ if ( d->mem_event->paging.ring_page ) {
+ /* Destroying the wait queue head means waking up all
+ * queued vcpus. This will drain the list, allowing
+ * the disable routine to complete. It will also drop
+ * all domain refs the wait-queued vcpus are holding.
+ * Finally, because this code path involves previously
+ * pausing the domain (domain_kill), unpausing the
+ * vcpus causes no harm. */
+ destroy_waitqueue_head(&d->mem_event->paging.wq);
+ (void)mem_event_disable(d, &d->mem_event->paging);
+ }
+ if ( d->mem_event->access.ring_page ) {
+ destroy_waitqueue_head(&d->mem_event->access.wq);
+ (void)mem_event_disable(d, &d->mem_event->access);
+ }
+ if ( d->mem_event->share.ring_page ) {
+ destroy_waitqueue_head(&d->mem_event->share.wq);
+ (void)mem_event_disable(d, &d->mem_event->share);
+ }
+}
+
+int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
+ XEN_GUEST_HANDLE_PARAM(void) u_domctl)
+{
+ int rc;
+
+ rc = xsm_mem_event_control(XSM_PRIV, d, mec->mode, mec->op);
+ if ( rc )
+ return rc;
+
+ if ( unlikely(d == current->domain) )
+ {
+ gdprintk(XENLOG_INFO, "Tried to do a memory event op on itself.\n");
+ return -EINVAL;
+ }
+
+ if ( unlikely(d->is_dying) )
+ {
+ gdprintk(XENLOG_INFO, "Ignoring memory event op on dying domain %u\n",
+ d->domain_id);
+ return 0;
+ }
+
+ if ( unlikely(d->vcpu == NULL) || unlikely(d->vcpu[0] == NULL) )
+ {
+ gdprintk(XENLOG_INFO,
+ "Memory event op on a domain (%u) with no vcpus\n",
+ d->domain_id);
+ return -EINVAL;
+ }
+
+ rc = -ENOSYS;
+
+ switch ( mec->mode )
+ {
+ case XEN_DOMCTL_MEM_EVENT_OP_PAGING:
+ {
+ struct mem_event_domain *med = &d->mem_event->paging;
+ rc = -EINVAL;
+
+ switch( mec->op )
+ {
+ case XEN_DOMCTL_MEM_EVENT_OP_PAGING_ENABLE:
+ {
+ struct p2m_domain *p2m = p2m_get_hostp2m(d);
+
+ rc = -EOPNOTSUPP;
+ /* pvh fixme: p2m_is_foreign types need addressing */
+ if ( is_pvh_vcpu(current) || is_pvh_domain(hardware_domain) )
+ break;
+
+ rc = -ENODEV;
+ /* Only HAP is supported */
+ if ( !hap_enabled(d) )
+ break;
+
+ /* No paging if iommu is used */
+ rc = -EMLINK;
+ if ( unlikely(need_iommu(d)) )
+ break;
+
+ rc = -EXDEV;
+ /* Disallow paging in a PoD guest */
+ if ( p2m->pod.entry_count )
+ break;
+
+ rc = mem_event_enable(d, mec, med, _VPF_mem_paging,
+ HVM_PARAM_PAGING_RING_PFN,
+ mem_paging_notification);
+ }
+ break;
+
+ case XEN_DOMCTL_MEM_EVENT_OP_PAGING_DISABLE:
+ {
+ if ( med->ring_page )
+ rc = mem_event_disable(d, med);
+ }
+ break;
+
+ default:
+ rc = -ENOSYS;
+ break;
+ }
+ }
+ break;
+
+ case XEN_DOMCTL_MEM_EVENT_OP_ACCESS:
+ {
+ struct mem_event_domain *med = &d->mem_event->access;
+ rc = -EINVAL;
+
+ switch( mec->op )
+ {
+ case XEN_DOMCTL_MEM_EVENT_OP_ACCESS_ENABLE:
+ {
+ rc = -ENODEV;
+ /* Only HAP is supported */
+ if ( !hap_enabled(d) )
+ break;
+
+ /* Currently only EPT is supported */
+ if ( !cpu_has_vmx )
+ break;
+
+ rc = mem_event_enable(d, mec, med, _VPF_mem_access,
+ HVM_PARAM_ACCESS_RING_PFN,
+ mem_access_notification);
+ }
+ break;
+
+ case XEN_DOMCTL_MEM_EVENT_OP_ACCESS_DISABLE:
+ {
+ if ( med->ring_page )
+ rc = mem_event_disable(d, med);
+ }
+ break;
+
+ default:
+ rc = -ENOSYS;
+ break;
+ }
+ }
+ break;
+
+ case XEN_DOMCTL_MEM_EVENT_OP_SHARING:
+ {
+ struct mem_event_domain *med = &d->mem_event->share;
+ rc = -EINVAL;
+
+ switch( mec->op )
+ {
+ case XEN_DOMCTL_MEM_EVENT_OP_SHARING_ENABLE:
+ {
+ rc = -EOPNOTSUPP;
+ /* pvh fixme: p2m_is_foreign types need addressing */
+ if ( is_pvh_vcpu(current) || is_pvh_domain(hardware_domain) )
+ break;
+
+ rc = -ENODEV;
+ /* Only HAP is supported */
+ if ( !hap_enabled(d) )
+ break;
+
+ rc = mem_event_enable(d, mec, med, _VPF_mem_sharing,
+ HVM_PARAM_SHARING_RING_PFN,
+ mem_sharing_notification);
+ }
+ break;
+
+ case XEN_DOMCTL_MEM_EVENT_OP_SHARING_DISABLE:
+ {
+ if ( med->ring_page )
+ rc = mem_event_disable(d, med);
+ }
+ break;
+
+ default:
+ rc = -ENOSYS;
+ break;
+ }
+ }
+ break;
+
+ default:
+ rc = -ENOSYS;
+ }
+
+ return rc;
+}
+
+void mem_event_vcpu_pause(struct vcpu *v)
+{
+ ASSERT(v == current);
+
+ atomic_inc(&v->mem_event_pause_count);
+ vcpu_pause_nosync(v);
+}
+
+void mem_event_vcpu_unpause(struct vcpu *v)
+{
+ int old, new, prev = v->mem_event_pause_count.counter;
+
+ /* All unpause requests as a result of toolstack responses. Prevent
+ * underflow of the vcpu pause count. */
+ do
+ {
+ old = prev;
+ new = old - 1;
+
+ if ( new < 0 )
+ {
+ printk(XENLOG_G_WARNING
+ "%pv mem_event: Too many unpause attempts\n", v);
+ return;
+ }
+
+ prev = cmpxchg(&v->mem_event_pause_count.counter, old, new);
+ } while ( prev != old );
+
+ vcpu_unpause(v);
+}
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/common/memory.c b/xen/common/memory.c
index 2e3225d..164f1b9 100644
--- a/xen/common/memory.c
+++ b/xen/common/memory.c
@@ -1104,6 +1104,69 @@ long do_memory_op(unsigned long cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
return rc;
}
+void destroy_ring_for_helper(
+ void **_va, struct page_info *page)
+{
+ void *va = *_va;
+
+ if ( va != NULL )
+ {
+ unmap_domain_page_global(va);
+ put_page_and_type(page);
+ *_va = NULL;
+ }
+}
+
+int prepare_ring_for_helper(
+ struct domain *d, unsigned long gmfn, struct page_info **_page,
+ void **_va)
+{
+ struct page_info *page;
+ p2m_type_t p2mt;
+ void *va;
+
+ page = get_page_from_gfn(d, gmfn, &p2mt, P2M_UNSHARE);
+
+#ifdef CONFIG_MEM_PAGING
+ if ( p2m_is_paging(p2mt) )
+ {
+ if ( page )
+ put_page(page);
+ p2m_mem_paging_populate(d, gmfn);
+ return -ENOENT;
+ }
+#endif
+#ifdef CONFIG_MEM_SHARING
+ if ( p2m_is_shared(p2mt) )
+ {
+ if ( page )
+ put_page(page);
+ return -ENOENT;
+ }
+#endif
+
+ if ( !page )
+ return -EINVAL;
+
+ if ( !get_page_type(page, PGT_writable_page) )
+ {
+ put_page(page);
+ return -EINVAL;
+ }
+
+ va = __map_domain_page_global(page);
+ if ( va == NULL )
+ {
+ put_page_and_type(page);
+ return -ENOMEM;
+ }
+
+ *_va = va;
+ *_page = page;
+
+ return 0;
+}
+
/*
* Local variables:
* mode: C
diff --git a/xen/include/asm-arm/mm.h b/xen/include/asm-arm/mm.h
index 9fa80a4..7fc3b97 100644
--- a/xen/include/asm-arm/mm.h
+++ b/xen/include/asm-arm/mm.h
@@ -301,7 +301,6 @@ struct page_info *get_page_from_gva(struct domain *d, vaddr_t va,
})
static inline void put_gfn(struct domain *d, unsigned long gfn) {}
-static inline void mem_event_cleanup(struct domain *d) {}
static inline int relinquish_shared_pages(struct domain *d)
{
return 0;
diff --git a/xen/include/asm-x86/config.h b/xen/include/asm-x86/config.h
index 210ff57..8a864ce 100644
--- a/xen/include/asm-x86/config.h
+++ b/xen/include/asm-x86/config.h
@@ -67,6 +67,9 @@
#define NR_CPUS 256
#endif
+#define CONFIG_MEM_SHARING 1
+#define CONFIG_MEM_PAGING 1
+
/* Maximum we can support with current vLAPIC ID mapping. */
#define MAX_HVM_VCPUS 128
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 1123857..74e66f8 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -226,12 +226,6 @@ int hvm_vcpu_cacheattr_init(struct vcpu *v);
void hvm_vcpu_cacheattr_destroy(struct vcpu *v);
void hvm_vcpu_reset_state(struct vcpu *v, uint16_t cs, uint16_t ip);
-/* Prepare/destroy a ring for a dom0 helper. Helper with talk
- * with Xen on behalf of this hvm domain. */
-int prepare_ring_for_helper(struct domain *d, unsigned long gmfn,
- struct page_info **_page, void **_va);
-void destroy_ring_for_helper(void **_va, struct page_info *page);
-
bool_t hvm_send_assist_req(ioreq_t *p);
void hvm_broadcast_assist_req(ioreq_t *p);
diff --git a/xen/include/asm-x86/mem_access.h b/xen/include/asm-x86/mem_access.h
deleted file mode 100644
index 5c7c5fd..0000000
--- a/xen/include/asm-x86/mem_access.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/******************************************************************************
- * include/asm-x86/mem_access.h
- *
- * Memory access support.
- *
- * Copyright (c) 2011 Virtuata, Inc.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- */
-
-#ifndef _XEN_ASM_MEM_ACCESS_H
-#define _XEN_ASM_MEM_ACCESS_H
-
-int mem_access_memop(unsigned long cmd,
- XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg);
-int mem_access_send_req(struct domain *d, mem_event_request_t *req);
-
-#endif /* _XEN_ASM_MEM_ACCESS_H */
-
-/*
- * Local variables:
- * mode: C
- * c-file-style: "BSD"
- * c-basic-offset: 4
- * indent-tabs-mode: nil
- * End:
- */
diff --git a/xen/include/asm-x86/mem_event.h b/xen/include/asm-x86/mem_event.h
deleted file mode 100644
index ed4481a..0000000
--- a/xen/include/asm-x86/mem_event.h
+++ /dev/null
@@ -1,82 +0,0 @@
-/******************************************************************************
- * include/asm-x86/mem_event.h
- *
- * Common interface for memory event support.
- *
- * Copyright (c) 2009 Citrix Systems, Inc. (Patrick Colp)
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- */
-
-
-#ifndef __MEM_EVENT_H__
-#define __MEM_EVENT_H__
-
-/* Returns whether a ring has been set up */
-bool_t mem_event_check_ring(struct mem_event_domain *med);
-
-/* Returns 0 on success, -ENOSYS if there is no ring, -EBUSY if there is no
- * available space and the caller is a foreign domain. If the guest itself
- * is the caller, -EBUSY is avoided by sleeping on a wait queue to ensure
- * that the ring does not lose future events.
- *
- * However, the allow_sleep flag can be set to false in cases in which it is ok
- * to lose future events, and thus -EBUSY can be returned to guest vcpus
- * (handle with care!).
- *
- * In general, you must follow a claim_slot() call with either put_request() or
- * cancel_slot(), both of which are guaranteed to
- * succeed.
- */
-int __mem_event_claim_slot(struct domain *d, struct mem_event_domain *med,
- bool_t allow_sleep);
-static inline int mem_event_claim_slot(struct domain *d,
- struct mem_event_domain *med)
-{
- return __mem_event_claim_slot(d, med, 1);
-}
-
-static inline int mem_event_claim_slot_nosleep(struct domain *d,
- struct mem_event_domain *med)
-{
- return __mem_event_claim_slot(d, med, 0);
-}
-
-void mem_event_cancel_slot(struct domain *d, struct mem_event_domain *med);
-
-void mem_event_put_request(struct domain *d, struct mem_event_domain *med,
- mem_event_request_t *req);
-
-int mem_event_get_response(struct domain *d, struct mem_event_domain *med,
- mem_event_response_t *rsp);
-
-int do_mem_event_op(int op, uint32_t domain, void *arg);
-int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
- XEN_GUEST_HANDLE_PARAM(void) u_domctl);
-
-void mem_event_vcpu_pause(struct vcpu *v);
-void mem_event_vcpu_unpause(struct vcpu *v);
-
-#endif /* __MEM_EVENT_H__ */
-
-
-/*
- * Local variables:
- * mode: C
- * c-file-style: "BSD"
- * c-basic-offset: 4
- * indent-tabs-mode: nil
- * End:
- */
diff --git a/xen/include/asm-x86/mm.h b/xen/include/asm-x86/mm.h
index 7b85865..ebd482d 100644
--- a/xen/include/asm-x86/mm.h
+++ b/xen/include/asm-x86/mm.h
@@ -611,8 +611,6 @@ unsigned int domain_clamp_alloc_bitsize(struct domain *d, unsigned int bits);
unsigned long domain_get_maximum_gpfn(struct domain *d);
-void mem_event_cleanup(struct domain *d);
-
extern struct domain *dom_xen, *dom_io, *dom_cow; /* for vmcoreinfo */
/* Definition of an mm lock: spinlock with extra fields for debugging */
diff --git a/xen/include/xen/mem_access.h b/xen/include/xen/mem_access.h
new file mode 100644
index 0000000..19d1a2d
--- /dev/null
+++ b/xen/include/xen/mem_access.h
@@ -0,0 +1,60 @@
+/******************************************************************************
+ * mem_access.h
+ *
+ * Memory access support.
+ *
+ * Copyright (c) 2011 Virtuata, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef _XEN_ASM_MEM_ACCESS_H
+#define _XEN_ASM_MEM_ACCESS_H
+
+#include <public/memory.h>
+
+#ifdef HAS_MEM_ACCESS
+
+int mem_access_memop(unsigned long cmd,
+ XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg);
+int mem_access_send_req(struct domain *d, mem_event_request_t *req);
+
+#else
+
+static inline
+int mem_access_memop(unsigned long cmd,
+ XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg)
+{
+ return -ENOSYS;
+}
+
+static inline
+int mem_access_send_req(struct domain *d, mem_event_request_t *req)
+{
+ return -ENOSYS;
+}
+
+#endif /* HAS_MEM_ACCESS */
+
+#endif /* _XEN_ASM_MEM_ACCESS_H */
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/include/xen/mem_event.h b/xen/include/xen/mem_event.h
new file mode 100644
index 0000000..8612b26
--- /dev/null
+++ b/xen/include/xen/mem_event.h
@@ -0,0 +1,143 @@
+/******************************************************************************
+ * mem_event.h
+ *
+ * Common interface for memory event support.
+ *
+ * Copyright (c) 2009 Citrix Systems, Inc. (Patrick Colp)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+
+#ifndef __MEM_EVENT_H__
+#define __MEM_EVENT_H__
+
+#include <xen/sched.h>
+
+#ifdef HAS_MEM_ACCESS
+
+/* Clean up on domain destruction */
+void mem_event_cleanup(struct domain *d);
+
+/* Returns whether a ring has been set up */
+bool_t mem_event_check_ring(struct mem_event_domain *med);
+
+/* Returns 0 on success, -ENOSYS if there is no ring, -EBUSY if there is no
+ * available space and the caller is a foreign domain. If the guest itself
+ * is the caller, -EBUSY is avoided by sleeping on a wait queue to ensure
+ * that the ring does not lose future events.
+ *
+ * However, the allow_sleep flag can be set to false in cases in which it is ok
+ * to lose future events, and thus -EBUSY can be returned to guest vcpus
+ * (handle with care!).
+ *
+ * In general, you must follow a claim_slot() call with either put_request() or
+ * cancel_slot(), both of which are guaranteed to
+ * succeed.
+ */
+int __mem_event_claim_slot(struct domain *d, struct mem_event_domain *med,
+ bool_t allow_sleep);
+static inline int mem_event_claim_slot(struct domain *d,
+ struct mem_event_domain *med)
+{
+ return __mem_event_claim_slot(d, med, 1);
+}
+
+static inline int mem_event_claim_slot_nosleep(struct domain *d,
+ struct mem_event_domain *med)
+{
+ return __mem_event_claim_slot(d, med, 0);
+}
+
+void mem_event_cancel_slot(struct domain *d, struct mem_event_domain *med);
+
+void mem_event_put_request(struct domain *d, struct mem_event_domain *med,
+ mem_event_request_t *req);
+
+int mem_event_get_response(struct domain *d, struct mem_event_domain *med,
+ mem_event_response_t *rsp);
+
+int do_mem_event_op(int op, uint32_t domain, void *arg);
+int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
+ XEN_GUEST_HANDLE_PARAM(void) u_domctl);
+
+void mem_event_vcpu_pause(struct vcpu *v);
+void mem_event_vcpu_unpause(struct vcpu *v);
+
+#else
+
+static inline void mem_event_cleanup(struct domain *d) {}
+
+static inline bool_t mem_event_check_ring(struct mem_event_domain *med)
+{
+ return 0;
+}
+
+static inline int mem_event_claim_slot(struct domain *d,
+ struct mem_event_domain *med)
+{
+ return -ENOSYS;
+}
+
+static inline int mem_event_claim_slot_nosleep(struct domain *d,
+ struct mem_event_domain *med)
+{
+ return -ENOSYS;
+}
+
+static inline
+void mem_event_cancel_slot(struct domain *d, struct mem_event_domain *med)
+{}
+
+static inline
+void mem_event_put_request(struct domain *d, struct mem_event_domain *med,
+ mem_event_request_t *req)
+{}
+
+static inline
+int mem_event_get_response(struct domain *d, struct mem_event_domain *med,
+ mem_event_response_t *rsp)
+{
+ return -ENOSYS;
+}
+
+static inline int do_mem_event_op(int op, uint32_t domain, void *arg)
+{
+ return -ENOSYS;
+}
+
+static inline
+int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
+ XEN_GUEST_HANDLE_PARAM(void) u_domctl)
+{
+ return -ENOSYS;
+}
+
+static inline void mem_event_vcpu_pause(struct vcpu *v) {}
+static inline void mem_event_vcpu_unpause(struct vcpu *v) {}
+
+#endif /* HAS_MEM_ACCESS */
+
+#endif /* __MEM_EVENT_H__ */
+
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/xen/include/xen/mm.h b/xen/include/xen/mm.h
index b183189..7c0efc7 100644
--- a/xen/include/xen/mm.h
+++ b/xen/include/xen/mm.h
@@ -371,4 +371,10 @@ int guest_remove_page(struct domain *d, unsigned long gmfn);
/* TRUE if the whole page at @mfn is of the requested RAM type(s) above. */
int page_is_ram_type(unsigned long mfn, unsigned long mem_type);
+/* Prepare/destroy a ring for a dom0 helper. Helper with talk
+ * with Xen on behalf of this domain. */
+int prepare_ring_for_helper(struct domain *d, unsigned long gmfn,
+ struct page_info **_page, void **_va);
+void destroy_ring_for_helper(void **_va, struct page_info *page);
+
#endif /* __XEN_MM_H__ */
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 02/21] xen: Relocate struct npfec definition into common
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 01/21] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 03/21] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
` (19 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Nested page fault exception code definitions can be reused on ARM as well.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Jan Beulich <jbeulich@suse.com>
---
xen/include/asm-x86/hvm/hvm.h | 2 +-
xen/include/asm-x86/mm.h | 21 ---------------------
xen/include/xen/mm.h | 21 +++++++++++++++++++++
3 files changed, 22 insertions(+), 22 deletions(-)
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 74e66f8..895f42b 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -27,7 +27,7 @@
#include <public/domctl.h>
#include <public/hvm/save.h>
#include <public/hvm/ioreq.h>
-#include <asm/mm.h>
+#include <xen/mm.h>
/* Interrupt acknowledgement sources. */
enum hvm_intsrc {
diff --git a/xen/include/asm-x86/mm.h b/xen/include/asm-x86/mm.h
index ebd482d..bafd28c 100644
--- a/xen/include/asm-x86/mm.h
+++ b/xen/include/asm-x86/mm.h
@@ -551,27 +551,6 @@ void audit_domains(void);
#endif
-/*
- * Extra fault info types which are used to further describe
- * the source of an access violation.
- */
-typedef enum {
- npfec_kind_unknown, /* must be first */
- npfec_kind_in_gpt, /* violation in guest page table */
- npfec_kind_with_gla /* violation with guest linear address */
-} npfec_kind_t;
-
-/*
- * Nested page fault exception codes.
- */
-struct npfec {
- unsigned int read_access:1;
- unsigned int write_access:1;
- unsigned int insn_fetch:1;
- unsigned int gla_valid:1;
- unsigned int kind:2; /* npfec_kind_t */
-};
-
int new_guest_cr3(unsigned long pfn);
void make_cr3(struct vcpu *v, unsigned long mfn);
void update_cr3(struct vcpu *v);
diff --git a/xen/include/xen/mm.h b/xen/include/xen/mm.h
index 7c0efc7..74a65a6 100644
--- a/xen/include/xen/mm.h
+++ b/xen/include/xen/mm.h
@@ -88,6 +88,27 @@ int assign_pages(
/* Dump info to serial console */
void arch_dump_shared_mem_info(void);
+/*
+ * Extra fault info types which are used to further describe
+ * the source of an access violation.
+ */
+typedef enum {
+ npfec_kind_unknown, /* must be first */
+ npfec_kind_in_gpt, /* violation in guest page table */
+ npfec_kind_with_gla /* violation with guest linear address */
+} npfec_kind_t;
+
+/*
+ * Nested page fault exception codes.
+ */
+struct npfec {
+ unsigned int read_access:1;
+ unsigned int write_access:1;
+ unsigned int insn_fetch:1;
+ unsigned int gla_valid:1;
+ unsigned int kind:2; /* npfec_kind_t */
+};
+
/* memflags: */
#define _MEMF_no_refcount 0
#define MEMF_no_refcount (1U<<_MEMF_no_refcount)
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 03/21] xen: Relocate mem_event_op domctl and access_op memop into common.
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 01/21] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 02/21] xen: Relocate struct npfec definition " Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl " Tamas K Lengyel
` (18 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Jan Beulich <jbeulich@suse.com>
---
v6: Grouping style fix of #includes in common/memory.c.
v5: Move memop compat into common as well.
Position domctl in switch relative to the domctl #.
v4: Don't remove memop handling from x86_64/compat and style fixes.
---
xen/arch/x86/domctl.c | 8 --------
xen/arch/x86/x86_64/compat/mm.c | 4 ----
xen/arch/x86/x86_64/mm.c | 4 ----
xen/common/compat/memory.c | 5 +++++
xen/common/domctl.c | 7 +++++++
xen/common/memory.c | 9 +++++++--
6 files changed, 19 insertions(+), 18 deletions(-)
diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index 26a3ea1..166cfb3 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -1131,14 +1131,6 @@ long arch_do_domctl(
}
break;
- case XEN_DOMCTL_mem_event_op:
- {
- ret = mem_event_domctl(d, &domctl->u.mem_event_op,
- guest_handle_cast(u_domctl, void));
- copyback = 1;
- }
- break;
-
case XEN_DOMCTL_mem_sharing_op:
{
ret = mem_sharing_domctl(d, &domctl->u.mem_sharing_op);
diff --git a/xen/arch/x86/x86_64/compat/mm.c b/xen/arch/x86/x86_64/compat/mm.c
index c079702..54f25b7 100644
--- a/xen/arch/x86/x86_64/compat/mm.c
+++ b/xen/arch/x86/x86_64/compat/mm.c
@@ -198,10 +198,6 @@ int compat_arch_memory_op(unsigned long cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
break;
}
- case XENMEM_access_op:
- rc = mem_access_memop(cmd, guest_handle_cast(arg, xen_mem_access_op_t));
- break;
-
case XENMEM_sharing_op:
{
xen_mem_sharing_op_t mso;
diff --git a/xen/arch/x86/x86_64/mm.c b/xen/arch/x86/x86_64/mm.c
index 0da6ddc..0aaa460 100644
--- a/xen/arch/x86/x86_64/mm.c
+++ b/xen/arch/x86/x86_64/mm.c
@@ -1048,10 +1048,6 @@ long subarch_memory_op(unsigned long cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
break;
}
- case XENMEM_access_op:
- rc = mem_access_memop(cmd, guest_handle_cast(arg, xen_mem_access_op_t));
- break;
-
case XENMEM_sharing_op:
{
xen_mem_sharing_op_t mso;
diff --git a/xen/common/compat/memory.c b/xen/common/compat/memory.c
index 25dc016..43d02bc 100644
--- a/xen/common/compat/memory.c
+++ b/xen/common/compat/memory.c
@@ -4,6 +4,7 @@
#include <xen/guest_access.h>
#include <xen/sched.h>
#include <xen/event.h>
+#include <xen/mem_access.h>
#include <asm/current.h>
#include <compat/memory.h>
@@ -381,6 +382,10 @@ int compat_memory_op(unsigned int cmd, XEN_GUEST_HANDLE_PARAM(void) compat)
break;
}
+ case XENMEM_access_op:
+ rc = mem_access_memop(cmd, guest_handle_cast(compat, xen_mem_access_op_t));
+ break;
+
case XENMEM_add_to_physmap_batch:
start_extent = end_extent;
break;
diff --git a/xen/common/domctl.c b/xen/common/domctl.c
index 1ad0729..c00a899 100644
--- a/xen/common/domctl.c
+++ b/xen/common/domctl.c
@@ -24,6 +24,7 @@
#include <xen/bitmap.h>
#include <xen/paging.h>
#include <xen/hypercall.h>
+#include <xen/mem_event.h>
#include <asm/current.h>
#include <asm/irq.h>
#include <asm/page.h>
@@ -1110,6 +1111,12 @@ long do_domctl(XEN_GUEST_HANDLE_PARAM(xen_domctl_t) u_domctl)
}
break;
+ case XEN_DOMCTL_mem_event_op:
+ ret = mem_event_domctl(d, &op->u.mem_event_op,
+ guest_handle_cast(u_domctl, void));
+ copyback = 1;
+ break;
+
case XEN_DOMCTL_disable_migrate:
{
d->disable_migrate = op->u.disable_migrate.disable;
diff --git a/xen/common/memory.c b/xen/common/memory.c
index 164f1b9..9c0d110 100644
--- a/xen/common/memory.c
+++ b/xen/common/memory.c
@@ -21,13 +21,14 @@
#include <xen/errno.h>
#include <xen/tmem.h>
#include <xen/tmem_xen.h>
+#include <xen/numa.h>
+#include <xen/mem_access.h>
+#include <xen/trace.h>
#include <asm/current.h>
#include <asm/hardirq.h>
#include <asm/p2m.h>
-#include <xen/numa.h>
#include <public/memory.h>
#include <xsm/xsm.h>
-#include <xen/trace.h>
struct memop_args {
/* INPUT */
@@ -939,6 +940,10 @@ long do_memory_op(unsigned long cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
break;
}
+ case XENMEM_access_op:
+ rc = mem_access_memop(cmd, guest_handle_cast(arg, xen_mem_access_op_t));
+ break;
+
case XENMEM_claim_pages:
if ( copy_from_guest(&reservation, arg, 1) )
return -EFAULT;
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (2 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 03/21] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-18 9:18 ` Jan Beulich
2014-09-17 20:51 ` [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order Tamas K Lengyel
` (17 subsequent siblings)
21 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/arch/x86/domctl.c | 14 --------------
xen/common/domctl.c | 15 +++++++++++++++
xen/include/asm-arm/p2m.h | 3 +++
3 files changed, 18 insertions(+), 14 deletions(-)
diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index 166cfb3..ec77555 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -1155,20 +1155,6 @@ long arch_do_domctl(
break;
#endif /* P2M_AUDIT */
- case XEN_DOMCTL_set_access_required:
- {
- struct p2m_domain* p2m;
-
- ret = -EPERM;
- if ( current->domain == d )
- break;
-
- ret = 0;
- p2m = p2m_get_hostp2m(d);
- p2m->access_required = domctl->u.access_required.access_required;
- }
- break;
-
case XEN_DOMCTL_set_broken_page_p2m:
{
p2m_type_t pt;
diff --git a/xen/common/domctl.c b/xen/common/domctl.c
index c00a899..fd8dd44 100644
--- a/xen/common/domctl.c
+++ b/xen/common/domctl.c
@@ -28,6 +28,7 @@
#include <asm/current.h>
#include <asm/irq.h>
#include <asm/page.h>
+#include <asm/p2m.h>
#include <public/domctl.h>
#include <xsm/xsm.h>
@@ -1123,6 +1124,20 @@ long do_domctl(XEN_GUEST_HANDLE_PARAM(xen_domctl_t) u_domctl)
}
break;
+ case XEN_DOMCTL_set_access_required:
+ {
+ struct p2m_domain* p2m;
+
+ ret = -EPERM;
+ if ( current->domain == d )
+ break;
+
+ ret = 0;
+ p2m = p2m_get_hostp2m(d);
+ p2m->access_required = op->u.access_required.access_required;
+ }
+ break;
+
case XEN_DOMCTL_set_virq_handler:
{
uint32_t virq = op->u.set_virq_handler.virq;
diff --git a/xen/include/asm-arm/p2m.h b/xen/include/asm-arm/p2m.h
index 97cbae4..bec0c9b 100644
--- a/xen/include/asm-arm/p2m.h
+++ b/xen/include/asm-arm/p2m.h
@@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain *d, unsigned long frame,
bool_t writeable);
int arch_grant_unmap_page_identity(struct domain *d, unsigned long frame);
+/* get host p2m table */
+#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
+
#endif /* _XEN_P2M_H */
/*
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (3 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl " Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-18 9:22 ` Jan Beulich
2014-09-17 20:51 ` [PATCH for-4.5 v7 06/21] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
` (16 subsequent siblings)
21 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
We swap the order of the enum of types n ... rwx, as to have rwx at 0. This allows
us to memset the permission store to 0, without having to specify rwx as the default
permission when memaccess is not in use. This also has performance benefits for
non-memaccess paths, as now comparison is to 0 when checking if memaccess is in use,
which is usually faster.
We fix one location in nested_hap where the order of the enum made a difference.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/arch/x86/mm/hap/nested_hap.c | 2 +-
xen/include/asm-x86/p2m.h | 28 ----------------------------
xen/include/xen/p2m-common.h | 29 +++++++++++++++++++++++++++++
3 files changed, 30 insertions(+), 29 deletions(-)
diff --git a/xen/arch/x86/mm/hap/nested_hap.c b/xen/arch/x86/mm/hap/nested_hap.c
index a4bb835..9c1ec11 100644
--- a/xen/arch/x86/mm/hap/nested_hap.c
+++ b/xen/arch/x86/mm/hap/nested_hap.c
@@ -264,7 +264,7 @@ nestedhvm_hap_nested_page_fault(struct vcpu *v, paddr_t *L2_gpa,
switch ( p2ma_10 )
{
- case p2m_access_n ... p2m_access_rwx:
+ case p2m_access_rwx ... p2m_access_n:
break;
case p2m_access_rx2rw:
p2ma_10 = p2m_access_rx;
diff --git a/xen/include/asm-x86/p2m.h b/xen/include/asm-x86/p2m.h
index bae669e..a2a6289 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -74,34 +74,6 @@ typedef enum {
p2m_map_foreign = 14, /* ram pages from foreign domain */
} p2m_type_t;
-/*
- * Additional access types, which are used to further restrict
- * the permissions given my the p2m_type_t memory type. Violations
- * caused by p2m_access_t restrictions are sent to the mem_event
- * interface.
- *
- * The access permissions are soft state: when any ambigious change of page
- * type or use occurs, or when pages are flushed, swapped, or at any other
- * convenient type, the access permissions can get reset to the p2m_domain
- * default.
- */
-typedef enum {
- p2m_access_n = 0, /* No access permissions allowed */
- p2m_access_r = 1,
- p2m_access_w = 2,
- p2m_access_rw = 3,
- p2m_access_x = 4,
- p2m_access_rx = 5,
- p2m_access_wx = 6,
- p2m_access_rwx = 7,
- p2m_access_rx2rw = 8, /* Special: page goes from RX to RW on write */
- p2m_access_n2rwx = 9, /* Special: page goes from N to RWX on access, *
- * generates an event but does not pause the
- * vcpu */
-
- /* NOTE: Assumed to be only 4 bits right now */
-} p2m_access_t;
-
/* Modifiers to the query */
typedef unsigned int p2m_query_t;
#define P2M_ALLOC (1u<<0) /* Populate PoD and paged-out entries */
diff --git a/xen/include/xen/p2m-common.h b/xen/include/xen/p2m-common.h
index 9f1b771..787fc43 100644
--- a/xen/include/xen/p2m-common.h
+++ b/xen/include/xen/p2m-common.h
@@ -1,6 +1,35 @@
#ifndef _XEN_P2M_COMMON_H
#define _XEN_P2M_COMMON_H
+/*
+ * Additional access types, which are used to further restrict
+ * the permissions given my the p2m_type_t memory type. Violations
+ * caused by p2m_access_t restrictions are sent to the mem_event
+ * interface.
+ *
+ * The access permissions are soft state: when any ambiguous change of page
+ * type or use occurs, or when pages are flushed, swapped, or at any other
+ * convenient type, the access permissions can get reset to the p2m_domain
+ * default.
+ */
+typedef enum {
+ p2m_access_rwx = 0, /* The default access type when not used. */
+ p2m_access_wx = 1,
+ p2m_access_rx = 2,
+ p2m_access_x = 3,
+ p2m_access_rw = 4,
+ p2m_access_w = 5,
+ p2m_access_r = 6,
+ p2m_access_n = 7, /* No access allowed. */
+
+ p2m_access_rx2rw = 8, /* Special: page goes from RX to RW on write */
+ p2m_access_n2rwx = 9, /* Special: page goes from N to RWX on access, *
+ * generates an event but does not pause the
+ * vcpu */
+
+ /* NOTE: Assumed to be only 4 bits right now on x86. */
+} p2m_access_t;
+
/* Map MMIO regions in the p2m: start_gfn and nr describe the range in
* * the guest physical address space to map, starting from the machine
* * frame number mfn. */
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 06/21] xen: Relocate p2m_mem_access_resume to mem_access common
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (4 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it Tamas K Lengyel
` (15 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
The p2m_mem_access resume function is not p2m specific and could be shared
across ARM/x86.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v6: Keep the comment describing the function.
v5: Style fix.
---
xen/arch/x86/mm/p2m.c | 24 ------------------------
xen/common/mem_access.c | 26 +++++++++++++++++++++++++-
xen/common/mem_event.c | 2 +-
xen/include/asm-x86/p2m.h | 2 --
xen/include/xen/mem_access.h | 5 +++++
5 files changed, 31 insertions(+), 28 deletions(-)
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index a9f120a..bf8e537 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -1427,30 +1427,6 @@ bool_t p2m_mem_access_check(paddr_t gpa, unsigned long gla,
return (p2ma == p2m_access_n2rwx);
}
-void p2m_mem_access_resume(struct domain *d)
-{
- mem_event_response_t rsp;
-
- /* Pull all responses off the ring */
- while( mem_event_get_response(d, &d->mem_event->access, &rsp) )
- {
- struct vcpu *v;
-
- if ( rsp.flags & MEM_EVENT_FLAG_DUMMY )
- continue;
-
- /* Validate the vcpu_id in the response. */
- if ( (rsp.vcpu_id >= d->max_vcpus) || !d->vcpu[rsp.vcpu_id] )
- continue;
-
- v = d->vcpu[rsp.vcpu_id];
-
- /* Unpause domain */
- if ( rsp.flags & MEM_EVENT_FLAG_VCPU_PAUSED )
- mem_event_vcpu_unpause(v);
- }
-}
-
/* Set access type for a region of pfns.
* If start_pfn == -1ul, sets the default access type */
long p2m_set_mem_access(struct domain *d, unsigned long pfn, uint32_t nr,
diff --git a/xen/common/mem_access.c b/xen/common/mem_access.c
index 9a8c1a9..1674d7a 100644
--- a/xen/common/mem_access.c
+++ b/xen/common/mem_access.c
@@ -29,6 +29,30 @@
#include <asm/p2m.h>
#include <xsm/xsm.h>
+void mem_access_resume(struct domain *d)
+{
+ mem_event_response_t rsp;
+
+ /* Pull all responses off the ring. */
+ while ( mem_event_get_response(d, &d->mem_event->access, &rsp) )
+ {
+ struct vcpu *v;
+
+ if ( rsp.flags & MEM_EVENT_FLAG_DUMMY )
+ continue;
+
+ /* Validate the vcpu_id in the response. */
+ if ( (rsp.vcpu_id >= d->max_vcpus) || !d->vcpu[rsp.vcpu_id] )
+ continue;
+
+ v = d->vcpu[rsp.vcpu_id];
+
+ /* Unpause domain. */
+ if ( rsp.flags & MEM_EVENT_FLAG_VCPU_PAUSED )
+ mem_event_vcpu_unpause(v);
+ }
+}
+
int mem_access_memop(unsigned long cmd,
XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg)
{
@@ -58,7 +82,7 @@ int mem_access_memop(unsigned long cmd,
switch ( mao.op )
{
case XENMEM_access_op_resume:
- p2m_mem_access_resume(d);
+ mem_access_resume(d);
rc = 0;
break;
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index 336b638..615a8b7 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -430,7 +430,7 @@ static void mem_paging_notification(struct vcpu *v, unsigned int port)
static void mem_access_notification(struct vcpu *v, unsigned int port)
{
if ( likely(v->domain->mem_event->access.ring_page != NULL) )
- p2m_mem_access_resume(v->domain);
+ mem_access_resume(v->domain);
}
/* Registered with Xen-bound event channel for incoming notifications. */
diff --git a/xen/include/asm-x86/p2m.h b/xen/include/asm-x86/p2m.h
index a2a6289..ff1ec97 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -573,8 +573,6 @@ void p2m_mem_paging_resume(struct domain *d);
bool_t p2m_mem_access_check(paddr_t gpa, unsigned long gla,
struct npfec npfec,
mem_event_request_t **req_ptr);
-/* Resumes the running of the VCPU, restarting the last instruction */
-void p2m_mem_access_resume(struct domain *d);
/* Set access type for a region of pfns.
* If start_pfn == -1ul, sets the default access type */
diff --git a/xen/include/xen/mem_access.h b/xen/include/xen/mem_access.h
index 19d1a2d..6ceb2a4 100644
--- a/xen/include/xen/mem_access.h
+++ b/xen/include/xen/mem_access.h
@@ -31,6 +31,9 @@ int mem_access_memop(unsigned long cmd,
XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg);
int mem_access_send_req(struct domain *d, mem_event_request_t *req);
+/* Resumes the running of the VCPU, restarting the last instruction */
+void mem_access_resume(struct domain *d);
+
#else
static inline
@@ -46,6 +49,8 @@ int mem_access_send_req(struct domain *d, mem_event_request_t *req)
return -ENOSYS;
}
+static inline void mem_access_resume(struct domain *d) {}
+
#endif /* HAS_MEM_ACCESS */
#endif /* _XEN_ASM_MEM_ACCESS_H */
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it.
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (5 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 06/21] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-18 9:28 ` Jan Beulich
2014-09-18 10:39 ` Tim Deegan
2014-09-17 20:51 ` [PATCH for-4.5 v7 08/21] x86/p2m: Typo fix for spelling ambiguous Tamas K Lengyel
` (14 subsequent siblings)
21 siblings, 2 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
The p2m_mem_access_check was used only to automatically promote the
special-case mem_access rights and to allocate the mem_access message.
In this patch we abstract the violation check into a common mem_access_check
function and change its return value to int.
Return with rc < 0 means the fault wasn't triggered by mem_access permissions.
Return with rc == 0 it was but rights were not automatically promoted.
Return with rc > 0 means mem_access rights had been automatically promoted.
As part of this patch, we relocate the x86/mm/mm-locks.h header into asm
as to be able to create a generic inlined p2m_gpfn_lock/unlock wrapper
that each arch can define based on its locking scheme.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/arch/x86/hvm/hvm.c | 59 ++------
xen/arch/x86/mm/mem_sharing.c | 2 +-
xen/arch/x86/mm/mm-locks.h | 299 -----------------------------------------
xen/arch/x86/mm/p2m-ept.c | 9 +-
xen/arch/x86/mm/p2m-pod.c | 2 +-
xen/arch/x86/mm/p2m-pt.c | 2 +-
xen/arch/x86/mm/p2m.c | 102 +-------------
xen/arch/x86/mm/paging.c | 4 +-
xen/common/mem_access.c | 143 +++++++++++++++++++-
xen/include/asm-x86/mm-locks.h | 299 +++++++++++++++++++++++++++++++++++++++++
xen/include/asm-x86/p2m.h | 12 ++
xen/include/asm-x86/page.h | 4 -
xen/include/xen/mem_access.h | 17 ++-
xen/include/xen/mm.h | 4 +
14 files changed, 492 insertions(+), 466 deletions(-)
delete mode 100644 xen/arch/x86/mm/mm-locks.h
create mode 100644 xen/include/asm-x86/mm-locks.h
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 7649d36..aab397a 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -2677,7 +2677,6 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla,
struct p2m_domain *p2m;
int rc, fall_through = 0, paged = 0;
int sharing_enomem = 0;
- mem_event_request_t *req_ptr = NULL;
/* On Nested Virtualization, walk the guest page table.
* If this succeeds, all is fine.
@@ -2745,50 +2744,21 @@ int hvm_hap_nested_page_fault(paddr_t gpa, unsigned long gla,
/* Check access permissions first, then handle faults */
if ( mfn_x(mfn) != INVALID_MFN )
{
- bool_t violation;
-
- /* If the access is against the permissions, then send to mem_event */
- switch (p2ma)
+ rc = mem_access_check(gpa, gla, npfec);
+ if ( rc < 0 )
{
- case p2m_access_n:
- case p2m_access_n2rwx:
- default:
- violation = npfec.read_access || npfec.write_access || npfec.insn_fetch;
- break;
- case p2m_access_r:
- violation = npfec.write_access || npfec.insn_fetch;
- break;
- case p2m_access_w:
- violation = npfec.read_access || npfec.insn_fetch;
- break;
- case p2m_access_x:
- violation = npfec.read_access || npfec.write_access;
- break;
- case p2m_access_rx:
- case p2m_access_rx2rw:
- violation = npfec.write_access;
- break;
- case p2m_access_wx:
- violation = npfec.read_access;
- break;
- case p2m_access_rw:
- violation = npfec.insn_fetch;
- break;
- case p2m_access_rwx:
- violation = 0;
- break;
+ /* Fault wasn't triggered by mem_access. */
}
-
- if ( violation )
+ else if ( rc > 0 )
{
- if ( p2m_mem_access_check(gpa, gla, npfec, &req_ptr) )
- {
- fall_through = 1;
- } else {
- /* Rights not promoted, vcpu paused, work here is done */
- rc = 1;
- goto out_put_gfn;
- }
+ /* Fault was triggered by mem_access and settings got promoted. */
+ fall_through = 1;
+ }
+ else
+ {
+ /* Rights not promoted, vcpu paused, work here is done */
+ rc = 1;
+ goto out_put_gfn;
}
}
@@ -2877,11 +2847,6 @@ out:
rc = 0;
}
}
- if ( req_ptr )
- {
- mem_access_send_req(v->domain, req_ptr);
- xfree(req_ptr);
- }
return rc;
}
diff --git a/xen/arch/x86/mm/mem_sharing.c b/xen/arch/x86/mm/mem_sharing.c
index 7c0fc7d..bcfcf6a 100644
--- a/xen/arch/x86/mm/mem_sharing.c
+++ b/xen/arch/x86/mm/mem_sharing.c
@@ -34,9 +34,9 @@
#include <asm/p2m.h>
#include <asm/atomic.h>
#include <asm/event.h>
+#include <asm/mm-locks.h>
#include <xsm/xsm.h>
-#include "mm-locks.h"
static shr_handle_t next_handle = 1;
diff --git a/xen/arch/x86/mm/mm-locks.h b/xen/arch/x86/mm/mm-locks.h
deleted file mode 100644
index 769f7bc..0000000
--- a/xen/arch/x86/mm/mm-locks.h
+++ /dev/null
@@ -1,299 +0,0 @@
-/******************************************************************************
- * arch/x86/mm/mm-locks.h
- *
- * Spinlocks used by the code in arch/x86/mm.
- *
- * Copyright (c) 2011 Citrix Systems, inc.
- * Copyright (c) 2007 Advanced Micro Devices (Wei Huang)
- * Copyright (c) 2006-2007 XenSource Inc.
- * Copyright (c) 2006 Michael A Fetterman
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
- */
-
-#ifndef _MM_LOCKS_H
-#define _MM_LOCKS_H
-
-#include <asm/mem_sharing.h>
-
-/* Per-CPU variable for enforcing the lock ordering */
-DECLARE_PER_CPU(int, mm_lock_level);
-#define __get_lock_level() (this_cpu(mm_lock_level))
-
-static inline void mm_lock_init(mm_lock_t *l)
-{
- spin_lock_init(&l->lock);
- l->locker = -1;
- l->locker_function = "nobody";
- l->unlock_level = 0;
-}
-
-static inline int mm_locked_by_me(mm_lock_t *l)
-{
- return (l->lock.recurse_cpu == current->processor);
-}
-
-/* If you see this crash, the numbers printed are lines in this file
- * where the offending locks are declared. */
-#define __check_lock_level(l) \
-do { \
- if ( unlikely(__get_lock_level() > (l)) ) \
- { \
- printk("mm locking order violation: %i > %i\n", \
- __get_lock_level(), (l)); \
- BUG(); \
- } \
-} while(0)
-
-#define __set_lock_level(l) \
-do { \
- __get_lock_level() = (l); \
-} while(0)
-
-static inline void _mm_lock(mm_lock_t *l, const char *func, int level, int rec)
-{
- if ( !((mm_locked_by_me(l)) && rec) )
- __check_lock_level(level);
- spin_lock_recursive(&l->lock);
- if ( l->lock.recurse_cnt == 1 )
- {
- l->locker_function = func;
- l->unlock_level = __get_lock_level();
- }
- else if ( (unlikely(!rec)) )
- panic("mm lock already held by %s", l->locker_function);
- __set_lock_level(level);
-}
-
-static inline void _mm_enforce_order_lock_pre(int level)
-{
- __check_lock_level(level);
-}
-
-static inline void _mm_enforce_order_lock_post(int level, int *unlock_level,
- unsigned short *recurse_count)
-{
- if ( recurse_count )
- {
- if ( (*recurse_count)++ == 0 )
- {
- *unlock_level = __get_lock_level();
- }
- } else {
- *unlock_level = __get_lock_level();
- }
- __set_lock_level(level);
-}
-
-
-static inline void mm_rwlock_init(mm_rwlock_t *l)
-{
- rwlock_init(&l->lock);
- l->locker = -1;
- l->locker_function = "nobody";
- l->unlock_level = 0;
-}
-
-static inline int mm_write_locked_by_me(mm_rwlock_t *l)
-{
- return (l->locker == get_processor_id());
-}
-
-static inline void _mm_write_lock(mm_rwlock_t *l, const char *func, int level)
-{
- if ( !mm_write_locked_by_me(l) )
- {
- __check_lock_level(level);
- write_lock(&l->lock);
- l->locker = get_processor_id();
- l->locker_function = func;
- l->unlock_level = __get_lock_level();
- __set_lock_level(level);
- }
- l->recurse_count++;
-}
-
-static inline void mm_write_unlock(mm_rwlock_t *l)
-{
- if ( --(l->recurse_count) != 0 )
- return;
- l->locker = -1;
- l->locker_function = "nobody";
- __set_lock_level(l->unlock_level);
- write_unlock(&l->lock);
-}
-
-static inline void _mm_read_lock(mm_rwlock_t *l, int level)
-{
- __check_lock_level(level);
- read_lock(&l->lock);
- /* There's nowhere to store the per-CPU unlock level so we can't
- * set the lock level. */
-}
-
-static inline void mm_read_unlock(mm_rwlock_t *l)
-{
- read_unlock(&l->lock);
-}
-
-/* This wrapper uses the line number to express the locking order below */
-#define declare_mm_lock(name) \
- static inline void mm_lock_##name(mm_lock_t *l, const char *func, int rec)\
- { _mm_lock(l, func, __LINE__, rec); }
-#define declare_mm_rwlock(name) \
- static inline void mm_write_lock_##name(mm_rwlock_t *l, const char *func) \
- { _mm_write_lock(l, func, __LINE__); } \
- static inline void mm_read_lock_##name(mm_rwlock_t *l) \
- { _mm_read_lock(l, __LINE__); }
-/* These capture the name of the calling function */
-#define mm_lock(name, l) mm_lock_##name(l, __func__, 0)
-#define mm_lock_recursive(name, l) mm_lock_##name(l, __func__, 1)
-#define mm_write_lock(name, l) mm_write_lock_##name(l, __func__)
-#define mm_read_lock(name, l) mm_read_lock_##name(l)
-
-/* This wrapper is intended for "external" locks which do not use
- * the mm_lock_t types. Such locks inside the mm code are also subject
- * to ordering constraints. */
-#define declare_mm_order_constraint(name) \
- static inline void mm_enforce_order_lock_pre_##name(void) \
- { _mm_enforce_order_lock_pre(__LINE__); } \
- static inline void mm_enforce_order_lock_post_##name( \
- int *unlock_level, unsigned short *recurse_count) \
- { _mm_enforce_order_lock_post(__LINE__, unlock_level, recurse_count); } \
-
-static inline void mm_unlock(mm_lock_t *l)
-{
- if ( l->lock.recurse_cnt == 1 )
- {
- l->locker_function = "nobody";
- __set_lock_level(l->unlock_level);
- }
- spin_unlock_recursive(&l->lock);
-}
-
-static inline void mm_enforce_order_unlock(int unlock_level,
- unsigned short *recurse_count)
-{
- if ( recurse_count )
- {
- BUG_ON(*recurse_count == 0);
- if ( (*recurse_count)-- == 1 )
- {
- __set_lock_level(unlock_level);
- }
- } else {
- __set_lock_level(unlock_level);
- }
-}
-
-/************************************************************************
- * *
- * To avoid deadlocks, these locks _MUST_ be taken in the order they're *
- * declared in this file. The locking functions will enforce this. *
- * *
- ************************************************************************/
-
-declare_mm_lock(nestedp2m)
-#define nestedp2m_lock(d) mm_lock(nestedp2m, &(d)->arch.nested_p2m_lock)
-#define nestedp2m_unlock(d) mm_unlock(&(d)->arch.nested_p2m_lock)
-
-/* P2M lock (per-p2m-table)
- *
- * This protects all queries and updates to the p2m table.
- * Queries may be made under the read lock but all modifications
- * need the main (write) lock.
- *
- * The write lock is recursive as it is common for a code path to look
- * up a gfn and later mutate it.
- */
-
-declare_mm_rwlock(p2m);
-#define p2m_lock(p) mm_write_lock(p2m, &(p)->lock);
-#define p2m_unlock(p) mm_write_unlock(&(p)->lock);
-#define gfn_lock(p,g,o) p2m_lock(p)
-#define gfn_unlock(p,g,o) p2m_unlock(p)
-#define p2m_read_lock(p) mm_read_lock(p2m, &(p)->lock)
-#define p2m_read_unlock(p) mm_read_unlock(&(p)->lock)
-#define p2m_locked_by_me(p) mm_write_locked_by_me(&(p)->lock)
-#define gfn_locked_by_me(p,g) p2m_locked_by_me(p)
-
-/* Sharing per page lock
- *
- * This is an external lock, not represented by an mm_lock_t. The memory
- * sharing lock uses it to protect addition and removal of (gfn,domain)
- * tuples to a shared page. We enforce order here against the p2m lock,
- * which is taken after the page_lock to change the gfn's p2m entry.
- *
- * The lock is recursive because during share we lock two pages. */
-
-declare_mm_order_constraint(per_page_sharing)
-#define page_sharing_mm_pre_lock() mm_enforce_order_lock_pre_per_page_sharing()
-#define page_sharing_mm_post_lock(l, r) \
- mm_enforce_order_lock_post_per_page_sharing((l), (r))
-#define page_sharing_mm_unlock(l, r) mm_enforce_order_unlock((l), (r))
-
-/* Nested P2M lock (per-domain)
- *
- * A per-domain lock that protects the mapping from nested-CR3 to
- * nested-p2m. In particular it covers:
- * - the array of nested-p2m tables, and all LRU activity therein; and
- * - setting the "cr3" field of any p2m table to a non-P2M_BASE_EAADR value.
- * (i.e. assigning a p2m table to be the shadow of that cr3 */
-
-/* PoD lock (per-p2m-table)
- *
- * Protects private PoD data structs: entry and cache
- * counts, page lists, sweep parameters. */
-
-declare_mm_lock(pod)
-#define pod_lock(p) mm_lock(pod, &(p)->pod.lock)
-#define pod_unlock(p) mm_unlock(&(p)->pod.lock)
-#define pod_locked_by_me(p) mm_locked_by_me(&(p)->pod.lock)
-
-/* Page alloc lock (per-domain)
- *
- * This is an external lock, not represented by an mm_lock_t. However,
- * pod code uses it in conjunction with the p2m lock, and expecting
- * the ordering which we enforce here.
- * The lock is not recursive. */
-
-declare_mm_order_constraint(page_alloc)
-#define page_alloc_mm_pre_lock() mm_enforce_order_lock_pre_page_alloc()
-#define page_alloc_mm_post_lock(l) mm_enforce_order_lock_post_page_alloc(&(l), NULL)
-#define page_alloc_mm_unlock(l) mm_enforce_order_unlock((l), NULL)
-
-/* Paging lock (per-domain)
- *
- * For shadow pagetables, this lock protects
- * - all changes to shadow page table pages
- * - the shadow hash table
- * - the shadow page allocator
- * - all changes to guest page table pages
- * - all changes to the page_info->tlbflush_timestamp
- * - the page_info->count fields on shadow pages
- *
- * For HAP, it protects the NPT/EPT tables and mode changes.
- *
- * It also protects the log-dirty bitmap from concurrent accesses (and
- * teardowns, etc). */
-
-declare_mm_lock(paging)
-#define paging_lock(d) mm_lock(paging, &(d)->arch.paging.lock)
-#define paging_lock_recursive(d) \
- mm_lock_recursive(paging, &(d)->arch.paging.lock)
-#define paging_unlock(d) mm_unlock(&(d)->arch.paging.lock)
-#define paging_locked_by_me(d) mm_locked_by_me(&(d)->arch.paging.lock)
-
-#endif /* _MM_LOCKS_H */
diff --git a/xen/arch/x86/mm/p2m-ept.c b/xen/arch/x86/mm/p2m-ept.c
index 15c6e83..1d11956 100644
--- a/xen/arch/x86/mm/p2m-ept.c
+++ b/xen/arch/x86/mm/p2m-ept.c
@@ -19,6 +19,9 @@
#include <xen/config.h>
#include <xen/domain_page.h>
#include <xen/sched.h>
+#include <xen/iommu.h>
+#include <xen/keyhandler.h>
+#include <xen/softirq.h>
#include <asm/current.h>
#include <asm/paging.h>
#include <asm/types.h>
@@ -26,13 +29,9 @@
#include <asm/p2m.h>
#include <asm/hvm/vmx/vmx.h>
#include <asm/hvm/vmx/vmcs.h>
-#include <xen/iommu.h>
#include <asm/mtrr.h>
#include <asm/hvm/cacheattr.h>
-#include <xen/keyhandler.h>
-#include <xen/softirq.h>
-
-#include "mm-locks.h"
+#include <asm/mm-locks.h>
#define atomic_read_ept_entry(__pepte) \
( (ept_entry_t) { .epte = read_atomic(&(__pepte)->epte) } )
diff --git a/xen/arch/x86/mm/p2m-pod.c b/xen/arch/x86/mm/p2m-pod.c
index 43f507c..1484aa3 100644
--- a/xen/arch/x86/mm/p2m-pod.c
+++ b/xen/arch/x86/mm/p2m-pod.c
@@ -32,8 +32,8 @@
#include <asm/mem_sharing.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/svm/amd-iommu-proto.h>
+#include <asm/mm-locks.h>
-#include "mm-locks.h"
/* Override macros from asm/page.h to make them work with mfn_t */
#undef mfn_to_page
diff --git a/xen/arch/x86/mm/p2m-pt.c b/xen/arch/x86/mm/p2m-pt.c
index e48b63a..a6bbb46 100644
--- a/xen/arch/x86/mm/p2m-pt.c
+++ b/xen/arch/x86/mm/p2m-pt.c
@@ -37,8 +37,8 @@
#include <asm/mem_sharing.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/svm/amd-iommu-proto.h>
+#include <asm/mm-locks.h>
-#include "mm-locks.h"
/* Override macros from asm/page.h to make them work with mfn_t */
#undef mfn_to_page
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index bf8e537..d192d6c 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -36,9 +36,9 @@
#include <asm/mem_sharing.h>
#include <asm/hvm/nestedhvm.h>
#include <asm/hvm/svm/amd-iommu-proto.h>
+#include <asm/mm-locks.h>
#include <xsm/xsm.h>
-#include "mm-locks.h"
/* turn on/off 1GB host page table support for hap, default on */
bool_t __read_mostly opt_hap_1gb = 1;
@@ -1327,106 +1327,6 @@ void p2m_mem_paging_resume(struct domain *d)
}
}
-bool_t p2m_mem_access_check(paddr_t gpa, unsigned long gla,
- struct npfec npfec,
- mem_event_request_t **req_ptr)
-{
- struct vcpu *v = current;
- unsigned long gfn = gpa >> PAGE_SHIFT;
- struct domain *d = v->domain;
- struct p2m_domain* p2m = p2m_get_hostp2m(d);
- mfn_t mfn;
- p2m_type_t p2mt;
- p2m_access_t p2ma;
- mem_event_request_t *req;
- int rc;
-
- /* First, handle rx2rw conversion automatically.
- * These calls to p2m->set_entry() must succeed: we have the gfn
- * locked and just did a successful get_entry(). */
- gfn_lock(p2m, gfn, 0);
- mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
-
- if ( npfec.write_access && p2ma == p2m_access_rx2rw )
- {
- rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K, p2mt, p2m_access_rw);
- ASSERT(rc == 0);
- gfn_unlock(p2m, gfn, 0);
- return 1;
- }
- else if ( p2ma == p2m_access_n2rwx )
- {
- ASSERT(npfec.write_access || npfec.read_access || npfec.insn_fetch);
- rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
- p2mt, p2m_access_rwx);
- ASSERT(rc == 0);
- }
- gfn_unlock(p2m, gfn, 0);
-
- /* Otherwise, check if there is a memory event listener, and send the message along */
- if ( !mem_event_check_ring(&d->mem_event->access) || !req_ptr )
- {
- /* No listener */
- if ( p2m->access_required )
- {
- gdprintk(XENLOG_INFO, "Memory access permissions failure, "
- "no mem_event listener VCPU %d, dom %d\n",
- v->vcpu_id, d->domain_id);
- domain_crash(v->domain);
- return 0;
- }
- else
- {
- gfn_lock(p2m, gfn, 0);
- mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
- if ( p2ma != p2m_access_n2rwx )
- {
- /* A listener is not required, so clear the access
- * restrictions. This set must succeed: we have the
- * gfn locked and just did a successful get_entry(). */
- rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
- p2mt, p2m_access_rwx);
- ASSERT(rc == 0);
- }
- gfn_unlock(p2m, gfn, 0);
- return 1;
- }
- }
-
- *req_ptr = NULL;
- req = xzalloc(mem_event_request_t);
- if ( req )
- {
- *req_ptr = req;
- req->reason = MEM_EVENT_REASON_VIOLATION;
-
- /* Pause the current VCPU */
- if ( p2ma != p2m_access_n2rwx )
- req->flags |= MEM_EVENT_FLAG_VCPU_PAUSED;
-
- /* Send request to mem event */
- req->gfn = gfn;
- req->offset = gpa & ((1 << PAGE_SHIFT) - 1);
- req->gla_valid = npfec.gla_valid;
- req->gla = gla;
- if ( npfec.kind == npfec_kind_with_gla )
- req->fault_with_gla = 1;
- else if ( npfec.kind == npfec_kind_in_gpt )
- req->fault_in_gpt = 1;
- req->access_r = npfec.read_access;
- req->access_w = npfec.write_access;
- req->access_x = npfec.insn_fetch;
- req->vcpu_id = v->vcpu_id;
- }
-
- /* Pause the current VCPU */
- if ( p2ma != p2m_access_n2rwx )
- mem_event_vcpu_pause(v);
-
- /* VCPU may be paused, return whether we promoted automatically */
- return (p2ma == p2m_access_n2rwx);
-}
-
/* Set access type for a region of pfns.
* If start_pfn == -1ul, sets the default access type */
long p2m_set_mem_access(struct domain *d, unsigned long pfn, uint32_t nr,
diff --git a/xen/arch/x86/mm/paging.c b/xen/arch/x86/mm/paging.c
index 455000d..eb7751e 100644
--- a/xen/arch/x86/mm/paging.c
+++ b/xen/arch/x86/mm/paging.c
@@ -22,15 +22,15 @@
#include <xen/init.h>
#include <xen/guest_access.h>
+#include <xen/numa.h>
#include <asm/paging.h>
#include <asm/shadow.h>
#include <asm/p2m.h>
#include <asm/hap.h>
#include <asm/hvm/nestedhvm.h>
-#include <xen/numa.h>
+#include <asm/mm-locks.h>
#include <xsm/xsm.h>
-#include "mm-locks.h"
/* Printouts */
#define PAGING_PRINTK(_f, _a...) \
diff --git a/xen/common/mem_access.c b/xen/common/mem_access.c
index 1674d7a..8fc2942 100644
--- a/xen/common/mem_access.c
+++ b/xen/common/mem_access.c
@@ -136,7 +136,7 @@ int mem_access_memop(unsigned long cmd,
return rc;
}
-int mem_access_send_req(struct domain *d, mem_event_request_t *req)
+static int mem_access_send_req(struct domain *d, mem_event_request_t *req)
{
int rc = mem_event_claim_slot(d, &d->mem_event->access);
if ( rc < 0 )
@@ -147,6 +147,147 @@ int mem_access_send_req(struct domain *d, mem_event_request_t *req)
return 0;
}
+int mem_access_check(paddr_t gpa, unsigned long gla, struct npfec npfec)
+{
+ bool_t violation;
+ mfn_t mfn;
+ p2m_access_t p2ma;
+ p2m_type_t p2mt;
+ mem_event_request_t *req;
+ int rc;
+ struct vcpu *v = current;
+ unsigned long gfn = gpa >> PAGE_SHIFT;
+ struct domain *d = v->domain;
+ struct p2m_domain* p2m = p2m_get_hostp2m(d);
+
+ /* First, handle rx2rw conversion automatically.
+ * These calls to p2m->set_entry() must succeed: we have the gfn
+ * locked and just did a successful get_entry(). */
+ p2m_gpfn_lock(p2m, gfn);
+ mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
+
+ /* Check if the access is against the permissions. */
+ switch ( p2ma )
+ {
+ case p2m_access_rwx:
+ violation = 0;
+ break;
+ case p2m_access_rw:
+ violation = npfec.insn_fetch;
+ break;
+ case p2m_access_wx:
+ violation = npfec.read_access;
+ break;
+ case p2m_access_rx:
+ case p2m_access_rx2rw:
+ violation = npfec.write_access;
+ break;
+ case p2m_access_x:
+ violation = npfec.read_access || npfec.write_access;
+ break;
+ case p2m_access_w:
+ violation = npfec.read_access || npfec.insn_fetch;
+ break;
+ case p2m_access_r:
+ violation = npfec.write_access || npfec.insn_fetch;
+ break;
+ case p2m_access_n:
+ case p2m_access_n2rwx:
+ default:
+ violation = npfec.read_access || npfec.write_access || npfec.insn_fetch;
+ break;
+ }
+
+ /* If no violation is found here, it needs to be reinjected. */
+ if ( !violation )
+ {
+ p2m_gpfn_unlock(p2m, gfn);
+ return -EFAULT;
+ }
+
+ /* Check for automatic setting promotion. */
+ if ( npfec.write_access && p2ma == p2m_access_rx2rw )
+ {
+ rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K, p2mt, p2m_access_rw);
+ ASSERT(rc == 0);
+ p2m_gpfn_unlock(p2m, gfn);
+ return 1;
+ }
+ else if ( p2ma == p2m_access_n2rwx )
+ {
+ ASSERT(npfec.write_access || npfec.read_access || npfec.insn_fetch);
+ rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
+ p2mt, p2m_access_rwx);
+ ASSERT(rc == 0);
+ }
+ p2m_gpfn_unlock(p2m, gfn);
+
+ /* Otherwise, check if there is a memory event listener, and send the message along */
+ if ( !mem_event_check_ring(&d->mem_event->access) )
+ {
+ /* No listener */
+ if ( p2m->access_required )
+ {
+ gdprintk(XENLOG_INFO, "Memory access permissions failure, "
+ "no mem_event listener VCPU %d, dom %d\n",
+ v->vcpu_id, d->domain_id);
+ domain_crash(v->domain);
+ return -EFAULT;
+ }
+ else
+ {
+ p2m_gpfn_lock(p2m, gfn);
+ mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
+ if ( p2ma != p2m_access_n2rwx )
+
+ {
+ /* A listener is not required, so clear the access
+ * restrictions. This set must succeed: we have the
+ * gfn locked and just did a successful get_entry(). */
+ rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
+ p2mt, p2m_access_rwx);
+ ASSERT(rc == 0);
+ }
+ p2m_gpfn_unlock(p2m, gfn);
+ return 1;
+ }
+ }
+
+ req = xzalloc(mem_event_request_t);
+ if ( req )
+ {
+ req->reason = MEM_EVENT_REASON_VIOLATION;
+
+ /* Pause the current VCPU */
+ if ( p2ma != p2m_access_n2rwx )
+ req->flags |= MEM_EVENT_FLAG_VCPU_PAUSED;
+
+ /* Send request to mem event */
+ req->gfn = gfn;
+ req->offset = gpa & ((1 << PAGE_SHIFT) - 1);
+ req->gla_valid = npfec.gla_valid;
+ req->gla = gla;
+ if ( npfec.kind == npfec_kind_with_gla )
+ req->fault_with_gla = 1;
+ else if ( npfec.kind == npfec_kind_in_gpt )
+ req->fault_in_gpt = 1;
+ req->access_r = npfec.read_access;
+ req->access_w = npfec.write_access;
+ req->access_x = npfec.insn_fetch;
+ req->vcpu_id = v->vcpu_id;
+
+ mem_access_send_req(v->domain, req);
+ xfree(req);
+ }
+
+ /* Pause the current VCPU */
+ if ( p2ma != p2m_access_n2rwx )
+ mem_event_vcpu_pause(v);
+
+ /* VCPU may be paused, return whether we promoted automatically */
+ return (p2ma == p2m_access_n2rwx);
+}
+
/*
* Local variables:
* mode: C
diff --git a/xen/include/asm-x86/mm-locks.h b/xen/include/asm-x86/mm-locks.h
new file mode 100644
index 0000000..7056a30
--- /dev/null
+++ b/xen/include/asm-x86/mm-locks.h
@@ -0,0 +1,299 @@
+/******************************************************************************
+ * mm-locks.h
+ *
+ * Spinlocks used by the code in arch/x86/mm.
+ *
+ * Copyright (c) 2011 Citrix Systems, inc.
+ * Copyright (c) 2007 Advanced Micro Devices (Wei Huang)
+ * Copyright (c) 2006-2007 XenSource Inc.
+ * Copyright (c) 2006 Michael A Fetterman
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef _MM_LOCKS_H
+#define _MM_LOCKS_H
+
+#include <asm/mem_sharing.h>
+
+/* Per-CPU variable for enforcing the lock ordering */
+DECLARE_PER_CPU(int, mm_lock_level);
+#define __get_lock_level() (this_cpu(mm_lock_level))
+
+static inline void mm_lock_init(mm_lock_t *l)
+{
+ spin_lock_init(&l->lock);
+ l->locker = -1;
+ l->locker_function = "nobody";
+ l->unlock_level = 0;
+}
+
+static inline int mm_locked_by_me(mm_lock_t *l)
+{
+ return (l->lock.recurse_cpu == current->processor);
+}
+
+/* If you see this crash, the numbers printed are lines in this file
+ * where the offending locks are declared. */
+#define __check_lock_level(l) \
+do { \
+ if ( unlikely(__get_lock_level() > (l)) ) \
+ { \
+ printk("mm locking order violation: %i > %i\n", \
+ __get_lock_level(), (l)); \
+ BUG(); \
+ } \
+} while(0)
+
+#define __set_lock_level(l) \
+do { \
+ __get_lock_level() = (l); \
+} while(0)
+
+static inline void _mm_lock(mm_lock_t *l, const char *func, int level, int rec)
+{
+ if ( !((mm_locked_by_me(l)) && rec) )
+ __check_lock_level(level);
+ spin_lock_recursive(&l->lock);
+ if ( l->lock.recurse_cnt == 1 )
+ {
+ l->locker_function = func;
+ l->unlock_level = __get_lock_level();
+ }
+ else if ( (unlikely(!rec)) )
+ panic("mm lock already held by %s", l->locker_function);
+ __set_lock_level(level);
+}
+
+static inline void _mm_enforce_order_lock_pre(int level)
+{
+ __check_lock_level(level);
+}
+
+static inline void _mm_enforce_order_lock_post(int level, int *unlock_level,
+ unsigned short *recurse_count)
+{
+ if ( recurse_count )
+ {
+ if ( (*recurse_count)++ == 0 )
+ {
+ *unlock_level = __get_lock_level();
+ }
+ } else {
+ *unlock_level = __get_lock_level();
+ }
+ __set_lock_level(level);
+}
+
+
+static inline void mm_rwlock_init(mm_rwlock_t *l)
+{
+ rwlock_init(&l->lock);
+ l->locker = -1;
+ l->locker_function = "nobody";
+ l->unlock_level = 0;
+}
+
+static inline int mm_write_locked_by_me(mm_rwlock_t *l)
+{
+ return (l->locker == get_processor_id());
+}
+
+static inline void _mm_write_lock(mm_rwlock_t *l, const char *func, int level)
+{
+ if ( !mm_write_locked_by_me(l) )
+ {
+ __check_lock_level(level);
+ write_lock(&l->lock);
+ l->locker = get_processor_id();
+ l->locker_function = func;
+ l->unlock_level = __get_lock_level();
+ __set_lock_level(level);
+ }
+ l->recurse_count++;
+}
+
+static inline void mm_write_unlock(mm_rwlock_t *l)
+{
+ if ( --(l->recurse_count) != 0 )
+ return;
+ l->locker = -1;
+ l->locker_function = "nobody";
+ __set_lock_level(l->unlock_level);
+ write_unlock(&l->lock);
+}
+
+static inline void _mm_read_lock(mm_rwlock_t *l, int level)
+{
+ __check_lock_level(level);
+ read_lock(&l->lock);
+ /* There's nowhere to store the per-CPU unlock level so we can't
+ * set the lock level. */
+}
+
+static inline void mm_read_unlock(mm_rwlock_t *l)
+{
+ read_unlock(&l->lock);
+}
+
+/* This wrapper uses the line number to express the locking order below */
+#define declare_mm_lock(name) \
+ static inline void mm_lock_##name(mm_lock_t *l, const char *func, int rec)\
+ { _mm_lock(l, func, __LINE__, rec); }
+#define declare_mm_rwlock(name) \
+ static inline void mm_write_lock_##name(mm_rwlock_t *l, const char *func) \
+ { _mm_write_lock(l, func, __LINE__); } \
+ static inline void mm_read_lock_##name(mm_rwlock_t *l) \
+ { _mm_read_lock(l, __LINE__); }
+/* These capture the name of the calling function */
+#define mm_lock(name, l) mm_lock_##name(l, __func__, 0)
+#define mm_lock_recursive(name, l) mm_lock_##name(l, __func__, 1)
+#define mm_write_lock(name, l) mm_write_lock_##name(l, __func__)
+#define mm_read_lock(name, l) mm_read_lock_##name(l)
+
+/* This wrapper is intended for "external" locks which do not use
+ * the mm_lock_t types. Such locks inside the mm code are also subject
+ * to ordering constraints. */
+#define declare_mm_order_constraint(name) \
+ static inline void mm_enforce_order_lock_pre_##name(void) \
+ { _mm_enforce_order_lock_pre(__LINE__); } \
+ static inline void mm_enforce_order_lock_post_##name( \
+ int *unlock_level, unsigned short *recurse_count) \
+ { _mm_enforce_order_lock_post(__LINE__, unlock_level, recurse_count); } \
+
+static inline void mm_unlock(mm_lock_t *l)
+{
+ if ( l->lock.recurse_cnt == 1 )
+ {
+ l->locker_function = "nobody";
+ __set_lock_level(l->unlock_level);
+ }
+ spin_unlock_recursive(&l->lock);
+}
+
+static inline void mm_enforce_order_unlock(int unlock_level,
+ unsigned short *recurse_count)
+{
+ if ( recurse_count )
+ {
+ BUG_ON(*recurse_count == 0);
+ if ( (*recurse_count)-- == 1 )
+ {
+ __set_lock_level(unlock_level);
+ }
+ } else {
+ __set_lock_level(unlock_level);
+ }
+}
+
+/************************************************************************
+ * *
+ * To avoid deadlocks, these locks _MUST_ be taken in the order they're *
+ * declared in this file. The locking functions will enforce this. *
+ * *
+ ************************************************************************/
+
+declare_mm_lock(nestedp2m)
+#define nestedp2m_lock(d) mm_lock(nestedp2m, &(d)->arch.nested_p2m_lock)
+#define nestedp2m_unlock(d) mm_unlock(&(d)->arch.nested_p2m_lock)
+
+/* P2M lock (per-p2m-table)
+ *
+ * This protects all queries and updates to the p2m table.
+ * Queries may be made under the read lock but all modifications
+ * need the main (write) lock.
+ *
+ * The write lock is recursive as it is common for a code path to look
+ * up a gfn and later mutate it.
+ */
+
+declare_mm_rwlock(p2m);
+#define p2m_lock(p) mm_write_lock(p2m, &(p)->lock);
+#define p2m_unlock(p) mm_write_unlock(&(p)->lock);
+#define gfn_lock(p,g,o) p2m_lock(p)
+#define gfn_unlock(p,g,o) p2m_unlock(p)
+#define p2m_read_lock(p) mm_read_lock(p2m, &(p)->lock)
+#define p2m_read_unlock(p) mm_read_unlock(&(p)->lock)
+#define p2m_locked_by_me(p) mm_write_locked_by_me(&(p)->lock)
+#define gfn_locked_by_me(p,g) p2m_locked_by_me(p)
+
+/* Sharing per page lock
+ *
+ * This is an external lock, not represented by an mm_lock_t. The memory
+ * sharing lock uses it to protect addition and removal of (gfn,domain)
+ * tuples to a shared page. We enforce order here against the p2m lock,
+ * which is taken after the page_lock to change the gfn's p2m entry.
+ *
+ * The lock is recursive because during share we lock two pages. */
+
+declare_mm_order_constraint(per_page_sharing)
+#define page_sharing_mm_pre_lock() mm_enforce_order_lock_pre_per_page_sharing()
+#define page_sharing_mm_post_lock(l, r) \
+ mm_enforce_order_lock_post_per_page_sharing((l), (r))
+#define page_sharing_mm_unlock(l, r) mm_enforce_order_unlock((l), (r))
+
+/* Nested P2M lock (per-domain)
+ *
+ * A per-domain lock that protects the mapping from nested-CR3 to
+ * nested-p2m. In particular it covers:
+ * - the array of nested-p2m tables, and all LRU activity therein; and
+ * - setting the "cr3" field of any p2m table to a non-P2M_BASE_EAADR value.
+ * (i.e. assigning a p2m table to be the shadow of that cr3 */
+
+/* PoD lock (per-p2m-table)
+ *
+ * Protects private PoD data structs: entry and cache
+ * counts, page lists, sweep parameters. */
+
+declare_mm_lock(pod)
+#define pod_lock(p) mm_lock(pod, &(p)->pod.lock)
+#define pod_unlock(p) mm_unlock(&(p)->pod.lock)
+#define pod_locked_by_me(p) mm_locked_by_me(&(p)->pod.lock)
+
+/* Page alloc lock (per-domain)
+ *
+ * This is an external lock, not represented by an mm_lock_t. However,
+ * pod code uses it in conjunction with the p2m lock, and expecting
+ * the ordering which we enforce here.
+ * The lock is not recursive. */
+
+declare_mm_order_constraint(page_alloc)
+#define page_alloc_mm_pre_lock() mm_enforce_order_lock_pre_page_alloc()
+#define page_alloc_mm_post_lock(l) mm_enforce_order_lock_post_page_alloc(&(l), NULL)
+#define page_alloc_mm_unlock(l) mm_enforce_order_unlock((l), NULL)
+
+/* Paging lock (per-domain)
+ *
+ * For shadow pagetables, this lock protects
+ * - all changes to shadow page table pages
+ * - the shadow hash table
+ * - the shadow page allocator
+ * - all changes to guest page table pages
+ * - all changes to the page_info->tlbflush_timestamp
+ * - the page_info->count fields on shadow pages
+ *
+ * For HAP, it protects the NPT/EPT tables and mode changes.
+ *
+ * It also protects the log-dirty bitmap from concurrent accesses (and
+ * teardowns, etc). */
+
+declare_mm_lock(paging)
+#define paging_lock(d) mm_lock(paging, &(d)->arch.paging.lock)
+#define paging_lock_recursive(d) \
+ mm_lock_recursive(paging, &(d)->arch.paging.lock)
+#define paging_unlock(d) mm_unlock(&(d)->arch.paging.lock)
+#define paging_locked_by_me(d) mm_locked_by_me(&(d)->arch.paging.lock)
+
+#endif /* _MM_LOCKS_H */
diff --git a/xen/include/asm-x86/p2m.h b/xen/include/asm-x86/p2m.h
index ff1ec97..f4c185d 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -32,6 +32,7 @@
#include <xen/p2m-common.h>
#include <asm/mem_sharing.h>
#include <asm/page.h> /* for pagetable_t */
+#include <asm/mm-locks.h> /* for gfn_lock */
extern bool_t opt_hap_1gb, opt_hap_2mb;
@@ -584,6 +585,17 @@ long p2m_set_mem_access(struct domain *d, unsigned long start_pfn, uint32_t nr,
int p2m_get_mem_access(struct domain *d, unsigned long pfn,
xenmem_access_t *access);
+/* Used by mem_access_check in mem_access common. */
+static inline void p2m_gpfn_lock(struct p2m_domain *p2m, unsigned long gpfn)
+{
+ gfn_lock(p2m, gpfn, 0);
+}
+
+static inline void p2m_gpfn_unlock(struct p2m_domain *p2m, unsigned long gpfn)
+{
+ gfn_unlock(p2m, gpfn, 0);
+}
+
/*
* Internal functions, only called by other p2m code
*/
diff --git a/xen/include/asm-x86/page.h b/xen/include/asm-x86/page.h
index ccc268d..66762c1 100644
--- a/xen/include/asm-x86/page.h
+++ b/xen/include/asm-x86/page.h
@@ -11,10 +11,6 @@
#define PAGE_MASK (~(PAGE_SIZE-1))
#define PAGE_FLAG_MASK (~0)
-#define PAGE_ORDER_4K 0
-#define PAGE_ORDER_2M 9
-#define PAGE_ORDER_1G 18
-
#ifndef __ASSEMBLY__
# include <asm/types.h>
# include <xen/lib.h>
diff --git a/xen/include/xen/mem_access.h b/xen/include/xen/mem_access.h
index 6ceb2a4..7d921a6 100644
--- a/xen/include/xen/mem_access.h
+++ b/xen/include/xen/mem_access.h
@@ -29,11 +29,20 @@
int mem_access_memop(unsigned long cmd,
XEN_GUEST_HANDLE_PARAM(xen_mem_access_op_t) arg);
-int mem_access_send_req(struct domain *d, mem_event_request_t *req);
/* Resumes the running of the VCPU, restarting the last instruction */
void mem_access_resume(struct domain *d);
+/*
+ * Return with rc < 0 means the fault wasn't triggered by mem_access.
+ * Return with rc == 0 means it was but rights were not automatically promoted.
+ * Return with rc > 0 means it was and rights were automatically promoted.
+ *
+ * If the fault was triggered by mem_access, this function automatically
+ * forwards the event to the listener.
+ */
+int mem_access_check(paddr_t gpa, unsigned long gla, struct npfec npfec);
+
#else
static inline
@@ -43,14 +52,14 @@ int mem_access_memop(unsigned long cmd,
return -ENOSYS;
}
+static inline void mem_access_resume(struct domain *d) {}
+
static inline
-int mem_access_send_req(struct domain *d, mem_event_request_t *req)
+int mem_access_check(paddr_t gpa, unsigned long gla, struct npfec npfec)
{
return -ENOSYS;
}
-static inline void mem_access_resume(struct domain *d) {}
-
#endif /* HAS_MEM_ACCESS */
#endif /* _XEN_ASM_MEM_ACCESS_H */
diff --git a/xen/include/xen/mm.h b/xen/include/xen/mm.h
index 74a65a6..bffd8e6 100644
--- a/xen/include/xen/mm.h
+++ b/xen/include/xen/mm.h
@@ -131,6 +131,10 @@ struct npfec {
#define MAX_ORDER 20 /* 2^20 contiguous pages */
#endif
+#define PAGE_ORDER_4K 0
+#define PAGE_ORDER_2M 9
+#define PAGE_ORDER_1G 18
+
#define page_list_entry list_head
#include <asm/mm.h>
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 08/21] x86/p2m: Typo fix for spelling ambiguous
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (6 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access Tamas K Lengyel
` (13 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/include/asm-x86/p2m.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xen/include/asm-x86/p2m.h b/xen/include/asm-x86/p2m.h
index f4c185d..e29743e 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -235,7 +235,7 @@ struct p2m_domain {
long (*audit_p2m)(struct p2m_domain *p2m);
/* Default P2M access type for each page in the the domain: new pages,
- * swapped in pages, cleared pages, and pages that are ambiquously
+ * swapped in pages, cleared pages, and pages that are ambiguously
* retyped get this access type. See definition of p2m_access_t. */
p2m_access_t default_access;
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (7 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 08/21] x86/p2m: Typo fix for spelling ambiguous Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-18 9:35 ` Jan Beulich
2014-09-17 20:51 ` [PATCH for-4.5 v7 10/21] xen/mem_event: Clean out superfluous white-spaces Tamas K Lengyel
` (12 subsequent siblings)
21 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/arch/x86/mm/p2m.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index d192d6c..d54c561 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -1402,7 +1402,7 @@ int p2m_get_mem_access(struct domain *d, unsigned long pfn,
mfn_t mfn;
static const xenmem_access_t memaccess[] = {
-#define ACCESS(ac) [XENMEM_access_##ac] = XENMEM_access_##ac
+#define ACCESS(ac) [p2m_access_##ac] = XENMEM_access_##ac
ACCESS(n),
ACCESS(r),
ACCESS(w),
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 10/21] xen/mem_event: Clean out superfluous white-spaces
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (8 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 11/21] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
` (11 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Tim Deegan <tim@xen.org>
---
v2: Clean the mem_event header as well.
---
xen/common/mem_event.c | 20 ++++++++++----------
xen/include/xen/mem_event.h | 8 ++++----
2 files changed, 14 insertions(+), 14 deletions(-)
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index 615a8b7..af0ffb2 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -57,7 +57,7 @@ static int mem_event_enable(
if ( med->ring_page )
return -EBUSY;
- /* The parameter defaults to zero, and it should be
+ /* The parameter defaults to zero, and it should be
* set to something */
if ( ring_gfn == 0 )
return -ENOSYS;
@@ -65,7 +65,7 @@ static int mem_event_enable(
mem_event_ring_lock_init(med);
mem_event_ring_lock(med);
- rc = prepare_ring_for_helper(d, ring_gfn, &med->ring_pg_struct,
+ rc = prepare_ring_for_helper(d, ring_gfn, &med->ring_pg_struct,
&med->ring_page);
if ( rc < 0 )
goto err;
@@ -97,7 +97,7 @@ static int mem_event_enable(
return 0;
err:
- destroy_ring_for_helper(&med->ring_page,
+ destroy_ring_for_helper(&med->ring_page,
med->ring_pg_struct);
mem_event_ring_unlock(med);
@@ -226,7 +226,7 @@ static int mem_event_disable(struct domain *d, struct mem_event_domain *med)
}
}
- destroy_ring_for_helper(&med->ring_page,
+ destroy_ring_for_helper(&med->ring_page,
med->ring_pg_struct);
mem_event_ring_unlock(med);
}
@@ -479,7 +479,7 @@ void mem_event_cleanup(struct domain *d)
* the disable routine to complete. It will also drop
* all domain refs the wait-queued vcpus are holding.
* Finally, because this code path involves previously
- * pausing the domain (domain_kill), unpausing the
+ * pausing the domain (domain_kill), unpausing the
* vcpus causes no harm. */
destroy_waitqueue_head(&d->mem_event->paging.wq);
(void)mem_event_disable(d, &d->mem_event->paging);
@@ -559,7 +559,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
if ( p2m->pod.entry_count )
break;
- rc = mem_event_enable(d, mec, med, _VPF_mem_paging,
+ rc = mem_event_enable(d, mec, med, _VPF_mem_paging,
HVM_PARAM_PAGING_RING_PFN,
mem_paging_notification);
}
@@ -579,7 +579,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
}
break;
- case XEN_DOMCTL_MEM_EVENT_OP_ACCESS:
+ case XEN_DOMCTL_MEM_EVENT_OP_ACCESS:
{
struct mem_event_domain *med = &d->mem_event->access;
rc = -EINVAL;
@@ -597,7 +597,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
if ( !cpu_has_vmx )
break;
- rc = mem_event_enable(d, mec, med, _VPF_mem_access,
+ rc = mem_event_enable(d, mec, med, _VPF_mem_access,
HVM_PARAM_ACCESS_RING_PFN,
mem_access_notification);
}
@@ -617,7 +617,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
}
break;
- case XEN_DOMCTL_MEM_EVENT_OP_SHARING:
+ case XEN_DOMCTL_MEM_EVENT_OP_SHARING:
{
struct mem_event_domain *med = &d->mem_event->share;
rc = -EINVAL;
@@ -636,7 +636,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
if ( !hap_enabled(d) )
break;
- rc = mem_event_enable(d, mec, med, _VPF_mem_sharing,
+ rc = mem_event_enable(d, mec, med, _VPF_mem_sharing,
HVM_PARAM_SHARING_RING_PFN,
mem_sharing_notification);
}
diff --git a/xen/include/xen/mem_event.h b/xen/include/xen/mem_event.h
index 8612b26..4f3ad8e 100644
--- a/xen/include/xen/mem_event.h
+++ b/xen/include/xen/mem_event.h
@@ -37,19 +37,19 @@ bool_t mem_event_check_ring(struct mem_event_domain *med);
/* Returns 0 on success, -ENOSYS if there is no ring, -EBUSY if there is no
* available space and the caller is a foreign domain. If the guest itself
* is the caller, -EBUSY is avoided by sleeping on a wait queue to ensure
- * that the ring does not lose future events.
+ * that the ring does not lose future events.
*
* However, the allow_sleep flag can be set to false in cases in which it is ok
* to lose future events, and thus -EBUSY can be returned to guest vcpus
- * (handle with care!).
+ * (handle with care!).
*
* In general, you must follow a claim_slot() call with either put_request() or
* cancel_slot(), both of which are guaranteed to
- * succeed.
+ * succeed.
*/
int __mem_event_claim_slot(struct domain *d, struct mem_event_domain *med,
bool_t allow_sleep);
-static inline int mem_event_claim_slot(struct domain *d,
+static inline int mem_event_claim_slot(struct domain *d,
struct mem_event_domain *med)
{
return __mem_event_claim_slot(d, med, 1);
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 11/21] xen/mem_event: Relax error condition on debug builds
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (9 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 10/21] xen/mem_event: Clean out superfluous white-spaces Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 12/21] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
` (10 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
A faulty tool stack can brick a debug hypervisor. Unpleasant while dev/test.
Suggested-by: Andres Lagar Cavilla <andres@lagarcavilla.org>
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v5: Make printout greppable.
v4: Add domain_id to the printout.
v3: Switch to gdprintk and print the vCPU id as well.
---
xen/common/mem_event.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index af0ffb2..9eca687 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -278,7 +278,11 @@ void mem_event_put_request(struct domain *d,
if ( current->domain != d )
{
req->flags |= MEM_EVENT_FLAG_FOREIGN;
- ASSERT( !(req->flags & MEM_EVENT_FLAG_VCPU_PAUSED) );
+#ifndef NDEBUG
+ if ( !(req->flags & MEM_EVENT_FLAG_VCPU_PAUSED) )
+ gdprintk(XENLOG_G_WARNING, "d%dv%d was not paused.\n",
+ d->domain_id, req->vcpu_id);
+#endif
}
mem_event_ring_lock(med);
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 12/21] xen/mem_event: Abstract architecture specific sanity checks
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (10 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 11/21] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 13/21] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
` (9 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Move architecture specific sanity checks into its own function
which is called when enabling mem_event.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v5: Style fix.
v4: Style fix.
v2: Move sanity check function into architecture specific p2m.h
---
xen/common/mem_event.c | 7 +------
xen/include/asm-x86/p2m.h | 6 ++++++
2 files changed, 7 insertions(+), 6 deletions(-)
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index 9eca687..4d0a27b 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -593,12 +593,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
case XEN_DOMCTL_MEM_EVENT_OP_ACCESS_ENABLE:
{
rc = -ENODEV;
- /* Only HAP is supported */
- if ( !hap_enabled(d) )
- break;
-
- /* Currently only EPT is supported */
- if ( !cpu_has_vmx )
+ if ( !p2m_mem_event_sanity_check(d) )
break;
rc = mem_event_enable(d, mec, med, _VPF_mem_access,
diff --git a/xen/include/asm-x86/p2m.h b/xen/include/asm-x86/p2m.h
index e29743e..11a99c3 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -585,6 +585,12 @@ long p2m_set_mem_access(struct domain *d, unsigned long start_pfn, uint32_t nr,
int p2m_get_mem_access(struct domain *d, unsigned long pfn,
xenmem_access_t *access);
+/* Sanity check for mem_event hardware support */
+static inline bool_t p2m_mem_event_sanity_check(struct domain *d)
+{
+ return hap_enabled(d) && cpu_has_vmx;
+}
+
/* Used by mem_access_check in mem_access common. */
static inline void p2m_gpfn_lock(struct p2m_domain *p2m, unsigned long gpfn)
{
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 13/21] xen/mem_access: Abstract architecture specific sanity check
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (11 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 12/21] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support Tamas K Lengyel
` (8 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v5: Style fix.
v4: Style fix.
v2: Move sanity check function into architecture specific p2m.h
---
xen/common/mem_access.c | 2 +-
xen/include/asm-x86/p2m.h | 6 ++++++
2 files changed, 7 insertions(+), 1 deletion(-)
diff --git a/xen/common/mem_access.c b/xen/common/mem_access.c
index 8fc2942..2bc4631 100644
--- a/xen/common/mem_access.c
+++ b/xen/common/mem_access.c
@@ -68,7 +68,7 @@ int mem_access_memop(unsigned long cmd,
return rc;
rc = -EINVAL;
- if ( !is_hvm_domain(d) )
+ if ( !p2m_mem_access_sanity_check(d) )
goto out;
rc = xsm_mem_event_op(XSM_DM_PRIV, d, XENMEM_access_op);
diff --git a/xen/include/asm-x86/p2m.h b/xen/include/asm-x86/p2m.h
index 11a99c3..3ed4a31 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -591,6 +591,12 @@ static inline bool_t p2m_mem_event_sanity_check(struct domain *d)
return hap_enabled(d) && cpu_has_vmx;
}
+/* Sanity check for mem_access hardware support */
+static inline bool_t p2m_mem_access_sanity_check(struct domain *d)
+{
+ return is_hvm_domain(d);
+}
+
/* Used by mem_access_check in mem_access common. */
static inline void p2m_gpfn_lock(struct p2m_domain *p2m, unsigned long gpfn)
{
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (12 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 13/21] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-22 12:20 ` Ian Campbell
2014-09-17 20:51 ` [PATCH for-4.5 v7 15/21] xen/arm: Implement domain_get_maximum_gpfn Tamas K Lengyel
` (7 subsequent siblings)
21 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Add p2m_access_t to struct page_info and add necessary changes for
page table construction routines to pass the default access information.
We store the p2m_access_t info in page_info as the PTE lacks enough
software programmable bits.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v7: - Remove radix tree init/destroy and move p2m_access_t store to page_info.
- Add p2m_gpfn_lock/unlock functions.
- Add bool_t lock input to p2m_lookup and apply_p2m_changes so the caller
can specify if locking should be performed. This is needed in order to
support mem_access_check from common.
v6: - Move mem_event header include to first patch that needs it.
v5: - #include grouping style-fix.
v4: - Move p2m_get_hostp2m definition here.
---
xen/arch/arm/p2m.c | 66 +++++++++++++++++++------------
xen/arch/arm/traps.c | 6 +--
xen/include/asm-arm/mm.h | 11 +++++-
xen/include/asm-arm/p2m.h | 98 ++++++++++++++++++++++++++++++++++-------------
4 files changed, 125 insertions(+), 56 deletions(-)
diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index a7dcdf5..92c3bd6 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -155,7 +155,7 @@ static lpae_t *p2m_map_first(struct p2m_domain *p2m, paddr_t addr)
* There are no processor functions to do a stage 2 only lookup therefore we
* do a a software walk.
*/
-paddr_t p2m_lookup(struct domain *d, paddr_t paddr, p2m_type_t *t)
+paddr_t p2m_lookup(struct domain *d, paddr_t paddr, p2m_type_t *t, bool_t lock)
{
struct p2m_domain *p2m = &d->arch.p2m;
lpae_t pte, *first = NULL, *second = NULL, *third = NULL;
@@ -168,7 +168,8 @@ paddr_t p2m_lookup(struct domain *d, paddr_t paddr, p2m_type_t *t)
*t = p2m_invalid;
- spin_lock(&p2m->lock);
+ if ( lock )
+ spin_lock(&p2m->lock);
first = p2m_map_first(p2m, paddr);
if ( !first )
@@ -209,7 +210,8 @@ done:
if (first) unmap_domain_page(first);
err:
- spin_unlock(&p2m->lock);
+ if ( lock )
+ spin_unlock(&p2m->lock);
return maddr;
}
@@ -229,7 +231,7 @@ int p2m_pod_decrease_reservation(struct domain *d,
}
static lpae_t mfn_to_p2m_entry(unsigned long mfn, unsigned int mattr,
- p2m_type_t t)
+ p2m_type_t t, p2m_access_t a)
{
paddr_t pa = ((paddr_t) mfn) << PAGE_SHIFT;
/* sh, xn and write bit will be defined in the following switches
@@ -347,7 +349,7 @@ static int p2m_create_table(struct domain *d, lpae_t *entry,
for ( i=0 ; i < LPAE_ENTRIES; i++ )
{
pte = mfn_to_p2m_entry(base_pfn + (i<<(level_shift-LPAE_SHIFT)),
- MATTR_MEM, t);
+ MATTR_MEM, t, p2m->default_access);
/*
* First and second level super pages set p2m.table = 0, but
@@ -367,7 +369,8 @@ static int p2m_create_table(struct domain *d, lpae_t *entry,
unmap_domain_page(p);
- pte = mfn_to_p2m_entry(page_to_mfn(page), MATTR_MEM, p2m_invalid);
+ pte = mfn_to_p2m_entry(page_to_mfn(page), MATTR_MEM, p2m_invalid,
+ p2m->default_access);
p2m_write_pte(entry, pte, flush_cache);
@@ -470,7 +473,8 @@ static int apply_one_level(struct domain *d,
paddr_t *maddr,
bool_t *flush,
int mattr,
- p2m_type_t t)
+ p2m_type_t t,
+ p2m_access_t a)
{
const paddr_t level_size = level_sizes[level];
const paddr_t level_mask = level_masks[level];
@@ -499,7 +503,7 @@ static int apply_one_level(struct domain *d,
page = alloc_domheap_pages(d, level_shift - PAGE_SHIFT, 0);
if ( page )
{
- pte = mfn_to_p2m_entry(page_to_mfn(page), mattr, t);
+ pte = mfn_to_p2m_entry(page_to_mfn(page), mattr, t, a);
if ( level < 3 )
pte.p2m.table = 0;
p2m_write_pte(entry, pte, flush_cache);
@@ -534,7 +538,7 @@ static int apply_one_level(struct domain *d,
(level == 3 || !p2m_table(orig_pte)) )
{
/* New mapping is superpage aligned, make it */
- pte = mfn_to_p2m_entry(*maddr >> PAGE_SHIFT, mattr, t);
+ pte = mfn_to_p2m_entry(*maddr >> PAGE_SHIFT, mattr, t, a);
if ( level < 3 )
pte.p2m.table = 0; /* Superpage entry */
@@ -713,7 +717,10 @@ static int apply_p2m_changes(struct domain *d,
paddr_t end_gpaddr,
paddr_t maddr,
int mattr,
- p2m_type_t t)
+ bool_t lock,
+ uint32_t mask,
+ p2m_type_t t,
+ p2m_access_t a)
{
int rc, ret;
struct p2m_domain *p2m = &d->arch.p2m;
@@ -733,7 +740,8 @@ static int apply_p2m_changes(struct domain *d,
*/
flush_pt = iommu_enabled && !iommu_has_feature(d, IOMMU_FEAT_COHERENT_WALK);
- spin_lock(&p2m->lock);
+ if ( lock )
+ spin_lock(&p2m->lock);
addr = start_gpaddr;
while ( addr < end_gpaddr )
@@ -780,7 +788,7 @@ static int apply_p2m_changes(struct domain *d,
level, flush_pt, op,
start_gpaddr, end_gpaddr,
&addr, &maddr, &flush,
- mattr, t);
+ mattr, t, a);
if ( ret < 0 ) { rc = ret ; goto out; }
count += ret;
if ( ret != P2M_ONE_DESCEND ) continue;
@@ -802,7 +810,7 @@ static int apply_p2m_changes(struct domain *d,
level, flush_pt, op,
start_gpaddr, end_gpaddr,
&addr, &maddr, &flush,
- mattr, t);
+ mattr, t, a);
if ( ret < 0 ) { rc = ret ; goto out; }
count += ret;
if ( ret != P2M_ONE_DESCEND ) continue;
@@ -822,7 +830,7 @@ static int apply_p2m_changes(struct domain *d,
level, flush_pt, op,
start_gpaddr, end_gpaddr,
&addr, &maddr, &flush,
- mattr, t);
+ mattr, t, a);
if ( ret < 0 ) { rc = ret ; goto out; }
/* L3 had better have done something! We cannot descend any further */
BUG_ON(ret == P2M_ONE_DESCEND);
@@ -865,10 +873,11 @@ out:
*/
apply_p2m_changes(d, REMOVE,
start_gpaddr, addr + level_sizes[level], orig_maddr,
- mattr, p2m_invalid);
+ mattr, 1, 0, p2m_invalid, d->arch.p2m.default_access);
}
- spin_unlock(&p2m->lock);
+ if ( lock )
+ spin_unlock(&p2m->lock);
return rc;
}
@@ -878,7 +887,8 @@ int p2m_populate_ram(struct domain *d,
paddr_t end)
{
return apply_p2m_changes(d, ALLOCATE, start, end,
- 0, MATTR_MEM, p2m_ram_rw);
+ 0, MATTR_MEM, 1, 0, p2m_ram_rw,
+ d->arch.p2m.default_access);
}
int map_mmio_regions(struct domain *d,
@@ -890,7 +900,8 @@ int map_mmio_regions(struct domain *d,
pfn_to_paddr(start_gfn),
pfn_to_paddr(start_gfn + nr),
pfn_to_paddr(mfn),
- MATTR_DEV, p2m_mmio_direct);
+ MATTR_DEV, 1, 0, p2m_mmio_direct,
+ d->arch.p2m.default_access);
}
int unmap_mmio_regions(struct domain *d,
@@ -902,7 +913,8 @@ int unmap_mmio_regions(struct domain *d,
pfn_to_paddr(start_gfn),
pfn_to_paddr(start_gfn + nr),
pfn_to_paddr(mfn),
- MATTR_DEV, p2m_invalid);
+ MATTR_DEV, 1, 0, p2m_invalid,
+ d->arch.p2m.default_access);
}
int guest_physmap_add_entry(struct domain *d,
@@ -914,7 +926,8 @@ int guest_physmap_add_entry(struct domain *d,
return apply_p2m_changes(d, INSERT,
pfn_to_paddr(gpfn),
pfn_to_paddr(gpfn + (1 << page_order)),
- pfn_to_paddr(mfn), MATTR_MEM, t);
+ pfn_to_paddr(mfn), MATTR_MEM, 1, 0, t,
+ d->arch.p2m.default_access);
}
void guest_physmap_remove_page(struct domain *d,
@@ -924,7 +937,8 @@ void guest_physmap_remove_page(struct domain *d,
apply_p2m_changes(d, REMOVE,
pfn_to_paddr(gpfn),
pfn_to_paddr(gpfn + (1<<page_order)),
- pfn_to_paddr(mfn), MATTR_MEM, p2m_invalid);
+ pfn_to_paddr(mfn), MATTR_MEM, 1, 0, p2m_invalid,
+ d->arch.p2m.default_access);
}
int arch_grant_map_page_identity(struct domain *d, unsigned long frame,
@@ -1078,6 +1092,8 @@ int p2m_init(struct domain *d)
p2m->max_mapped_gfn = 0;
p2m->lowest_mapped_gfn = ULONG_MAX;
+ p2m->default_access = p2m_access_rwx;
+
err:
spin_unlock(&p2m->lock);
@@ -1092,7 +1108,8 @@ int relinquish_p2m_mapping(struct domain *d)
pfn_to_paddr(p2m->lowest_mapped_gfn),
pfn_to_paddr(p2m->max_mapped_gfn),
pfn_to_paddr(INVALID_MFN),
- MATTR_MEM, p2m_invalid);
+ MATTR_MEM, 1, 0, p2m_invalid,
+ d->arch.p2m.default_access);
}
int p2m_cache_flush(struct domain *d, xen_pfn_t start_mfn, xen_pfn_t end_mfn)
@@ -1106,12 +1123,13 @@ int p2m_cache_flush(struct domain *d, xen_pfn_t start_mfn, xen_pfn_t end_mfn)
pfn_to_paddr(start_mfn),
pfn_to_paddr(end_mfn),
pfn_to_paddr(INVALID_MFN),
- MATTR_MEM, p2m_invalid);
+ MATTR_MEM, 1, 0, p2m_invalid,
+ d->arch.p2m.default_access);
}
unsigned long gmfn_to_mfn(struct domain *d, unsigned long gpfn)
{
- paddr_t p = p2m_lookup(d, pfn_to_paddr(gpfn), NULL);
+ paddr_t p = p2m_lookup(d, pfn_to_paddr(gpfn), NULL, 1);
return p >> PAGE_SHIFT;
}
diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
index 25fa8a0..ae7458c 100644
--- a/xen/arch/arm/traps.c
+++ b/xen/arch/arm/traps.c
@@ -1795,7 +1795,7 @@ void dump_guest_s1_walk(struct domain *d, vaddr_t addr)
printk("dom%d VA 0x%08"PRIvaddr"\n", d->domain_id, addr);
printk(" TTBCR: 0x%08"PRIregister"\n", ttbcr);
printk(" TTBR0: 0x%016"PRIx64" = 0x%"PRIpaddr"\n",
- ttbr0, p2m_lookup(d, ttbr0 & PAGE_MASK, NULL));
+ ttbr0, p2m_lookup(d, ttbr0 & PAGE_MASK, NULL, 1));
if ( ttbcr & TTBCR_EAE )
{
@@ -1808,7 +1808,7 @@ void dump_guest_s1_walk(struct domain *d, vaddr_t addr)
return;
}
- paddr = p2m_lookup(d, ttbr0 & PAGE_MASK, NULL);
+ paddr = p2m_lookup(d, ttbr0 & PAGE_MASK, NULL, 1);
if ( paddr == INVALID_PADDR )
{
printk("Failed TTBR0 maddr lookup\n");
@@ -1823,7 +1823,7 @@ void dump_guest_s1_walk(struct domain *d, vaddr_t addr)
!(first[offset] & 0x2) )
goto done;
- paddr = p2m_lookup(d, first[offset] & PAGE_MASK, NULL);
+ paddr = p2m_lookup(d, first[offset] & PAGE_MASK, NULL, 1);
if ( paddr == INVALID_PADDR )
{
diff --git a/xen/include/asm-arm/mm.h b/xen/include/asm-arm/mm.h
index 7fc3b97..a47afcb 100644
--- a/xen/include/asm-arm/mm.h
+++ b/xen/include/asm-arm/mm.h
@@ -3,9 +3,10 @@
#include <xen/config.h>
#include <xen/kernel.h>
-#include <asm/page.h>
-#include <public/xen.h>
#include <xen/domain_page.h>
+#include <xen/p2m-common.h>
+#include <public/xen.h>
+#include <asm/page.h>
/* Align Xen to a 2 MiB boundary. */
#define XEN_PADDR_ALIGN (1 << 21)
@@ -64,6 +65,10 @@ struct page_info
*/
u32 tlbflush_timestamp;
};
+
+ /* Mem access permission for this page. */
+ p2m_access_t a;
+
u64 pad;
};
@@ -348,6 +353,8 @@ static inline void put_page_and_type(struct page_info *page)
void clear_and_clean_page(struct page_info *page);
+typedef unsigned long mfn_t;
+
#endif /* __ARCH_ARM_MM__ */
/*
* Local variables:
diff --git a/xen/include/asm-arm/p2m.h b/xen/include/asm-arm/p2m.h
index bec0c9b..0049538 100644
--- a/xen/include/asm-arm/p2m.h
+++ b/xen/include/asm-arm/p2m.h
@@ -11,6 +11,31 @@ struct domain;
extern void memory_type_changed(struct domain *);
+/* List of possible type for each page in the p2m entry.
+ * The number of available bit per page in the pte for this purpose is 4 bits.
+ * So it's possible to only have 16 fields. If we run out of value in the
+ * future, it's possible to use higher value for pseudo-type and don't store
+ * them in the p2m entry.
+ */
+typedef enum {
+ p2m_invalid = 0, /* Nothing mapped here */
+ p2m_ram_rw, /* Normal read/write guest RAM */
+ p2m_ram_ro, /* Read-only; writes are silently dropped */
+ p2m_mmio_direct, /* Read/write mapping of genuine MMIO area */
+ p2m_map_foreign, /* Ram pages from foreign domain */
+ p2m_grant_map_rw, /* Read/write grant mapping */
+ p2m_grant_map_ro, /* Read-only grant mapping */
+ /* The types below are only used to decide the page attribute in the P2M */
+ p2m_iommu_map_rw, /* Read/write iommu mapping */
+ p2m_iommu_map_ro, /* Read-only iommu mapping */
+ p2m_max_real_type, /* Types after this won't be store in the p2m */
+} p2m_type_t;
+
+/* Look up a GFN and take a reference count on the backing page. */
+typedef unsigned int p2m_query_t;
+#define P2M_ALLOC (1u<<0) /* Populate PoD and paged-out entries */
+#define P2M_UNSHARE (1u<<1) /* Break CoW sharing */
+
/* Per-p2m-table state */
struct p2m_domain {
/* Lock that protects updates to the p2m */
@@ -44,27 +69,30 @@ struct p2m_domain {
* at each p2m tree level. */
unsigned long shattered[4];
} stats;
-};
-/* List of possible type for each page in the p2m entry.
- * The number of available bit per page in the pte for this purpose is 4 bits.
- * So it's possible to only have 16 fields. If we run out of value in the
- * future, it's possible to use higher value for pseudo-type and don't store
- * them in the p2m entry.
- */
-typedef enum {
- p2m_invalid = 0, /* Nothing mapped here */
- p2m_ram_rw, /* Normal read/write guest RAM */
- p2m_ram_ro, /* Read-only; writes are silently dropped */
- p2m_mmio_direct, /* Read/write mapping of genuine MMIO area */
- p2m_map_foreign, /* Ram pages from foreign domain */
- p2m_grant_map_rw, /* Read/write grant mapping */
- p2m_grant_map_ro, /* Read-only grant mapping */
- /* The types below are only used to decide the page attribute in the P2M */
- p2m_iommu_map_rw, /* Read/write iommu mapping */
- p2m_iommu_map_ro, /* Read-only iommu mapping */
- p2m_max_real_type, /* Types after this won't be store in the p2m */
-} p2m_type_t;
+ /* Default P2M access type for each page in the the domain: new pages,
+ * swapped in pages, cleared pages, and pages that are ambiguously
+ * retyped get this access type. See definition of p2m_access_t. */
+ p2m_access_t default_access;
+
+ /* If true, and an access fault comes in and there is no mem_event listener,
+ * pause domain. Otherwise, remove access restrictions. */
+ bool_t access_required;
+
+ /* Compat functions for mem_access_check. */
+ int (*set_entry )(struct p2m_domain *p2m,
+ unsigned long gfn,
+ mfn_t mfn, unsigned int page_order,
+ p2m_type_t p2mt,
+ p2m_access_t p2ma);
+ mfn_t (*get_entry )(struct p2m_domain *p2m,
+ unsigned long gfn,
+ p2m_type_t *p2mt,
+ p2m_access_t *p2ma,
+ p2m_query_t q,
+ unsigned int *page_order);
+
+};
#define p2m_is_foreign(_t) ((_t) == p2m_map_foreign)
#define p2m_is_ram(_t) ((_t) == p2m_ram_rw || (_t) == p2m_ram_ro)
@@ -98,7 +126,7 @@ void p2m_restore_state(struct vcpu *n);
void p2m_dump_info(struct domain *d);
/* Look up the MFN corresponding to a domain's PFN. */
-paddr_t p2m_lookup(struct domain *d, paddr_t gpfn, p2m_type_t *t);
+paddr_t p2m_lookup(struct domain *d, paddr_t gpfn, p2m_type_t *t, bool_t lock);
/* Clean & invalidate caches corresponding to a region of guest address space */
int p2m_cache_flush(struct domain *d, xen_pfn_t start_mfn, xen_pfn_t end_mfn);
@@ -138,17 +166,12 @@ p2m_pod_decrease_reservation(struct domain *d,
xen_pfn_t gpfn,
unsigned int order);
-/* Look up a GFN and take a reference count on the backing page. */
-typedef unsigned int p2m_query_t;
-#define P2M_ALLOC (1u<<0) /* Populate PoD and paged-out entries */
-#define P2M_UNSHARE (1u<<1) /* Break CoW sharing */
-
static inline struct page_info *get_page_from_gfn(
struct domain *d, unsigned long gfn, p2m_type_t *t, p2m_query_t q)
{
struct page_info *page;
p2m_type_t p2mt;
- paddr_t maddr = p2m_lookup(d, pfn_to_paddr(gfn), &p2mt);
+ paddr_t maddr = p2m_lookup(d, pfn_to_paddr(gfn), &p2mt, 1);
unsigned long mfn = maddr >> PAGE_SHIFT;
if (t)
@@ -201,6 +224,27 @@ int arch_grant_unmap_page_identity(struct domain *d, unsigned long frame);
/* get host p2m table */
#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
+/* mem_event and mem_access are supported on any ARM guest */
+static inline bool_t p2m_mem_access_sanity_check(struct domain *d)
+{
+ return 1;
+}
+
+static inline bool_t p2m_mem_event_sanity_check(struct domain *d)
+{
+ return 1;
+}
+
+static inline void p2m_gpfn_lock(struct p2m_domain *p2m, unsigned long gpfn)
+{
+ spin_lock(&p2m->lock);
+}
+
+static inline void p2m_gpfn_unlock(struct p2m_domain *p2m, unsigned long gpfn)
+{
+ spin_unlock(&p2m->lock);
+}
+
#endif /* _XEN_P2M_H */
/*
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 15/21] xen/arm: Implement domain_get_maximum_gpfn
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (13 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers Tamas K Lengyel
` (6 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra
From: Julien Grall <julien.grall@linaro.org>
The function domain_get_maximum_gpfn is returning the maximum gpfn ever
mapped in the guest. We can use d->arch.p2m.max_mapped_gfn for this purpose.
We use this in xenaccess as to avoid the user attempting to set page
permissions on pages which don't exist for the domain, as a non-arch specific
sanity check.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
---
xen/arch/arm/mm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xen/arch/arm/mm.c b/xen/arch/arm/mm.c
index 0a243b0..e4a1e5e 100644
--- a/xen/arch/arm/mm.c
+++ b/xen/arch/arm/mm.c
@@ -954,7 +954,7 @@ int page_is_ram_type(unsigned long mfn, unsigned long mem_type)
unsigned long domain_get_maximum_gpfn(struct domain *d)
{
- return -ENOSYS;
+ return d->arch.p2m.max_mapped_gfn;
}
void share_xen_page_with_guest(struct page_info *page,
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers.
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (14 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 15/21] xen/arm: Implement domain_get_maximum_gpfn Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-22 12:24 ` Ian Campbell
2014-09-17 20:51 ` [PATCH for-4.5 v7 17/21] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
` (5 subsequent siblings)
21 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/arch/arm/p2m.c | 138 ++++++++++++++++++++++++++++++++++++-----------------
1 file changed, 95 insertions(+), 43 deletions(-)
diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index 92c3bd6..b8efeae 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -230,6 +230,76 @@ int p2m_pod_decrease_reservation(struct domain *d,
return -ENOSYS;
}
+static void p2m_set_permission(lpae_t *e, p2m_type_t t, p2m_access_t a)
+{
+ /* First apply type permissions */
+ switch ( t )
+ {
+ case p2m_ram_rw:
+ e->p2m.xn = 0;
+ e->p2m.write = 1;
+ break;
+
+ case p2m_ram_ro:
+ e->p2m.xn = 0;
+ e->p2m.write = 0;
+ break;
+
+ case p2m_iommu_map_rw:
+ case p2m_map_foreign:
+ case p2m_grant_map_rw:
+ case p2m_mmio_direct:
+ e->p2m.xn = 1;
+ e->p2m.write = 1;
+ break;
+
+ case p2m_iommu_map_ro:
+ case p2m_grant_map_ro:
+ case p2m_invalid:
+ e->p2m.xn = 1;
+ e->p2m.write = 0;
+ break;
+
+ case p2m_max_real_type:
+ BUG();
+ break;
+ }
+
+ /* Then restrict with access permissions */
+ switch ( a )
+ {
+ case p2m_access_rwx:
+ break;
+ case p2m_access_wx:
+ e->p2m.read = 0;
+ break;
+ case p2m_access_rw:
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_w:
+ e->p2m.read = 0;
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_rx:
+ case p2m_access_rx2rw:
+ e->p2m.write = 0;
+ break;
+ case p2m_access_x:
+ e->p2m.write = 0;
+ e->p2m.read = 0;
+ break;
+ case p2m_access_r:
+ e->p2m.write = 0;
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_n:
+ case p2m_access_n2rwx:
+ e->p2m.read = e->p2m.write = 0;
+ e->p2m.xn = 1;
+ break;
+ }
+}
+
static lpae_t mfn_to_p2m_entry(unsigned long mfn, unsigned int mattr,
p2m_type_t t, p2m_access_t a)
{
@@ -261,37 +331,7 @@ static lpae_t mfn_to_p2m_entry(unsigned long mfn, unsigned int mattr,
break;
}
- switch (t)
- {
- case p2m_ram_rw:
- e.p2m.xn = 0;
- e.p2m.write = 1;
- break;
-
- case p2m_ram_ro:
- e.p2m.xn = 0;
- e.p2m.write = 0;
- break;
-
- case p2m_iommu_map_rw:
- case p2m_map_foreign:
- case p2m_grant_map_rw:
- case p2m_mmio_direct:
- e.p2m.xn = 1;
- e.p2m.write = 1;
- break;
-
- case p2m_iommu_map_ro:
- case p2m_grant_map_ro:
- case p2m_invalid:
- e.p2m.xn = 1;
- e.p2m.write = 0;
- break;
-
- case p2m_max_real_type:
- BUG();
- break;
- }
+ p2m_set_permission(&e, t, a);
ASSERT(!(pa & ~PAGE_MASK));
ASSERT(!(pa & ~PADDR_MASK));
@@ -443,6 +483,26 @@ static bool_t is_mapping_aligned(const paddr_t start_gpaddr,
return true;
}
+static int p2m_shatter_page(struct domain *d,
+ lpae_t *entry,
+ unsigned int level,
+ paddr_t level_shift,
+ bool_t flush_cache)
+{
+ int rc = p2m_create_table(d, entry,
+ level_shift - PAGE_SHIFT, flush_cache);
+
+ if ( !rc )
+ {
+ struct p2m_domain *p2m = &d->arch.p2m;
+ p2m->stats.shattered[level]++;
+ p2m->stats.mappings[level]--;
+ p2m->stats.mappings[level+1] += LPAE_ENTRIES;
+ }
+
+ return rc;
+}
+
#define P2M_ONE_DESCEND 0
#define P2M_ONE_PROGRESS_NOP 0x1
#define P2M_ONE_PROGRESS 0x10
@@ -587,14 +647,10 @@ static int apply_one_level(struct domain *d,
if ( p2m_mapping(orig_pte) )
{
*flush = true;
- rc = p2m_create_table(d, entry,
- level_shift - PAGE_SHIFT, flush_cache);
+ rc = p2m_shatter_page(d, entry, level, level_shift,
+ flush_cache);
if ( rc < 0 )
return rc;
-
- p2m->stats.shattered[level]++;
- p2m->stats.mappings[level]--;
- p2m->stats.mappings[level+1] += LPAE_ENTRIES;
} /* else: an existing table mapping -> descend */
BUG_ON(!p2m_table(*entry));
@@ -629,15 +685,11 @@ static int apply_one_level(struct domain *d,
* and descend.
*/
*flush = true;
- rc = p2m_create_table(d, entry,
- level_shift - PAGE_SHIFT, flush_cache);
+ rc = p2m_shatter_page(d, entry, level, level_shift,
+ flush_cache);
if ( rc < 0 )
return rc;
- p2m->stats.shattered[level]++;
- p2m->stats.mappings[level]--;
- p2m->stats.mappings[level+1] += LPAE_ENTRIES;
-
return P2M_ONE_DESCEND;
}
}
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 17/21] xen/arm: Data abort exception (R/W) mem_events.
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (15 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 18/21] xen/arm: Instruction prefetch abort (X) mem_event handling Tamas K Lengyel
` (4 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
This patch enables to store, set, check and deliver LPAE R/W mem_events.
As the LPAE PTE's lack enough available software programmable bits,
we store the permissions in the extended struct page_info. This enables
the lookup mechanism in the trap handler to use the hardware translation,
so we can rapidly locate the custom page permissions (if there are any).
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v7: - Removed p2m_shatter_page and p2m_set_permission into separate
patch.
- Replace Radix tree settings store with extended struct page_info
approach. This way the trap handlers can use the MMU directly to
locate the permission store instead of having to do a tree-lookup.
- Add p2m_get_entry/set_entry compat functions which are required by
the common mem_access_check function.
- Typo fixes.
v6: - Add helper function p2m_shatter_page.
- Only allocate 4k pages when mem_access is in use.
- If no setting was found in radix tree but PTE exists,
return rwx as permission.
- Move the inclusion of various headers into this patch.
- Make npfec a const.
v5: - Move p2m_set_entry's logic into apply_one_level via
a new p2m_op, MEMACCESS.
v4: - Add p2m_mem_access_radix_set function to be called
when inserting new PTE's and when updating existing entries.
- Switch p2m_mem_access_check to return bool_t.
- Use new struct npfec to pass violation info.
v3: - Add new function for updating the PTE entries, p2m_set_entry.
- Use the new struct npfec to pass violation information.
- Implement n2rwx, rx2rw and listener required routines.
v2: - Patch been split to ease the review process.
- Add definitions of data abort data fetch status codes (enum dabt_dfsc)
and only call p2m_mem_access_check for traps caused by permission violations.
- Only call p2m_write_pte in p2m_lookup if the PTE permission actually changed.
- Properly save settings in the Radix tree and pause the VCPU with
mem_event_vcpu_pause.
---
xen/arch/arm/p2m.c | 215 +++++++++++++++++++++++++++++++++++++++-
xen/arch/arm/traps.c | 29 +++++-
xen/include/asm-arm/p2m.h | 11 ++
xen/include/asm-arm/processor.h | 30 ++++++
4 files changed, 278 insertions(+), 7 deletions(-)
diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index b8efeae..0b52a72 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -5,6 +5,9 @@
#include <xen/errno.h>
#include <xen/domain_page.h>
#include <xen/bitops.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.h>
+#include <public/mem_event.h>
#include <asm/flushtlb.h>
#include <asm/gic.h>
#include <asm/event.h>
@@ -423,6 +426,7 @@ enum p2m_operation {
REMOVE,
RELINQUISH,
CACHEFLUSH,
+ MEMACCESS,
};
/* Put any references on the single 4K page referenced by pte. TODO:
@@ -556,7 +560,11 @@ static int apply_one_level(struct domain *d,
if ( p2m_valid(orig_pte) )
return P2M_ONE_DESCEND;
- if ( is_mapping_aligned(*addr, end_gpaddr, 0, level_size) )
+ if ( level < 3 && p2m_access_rwx != a )
+ {
+ /* We create only 4k pages when mem_access is in use. */
+ }
+ else if ( is_mapping_aligned(*addr, end_gpaddr, 0, level_size) )
{
struct page_info *page;
@@ -567,6 +575,7 @@ static int apply_one_level(struct domain *d,
if ( level < 3 )
pte.p2m.table = 0;
p2m_write_pte(entry, pte, flush_cache);
+ page->a = a;
p2m->stats.mappings[level]++;
*addr += level_size;
@@ -583,8 +592,8 @@ static int apply_one_level(struct domain *d,
/*
* If we get here then we failed to allocate a sufficiently
* large contiguous region for this level (which can't be
- * L3). Create a page table and continue to descend so we try
- * smaller allocations.
+ * L3) or mem_access is in use. Create a page table and
+ * continue to descend so we try smaller allocations.
*/
rc = p2m_create_table(d, entry, 0, flush_cache);
if ( rc < 0 )
@@ -604,6 +613,13 @@ static int apply_one_level(struct domain *d,
p2m_write_pte(entry, pte, flush_cache);
+ if ( mfn_valid(pte.p2m.base) )
+ {
+ struct page_info *page = mfn_to_page(pte.p2m.base);
+ if ( page )
+ page->a = a;
+ }
+
*flush |= p2m_valid(orig_pte);
*addr += level_size;
@@ -758,6 +774,49 @@ static int apply_one_level(struct domain *d,
*addr += PAGE_SIZE;
return P2M_ONE_PROGRESS_NOP;
}
+
+ case MEMACCESS:
+ if ( level < 3 )
+ {
+ if ( !p2m_valid(orig_pte) )
+ {
+ *addr += level_size;
+ return P2M_ONE_PROGRESS_NOP;
+ }
+
+ /* Shatter large pages as we descend */
+ if ( p2m_mapping(orig_pte) )
+ {
+ rc = p2m_shatter_page(d, entry, level, level_shift, flush_cache);
+
+ if ( rc < 0 )
+ return rc;
+ } /* else: an existing table mapping -> descend */
+
+ return P2M_ONE_DESCEND;
+ }
+ else
+ {
+ pte = orig_pte;
+
+ if ( !p2m_table(pte) )
+ pte.bits = 0;
+
+ if ( p2m_valid(pte) )
+ {
+ struct page_info *page;
+ ASSERT(pte.p2m.type != p2m_invalid);
+
+ page = mfn_to_page(pte.p2m.base);
+ page->a = a;
+
+ p2m_set_permission(&pte, pte.p2m.type, a);
+ p2m_write_pte(entry, pte, flush_cache);
+ }
+
+ *addr += level_size;
+ return P2M_ONE_PROGRESS;
+ }
}
BUG(); /* Should never get here */
@@ -781,7 +840,9 @@ static int apply_p2m_changes(struct domain *d,
unsigned int level = 0;
unsigned long cur_first_page = ~0,
cur_first_offset = ~0,
- cur_second_offset = ~0;
+ cur_second_offset = ~0,
+ start_gpfn = paddr_to_pfn(start_gpaddr),
+ end_gpfn = paddr_to_pfn(end_gpaddr);
unsigned long count = 0;
bool_t flush = false;
bool_t flush_pt;
@@ -817,6 +878,20 @@ static int apply_p2m_changes(struct domain *d,
count = 0;
}
+ /*
+ * Preempt setting mem_access permissions as required by XSA-89,
+ * if it's not the last iteration. */
+ if ( op == MEMACCESS && count )
+ {
+ int progress = paddr_to_pfn(addr) - start_gpfn + 1;
+ if ( (end_gpfn-start_gpfn) > progress && !(progress & mask)
+ && hypercall_preempt_check() )
+ {
+ rc = progress;
+ goto out;
+ }
+ }
+
if ( cur_first_page != p2m_first_level_index(addr) )
{
if ( first ) unmap_domain_page(first);
@@ -1103,6 +1178,33 @@ static void p2m_free_vmid(struct domain *d)
spin_unlock(&vmid_alloc_lock);
}
+static int
+p2m_set_entry(struct p2m_domain *p2m, unsigned long gfn, mfn_t mfn,
+ unsigned int order, p2m_type_t p2mt, p2m_access_t p2ma)
+{
+ return apply_p2m_changes(current->domain, MEMACCESS,
+ pfn_to_paddr(gfn), 1,
+ 0, MATTR_MEM, 0, ~0, p2mt, p2ma);
+}
+
+static mfn_t p2m_get_entry(struct p2m_domain *p2m, unsigned long gfn,
+ p2m_type_t *t, p2m_access_t* a, p2m_query_t q,
+ unsigned int *page_order)
+{
+ p2m_type_t _t;
+ paddr_t maddr = p2m_lookup(current->domain, pfn_to_paddr(gfn), &_t, 0);
+ mfn_t mfn = maddr >> PAGE_SHIFT;
+ if ( mfn_valid(mfn) )
+ {
+ struct page_info *page = mfn_to_page(mfn);
+ *a = page->a;
+ *t = _t;
+ return mfn;
+ }
+
+ return INVALID_MFN;
+}
+
void p2m_teardown(struct domain *d)
{
struct p2m_domain *p2m = &d->arch.p2m;
@@ -1145,6 +1247,8 @@ int p2m_init(struct domain *d)
p2m->lowest_mapped_gfn = ULONG_MAX;
p2m->default_access = p2m_access_rwx;
+ p2m->set_entry = p2m_set_entry;
+ p2m->get_entry = p2m_get_entry;
err:
spin_unlock(&p2m->lock);
@@ -1213,6 +1317,109 @@ err:
return page;
}
+/* Set access type for a region of pfns.
+ * If start_pfn == -1ul, sets the default access type */
+long p2m_set_mem_access(struct domain *d, unsigned long pfn, uint32_t nr,
+ uint32_t start, uint32_t mask, xenmem_access_t access)
+{
+ struct p2m_domain *p2m = p2m_get_hostp2m(d);
+ p2m_access_t a;
+ long rc = 0;
+
+ static const p2m_access_t memaccess[] = {
+#define ACCESS(ac) [XENMEM_access_##ac] = p2m_access_##ac
+ ACCESS(n),
+ ACCESS(r),
+ ACCESS(w),
+ ACCESS(rw),
+ ACCESS(x),
+ ACCESS(rx),
+ ACCESS(wx),
+ ACCESS(rwx),
+ ACCESS(rx2rw),
+ ACCESS(n2rwx),
+#undef ACCESS
+ };
+
+ switch ( access )
+ {
+ case 0 ... ARRAY_SIZE(memaccess) - 1:
+ a = memaccess[access];
+ break;
+ case XENMEM_access_default:
+ a = p2m->default_access;
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ /* If request to set default access. */
+ if ( pfn == ~0ul )
+ {
+ p2m->default_access = a;
+ return 0;
+ }
+
+ rc = apply_p2m_changes(d, MEMACCESS,
+ pfn_to_paddr(pfn+start), pfn_to_paddr(pfn+nr),
+ 0, MATTR_MEM, 1, mask, 0, a);
+
+ if ( rc < 0 )
+ return rc;
+ else if ( rc > 0 )
+ return start+rc;
+
+ flush_tlb_domain(d);
+ return 0;
+}
+
+int p2m_get_mem_access(struct domain *d, unsigned long gpfn,
+ xenmem_access_t *access)
+{
+ struct p2m_domain *p2m = p2m_get_hostp2m(d);
+ struct page_info *page;
+ unsigned long mfn;
+
+ static const xenmem_access_t memaccess[] = {
+#define ACCESS(ac) [p2m_access_##ac] = XENMEM_access_##ac
+ ACCESS(n),
+ ACCESS(r),
+ ACCESS(w),
+ ACCESS(rw),
+ ACCESS(x),
+ ACCESS(rx),
+ ACCESS(wx),
+ ACCESS(rwx),
+ ACCESS(rx2rw),
+ ACCESS(n2rwx),
+#undef ACCESS
+ };
+
+ /* If request to get default access */
+ if ( gpfn == ~0ull )
+ {
+ *access = memaccess[p2m->default_access];
+ return 0;
+ }
+
+ mfn = gmfn_to_mfn(d, gpfn);
+ if ( !mfn_valid(mfn) )
+ {
+ return -ESRCH;
+ }
+
+ page = mfn_to_page(mfn);
+ ASSERT(page);
+
+ if ( page->a >= ARRAY_SIZE(memaccess) )
+ return -ERANGE;
+
+ *access = memaccess[page->a];
+
+ return 0;
+}
+
+
/*
* Local variables:
* mode: C
diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
index ae7458c..75ce3dc 100644
--- a/xen/arch/arm/traps.c
+++ b/xen/arch/arm/traps.c
@@ -30,6 +30,7 @@
#include <xen/hypercall.h>
#include <xen/softirq.h>
#include <xen/domain_page.h>
+#include <xen/mem_access.h>
#include <public/sched.h>
#include <public/xen.h>
#include <asm/event.h>
@@ -1867,11 +1868,33 @@ static void do_trap_data_abort_guest(struct cpu_user_regs *regs,
info.gva = READ_SYSREG64(FAR_EL2);
#endif
- if (dabt.s1ptw)
+ rc = gva_to_ipa(info.gva, &info.gpa);
+ if ( -EFAULT == rc )
goto bad_data_abort;
- rc = gva_to_ipa(info.gva, &info.gpa);
- if ( rc == -EFAULT )
+ switch ( dabt.dfsc )
+ {
+ case DABT_DFSC_PERMISSION_1:
+ case DABT_DFSC_PERMISSION_2:
+ case DABT_DFSC_PERMISSION_3:
+ {
+ const struct npfec npfec = {
+ .read_access = 1,
+ .write_access = dabt.write,
+ .gla_valid = 1,
+ .kind = dabt.s1ptw ? npfec_kind_in_gpt : npfec_kind_with_gla
+ };
+
+ rc = mem_access_check(info.gpa, info.gva, npfec);
+
+ /* Trap was triggered by mem_access, work here is done */
+ if ( rc >= 0 )
+ return;
+ }
+ break;
+ }
+
+ if ( dabt.s1ptw )
goto bad_data_abort;
/* XXX: Decode the instruction if ISS is not valid */
diff --git a/xen/include/asm-arm/p2m.h b/xen/include/asm-arm/p2m.h
index 0049538..b5c71c5 100644
--- a/xen/include/asm-arm/p2m.h
+++ b/xen/include/asm-arm/p2m.h
@@ -2,6 +2,7 @@
#define _XEN_P2M_H
#include <xen/mm.h>
+#include <public/memory.h>
#include <xen/p2m-common.h>
@@ -245,6 +246,16 @@ static inline void p2m_gpfn_unlock(struct p2m_domain *p2m, unsigned long gpfn)
spin_unlock(&p2m->lock);
}
+/* Set access type for a region of pfns.
+ * If start_pfn == -1ul, sets the default access type */
+long p2m_set_mem_access(struct domain *d, unsigned long start_pfn, uint32_t nr,
+ uint32_t start, uint32_t mask, xenmem_access_t access);
+
+/* Get access type for a pfn
+ * If pfn == -1ul, gets the default access type */
+int p2m_get_mem_access(struct domain *d, unsigned long pfn,
+ xenmem_access_t *access);
+
#endif /* _XEN_P2M_H */
/*
diff --git a/xen/include/asm-arm/processor.h b/xen/include/asm-arm/processor.h
index 4aab823..34ffd2d 100644
--- a/xen/include/asm-arm/processor.h
+++ b/xen/include/asm-arm/processor.h
@@ -275,6 +275,36 @@ enum dabt_size {
DABT_DOUBLE_WORD = 3,
};
+/* Data abort data fetch status codes */
+enum dabt_dfsc {
+ DABT_DFSC_ADDR_SIZE_0 = 0b000000,
+ DABT_DFSC_ADDR_SIZE_1 = 0b000001,
+ DABT_DFSC_ADDR_SIZE_2 = 0b000010,
+ DABT_DFSC_ADDR_SIZE_3 = 0b000011,
+ DABT_DFSC_TRANSLATION_0 = 0b000100,
+ DABT_DFSC_TRANSLATION_1 = 0b000101,
+ DABT_DFSC_TRANSLATION_2 = 0b000110,
+ DABT_DFSC_TRANSLATION_3 = 0b000111,
+ DABT_DFSC_ACCESS_1 = 0b001001,
+ DABT_DFSC_ACCESS_2 = 0b001010,
+ DABT_DFSC_ACCESS_3 = 0b001011,
+ DABT_DFSC_PERMISSION_1 = 0b001101,
+ DABT_DFSC_PERMISSION_2 = 0b001110,
+ DABT_DFSC_PERMISSION_3 = 0b001111,
+ DABT_DFSC_SYNC_EXT = 0b010000,
+ DABT_DFSC_SYNC_PARITY = 0b011000,
+ DABT_DFSC_SYNC_EXT_TTW_0 = 0b010100,
+ DABT_DFSC_SYNC_EXT_TTW_1 = 0b010101,
+ DABT_DFSC_SYNC_EXT_TTW_2 = 0b010110,
+ DABT_DFSC_SYNC_EXT_TTW_3 = 0b010111,
+ DABT_DFSC_SYNC_PARITY_TTW_0 = 0b011100,
+ DABT_DFSC_SYNC_PARITY_TTW_1 = 0b011101,
+ DABT_DFSC_SYNC_PARITY_TTW_2 = 0b011110,
+ DABT_DFSC_SYNC_PARITY_TTW_3 = 0b011111,
+ DABT_DFSC_ALIGNMENT = 0b100001,
+ DABT_DFSC_TLB_CONFLICT = 0b110000,
+};
+
union hsr {
uint32_t bits;
struct {
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 18/21] xen/arm: Instruction prefetch abort (X) mem_event handling
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (16 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 17/21] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 19/21] xen/arm: Enable the compilation of mem_access and mem_event on ARM Tamas K Lengyel
` (3 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Add missing structure definition for iabt and update the trap handling
mechanism to only inject the exception if the mem_access checker
decides to do so.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v7: - Use the new common mem_access_check.
v6: - Make npfec a const.
v4: - Don't mark instruction fetch violation as read violation.
- Use new struct npfec to pass violation info.
v2: - Add definition for instruction abort instruction fetch status codes
(enum iabt_ifsc) and only call p2m_mem_access_check for traps triggered
for permission violations.
---
xen/arch/arm/traps.c | 39 ++++++++++++++++++++++++++++++++++++++-
xen/include/asm-arm/processor.h | 40 +++++++++++++++++++++++++++++++++++++++-
2 files changed, 77 insertions(+), 2 deletions(-)
diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
index 75ce3dc..c048016 100644
--- a/xen/arch/arm/traps.c
+++ b/xen/arch/arm/traps.c
@@ -1844,7 +1844,44 @@ done:
static void do_trap_instr_abort_guest(struct cpu_user_regs *regs,
union hsr hsr)
{
- register_t addr = READ_SYSREG(FAR_EL2);
+ struct hsr_iabt iabt = hsr.iabt;
+ int rc;
+ register_t addr;
+ vaddr_t gva;
+ paddr_t gpa;
+
+#ifdef CONFIG_ARM_32
+ gva = READ_CP32(HIFAR);
+#else
+ gva = READ_SYSREG64(FAR_EL2);
+#endif
+
+ rc = gva_to_ipa(gva, &gpa);
+ if ( -EFAULT == rc )
+ return;
+
+ switch ( iabt.ifsc )
+ {
+ case IABT_IFSC_PERMISSION_1:
+ case IABT_IFSC_PERMISSION_2:
+ case IABT_IFSC_PERMISSION_3:
+ {
+ const struct npfec npfec = {
+ .insn_fetch = 1,
+ .gla_valid = 1,
+ .kind = iabt.s1ptw ? npfec_kind_in_gpt : npfec_kind_with_gla
+ };
+
+ rc = mem_access_check(gpa, gva, npfec);
+
+ /* Trap was triggered by mem_access, work here is done */
+ if ( rc >= 0 )
+ return;
+ }
+ break;
+ }
+
+ addr = READ_SYSREG(FAR_EL2);
inject_iabt_exception(regs, addr, hsr.len);
}
diff --git a/xen/include/asm-arm/processor.h b/xen/include/asm-arm/processor.h
index 34ffd2d..9fe52e3 100644
--- a/xen/include/asm-arm/processor.h
+++ b/xen/include/asm-arm/processor.h
@@ -305,6 +305,36 @@ enum dabt_dfsc {
DABT_DFSC_TLB_CONFLICT = 0b110000,
};
+/* Instruction abort instruction fault status codes */
+enum iabt_ifsc {
+ IABT_IFSC_ADDR_SIZE_0 = 0b000000,
+ IABT_IFSC_ADDR_SIZE_1 = 0b000001,
+ IABT_IFSC_ADDR_SIZE_2 = 0b000010,
+ IABT_IFSC_ADDR_SIZE_3 = 0b000011,
+ IABT_IFSC_TRANSLATION_0 = 0b000100,
+ IABT_IFSC_TRANSLATION_1 = 0b000101,
+ IABT_IFSC_TRANSLATION_2 = 0b000110,
+ IABT_IFSC_TRANSLATION_3 = 0b000111,
+ IABT_IFSC_ACCESS_1 = 0b001001,
+ IABT_IFSC_ACCESS_2 = 0b001010,
+ IABT_IFSC_ACCESS_3 = 0b001011,
+ IABT_IFSC_PERMISSION_1 = 0b001101,
+ IABT_IFSC_PERMISSION_2 = 0b001110,
+ IABT_IFSC_PERMISSION_3 = 0b001111,
+ IABT_IFSC_SYNC_EXT = 0b010000,
+ IABT_IFSC_SYNC_PARITY = 0b011000,
+ IABT_IFSC_SYNC_EXT_TTW_0 = 0b010100,
+ IABT_IFSC_SYNC_EXT_TTW_1 = 0b010101,
+ IABT_IFSC_SYNC_EXT_TTW_2 = 0b010110,
+ IABT_IFSC_SYNC_EXT_TTW_3 = 0b010111,
+ IABT_IFSC_SYNC_PARITY_TTW_0 = 0b011100,
+ IABT_IFSC_SYNC_PARITY_TTW_1 = 0b011101,
+ IABT_IFSC_SYNC_PARITY_TTW_2 = 0b011110,
+ IABT_IFSC_SYNC_PARITY_TTW_3 = 0b011111,
+ IABT_IFSC_ALIGNMENT = 0b100001,
+ IABT_IFSC_TLB_CONFLICT = 0b110000,
+};
+
union hsr {
uint32_t bits;
struct {
@@ -384,10 +414,18 @@ union hsr {
} sysreg; /* HSR_EC_SYSREG */
#endif
+ struct hsr_iabt {
+ unsigned long ifsc:6; /* Instruction fault status code */
+ unsigned long res0:1;
+ unsigned long s1ptw:1; /* Fault during a stage 1 translation table walk */
+ unsigned long res1:1;
+ unsigned long ea:1; /* External abort type */
+ } iabt; /* HSR_EC_INSTR_ABORT_* */
+
struct hsr_dabt {
unsigned long dfsc:6; /* Data Fault Status Code */
unsigned long write:1; /* Write / not Read */
- unsigned long s1ptw:1; /* */
+ unsigned long s1ptw:1; /* Fault during a stage 1 translation table walk */
unsigned long cache:1; /* Cache Maintenance */
unsigned long eat:1; /* External Abort Type */
#ifdef CONFIG_ARM_32
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 19/21] xen/arm: Enable the compilation of mem_access and mem_event on ARM.
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (17 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 18/21] xen/arm: Instruction prefetch abort (X) mem_event handling Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 20/21] tools/libxc: Allocate magic page for mem access " Tamas K Lengyel
` (2 subsequent siblings)
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
This patch sets up the infrastructure to support mem_access and mem_event
on ARM and turns on compilation. We define the required XSM functions.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
---
v3: Wrap mem_event related functions in XSM into #ifdef HAS_MEM_ACCESS
blocks.
Update XSM hooks in flask to properly wire it up on ARM.
v2: Add CONFIG_MEM_PAGING and CONFIG_MEM_SHARING definitions and
use them instead of CONFIG_X86.
Split domctl copy-back and p2m type definitions into separate
patches and move this patch to the end of the series.
---
xen/arch/arm/Rules.mk | 1 +
xen/common/mem_event.c | 19 +++++++++++++++++++
xen/include/asm-x86/config.h | 3 +++
xen/include/xsm/dummy.h | 26 ++++++++++++++------------
xen/include/xsm/xsm.h | 29 +++++++++++++++++------------
xen/xsm/dummy.c | 7 +++++--
xen/xsm/flask/hooks.c | 33 ++++++++++++++++++++-------------
7 files changed, 79 insertions(+), 39 deletions(-)
diff --git a/xen/arch/arm/Rules.mk b/xen/arch/arm/Rules.mk
index 8658176..f6781b5 100644
--- a/xen/arch/arm/Rules.mk
+++ b/xen/arch/arm/Rules.mk
@@ -10,6 +10,7 @@ HAS_DEVICE_TREE := y
HAS_VIDEO := y
HAS_ARM_HDLCD := y
HAS_PASSTHROUGH := y
+HAS_MEM_ACCESS := y
CFLAGS += -I$(BASEDIR)/include
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index 4d0a27b..59edf07 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -27,8 +27,15 @@
#include <xen/mem_event.h>
#include <xen/mem_access.h>
#include <asm/p2m.h>
+
+#ifdef CONFIG_MEM_PAGING
#include <asm/mem_paging.h>
+#endif
+
+#ifdef CONFIG_MEM_SHARING
#include <asm/mem_sharing.h>
+#endif
+
#include <xsm/xsm.h>
/* for public/io/ring.h macros */
@@ -423,12 +430,14 @@ int __mem_event_claim_slot(struct domain *d, struct mem_event_domain *med,
return mem_event_grab_slot(med, (current->domain != d));
}
+#ifdef CONFIG_MEM_PAGING
/* Registered with Xen-bound event channel for incoming notifications. */
static void mem_paging_notification(struct vcpu *v, unsigned int port)
{
if ( likely(v->domain->mem_event->paging.ring_page != NULL) )
p2m_mem_paging_resume(v->domain);
}
+#endif
/* Registered with Xen-bound event channel for incoming notifications. */
static void mem_access_notification(struct vcpu *v, unsigned int port)
@@ -437,15 +446,20 @@ static void mem_access_notification(struct vcpu *v, unsigned int port)
mem_access_resume(v->domain);
}
+#ifdef CONFIG_MEM_SHARING
/* Registered with Xen-bound event channel for incoming notifications. */
static void mem_sharing_notification(struct vcpu *v, unsigned int port)
{
if ( likely(v->domain->mem_event->share.ring_page != NULL) )
mem_sharing_sharing_resume(v->domain);
}
+#endif
int do_mem_event_op(int op, uint32_t domain, void *arg)
{
+#if !defined(CONFIG_MEM_PAGING) && !defined(CONFIG_MEM_SHARING)
+ return -ENOSYS;
+#else
int ret;
struct domain *d;
@@ -472,6 +486,7 @@ int do_mem_event_op(int op, uint32_t domain, void *arg)
out:
rcu_unlock_domain(d);
return ret;
+#endif
}
/* Clean up on domain destruction */
@@ -532,6 +547,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
switch ( mec->mode )
{
+#ifdef CONFIG_MEM_PAGING
case XEN_DOMCTL_MEM_EVENT_OP_PAGING:
{
struct mem_event_domain *med = &d->mem_event->paging;
@@ -582,6 +598,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
}
}
break;
+#endif
case XEN_DOMCTL_MEM_EVENT_OP_ACCESS:
{
@@ -616,6 +633,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
}
break;
+#ifdef CONFIG_MEM_SHARING
case XEN_DOMCTL_MEM_EVENT_OP_SHARING:
{
struct mem_event_domain *med = &d->mem_event->share;
@@ -654,6 +672,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
}
}
break;
+#endif
default:
rc = -ENOSYS;
diff --git a/xen/include/asm-x86/config.h b/xen/include/asm-x86/config.h
index 8a864ce..f8ef043 100644
--- a/xen/include/asm-x86/config.h
+++ b/xen/include/asm-x86/config.h
@@ -57,6 +57,9 @@
#define CONFIG_LATE_HWDOM 1
#endif
+#define CONFIG_MEM_SHARING 1
+#define CONFIG_MEM_PAGING 1
+
#define HZ 100
#define OPT_CONSOLE_STR "vga"
diff --git a/xen/include/xsm/dummy.h b/xen/include/xsm/dummy.h
index df55e70..f20e89c 100644
--- a/xen/include/xsm/dummy.h
+++ b/xen/include/xsm/dummy.h
@@ -513,6 +513,20 @@ static XSM_INLINE int xsm_hvm_param_nested(XSM_DEFAULT_ARG struct domain *d)
return xsm_default_action(action, current->domain, d);
}
+#ifdef HAS_MEM_ACCESS
+static XSM_INLINE int xsm_mem_event_control(XSM_DEFAULT_ARG struct domain *d, int mode, int op)
+{
+ XSM_ASSERT_ACTION(XSM_PRIV);
+ return xsm_default_action(action, current->domain, d);
+}
+
+static XSM_INLINE int xsm_mem_event_op(XSM_DEFAULT_ARG struct domain *d, int op)
+{
+ XSM_ASSERT_ACTION(XSM_DM_PRIV);
+ return xsm_default_action(action, current->domain, d);
+}
+#endif
+
#ifdef CONFIG_X86
static XSM_INLINE int xsm_do_mca(XSM_DEFAULT_VOID)
{
@@ -556,18 +570,6 @@ static XSM_INLINE int xsm_hvm_ioreq_server(XSM_DEFAULT_ARG struct domain *d, int
return xsm_default_action(action, current->domain, d);
}
-static XSM_INLINE int xsm_mem_event_control(XSM_DEFAULT_ARG struct domain *d, int mode, int op)
-{
- XSM_ASSERT_ACTION(XSM_PRIV);
- return xsm_default_action(action, current->domain, d);
-}
-
-static XSM_INLINE int xsm_mem_event_op(XSM_DEFAULT_ARG struct domain *d, int op)
-{
- XSM_ASSERT_ACTION(XSM_DM_PRIV);
- return xsm_default_action(action, current->domain, d);
-}
-
static XSM_INLINE int xsm_mem_sharing_op(XSM_DEFAULT_ARG struct domain *d, struct domain *cd, int op)
{
XSM_ASSERT_ACTION(XSM_DM_PRIV);
diff --git a/xen/include/xsm/xsm.h b/xen/include/xsm/xsm.h
index 6c1c079..4ce089f 100644
--- a/xen/include/xsm/xsm.h
+++ b/xen/include/xsm/xsm.h
@@ -141,6 +141,11 @@ struct xsm_operations {
int (*hvm_param_nested) (struct domain *d);
int (*get_vnumainfo) (struct domain *d);
+#ifdef HAS_MEM_ACCESS
+ int (*mem_event_control) (struct domain *d, int mode, int op);
+ int (*mem_event_op) (struct domain *d, int op);
+#endif
+
#ifdef CONFIG_X86
int (*do_mca) (void);
int (*shadow_control) (struct domain *d, uint32_t op);
@@ -149,8 +154,6 @@ struct xsm_operations {
int (*hvm_set_pci_link_route) (struct domain *d);
int (*hvm_inject_msi) (struct domain *d);
int (*hvm_ioreq_server) (struct domain *d, int op);
- int (*mem_event_control) (struct domain *d, int mode, int op);
- int (*mem_event_op) (struct domain *d, int op);
int (*mem_sharing_op) (struct domain *d, struct domain *cd, int op);
int (*apic) (struct domain *d, int cmd);
int (*memtype) (uint32_t access);
@@ -540,6 +543,18 @@ static inline int xsm_get_vnumainfo (xsm_default_t def, struct domain *d)
return xsm_ops->get_vnumainfo(d);
}
+#ifdef HAS_MEM_ACCESS
+static inline int xsm_mem_event_control (xsm_default_t def, struct domain *d, int mode, int op)
+{
+ return xsm_ops->mem_event_control(d, mode, op);
+}
+
+static inline int xsm_mem_event_op (xsm_default_t def, struct domain *d, int op)
+{
+ return xsm_ops->mem_event_op(d, op);
+}
+#endif
+
#ifdef CONFIG_X86
static inline int xsm_do_mca(xsm_default_t def)
{
@@ -576,16 +591,6 @@ static inline int xsm_hvm_ioreq_server (xsm_default_t def, struct domain *d, int
return xsm_ops->hvm_ioreq_server(d, op);
}
-static inline int xsm_mem_event_control (xsm_default_t def, struct domain *d, int mode, int op)
-{
- return xsm_ops->mem_event_control(d, mode, op);
-}
-
-static inline int xsm_mem_event_op (xsm_default_t def, struct domain *d, int op)
-{
- return xsm_ops->mem_event_op(d, op);
-}
-
static inline int xsm_mem_sharing_op (xsm_default_t def, struct domain *d, struct domain *cd, int op)
{
return xsm_ops->mem_sharing_op(d, cd, op);
diff --git a/xen/xsm/dummy.c b/xen/xsm/dummy.c
index 0826a8b..8eb3050 100644
--- a/xen/xsm/dummy.c
+++ b/xen/xsm/dummy.c
@@ -118,6 +118,11 @@ void xsm_fixup_ops (struct xsm_operations *ops)
set_to_dummy_if_null(ops, remove_from_physmap);
set_to_dummy_if_null(ops, map_gmfn_foreign);
+#ifdef HAS_MEM_ACCESS
+ set_to_dummy_if_null(ops, mem_event_control);
+ set_to_dummy_if_null(ops, mem_event_op);
+#endif
+
#ifdef CONFIG_X86
set_to_dummy_if_null(ops, do_mca);
set_to_dummy_if_null(ops, shadow_control);
@@ -126,8 +131,6 @@ void xsm_fixup_ops (struct xsm_operations *ops)
set_to_dummy_if_null(ops, hvm_set_pci_link_route);
set_to_dummy_if_null(ops, hvm_inject_msi);
set_to_dummy_if_null(ops, hvm_ioreq_server);
- set_to_dummy_if_null(ops, mem_event_control);
- set_to_dummy_if_null(ops, mem_event_op);
set_to_dummy_if_null(ops, mem_sharing_op);
set_to_dummy_if_null(ops, apic);
set_to_dummy_if_null(ops, platform_op);
diff --git a/xen/xsm/flask/hooks.c b/xen/xsm/flask/hooks.c
index df05566..8de5e49 100644
--- a/xen/xsm/flask/hooks.c
+++ b/xen/xsm/flask/hooks.c
@@ -577,6 +577,9 @@ static int flask_domctl(struct domain *d, int cmd)
case XEN_DOMCTL_iomem_permission:
case XEN_DOMCTL_memory_mapping:
case XEN_DOMCTL_set_target:
+#ifdef HAS_MEM_ACCESS
+ case XEN_DOMCTL_mem_event_op:
+#endif
#ifdef CONFIG_X86
/* These have individual XSM hooks (arch/x86/domctl.c) */
case XEN_DOMCTL_shadow_op:
@@ -584,7 +587,6 @@ static int flask_domctl(struct domain *d, int cmd)
case XEN_DOMCTL_bind_pt_irq:
case XEN_DOMCTL_unbind_pt_irq:
case XEN_DOMCTL_ioport_mapping:
- case XEN_DOMCTL_mem_event_op:
/* These have individual XSM hooks (drivers/passthrough/iommu.c) */
case XEN_DOMCTL_get_device_group:
case XEN_DOMCTL_test_assign_device:
@@ -1189,6 +1191,18 @@ static int flask_deassign_device(struct domain *d, uint32_t machine_bdf)
}
#endif /* HAS_PASSTHROUGH && HAS_PCI */
+#ifdef HAS_MEM_ACCESS
+static int flask_mem_event_control(struct domain *d, int mode, int op)
+{
+ return current_has_perm(d, SECCLASS_HVM, HVM__MEM_EVENT);
+}
+
+static int flask_mem_event_op(struct domain *d, int op)
+{
+ return current_has_perm(d, SECCLASS_HVM, HVM__MEM_EVENT);
+}
+#endif /* HAS_MEM_ACCESS */
+
#ifdef CONFIG_X86
static int flask_do_mca(void)
{
@@ -1299,16 +1313,6 @@ static int flask_hvm_ioreq_server(struct domain *d, int op)
return current_has_perm(d, SECCLASS_HVM, HVM__HVMCTL);
}
-static int flask_mem_event_control(struct domain *d, int mode, int op)
-{
- return current_has_perm(d, SECCLASS_HVM, HVM__MEM_EVENT);
-}
-
-static int flask_mem_event_op(struct domain *d, int op)
-{
- return current_has_perm(d, SECCLASS_HVM, HVM__MEM_EVENT);
-}
-
static int flask_mem_sharing_op(struct domain *d, struct domain *cd, int op)
{
int rc = current_has_perm(cd, SECCLASS_HVM, HVM__MEM_SHARING);
@@ -1577,6 +1581,11 @@ static struct xsm_operations flask_ops = {
.deassign_device = flask_deassign_device,
#endif
+#ifdef HAS_MEM_ACCESS
+ .mem_event_control = flask_mem_event_control,
+ .mem_event_op = flask_mem_event_op,
+#endif
+
#ifdef CONFIG_X86
.do_mca = flask_do_mca,
.shadow_control = flask_shadow_control,
@@ -1585,8 +1594,6 @@ static struct xsm_operations flask_ops = {
.hvm_set_pci_link_route = flask_hvm_set_pci_link_route,
.hvm_inject_msi = flask_hvm_inject_msi,
.hvm_ioreq_server = flask_hvm_ioreq_server,
- .mem_event_control = flask_mem_event_control,
- .mem_event_op = flask_mem_event_op,
.mem_sharing_op = flask_mem_sharing_op,
.apic = flask_apic,
.platform_op = flask_platform_op,
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 20/21] tools/libxc: Allocate magic page for mem access on ARM
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (18 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 19/21] xen/arm: Enable the compilation of mem_access and mem_event on ARM Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access " Tamas K Lengyel
2014-09-18 9:21 ` [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tim Deegan
21 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Reviewed-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
---
tools/libxc/xc_dom_arm.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/tools/libxc/xc_dom_arm.c b/tools/libxc/xc_dom_arm.c
index 9b31b1f..13e881e 100644
--- a/tools/libxc/xc_dom_arm.c
+++ b/tools/libxc/xc_dom_arm.c
@@ -26,9 +26,10 @@
#include "xg_private.h"
#include "xc_dom.h"
-#define NR_MAGIC_PAGES 2
+#define NR_MAGIC_PAGES 3
#define CONSOLE_PFN_OFFSET 0
#define XENSTORE_PFN_OFFSET 1
+#define MEMACCESS_PFN_OFFSET 2
#define LPAE_SHIFT 9
@@ -87,10 +88,13 @@ static int alloc_magic_pages(struct xc_dom_image *dom)
xc_clear_domain_page(dom->xch, dom->guest_domid, dom->console_pfn);
xc_clear_domain_page(dom->xch, dom->guest_domid, dom->xenstore_pfn);
+ xc_clear_domain_page(dom->xch, dom->guest_domid, base + MEMACCESS_PFN_OFFSET);
xc_hvm_param_set(dom->xch, dom->guest_domid, HVM_PARAM_CONSOLE_PFN,
dom->console_pfn);
xc_hvm_param_set(dom->xch, dom->guest_domid, HVM_PARAM_STORE_PFN,
dom->xenstore_pfn);
+ xc_hvm_param_set(dom->xch, dom->guest_domid, HVM_PARAM_ACCESS_RING_PFN,
+ base + MEMACCESS_PFN_OFFSET);
/* allocated by toolstack */
xc_hvm_param_set(dom->xch, dom->guest_domid, HVM_PARAM_CONSOLE_EVTCHN,
dom->console_evtchn);
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access on ARM
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (19 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 20/21] tools/libxc: Allocate magic page for mem access " Tamas K Lengyel
@ 2014-09-17 20:51 ` Tamas K Lengyel
2014-09-22 12:29 ` Ian Campbell
2014-09-18 9:21 ` [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tim Deegan
21 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-17 20:51 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Define the ARM specific test_and_set_bit functions and switch
to use maximum gpfn as the limit to setting permissions. Also,
move HAS_MEM_ACCESS definition into config.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v6: - Just use xc_domain_maximum_gpfn to get max_gpfn.
v5: - Use the new information returned by getdomaininfo, max_gpfn, to
set access permissions. On ARM this will include the potential
memory hole as well which the hypervisor just loops over.
v4: - Take into account multiple guest ram banks on ARM.
- Move HAS_MEM_ACCESS definition into config/*.mk and only compile
xen-access when it is defined.
- Pass CONFIG_X86/CONFIG_ARM flags during compilation in xen-access
Makefile.
---
config/arm32.mk | 1 +
config/arm64.mk | 1 +
config/x86_32.mk | 2 +
config/x86_64.mk | 2 +
tools/tests/xen-access/Makefile | 9 ++++-
tools/tests/xen-access/xen-access.c | 79 ++++++++++++++++++++++++-------------
xen/arch/arm/Rules.mk | 1 -
xen/arch/x86/Rules.mk | 1 -
8 files changed, 65 insertions(+), 31 deletions(-)
diff --git a/config/arm32.mk b/config/arm32.mk
index aa79d22..4a7c259 100644
--- a/config/arm32.mk
+++ b/config/arm32.mk
@@ -13,6 +13,7 @@ HAS_PL011 := y
HAS_EXYNOS4210 := y
HAS_OMAP := y
HAS_NS16550 := y
+HAS_MEM_ACCESS := y
# Use only if calling $(LD) directly.
LDFLAGS_DIRECT += -EL
diff --git a/config/arm64.mk b/config/arm64.mk
index 15b57a4..ea6558d 100644
--- a/config/arm64.mk
+++ b/config/arm64.mk
@@ -8,6 +8,7 @@ CFLAGS += #-marm -march= -mcpu= etc
HAS_PL011 := y
HAS_NS16550 := y
+HAS_MEM_ACCESS := y
# Use only if calling $(LD) directly.
LDFLAGS_DIRECT += -EL
diff --git a/config/x86_32.mk b/config/x86_32.mk
index 7f76b25..5906e60 100644
--- a/config/x86_32.mk
+++ b/config/x86_32.mk
@@ -6,6 +6,8 @@ CONFIG_HVM := y
CONFIG_MIGRATE := y
CONFIG_XCUTILS := y
+HAS_MEM_ACCESS := y
+
CFLAGS += -m32 -march=i686
# Use only if calling $(LD) directly.
diff --git a/config/x86_64.mk b/config/x86_64.mk
index 11104bd..b000c4e 100644
--- a/config/x86_64.mk
+++ b/config/x86_64.mk
@@ -7,6 +7,8 @@ CONFIG_HVM := y
CONFIG_MIGRATE := y
CONFIG_XCUTILS := y
+HAS_MEM_ACCESS := y
+
CONFIG_XEN_INSTALL_SUFFIX := .gz
CFLAGS += -m64
diff --git a/tools/tests/xen-access/Makefile b/tools/tests/xen-access/Makefile
index 65eef99..5056972 100644
--- a/tools/tests/xen-access/Makefile
+++ b/tools/tests/xen-access/Makefile
@@ -7,8 +7,13 @@ CFLAGS += $(CFLAGS_libxenctrl)
CFLAGS += $(CFLAGS_libxenguest)
CFLAGS += $(CFLAGS_xeninclude)
-TARGETS-y :=
-TARGETS-$(CONFIG_X86) += xen-access
+CFLAGS-y :=
+CFLAGS-$(CONFIG_X86) := -DCONFIG_X86
+CFLAGS-$(CONFIG_ARM) := -DCONFIG_ARM
+CFLAGS += $(CFLAGS-y)
+
+TARGETS-y :=
+TARGETS-$(HAS_MEM_ACCESS) := xen-access
TARGETS := $(TARGETS-y)
.PHONY: all
diff --git a/tools/tests/xen-access/xen-access.c b/tools/tests/xen-access/xen-access.c
index 6cb382d..40a7143 100644
--- a/tools/tests/xen-access/xen-access.c
+++ b/tools/tests/xen-access/xen-access.c
@@ -41,22 +41,16 @@
#include <xenctrl.h>
#include <xen/mem_event.h>
-#define DPRINTF(a, b...) fprintf(stderr, a, ## b)
-#define ERROR(a, b...) fprintf(stderr, a "\n", ## b)
-#define PERROR(a, b...) fprintf(stderr, a ": %s\n", ## b, strerror(errno))
-
-/* Spinlock and mem event definitions */
-
-#define SPIN_LOCK_UNLOCKED 0
+#ifdef CONFIG_X86
+#define START_PFN 0ULL
#define ADDR (*(volatile long *) addr)
+
/**
* test_and_set_bit - Set a bit and return its old value
* @nr: Bit to set
* @addr: Address to count from
*
- * This operation is atomic and cannot be reordered.
- * It also implies a memory barrier.
*/
static inline int test_and_set_bit(int nr, volatile void *addr)
{
@@ -69,6 +63,43 @@ static inline int test_and_set_bit(int nr, volatile void *addr)
return oldbit;
}
+#elif CONFIG_ARM
+
+#include <xen/arch-arm.h>
+
+#define PAGE_SHIFT 12
+#define START_PFN (GUEST_RAM0_BASE >> PAGE_SHIFT)
+#define BITS_PER_WORD 32
+#define BIT_MASK(nr) (1UL << ((nr) % BITS_PER_WORD))
+#define BIT_WORD(nr) ((nr) / BITS_PER_WORD)
+
+/**
+ * test_and_set_bit - Set a bit and return its old value
+ * @nr: Bit to set
+ * @addr: Address to count from
+ *
+ */
+static inline int test_and_set_bit(int nr, volatile void *addr)
+{
+ unsigned int mask = BIT_MASK(nr);
+ volatile unsigned int *p =
+ ((volatile unsigned int *)addr) + BIT_WORD(nr);
+ unsigned int old = *p;
+
+ *p = old | mask;
+ return (old & mask) != 0;
+}
+
+#endif
+
+#define DPRINTF(a, b...) fprintf(stderr, a, ## b)
+#define ERROR(a, b...) fprintf(stderr, a "\n", ## b)
+#define PERROR(a, b...) fprintf(stderr, a ": %s\n", ## b, strerror(errno))
+
+/* Spinlock and mem event definitions */
+
+#define SPIN_LOCK_UNLOCKED 0
+
typedef int spinlock_t;
static inline void spin_lock(spinlock_t *lock)
@@ -108,7 +139,7 @@ typedef struct mem_event {
typedef struct xenaccess {
xc_interface *xc_handle;
- xc_domaininfo_t *domain_info;
+ int max_gpfn;
mem_event_t mem_event;
} xenaccess_t;
@@ -212,7 +243,6 @@ int xenaccess_teardown(xc_interface *xch, xenaccess_t *xenaccess)
}
xenaccess->xc_handle = NULL;
- free(xenaccess->domain_info);
free(xenaccess);
return 0;
@@ -293,23 +323,17 @@ xenaccess_t *xenaccess_init(xc_interface **xch_r, domid_t domain_id)
(mem_event_sring_t *)xenaccess->mem_event.ring_page,
XC_PAGE_SIZE);
- /* Get domaininfo */
- xenaccess->domain_info = malloc(sizeof(xc_domaininfo_t));
- if ( xenaccess->domain_info == NULL )
- {
- ERROR("Error allocating memory for domain info");
- goto err;
- }
+ /* Get max_gpfn */
+ xenaccess->max_gpfn = xc_domain_maximum_gpfn(xenaccess->xc_handle,
+ xenaccess->mem_event.domain_id);
- rc = xc_domain_getinfolist(xenaccess->xc_handle, domain_id, 1,
- xenaccess->domain_info);
- if ( rc != 1 )
+ if ( xenaccess->max_gpfn < 0 )
{
- ERROR("Error getting domain info");
+ ERROR("Failed to get max gpfn");
goto err;
}
- DPRINTF("max_pages = %"PRIx64"\n", xenaccess->domain_info->max_pages);
+ DPRINTF("max_gpfn = %"PRIx32"\n", xenaccess->max_gpfn);
return xenaccess;
@@ -492,8 +516,9 @@ int main(int argc, char *argv[])
goto exit;
}
- rc = xc_set_mem_access(xch, domain_id, default_access, 0,
- xenaccess->domain_info->max_pages);
+ rc = xc_set_mem_access(xch, domain_id, default_access, START_PFN,
+ (xenaccess->max_gpfn - START_PFN) );
+
if ( rc < 0 )
{
ERROR("Error %d setting all memory to access type %d\n", rc,
@@ -520,8 +545,8 @@ int main(int argc, char *argv[])
/* Unregister for every event */
rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx, ~0ull, 0);
- rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx, 0,
- xenaccess->domain_info->max_pages);
+ rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx, START_PFN,
+ (xenaccess->max_gpfn - START_PFN) );
rc = xc_hvm_param_set(xch, domain_id, HVM_PARAM_MEMORY_EVENT_INT3, HVMPME_mode_disabled);
shutting_down = 1;
diff --git a/xen/arch/arm/Rules.mk b/xen/arch/arm/Rules.mk
index f6781b5..8658176 100644
--- a/xen/arch/arm/Rules.mk
+++ b/xen/arch/arm/Rules.mk
@@ -10,7 +10,6 @@ HAS_DEVICE_TREE := y
HAS_VIDEO := y
HAS_ARM_HDLCD := y
HAS_PASSTHROUGH := y
-HAS_MEM_ACCESS := y
CFLAGS += -I$(BASEDIR)/include
diff --git a/xen/arch/x86/Rules.mk b/xen/arch/x86/Rules.mk
index bd4e342..576985e 100644
--- a/xen/arch/x86/Rules.mk
+++ b/xen/arch/x86/Rules.mk
@@ -12,7 +12,6 @@ HAS_NS16550 := y
HAS_EHCI := y
HAS_KEXEC := y
HAS_GDBSX := y
-HAS_MEM_ACCESS := y
xenoprof := y
#
--
2.1.0
^ permalink raw reply related [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-17 20:51 ` [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl " Tamas K Lengyel
@ 2014-09-18 9:18 ` Jan Beulich
2014-09-18 10:36 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Jan Beulich @ 2014-09-18 9:18 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Jan Beulich <jbeulich@suse.com>
with one minor remark:
> --- a/xen/include/asm-arm/p2m.h
> +++ b/xen/include/asm-arm/p2m.h
> @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain *d, unsigned long frame,
> bool_t writeable);
> int arch_grant_unmap_page_identity(struct domain *d, unsigned long frame);
>
> +/* get host p2m table */
> +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
There's a superfluous pair of parentheses here.
Jan
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
` (20 preceding siblings ...)
2014-09-17 20:51 ` [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access " Tamas K Lengyel
@ 2014-09-18 9:21 ` Tim Deegan
21 siblings, 0 replies; 46+ messages in thread
From: Tim Deegan @ 2014-09-18 9:21 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, jbeulich, dgdegra
At 22:51 +0200 on 17 Sep (1410990663), Tamas K Lengyel wrote:
> The ARM virtualization extension provides 2-stage paging, a similar mechanisms
> to Intel's EPT, which can be used to trace the memory accesses performed by
> the guest systems. This series moves the mem_access and mem_event codebase
> into Xen common, performs some code cleanup and architecture specific division
> of components, then sets up the necessary infrastructure in the ARM code
> to deliver the event on R/W/X traps. Finally, we turn on the compilation of
> mem_access and mem_event on ARM and perform the necessary changes to the tools side.
>
> This version of the series has been fully tested and is functional on both an
> Arndale board and on Intel hardware.
>
> This PATCH series is also available at:
> https://github.com/tklengyel/xen/tree/arm_memaccess7
Patches 1-6, 8-13 Acked-by: Tim Deegan <tim@xen.org>
I'll respond to patch 7 separately.
Cheers,
Tim.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order
2014-09-17 20:51 ` [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order Tamas K Lengyel
@ 2014-09-18 9:22 ` Jan Beulich
2014-09-18 10:52 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Jan Beulich @ 2014-09-18 9:22 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> We swap the order of the enum of types n ... rwx, as to have rwx at 0. This
> allows
> us to memset the permission store to 0, without having to specify rwx as the
> default
> permission when memaccess is not in use. This also has performance benefits
> for
> non-memaccess paths, as now comparison is to 0 when checking if memaccess is
> in use,
> which is usually faster.
I'd say often instead of usually here.
And then the description leaves open whether any dependency on
rwx now being first is getting introduced (no memset() is being added
here, yet wherever an explicit one would get added a respective
BUILD_BUG_ON() should be put along with it).
Jan
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it.
2014-09-17 20:51 ` [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it Tamas K Lengyel
@ 2014-09-18 9:28 ` Jan Beulich
2014-09-18 10:39 ` Tim Deegan
1 sibling, 0 replies; 46+ messages in thread
From: Jan Beulich @ 2014-09-18 9:28 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> --- a/xen/include/xen/mm.h
> +++ b/xen/include/xen/mm.h
> @@ -131,6 +131,10 @@ struct npfec {
> #define MAX_ORDER 20 /* 2^20 contiguous pages */
> #endif
>
> +#define PAGE_ORDER_4K 0
> +#define PAGE_ORDER_2M 9
> +#define PAGE_ORDER_1G 18
> +
Are these really defines useful in a architecture independent manner?
I think if specific page sizes warrant such definitions, they ought to
e.g. live in the per-arch config.h. And their use in common code is
of course very bogus (implying a certain granularity of the underlying
(v)MMU).
Jan
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access
2014-09-17 20:51 ` [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access Tamas K Lengyel
@ 2014-09-18 9:35 ` Jan Beulich
2014-09-18 9:41 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Jan Beulich @ 2014-09-18 9:35 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
and I think this should really be committed as soon as possible, without
regard to the rest of this series.
Jan
> --- a/xen/arch/x86/mm/p2m.c
> +++ b/xen/arch/x86/mm/p2m.c
> @@ -1402,7 +1402,7 @@ int p2m_get_mem_access(struct domain *d, unsigned long pfn,
> mfn_t mfn;
>
> static const xenmem_access_t memaccess[] = {
> -#define ACCESS(ac) [XENMEM_access_##ac] = XENMEM_access_##ac
> +#define ACCESS(ac) [p2m_access_##ac] = XENMEM_access_##ac
> ACCESS(n),
> ACCESS(r),
> ACCESS(w),
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access
2014-09-18 9:35 ` Jan Beulich
@ 2014-09-18 9:41 ` Tamas K Lengyel
0 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 9:41 UTC (permalink / raw)
To: Jan Beulich
Cc: Ian Campbell, Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Daniel De Graaf,
Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 1068 bytes --]
On Thu, Sep 18, 2014 at 11:35 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
>
> Reviewed-by: Jan Beulich <jbeulich@suse.com>
>
> and I think this should really be committed as soon as possible, without
> regard to the rest of this series.
>
> Jan
>
Agree. It's benign right now as p2m_access is 1:1 mapped with
XENMEM_access, but that's just by luck.
Tamas
>
> > --- a/xen/arch/x86/mm/p2m.c
> > +++ b/xen/arch/x86/mm/p2m.c
> > @@ -1402,7 +1402,7 @@ int p2m_get_mem_access(struct domain *d, unsigned
> long pfn,
> > mfn_t mfn;
> >
> > static const xenmem_access_t memaccess[] = {
> > -#define ACCESS(ac) [XENMEM_access_##ac] = XENMEM_access_##ac
> > +#define ACCESS(ac) [p2m_access_##ac] = XENMEM_access_##ac
> > ACCESS(n),
> > ACCESS(r),
> > ACCESS(w),
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
>
[-- Attachment #1.2: Type: text/html, Size: 2142 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-18 9:18 ` Jan Beulich
@ 2014-09-18 10:36 ` Tamas K Lengyel
2014-09-18 10:37 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 10:36 UTC (permalink / raw)
To: Jan Beulich
Cc: Ian Campbell, Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Daniel De Graaf,
Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 827 bytes --]
On Thu, Sep 18, 2014 at 11:18 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
>
> Acked-by: Jan Beulich <jbeulich@suse.com>
> with one minor remark:
>
> > --- a/xen/include/asm-arm/p2m.h
> > +++ b/xen/include/asm-arm/p2m.h
> > @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain *d,
> unsigned long frame,
> > bool_t writeable);
> > int arch_grant_unmap_page_identity(struct domain *d, unsigned long
> frame);
> >
> > +/* get host p2m table */
> > +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
>
> There's a superfluous pair of parentheses here.
>
> Jan
>
Ack, I find it more readable to have &() when dereferencing a pointer
within the parenthesis.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1530 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-18 10:36 ` Tamas K Lengyel
@ 2014-09-18 10:37 ` Tamas K Lengyel
2014-09-18 11:17 ` Tim Deegan
2014-09-18 12:34 ` Jan Beulich
0 siblings, 2 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 10:37 UTC (permalink / raw)
To: Jan Beulich
Cc: Ian Campbell, Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Daniel De Graaf,
Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 1000 bytes --]
On Thu, Sep 18, 2014 at 12:36 PM, Tamas K Lengyel <
tamas.lengyel@zentific.com> wrote:
>
>
> On Thu, Sep 18, 2014 at 11:18 AM, Jan Beulich <JBeulich@suse.com> wrote:
>
>> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
>> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
>>
>> Acked-by: Jan Beulich <jbeulich@suse.com>
>> with one minor remark:
>>
>> > --- a/xen/include/asm-arm/p2m.h
>> > +++ b/xen/include/asm-arm/p2m.h
>> > @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain *d,
>> unsigned long frame,
>> > bool_t writeable);
>> > int arch_grant_unmap_page_identity(struct domain *d, unsigned long
>> frame);
>> >
>> > +/* get host p2m table */
>> > +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
>>
>> There's a superfluous pair of parentheses here.
>>
>> Jan
>>
>
> Ack, I find it more readable to have &() when dereferencing a pointer
> within the parenthesis.
>
> Tamas
>
Ah, you mean around (d), I see it now.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 2140 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it.
2014-09-17 20:51 ` [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it Tamas K Lengyel
2014-09-18 9:28 ` Jan Beulich
@ 2014-09-18 10:39 ` Tim Deegan
2014-09-18 10:47 ` Tamas K Lengyel
1 sibling, 1 reply; 46+ messages in thread
From: Tim Deegan @ 2014-09-18 10:39 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, jbeulich, dgdegra
Hi,
Thanks for the patch. Tidying this up is a great idea but I'm afraid
it can't be done quite this way.
At 22:51 +0200 on 17 Sep (1410990670), Tamas K Lengyel wrote:
> The p2m_mem_access_check was used only to automatically promote the
> special-case mem_access rights and to allocate the mem_access message.
> In this patch we abstract the violation check into a common mem_access_check
> function and change its return value to int.
>
> Return with rc < 0 means the fault wasn't triggered by mem_access permissions.
> Return with rc == 0 it was but rights were not automatically promoted.
> Return with rc > 0 means mem_access rights had been automatically promoted.
>
> As part of this patch, we relocate the x86/mm/mm-locks.h header into asm
> as to be able to create a generic inlined p2m_gpfn_lock/unlock wrapper
> that each arch can define based on its locking scheme.
I'm afraid not. mm-locks.h is hidden away in arch/x86/mm
deliberately, to stop p2m internals from leaking out into the rest of
the code and making it harder to reason about. So the fact that you
had to move it suggests that we need to think about using cleaner
interfaces between the code you're extracting and the rest of the p2m
code...
> +int mem_access_check(paddr_t gpa, unsigned long gla, struct npfec npfec)
> +{
> + bool_t violation;
> + mfn_t mfn;
> + p2m_access_t p2ma;
> + p2m_type_t p2mt;
> + mem_event_request_t *req;
> + int rc;
> + struct vcpu *v = current;
> + unsigned long gfn = gpa >> PAGE_SHIFT;
> + struct domain *d = v->domain;
> + struct p2m_domain* p2m = p2m_get_hostp2m(d);
> +
> + /* First, handle rx2rw conversion automatically.
> + * These calls to p2m->set_entry() must succeed: we have the gfn
> + * locked and just did a successful get_entry(). */
> + p2m_gpfn_lock(p2m, gfn);
> + mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
Neither of these (direct access to a p2m lock or direct call of a p2m
function pointer) should happen outside the p2m code.
>From an outside caller you _could_ use get_gfn_type_access(p2m, gfn,
&p2mt, &p2ma, 0, NULL) instead (and put_gfn() instead of
p2m_gpfn_unlock()). But directly calling p2m->set_entry() (or
p2m_set_entry()) is still not OK.
I think it would be better to leave an arch-specific
"bool_t p2m_mem_access_check(p2m, gfn, npfec, &p2ma)" that does the
lookup and the automatic promotions and returns the access type. It
should also do the check for missing listeners and automatically reset
p2m entries (i.e. the second call to ->set_entry() below). Then this
function won't need to take any locks or get_gfn calls at all -- it
will just have the check for non-vioating faults, the call to
p2m_mem_access_check() that might ajdust entries, and the code to
prepare a req.
> +
> + /* Check if the access is against the permissions. */
> + switch ( p2ma )
> + {
> + case p2m_access_rwx:
> + violation = 0;
> + break;
> + case p2m_access_rw:
> + violation = npfec.insn_fetch;
> + break;
> + case p2m_access_wx:
> + violation = npfec.read_access;
> + break;
> + case p2m_access_rx:
> + case p2m_access_rx2rw:
> + violation = npfec.write_access;
> + break;
> + case p2m_access_x:
> + violation = npfec.read_access || npfec.write_access;
> + break;
> + case p2m_access_w:
> + violation = npfec.read_access || npfec.insn_fetch;
> + break;
> + case p2m_access_r:
> + violation = npfec.write_access || npfec.insn_fetch;
> + break;
> + case p2m_access_n:
> + case p2m_access_n2rwx:
> + default:
> + violation = npfec.read_access || npfec.write_access || npfec.insn_fetch;
> + break;
> + }
In any case this block, which doesn't depend on the p2m contents at
all, should happen before any locks or lookups. That way you can
avoid taking the locks at all for non-violating faults (as the
original code does).
> + /* If no violation is found here, it needs to be reinjected. */
> + if ( !violation )
> + {
> + p2m_gpfn_unlock(p2m, gfn);
> + return -EFAULT;
> + }
> +
> + /* Check for automatic setting promotion. */
> + if ( npfec.write_access && p2ma == p2m_access_rx2rw )
> + {
> + rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K, p2mt, p2m_access_rw);
> + ASSERT(rc == 0);
> + p2m_gpfn_unlock(p2m, gfn);
> + return 1;
> + }
> + else if ( p2ma == p2m_access_n2rwx )
> + {
> + ASSERT(npfec.write_access || npfec.read_access || npfec.insn_fetch);
> + rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
> + p2mt, p2m_access_rwx);
> + ASSERT(rc == 0);
> + }
> + p2m_gpfn_unlock(p2m, gfn);
> +
> + /* Otherwise, check if there is a memory event listener, and send the message along */
> + if ( !mem_event_check_ring(&d->mem_event->access) )
> + {
> + /* No listener */
> + if ( p2m->access_required )
> + {
> + gdprintk(XENLOG_INFO, "Memory access permissions failure, "
> + "no mem_event listener VCPU %d, dom %d\n",
> + v->vcpu_id, d->domain_id);
> + domain_crash(v->domain);
> + return -EFAULT;
> + }
> + else
> + {
> + p2m_gpfn_lock(p2m, gfn);
> + mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
> + if ( p2ma != p2m_access_n2rwx )
> +
> + {
> + /* A listener is not required, so clear the access
> + * restrictions. This set must succeed: we have the
> + * gfn locked and just did a successful get_entry(). */
> + rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
> + p2mt, p2m_access_rwx);
> + ASSERT(rc == 0);
> + }
> + p2m_gpfn_unlock(p2m, gfn);
> + return 1;
> + }
> + }
> +
> + req = xzalloc(mem_event_request_t);
> + if ( req )
> + {
> + req->reason = MEM_EVENT_REASON_VIOLATION;
> +
> + /* Pause the current VCPU */
> + if ( p2ma != p2m_access_n2rwx )
> + req->flags |= MEM_EVENT_FLAG_VCPU_PAUSED;
> +
> + /* Send request to mem event */
> + req->gfn = gfn;
> + req->offset = gpa & ((1 << PAGE_SHIFT) - 1);
> + req->gla_valid = npfec.gla_valid;
> + req->gla = gla;
> + if ( npfec.kind == npfec_kind_with_gla )
> + req->fault_with_gla = 1;
> + else if ( npfec.kind == npfec_kind_in_gpt )
> + req->fault_in_gpt = 1;
> + req->access_r = npfec.read_access;
> + req->access_w = npfec.write_access;
> + req->access_x = npfec.insn_fetch;
> + req->vcpu_id = v->vcpu_id;
> +
> + mem_access_send_req(v->domain, req);
The original function returned the req for the caller to send, and the
caller delayed sending it until the end of hvm_hap_nested_page_fault(),
when the last locks are dropped. You need to keep this behaviour or
the system can deadlock.
I wonder whether we could add an ASSERT_NOT_IN_ATOMIC() at the top of
mem_event_wait_slot() to make this kind of thing easier to spot. At
the moment it will only fail if it actually has to wait.
Cheers,
Tim.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it.
2014-09-18 10:39 ` Tim Deegan
@ 2014-09-18 10:47 ` Tamas K Lengyel
2014-09-18 11:13 ` Tim Deegan
2014-09-22 12:08 ` Ian Campbell
0 siblings, 2 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 10:47 UTC (permalink / raw)
To: Tim Deegan
Cc: Ian Campbell, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 8135 bytes --]
On Thu, Sep 18, 2014 at 12:39 PM, Tim Deegan <tim@xen.org> wrote:
> Hi,
>
> Thanks for the patch. Tidying this up is a great idea but I'm afraid
> it can't be done quite this way.
>
> At 22:51 +0200 on 17 Sep (1410990670), Tamas K Lengyel wrote:
> > The p2m_mem_access_check was used only to automatically promote the
> > special-case mem_access rights and to allocate the mem_access message.
> > In this patch we abstract the violation check into a common
> mem_access_check
> > function and change its return value to int.
> >
> > Return with rc < 0 means the fault wasn't triggered by mem_access
> permissions.
> > Return with rc == 0 it was but rights were not automatically promoted.
> > Return with rc > 0 means mem_access rights had been automatically
> promoted.
> >
> > As part of this patch, we relocate the x86/mm/mm-locks.h header into asm
> > as to be able to create a generic inlined p2m_gpfn_lock/unlock wrapper
> > that each arch can define based on its locking scheme.
>
> I'm afraid not. mm-locks.h is hidden away in arch/x86/mm
> deliberately, to stop p2m internals from leaking out into the rest of
> the code and making it harder to reason about. So the fact that you
> had to move it suggests that we need to think about using cleaner
> interfaces between the code you're extracting and the rest of the p2m
> code...
>
> > +int mem_access_check(paddr_t gpa, unsigned long gla, struct npfec npfec)
> > +{
> > + bool_t violation;
> > + mfn_t mfn;
> > + p2m_access_t p2ma;
> > + p2m_type_t p2mt;
> > + mem_event_request_t *req;
> > + int rc;
> > + struct vcpu *v = current;
> > + unsigned long gfn = gpa >> PAGE_SHIFT;
> > + struct domain *d = v->domain;
> > + struct p2m_domain* p2m = p2m_get_hostp2m(d);
> > +
> > + /* First, handle rx2rw conversion automatically.
> > + * These calls to p2m->set_entry() must succeed: we have the gfn
> > + * locked and just did a successful get_entry(). */
> > + p2m_gpfn_lock(p2m, gfn);
> > + mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
>
> Neither of these (direct access to a p2m lock or direct call of a p2m
> function pointer) should happen outside the p2m code.
>
> From an outside caller you _could_ use get_gfn_type_access(p2m, gfn,
> &p2mt, &p2ma, 0, NULL) instead (and put_gfn() instead of
> p2m_gpfn_unlock()). But directly calling p2m->set_entry() (or
> p2m_set_entry()) is still not OK.
>
> I think it would be better to leave an arch-specific
> "bool_t p2m_mem_access_check(p2m, gfn, npfec, &p2ma)" that does the
> lookup and the automatic promotions and returns the access type. It
> should also do the check for missing listeners and automatically reset
> p2m entries (i.e. the second call to ->set_entry() below). Then this
> function won't need to take any locks or get_gfn calls at all -- it
> will just have the check for non-vioating faults, the call to
> p2m_mem_access_check() that might ajdust entries, and the code to
> prepare a req.
>
> > +
> > + /* Check if the access is against the permissions. */
> > + switch ( p2ma )
> > + {
> > + case p2m_access_rwx:
> > + violation = 0;
> > + break;
> > + case p2m_access_rw:
> > + violation = npfec.insn_fetch;
> > + break;
> > + case p2m_access_wx:
> > + violation = npfec.read_access;
> > + break;
> > + case p2m_access_rx:
> > + case p2m_access_rx2rw:
> > + violation = npfec.write_access;
> > + break;
> > + case p2m_access_x:
> > + violation = npfec.read_access || npfec.write_access;
> > + break;
> > + case p2m_access_w:
> > + violation = npfec.read_access || npfec.insn_fetch;
> > + break;
> > + case p2m_access_r:
> > + violation = npfec.write_access || npfec.insn_fetch;
> > + break;
> > + case p2m_access_n:
> > + case p2m_access_n2rwx:
> > + default:
> > + violation = npfec.read_access || npfec.write_access ||
> npfec.insn_fetch;
> > + break;
> > + }
>
> In any case this block, which doesn't depend on the p2m contents at
> all, should happen before any locks or lookups. That way you can
> avoid taking the locks at all for non-violating faults (as the
> original code does).
>
> > + /* If no violation is found here, it needs to be reinjected. */
> > + if ( !violation )
> > + {
> > + p2m_gpfn_unlock(p2m, gfn);
> > + return -EFAULT;
> > + }
> > +
> > + /* Check for automatic setting promotion. */
> > + if ( npfec.write_access && p2ma == p2m_access_rx2rw )
> > + {
> > + rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K, p2mt,
> p2m_access_rw);
> > + ASSERT(rc == 0);
> > + p2m_gpfn_unlock(p2m, gfn);
> > + return 1;
> > + }
> > + else if ( p2ma == p2m_access_n2rwx )
> > + {
> > + ASSERT(npfec.write_access || npfec.read_access ||
> npfec.insn_fetch);
> > + rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
> > + p2mt, p2m_access_rwx);
> > + ASSERT(rc == 0);
> > + }
> > + p2m_gpfn_unlock(p2m, gfn);
> > +
> > + /* Otherwise, check if there is a memory event listener, and send
> the message along */
> > + if ( !mem_event_check_ring(&d->mem_event->access) )
> > + {
> > + /* No listener */
> > + if ( p2m->access_required )
> > + {
> > + gdprintk(XENLOG_INFO, "Memory access permissions failure, "
> > + "no mem_event listener VCPU %d, dom
> %d\n",
> > + v->vcpu_id, d->domain_id);
> > + domain_crash(v->domain);
> > + return -EFAULT;
> > + }
> > + else
> > + {
> > + p2m_gpfn_lock(p2m, gfn);
> > + mfn = p2m->get_entry(p2m, gfn, &p2mt, &p2ma, 0, NULL);
> > + if ( p2ma != p2m_access_n2rwx )
> > +
> > + {
> > + /* A listener is not required, so clear the access
> > + * restrictions. This set must succeed: we have the
> > + * gfn locked and just did a successful get_entry(). */
> > + rc = p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K,
> > + p2mt, p2m_access_rwx);
> > + ASSERT(rc == 0);
> > + }
> > + p2m_gpfn_unlock(p2m, gfn);
> > + return 1;
> > + }
> > + }
> > +
> > + req = xzalloc(mem_event_request_t);
> > + if ( req )
> > + {
> > + req->reason = MEM_EVENT_REASON_VIOLATION;
> > +
> > + /* Pause the current VCPU */
> > + if ( p2ma != p2m_access_n2rwx )
> > + req->flags |= MEM_EVENT_FLAG_VCPU_PAUSED;
> > +
> > + /* Send request to mem event */
> > + req->gfn = gfn;
> > + req->offset = gpa & ((1 << PAGE_SHIFT) - 1);
> > + req->gla_valid = npfec.gla_valid;
> > + req->gla = gla;
> > + if ( npfec.kind == npfec_kind_with_gla )
> > + req->fault_with_gla = 1;
> > + else if ( npfec.kind == npfec_kind_in_gpt )
> > + req->fault_in_gpt = 1;
> > + req->access_r = npfec.read_access;
> > + req->access_w = npfec.write_access;
> > + req->access_x = npfec.insn_fetch;
> > + req->vcpu_id = v->vcpu_id;
> > +
> > + mem_access_send_req(v->domain, req);
>
> The original function returned the req for the caller to send, and the
> caller delayed sending it until the end of hvm_hap_nested_page_fault(),
> when the last locks are dropped. You need to keep this behaviour or
> the system can deadlock.
>
> I wonder whether we could add an ASSERT_NOT_IN_ATOMIC() at the top of
> mem_event_wait_slot() to make this kind of thing easier to spot. At
> the moment it will only fail if it actually has to wait.
>
> Cheers,
>
> Tim.
>
Ack. I think I'll drop this patch for the time being as the refactoring
work required for it to be in common mem_access is quite significant, both
in the x86 and the ARM side. And it would only result in a bit of
code-deduplication. We could pick-it up later if there is a need for it /
value in it.
Thanks,
Tamas
[-- Attachment #1.2: Type: text/html, Size: 9785 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order
2014-09-18 9:22 ` Jan Beulich
@ 2014-09-18 10:52 ` Tamas K Lengyel
0 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 10:52 UTC (permalink / raw)
To: Jan Beulich
Cc: Ian Campbell, Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Daniel De Graaf,
Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 1139 bytes --]
On Thu, Sep 18, 2014 at 11:22 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> > We swap the order of the enum of types n ... rwx, as to have rwx at 0.
> This
> > allows
> > us to memset the permission store to 0, without having to specify rwx as
> the
> > default
> > permission when memaccess is not in use. This also has performance
> benefits
> > for
> > non-memaccess paths, as now comparison is to 0 when checking if
> memaccess is
> > in use,
> > which is usually faster.
>
> I'd say often instead of usually here.
>
Ack.
>
> And then the description leaves open whether any dependency on
> rwx now being first is getting introduced (no memset() is being added
> here, yet wherever an explicit one would get added a respective
> BUILD_BUG_ON() should be put along with it).
>
> Jan
>
I guess the commit message is a bit ambiguous, the paths I see all set
mem_access explicitly afterwards to default_access, so there is no reliance
on pure memset initializing it right now. This patch really just opens the
door to have such an approach in the future (if needed).
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1840 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it.
2014-09-18 10:47 ` Tamas K Lengyel
@ 2014-09-18 11:13 ` Tim Deegan
2014-09-22 12:08 ` Ian Campbell
1 sibling, 0 replies; 46+ messages in thread
From: Tim Deegan @ 2014-09-18 11:13 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: Ian Campbell, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
At 12:47 +0200 on 18 Sep (1411040839), Tamas K Lengyel wrote:
> Ack. I think I'll drop this patch for the time being as the refactoring
> work required for it to be in common mem_access is quite significant, both
> in the x86 and the ARM side. And it would only result in a bit of
> code-deduplication. We could pick-it up later if there is a need for it /
> value in it.
OK, thanks.
Tim.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-18 10:37 ` Tamas K Lengyel
@ 2014-09-18 11:17 ` Tim Deegan
2014-09-18 11:21 ` Tamas K Lengyel
2014-09-18 12:34 ` Jan Beulich
1 sibling, 1 reply; 46+ messages in thread
From: Tim Deegan @ 2014-09-18 11:17 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: Ian Campbell, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
At 12:37 +0200 on 18 Sep (1411040252), Tamas K Lengyel wrote:
> On Thu, Sep 18, 2014 at 12:36 PM, Tamas K Lengyel <
> tamas.lengyel@zentific.com> wrote:
>
> >
> >
> > On Thu, Sep 18, 2014 at 11:18 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >
> >> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> >> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> >>
> >> Acked-by: Jan Beulich <jbeulich@suse.com>
> >> with one minor remark:
> >>
> >> > --- a/xen/include/asm-arm/p2m.h
> >> > +++ b/xen/include/asm-arm/p2m.h
> >> > @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain *d,
> >> unsigned long frame,
> >> > bool_t writeable);
> >> > int arch_grant_unmap_page_identity(struct domain *d, unsigned long
> >> frame);
> >> >
> >> > +/* get host p2m table */
> >> > +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
> >>
> >> There's a superfluous pair of parentheses here.
> >>
> >> Jan
> >>
> >
> > Ack, I find it more readable to have &() when dereferencing a pointer
> > within the parenthesis.
> >
> > Tamas
> >
>
> Ah, you mean around (d), I see it now.
No, the parens around 'd' are necessary since this is a macro
-- e.g. consider 'p2m_get_hostp2m(pred ? d1 : d2)'.
Tim.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-18 11:17 ` Tim Deegan
@ 2014-09-18 11:21 ` Tamas K Lengyel
0 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 11:21 UTC (permalink / raw)
To: Tim Deegan
Cc: Ian Campbell, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 1413 bytes --]
On Thu, Sep 18, 2014 at 1:17 PM, Tim Deegan <tim@xen.org> wrote:
> At 12:37 +0200 on 18 Sep (1411040252), Tamas K Lengyel wrote:
> > On Thu, Sep 18, 2014 at 12:36 PM, Tamas K Lengyel <
> > tamas.lengyel@zentific.com> wrote:
> >
> > >
> > >
> > > On Thu, Sep 18, 2014 at 11:18 AM, Jan Beulich <JBeulich@suse.com>
> wrote:
> > >
> > >> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> > >> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> > >>
> > >> Acked-by: Jan Beulich <jbeulich@suse.com>
> > >> with one minor remark:
> > >>
> > >> > --- a/xen/include/asm-arm/p2m.h
> > >> > +++ b/xen/include/asm-arm/p2m.h
> > >> > @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain
> *d,
> > >> unsigned long frame,
> > >> > bool_t writeable);
> > >> > int arch_grant_unmap_page_identity(struct domain *d, unsigned long
> > >> frame);
> > >> >
> > >> > +/* get host p2m table */
> > >> > +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
> > >>
> > >> There's a superfluous pair of parentheses here.
> > >>
> > >> Jan
> > >>
> > >
> > > Ack, I find it more readable to have &() when dereferencing a pointer
> > > within the parenthesis.
> > >
> > > Tamas
> > >
> >
> > Ah, you mean around (d), I see it now.
>
> No, the parens around 'd' are necessary since this is a macro
> -- e.g. consider 'p2m_get_hostp2m(pred ? d1 : d2)'.
>
> Tim.
>
Ack.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 2541 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-18 10:37 ` Tamas K Lengyel
2014-09-18 11:17 ` Tim Deegan
@ 2014-09-18 12:34 ` Jan Beulich
2014-09-18 13:02 ` Tamas K Lengyel
1 sibling, 1 reply; 46+ messages in thread
From: Jan Beulich @ 2014-09-18 12:34 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: Ian Campbell, Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Daniel De Graaf,
Tamas K Lengyel
>>> On 18.09.14 at 12:37, <tamas.lengyel@zentific.com> wrote:
> On Thu, Sep 18, 2014 at 12:36 PM, Tamas K Lengyel <
> tamas.lengyel@zentific.com> wrote:
>
>>
>>
>> On Thu, Sep 18, 2014 at 11:18 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>
>>> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
>>> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
>>>
>>> Acked-by: Jan Beulich <jbeulich@suse.com>
>>> with one minor remark:
>>>
>>> > --- a/xen/include/asm-arm/p2m.h
>>> > +++ b/xen/include/asm-arm/p2m.h
>>> > @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain *d,
>>> unsigned long frame,
>>> > bool_t writeable);
>>> > int arch_grant_unmap_page_identity(struct domain *d, unsigned long
>>> frame);
>>> >
>>> > +/* get host p2m table */
>>> > +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
>>>
>>> There's a superfluous pair of parentheses here.
>>
>> Ack, I find it more readable to have &() when dereferencing a pointer
>> within the parenthesis.
>
> Ah, you mean around (d), I see it now.
No, that one absolutely has to stay.
Jan
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl into common
2014-09-18 12:34 ` Jan Beulich
@ 2014-09-18 13:02 ` Tamas K Lengyel
0 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-18 13:02 UTC (permalink / raw)
To: Jan Beulich
Cc: Ian Campbell, Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Daniel De Graaf,
Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 1300 bytes --]
On Thu, Sep 18, 2014 at 2:34 PM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 18.09.14 at 12:37, <tamas.lengyel@zentific.com> wrote:
> > On Thu, Sep 18, 2014 at 12:36 PM, Tamas K Lengyel <
> > tamas.lengyel@zentific.com> wrote:
> >
> >>
> >>
> >> On Thu, Sep 18, 2014 at 11:18 AM, Jan Beulich <JBeulich@suse.com>
> wrote:
> >>
> >>> >>> On 17.09.14 at 22:51, <tklengyel@sec.in.tum.de> wrote:
> >>> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> >>>
> >>> Acked-by: Jan Beulich <jbeulich@suse.com>
> >>> with one minor remark:
> >>>
> >>> > --- a/xen/include/asm-arm/p2m.h
> >>> > +++ b/xen/include/asm-arm/p2m.h
> >>> > @@ -198,6 +198,9 @@ int arch_grant_map_page_identity(struct domain
> *d,
> >>> unsigned long frame,
> >>> > bool_t writeable);
> >>> > int arch_grant_unmap_page_identity(struct domain *d, unsigned long
> >>> frame);
> >>> >
> >>> > +/* get host p2m table */
> >>> > +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
> >>>
> >>> There's a superfluous pair of parentheses here.
> >>
> >> Ack, I find it more readable to have &() when dereferencing a pointer
> >> within the parenthesis.
> >
> > Ah, you mean around (d), I see it now.
>
> No, that one absolutely has to stay.
>
> Jan
>
Ack, Tim has already pointed that out.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 2434 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it.
2014-09-18 10:47 ` Tamas K Lengyel
2014-09-18 11:13 ` Tim Deegan
@ 2014-09-22 12:08 ` Ian Campbell
1 sibling, 0 replies; 46+ messages in thread
From: Ian Campbell @ 2014-09-22 12:08 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
On Thu, 2014-09-18 at 12:47 +0200, Tamas K Lengyel wrote:
> Ack. I think I'll drop this patch for the time being as the
> refactoring work required for it to be in common mem_access is quite
> significant, both in the x86 and the ARM side. And it would only
> result in a bit of code-deduplication. We could pick-it up later if
> there is a need for it / value in it.
As the one who asked if this could be made common this is fine with me
after seeing this subthread.
Ian.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support
2014-09-17 20:51 ` [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support Tamas K Lengyel
@ 2014-09-22 12:20 ` Ian Campbell
2014-09-22 17:16 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Ian Campbell @ 2014-09-22 12:20 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: tim, julien.grall, ian.jackson, xen-devel, stefano.stabellini,
andres, jbeulich, dgdegra
On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> Add p2m_access_t to struct page_info and add necessary changes for
> page table construction routines to pass the default access information.
> We store the p2m_access_t info in page_info as the PTE lacks enough
> software programmable bits.
>
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> ---
> v7: - Remove radix tree init/destroy and move p2m_access_t store to page_info.
I'm afraid I don't think this will work, since a single page can be
mapped into multiple guests (e.g. granted), possibly multiple times per
guest (less likely I admit) and AIUI the access type is at least per
guest if not per p2m mapping.
The radix tree would have been tolerable if there was some sort of quick
to check flag in p2m->foo to indicate if it is in use for a given
domain. (I previously suggested a sentinal p2m_access_t value for
example, but a bool etc would work too).
I don't especially like trusting radix_lookup to be cheap for an empty
tree in such a hotpatch as p2m lookup since a) I couldn't trivially
determine that it was cheap in such cases by inspection and b) we then
become dependent on the radix tree implementation.
> - Add p2m_gpfn_lock/unlock functions.
> - Add bool_t lock input to p2m_lookup and apply_p2m_changes so the caller
> can specify if locking should be performed. This is needed in order to
> support mem_access_check from common.
Based on the review of the patch which added mem_access_check resulting
in it not being refactored I expect most of this patch will be changing
again, so I won't look at it further unless you ask me to. (except to
note for future reference that if you have a bool_t parameter you should
genherally pass true/false/bool-expr not 0/1)
Ian.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers.
2014-09-17 20:51 ` [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers Tamas K Lengyel
@ 2014-09-22 12:24 ` Ian Campbell
2014-09-22 18:24 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Ian Campbell @ 2014-09-22 12:24 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: tim, julien.grall, ian.jackson, xen-devel, stefano.stabellini,
andres, jbeulich, dgdegra
On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Although you could consider looking up level_shift in p2m_shatter_page
rather than passing it. Either way the ack stands.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access on ARM
2014-09-17 20:51 ` [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access " Tamas K Lengyel
@ 2014-09-22 12:29 ` Ian Campbell
2014-09-22 17:11 ` Tamas K Lengyel
0 siblings, 1 reply; 46+ messages in thread
From: Ian Campbell @ 2014-09-22 12:29 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: tim, julien.grall, ian.jackson, xen-devel, stefano.stabellini,
andres, jbeulich, dgdegra
On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> diff --git a/config/arm32.mk b/config/arm32.mk
> index aa79d22..4a7c259 100644
> --- a/config/arm32.mk
> +++ b/config/arm32.mk
> @@ -13,6 +13,7 @@ HAS_PL011 := y
> HAS_EXYNOS4210 := y
> HAS_OMAP := y
> HAS_NS16550 := y
> +HAS_MEM_ACCESS := y
I think this (and the others) duplicate the one added to arch/*/Rules.mk
in a previous patch. I think you can get away with only this one (since
it is ultimately included I think).
For x86 I'd suggest doing the MEM_PAGING one here too for consistency.
Ian.
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access on ARM
2014-09-22 12:29 ` Ian Campbell
@ 2014-09-22 17:11 ` Tamas K Lengyel
2014-09-23 8:37 ` Ian Campbell
0 siblings, 1 reply; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-22 17:11 UTC (permalink / raw)
To: Ian Campbell
Cc: Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 761 bytes --]
On Mon, Sep 22, 2014 at 2:29 PM, Ian Campbell <Ian.Campbell@citrix.com>
wrote:
> On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> > diff --git a/config/arm32.mk b/config/arm32.mk
> > index aa79d22..4a7c259 100644
> > --- a/config/arm32.mk
> > +++ b/config/arm32.mk
> > @@ -13,6 +13,7 @@ HAS_PL011 := y
> > HAS_EXYNOS4210 := y
> > HAS_OMAP := y
> > HAS_NS16550 := y
> > +HAS_MEM_ACCESS := y
>
> I think this (and the others) duplicate the one added to arch/*/Rules.mk
> in a previous patch. I think you can get away with only this one (since
> it is ultimately included I think).
>
> For x86 I'd suggest doing the MEM_PAGING one here too for consistency.
>
> Ian.
Ack, what about MEM_SHARING? I think that should be defined the same way.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1448 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support
2014-09-22 12:20 ` Ian Campbell
@ 2014-09-22 17:16 ` Tamas K Lengyel
0 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-22 17:16 UTC (permalink / raw)
To: Ian Campbell
Cc: Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 2333 bytes --]
On Mon, Sep 22, 2014 at 2:20 PM, Ian Campbell <Ian.Campbell@citrix.com>
wrote:
> On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> > Add p2m_access_t to struct page_info and add necessary changes for
> > page table construction routines to pass the default access information.
> > We store the p2m_access_t info in page_info as the PTE lacks enough
> > software programmable bits.
> >
> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> > ---
> > v7: - Remove radix tree init/destroy and move p2m_access_t store to
> page_info.
>
> I'm afraid I don't think this will work, since a single page can be
> mapped into multiple guests (e.g. granted), possibly multiple times per
> guest (less likely I admit) and AIUI the access type is at least per
> guest if not per p2m mapping.
>
Ah, good point. I'll bring back the radix tree for now and we might revisit
this approach later. I really liked the fast hardware-based lookup so some
solution may still be viable in this general area.
>
> The radix tree would have been tolerable if there was some sort of quick
> to check flag in p2m->foo to indicate if it is in use for a given
> domain. (I previously suggested a sentinal p2m_access_t value for
> example, but a bool etc would work too).
>
> I don't especially like trusting radix_lookup to be cheap for an empty
> tree in such a hotpatch as p2m lookup since a) I couldn't trivially
> determine that it was cheap in such cases by inspection and b) we then
> become dependent on the radix tree implementation.
>
Ack, the boolean value should be acceptable to indicate if the tree is
empty or not.
>
> > - Add p2m_gpfn_lock/unlock functions.
> > - Add bool_t lock input to p2m_lookup and apply_p2m_changes so the
> caller
> > can specify if locking should be performed. This is needed in
> order to
> > support mem_access_check from common.
>
> Based on the review of the patch which added mem_access_check resulting
> in it not being refactored I expect most of this patch will be changing
> again, so I won't look at it further unless you ask me to. (except to
> note for future reference that if you have a bool_t parameter you should
> genherally pass true/false/bool-expr not 0/1)
>
> Ian.
>
Thanks, yes, this patch will change soon, expecting to send new revision
tomorrow.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 3306 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers.
2014-09-22 12:24 ` Ian Campbell
@ 2014-09-22 18:24 ` Tamas K Lengyel
0 siblings, 0 replies; 46+ messages in thread
From: Tamas K Lengyel @ 2014-09-22 18:24 UTC (permalink / raw)
To: Ian Campbell
Cc: Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 433 bytes --]
On Mon, Sep 22, 2014 at 2:24 PM, Ian Campbell <Ian.Campbell@citrix.com>
wrote:
> On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
>
> Acked-by: Ian Campbell <ian.campbell@citrix.com>
>
> Although you could consider looking up level_shift in p2m_shatter_page
> rather than passing it. Either way the ack stands.
>
Thanks, will do so in the next revision!
Tamas
[-- Attachment #1.2: Type: text/html, Size: 920 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 46+ messages in thread
* Re: [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access on ARM
2014-09-22 17:11 ` Tamas K Lengyel
@ 2014-09-23 8:37 ` Ian Campbell
0 siblings, 0 replies; 46+ messages in thread
From: Ian Campbell @ 2014-09-23 8:37 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: Tim Deegan, Julien Grall, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
On Mon, 2014-09-22 at 19:11 +0200, Tamas K Lengyel wrote:
>
>
> On Mon, Sep 22, 2014 at 2:29 PM, Ian Campbell
> <Ian.Campbell@citrix.com> wrote:
> On Wed, 2014-09-17 at 22:51 +0200, Tamas K Lengyel wrote:
> > diff --git a/config/arm32.mk b/config/arm32.mk
> > index aa79d22..4a7c259 100644
> > --- a/config/arm32.mk
> > +++ b/config/arm32.mk
> > @@ -13,6 +13,7 @@ HAS_PL011 := y
> > HAS_EXYNOS4210 := y
> > HAS_OMAP := y
> > HAS_NS16550 := y
> > +HAS_MEM_ACCESS := y
>
> I think this (and the others) duplicate the one added to
> arch/*/Rules.mk
> in a previous patch. I think you can get away with only this
> one (since
> it is ultimately included I think).
>
> For x86 I'd suggest doing the MEM_PAGING one here too for
> consistency.
>
> Ian.
>
>
> Ack, what about MEM_SHARING? I think that should be defined the same
> way.
I forgot that one, I agree they should be all in the same place.
Ian.
^ permalink raw reply [flat|nested] 46+ messages in thread
end of thread, other threads:[~2014-09-23 8:37 UTC | newest]
Thread overview: 46+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-09-17 20:51 [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 01/21] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 02/21] xen: Relocate struct npfec definition " Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 03/21] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 04/21] xen: Relocate set_access_required domctl " Tamas K Lengyel
2014-09-18 9:18 ` Jan Beulich
2014-09-18 10:36 ` Tamas K Lengyel
2014-09-18 10:37 ` Tamas K Lengyel
2014-09-18 11:17 ` Tim Deegan
2014-09-18 11:21 ` Tamas K Lengyel
2014-09-18 12:34 ` Jan Beulich
2014-09-18 13:02 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 05/21] xen: Relocate p2m_access_t into common and swap the order Tamas K Lengyel
2014-09-18 9:22 ` Jan Beulich
2014-09-18 10:52 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 06/21] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 07/21] xen: Relocate p2m_mem_access_check into common and refactor it Tamas K Lengyel
2014-09-18 9:28 ` Jan Beulich
2014-09-18 10:39 ` Tim Deegan
2014-09-18 10:47 ` Tamas K Lengyel
2014-09-18 11:13 ` Tim Deegan
2014-09-22 12:08 ` Ian Campbell
2014-09-17 20:51 ` [PATCH for-4.5 v7 08/21] x86/p2m: Typo fix for spelling ambiguous Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 09/21] x86/p2m: Fix conversion macro of p2m_access to XENMEM_access Tamas K Lengyel
2014-09-18 9:35 ` Jan Beulich
2014-09-18 9:41 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 10/21] xen/mem_event: Clean out superfluous white-spaces Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 11/21] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 12/21] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 13/21] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 14/21] xen/arm: p2m changes for mem_access support Tamas K Lengyel
2014-09-22 12:20 ` Ian Campbell
2014-09-22 17:16 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 15/21] xen/arm: Implement domain_get_maximum_gpfn Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 16/21] xen/arm: Add p2m_set_permission and p2m_shatter_page helpers Tamas K Lengyel
2014-09-22 12:24 ` Ian Campbell
2014-09-22 18:24 ` Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 17/21] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 18/21] xen/arm: Instruction prefetch abort (X) mem_event handling Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 19/21] xen/arm: Enable the compilation of mem_access and mem_event on ARM Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 20/21] tools/libxc: Allocate magic page for mem access " Tamas K Lengyel
2014-09-17 20:51 ` [PATCH for-4.5 v7 21/21] tools/tests: Enable xen-access " Tamas K Lengyel
2014-09-22 12:29 ` Ian Campbell
2014-09-22 17:11 ` Tamas K Lengyel
2014-09-23 8:37 ` Ian Campbell
2014-09-18 9:21 ` [PATCH for-4.5 v7 00/21] Mem_event and mem_access for ARM Tim Deegan
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.