* [PATCH v4 00/16] Mem_event and mem_access for ARM
@ 2014-09-05 8:58 Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
` (15 more replies)
0 siblings, 16 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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 an
Arndale board.
This PATCH series is also available at:
https://github.com/tklengyel/xen/tree/arm_memaccess4
Tamas K Lengyel (16):
xen: Relocate mem_access and mem_event into common.
xen: Relocate p2m_mem_access_resume to mem_access common
xen: Relocate struct npfec definition into common
xen: Relocate mem_event_op domctl and access_op memop into common.
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 type definitions and changes
xen/arm: Add set access required domctl
xen/arm: Data abort exception (R/W) mem_events.
xen/arm: Instruction prefetch abort (X) mem_event handling
xen/arm: Shatter large pages when using mem_acces
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 | 94 ++++-
xen/Rules.mk | 1 +
xen/arch/arm/domctl.c | 13 +
xen/arch/arm/p2m.c | 509 ++++++++++++++++++++++---
xen/arch/arm/traps.c | 73 +++-
xen/arch/x86/domctl.c | 10 +-
xen/arch/x86/hvm/hvm.c | 61 +--
xen/arch/x86/mm/Makefile | 2 -
xen/arch/x86/mm/hap/nested_ept.c | 2 +-
xen/arch/x86/mm/hap/nested_hap.c | 2 +-
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 | 2 +-
xen/arch/x86/mm/p2m-pod.c | 2 +-
xen/arch/x86/mm/p2m-pt.c | 2 +-
xen/arch/x86/mm/p2m.c | 26 +-
xen/arch/x86/x86_64/compat/mm.c | 4 +-
xen/arch/x86/x86_64/mm.c | 8 +-
xen/common/Makefile | 2 +
xen/common/domain.c | 1 +
xen/common/domctl.c | 7 +
xen/common/mem_access.c | 157 ++++++++
xen/common/mem_event.c | 724 ++++++++++++++++++++++++++++++++++++
xen/common/memory.c | 68 ++++
xen/include/asm-arm/mm.h | 1 -
xen/include/asm-arm/p2m.h | 109 +++++-
xen/include/asm-arm/processor.h | 70 +++-
xen/include/asm-x86/config.h | 6 +
xen/include/asm-x86/hvm/hvm.h | 7 +-
xen/include/asm-x86/mem_access.h | 39 --
xen/include/asm-x86/mem_event.h | 82 ----
xen/include/asm-x86/mm.h | 23 --
xen/include/asm-x86/p2m.h | 24 +-
xen/include/xen/mem_access.h | 63 ++++
xen/include/xen/mem_event.h | 141 +++++++
xen/include/xen/mm.h | 27 ++
xen/include/xsm/dummy.h | 26 +-
xen/include/xsm/xsm.h | 29 +-
xen/xsm/dummy.c | 7 +-
xen/xsm/flask/hooks.c | 33 +-
48 files changed, 2100 insertions(+), 1224 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
--
2.1.0
^ permalink raw reply [flat|nested] 45+ messages in thread
* [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 9:14 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
` (14 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
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 | 61 +---
xen/arch/x86/mm/Makefile | 2 -
xen/arch/x86/mm/hap/nested_ept.c | 2 +-
xen/arch/x86/mm/hap/nested_hap.c | 2 +-
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 | 2 +-
xen/arch/x86/mm/p2m-pod.c | 2 +-
xen/arch/x86/mm/p2m-pt.c | 2 +-
xen/arch/x86/mm/p2m.c | 2 +-
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 | 141 ++++++++
xen/include/xen/mm.h | 6 +
31 files changed, 1136 insertions(+), 1041 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 d1517c4..3aeb79d 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 83e6fae..3569481 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -63,8 +63,8 @@
#include <public/hvm/ioreq.h>
#include <public/version.h>
#include <public/memory.h>
-#include <asm/mem_event.h>
-#include <asm/mem_access.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.h>
#include <public/mem_event.h>
#include <xen/rangeset.h>
#include <public/arch-x86/cpuid.h>
@@ -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..704bb66 100644
--- a/xen/arch/x86/mm/hap/nested_ept.c
+++ b/xen/arch/x86/mm/hap/nested_ept.c
@@ -21,7 +21,7 @@
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
#include <public/mem_event.h>
#include <asm/mem_sharing.h>
#include <xen/event.h>
diff --git a/xen/arch/x86/mm/hap/nested_hap.c b/xen/arch/x86/mm/hap/nested_hap.c
index 137a87c..f6becd4 100644
--- a/xen/arch/x86/mm/hap/nested_hap.c
+++ b/xen/arch/x86/mm/hap/nested_hap.c
@@ -23,7 +23,7 @@
#include <asm/page.h>
#include <asm/paging.h>
#include <asm/p2m.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
#include <public/mem_event.h>
#include <asm/mem_sharing.h>
#include <xen/event.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..fa845fd 100644
--- a/xen/arch/x86/mm/mem_sharing.c
+++ b/xen/arch/x86/mm/mem_sharing.c
@@ -30,7 +30,7 @@
#include <asm/page.h>
#include <asm/string.h>
#include <asm/p2m.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
#include <asm/atomic.h>
#include <xen/rcupdate.h>
#include <asm/event.h>
diff --git a/xen/arch/x86/mm/p2m-pod.c b/xen/arch/x86/mm/p2m-pod.c
index bd4c7c8..881259a 100644
--- a/xen/arch/x86/mm/p2m-pod.c
+++ b/xen/arch/x86/mm/p2m-pod.c
@@ -26,7 +26,7 @@
#include <asm/p2m.h>
#include <asm/hvm/vmx/vmx.h> /* ept_p2m_init() */
#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 <xen/event.h>
diff --git a/xen/arch/x86/mm/p2m-pt.c b/xen/arch/x86/mm/p2m-pt.c
index 085ab6f..46231cf 100644
--- a/xen/arch/x86/mm/p2m-pt.c
+++ b/xen/arch/x86/mm/p2m-pt.c
@@ -30,7 +30,7 @@
#include <asm/paging.h>
#include <asm/p2m.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 <xen/event.h>
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index c2e89e1..ac30e9c 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -30,7 +30,7 @@
#include <asm/p2m.h>
#include <asm/hvm/vmx/vmx.h> /* ept_p2m_init() */
#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 <xen/event.h>
diff --git a/xen/arch/x86/x86_64/compat/mm.c b/xen/arch/x86/x86_64/compat/mm.c
index 69c6195..203c6b4 100644
--- a/xen/arch/x86/x86_64/compat/mm.c
+++ b/xen/arch/x86/x86_64/compat/mm.c
@@ -2,9 +2,9 @@
#include <xen/multicall.h>
#include <compat/memory.h>
#include <compat/xen.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
#include <asm/mem_sharing.h>
-#include <asm/mem_access.h>
+#include <xen/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..1f9702d 100644
--- a/xen/arch/x86/x86_64/mm.c
+++ b/xen/arch/x86/x86_64/mm.c
@@ -35,9 +35,9 @@
#include <asm/msr.h>
#include <asm/setup.h>
#include <asm/numa.h>
-#include <asm/mem_event.h>
+#include <xen/mem_event.h>
#include <asm/mem_sharing.h>
-#include <asm/mem_access.h>
+#include <xen/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 1952070..6f51311 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..07161a2
--- /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 <asm/p2m.h>
+#include <public/memory.h>
+#include <xen/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/common/mem_event.c b/xen/common/mem_event.c
new file mode 100644
index 0000000..b4a23fd
--- /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 <asm/p2m.h>
+#include <xen/mem_event.h>
+#include <xen/mem_access.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 c2dd31b..cc8a3d0 100644
--- a/xen/common/memory.c
+++ b/xen/common/memory.c
@@ -977,6 +977,69 @@ 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..5e3963e
--- /dev/null
+++ b/xen/include/xen/mem_event.h
@@ -0,0 +1,141 @@
+/******************************************************************************
+ * 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__
+
+#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] 45+ messages in thread
* [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 9:16 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 03/16] xen: Relocate struct npfec definition into common Tamas K Lengyel
` (13 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
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 | 3 +++
5 files changed, 29 insertions(+), 28 deletions(-)
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index ac30e9c..e6dea90 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -1423,30 +1423,6 @@ 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 07161a2..da5f363 100644
--- a/xen/common/mem_access.c
+++ b/xen/common/mem_access.c
@@ -29,6 +29,30 @@
#include <xen/mem_event.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 b4a23fd..abafd23 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 3975e32..ae755bc 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -600,8 +600,6 @@ 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..a58a41c 100644
--- a/xen/include/xen/mem_access.h
+++ b/xen/include/xen/mem_access.h
@@ -30,6 +30,7 @@
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);
+void mem_access_resume(struct domain *d);
#else
@@ -46,6 +47,8 @@ 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] 45+ messages in thread
* [PATCH v4 03/16] xen: Relocate struct npfec definition into common
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
` (12 subsequent siblings)
15 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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 | 1 +
xen/include/asm-x86/mm.h | 21 ---------------------
xen/include/xen/mm.h | 21 +++++++++++++++++++++
3 files changed, 22 insertions(+), 21 deletions(-)
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 74e66f8..2701e35 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -27,6 +27,7 @@
#include <public/domctl.h>
#include <public/hvm/save.h>
#include <public/hvm/ioreq.h>
+#include <xen/mm.h>
#include <asm/mm.h>
/* Interrupt acknowledgement sources. */
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] 45+ messages in thread
* [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop into common.
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (2 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 03/16] xen: Relocate struct npfec definition into common Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 9:23 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 05/16] xen/mem_event: Clean out superfluous white-spaces Tamas K Lengyel
` (11 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
v4: Don't remove memop handling from x86_64/compat and style fixes.
---
xen/arch/x86/domctl.c | 8 --------
xen/arch/x86/x86_64/mm.c | 4 ----
xen/common/domctl.c | 7 +++++++
xen/common/memory.c | 5 +++++
4 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index 3aeb79d..55a9495 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -1207,14 +1207,6 @@ 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/mm.c b/xen/arch/x86/x86_64/mm.c
index 1f9702d..c8272e9 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/domctl.c b/xen/common/domctl.c
index c326aba..bd9cec2 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>
@@ -967,6 +968,12 @@ 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;
+
default:
ret = arch_do_domctl(op, d, u_domctl);
break;
diff --git a/xen/common/memory.c b/xen/common/memory.c
index cc8a3d0..4e530bf 100644
--- a/xen/common/memory.c
+++ b/xen/common/memory.c
@@ -25,6 +25,7 @@
#include <asm/hardirq.h>
#include <asm/p2m.h>
#include <xen/numa.h>
+#include <xen/mem_access.h>
#include <public/memory.h>
#include <xsm/xsm.h>
#include <xen/trace.h>
@@ -969,6 +970,10 @@ 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;
+
default:
rc = arch_memory_op(cmd, arg);
break;
--
2.1.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v4 05/16] xen/mem_event: Clean out superfluous white-spaces
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (3 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
` (10 subsequent siblings)
15 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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 abafd23..bf1c415 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 5e3963e..33c9e7c 100644
--- a/xen/include/xen/mem_event.h
+++ b/xen/include/xen/mem_event.h
@@ -35,19 +35,19 @@ 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] 45+ messages in thread
* [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (4 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 05/16] xen/mem_event: Clean out superfluous white-spaces Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 9:25 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
` (9 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
v4: Add domain_id to the printout.
v3: Switch to gdprintk and print the vCPU id as well.
---
xen/common/mem_event.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index bf1c415..b31bb9d 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -278,7 +278,12 @@ 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,
+ "Dom%"PRIu32" VCPU %"PRIu32" 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] 45+ messages in thread
* [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (5 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 9:28 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
` (8 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
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 | 14 ++++++++++++++
2 files changed, 15 insertions(+), 6 deletions(-)
diff --git a/xen/common/mem_event.c b/xen/common/mem_event.c
index b31bb9d..fbbed43 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -594,12 +594,7 @@ 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 ae755bc..14be50f 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -611,6 +611,20 @@ 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)
+{
+ /* Only HAP is supported */
+ if ( !hap_enabled(d) )
+ return 0;
+
+ /* Currently only EPT is supported */
+ if ( !cpu_has_vmx )
+ return 0;
+
+ return 1;
+}
+
/*
* Internal functions, only called by other p2m code
*/
--
2.1.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (6 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 9:29 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 09/16] xen/arm: p2m type definitions and changes Tamas K Lengyel
` (7 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
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 | 8 ++++++++
2 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/xen/common/mem_access.c b/xen/common/mem_access.c
index da5f363..80ccc30 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 14be50f..2789723 100644
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -625,6 +625,14 @@ static inline bool_t p2m_mem_event_sanity_check(struct domain *d)
return 1;
}
+/* Sanity check for mem_access hardware support */
+static inline bool_t p2m_mem_access_sanity_check(struct domain *d)
+{
+ if ( !is_hvm_domain(d) )
+ return 0;
+ return 1;
+}
+
/*
* Internal functions, only called by other p2m code
*/
--
2.1.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v4 09/16] xen/arm: p2m type definitions and changes
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (7 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-09 0:06 ` Julien Grall
2014-09-05 8:58 ` [PATCH v4 10/16] xen/arm: Add set access required domctl Tamas K Lengyel
` (6 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
Define p2m_access_t in ARM and add necessary changes for page table
construction routines to pass the default access information. Also,
define the Radix tree that will hold access permission settings as
the PTE's don't have enough software programmable bits available.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
v4: move p2m_get_hostp2m definition here.
---
xen/arch/arm/p2m.c | 49 ++++++++++++++++---------
xen/include/asm-arm/p2m.h | 92 ++++++++++++++++++++++++++++++++++++-----------
2 files changed, 105 insertions(+), 36 deletions(-)
diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index 143199b..29e03b6 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -10,6 +10,9 @@
#include <asm/event.h>
#include <asm/hardirq.h>
#include <asm/page.h>
+#include <xen/mem_event.h>
+#include <public/mem_event.h>
+#include <xen/mem_access.h>
/* First level P2M is 2 consecutive pages */
#define P2M_FIRST_ORDER 1
@@ -225,7 +228,7 @@ 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
@@ -343,7 +346,7 @@ 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
@@ -363,7 +366,8 @@ 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);
@@ -458,7 +462,8 @@ 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)
{
/* Helpers to lookup the properties of each level */
const paddr_t level_sizes[] =
@@ -494,7 +499,7 @@ 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);
@@ -529,7 +534,7 @@ 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 */
@@ -690,7 +695,8 @@ static int apply_p2m_changes(struct domain *d,
paddr_t end_gpaddr,
paddr_t maddr,
int mattr,
- p2m_type_t t)
+ p2m_type_t t,
+ p2m_access_t a)
{
int rc, ret;
struct p2m_domain *p2m = &d->arch.p2m;
@@ -755,7 +761,7 @@ static int apply_p2m_changes(struct domain *d,
1, 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;
@@ -776,7 +782,7 @@ static int apply_p2m_changes(struct domain *d,
2, 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;
@@ -795,7 +801,7 @@ static int apply_p2m_changes(struct domain *d,
3, 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);
@@ -837,7 +843,8 @@ 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, p2m_ram_rw,
+ d->arch.p2m.default_access);
}
int map_mmio_regions(struct domain *d,
@@ -849,7 +856,8 @@ int map_mmio_regions(struct domain *d,
pfn_to_paddr(start_gfn),
pfn_to_paddr(start_gfn + nr_mfns),
pfn_to_paddr(mfn),
- MATTR_DEV, p2m_mmio_direct);
+ MATTR_DEV, p2m_mmio_direct,
+ d->arch.p2m.default_access);
}
int guest_physmap_add_entry(struct domain *d,
@@ -861,7 +869,8 @@ 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, t,
+ d->arch.p2m.default_access);
}
void guest_physmap_remove_page(struct domain *d,
@@ -871,7 +880,8 @@ 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, p2m_invalid,
+ d->arch.p2m.default_access);
}
int p2m_alloc_table(struct domain *d)
@@ -974,6 +984,8 @@ void p2m_teardown(struct domain *d)
p2m_free_vmid(d);
+ radix_tree_destroy(&p2m->mem_access_settings, NULL);
+
spin_unlock(&p2m->lock);
}
@@ -999,6 +1011,9 @@ int p2m_init(struct domain *d)
p2m->max_mapped_gfn = 0;
p2m->lowest_mapped_gfn = ULONG_MAX;
+ p2m->default_access = p2m_access_rwx;
+ radix_tree_init(&p2m->mem_access_settings);
+
err:
spin_unlock(&p2m->lock);
@@ -1013,7 +1028,8 @@ 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, p2m_invalid,
+ d->arch.p2m.default_access);
}
int p2m_cache_flush(struct domain *d, xen_pfn_t start_mfn, xen_pfn_t end_mfn)
@@ -1027,7 +1043,8 @@ 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, p2m_invalid,
+ d->arch.p2m.default_access);
}
unsigned long gmfn_to_mfn(struct domain *d, unsigned long gpfn)
diff --git a/xen/include/asm-arm/p2m.h b/xen/include/asm-arm/p2m.h
index 06c93a0..b2009ee 100644
--- a/xen/include/asm-arm/p2m.h
+++ b/xen/include/asm-arm/p2m.h
@@ -2,9 +2,54 @@
#define _XEN_P2M_H
#include <xen/mm.h>
+#include <xen/radix-tree.h>
+#include <public/memory.h>
+#include <public/mem_event.h>
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;
+
+/*
+ * Additional access types, which are used to further restrict
+ * the permissions given by 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_t;
+
/* Per-p2m-table state */
struct p2m_domain {
/* Lock that protects updates to the p2m */
@@ -38,27 +83,20 @@ 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 ambiquously
+ * 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;
+
+ /* Radix tree to store the p2m_access_t settings as the pte's don't have
+ * enough available bits to store this information. */
+ struct radix_tree_root mem_access_settings;
+};
#define p2m_is_foreign(_t) ((_t) == p2m_map_foreign)
#define p2m_is_ram(_t) ((_t) == p2m_ram_rw || (_t) == p2m_ram_ro)
@@ -195,6 +233,20 @@ static inline int get_page_and_type(struct page_info *page,
return rc;
}
+/* get host p2m table */
+#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
+
+/* mem_event and mem_access are supported on all ARM */
+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;
+}
+
#endif /* _XEN_P2M_H */
/*
--
2.1.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v4 10/16] xen/arm: Add set access required domctl
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (8 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 09/16] xen/arm: p2m type definitions and changes Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-09 0:07 ` Julien Grall
2014-09-05 8:58 ` [PATCH v4 11/16] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
` (5 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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/domctl.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/xen/arch/arm/domctl.c b/xen/arch/arm/domctl.c
index 45974e7..7cf719c 100644
--- a/xen/arch/arm/domctl.c
+++ b/xen/arch/arm/domctl.c
@@ -31,6 +31,19 @@ long arch_do_domctl(struct xen_domctl *domctl, struct domain *d,
return p2m_cache_flush(d, s, e);
}
+ case XEN_DOMCTL_set_access_required:
+ {
+ struct p2m_domain* p2m;
+
+ if ( current->domain == d )
+ return -EPERM;
+
+ p2m = p2m_get_hostp2m(d);
+ p2m->access_required = domctl->u.access_required.access_required;
+ return 0;
+ }
+ break;
+
default:
return subarch_do_domctl(domctl, d, u_domctl);
}
--
2.1.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v4 11/16] xen/arm: Data abort exception (R/W) mem_events.
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (9 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 10/16] xen/arm: Add set access required domctl Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-09 0:25 ` Julien Grall
2014-09-05 8:58 ` [PATCH v4 12/16] xen/arm: Instruction prefetch abort (X) mem_event handling Tamas K Lengyel
` (4 subsequent siblings)
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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 a Radix tree, where the key is the pfn
of a 4k page. Only settings other than p2m_access_rwx are saved
in the Radix tree.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
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 | 443 +++++++++++++++++++++++++++++++++++++---
xen/arch/arm/traps.c | 31 ++-
xen/include/asm-arm/p2m.h | 17 ++
xen/include/asm-arm/processor.h | 30 +++
4 files changed, 487 insertions(+), 34 deletions(-)
diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index 29e03b6..c723f9a 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -10,6 +10,7 @@
#include <asm/event.h>
#include <asm/hardirq.h>
#include <asm/page.h>
+#include <xen/radix-tree.h>
#include <xen/mem_event.h>
#include <public/mem_event.h>
#include <xen/mem_access.h>
@@ -148,6 +149,74 @@ static lpae_t *p2m_map_first(struct p2m_domain *p2m, paddr_t addr)
return __map_domain_page(page);
}
+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_n:
+ e->p2m.read = e->p2m.write = 0;
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_r:
+ e->p2m.write = 0;
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_x:
+ e->p2m.write = 0;
+ e->p2m.read = 0;
+ break;
+ case p2m_access_rx:
+ e->p2m.write = 0;
+ break;
+ case p2m_access_w:
+ e->p2m.read = 0;
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_rw:
+ e->p2m.xn = 1;
+ break;
+ case p2m_access_wx:
+ e->p2m.read = 0;
+ break;
+ case p2m_access_rwx:
+ break;
+ }
+}
+
/*
* Lookup the MFN corresponding to a domain's PFN.
*
@@ -258,37 +327,7 @@ 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));
@@ -440,6 +479,33 @@ static bool_t is_mapping_aligned(const paddr_t start_gpaddr,
return true;
}
+static long p2m_mem_access_radix_set(struct p2m_domain *p2m, unsigned long pfn,
+ p2m_access_t a)
+{
+ long rc;
+
+ if ( p2m_access_rwx == a )
+ {
+ radix_tree_delete(&p2m->mem_access_settings, pfn);
+ return 0;
+ }
+
+ rc = radix_tree_insert(&p2m->mem_access_settings, pfn,
+ radix_tree_int_to_ptr(a));
+
+ if ( -EEXIST == rc )
+ {
+ /* If a setting existed already, change it to the new one */
+ radix_tree_replace_slot(
+ radix_tree_lookup_slot(
+ &p2m->mem_access_settings, pfn),
+ radix_tree_int_to_ptr(a));
+ rc = 0;
+ }
+
+ return rc;
+}
+
#define P2M_ONE_DESCEND 0
#define P2M_ONE_PROGRESS_NOP 0x1
#define P2M_ONE_PROGRESS 0x10
@@ -502,6 +568,14 @@ static int apply_one_level(struct domain *d,
pte = mfn_to_p2m_entry(page_to_mfn(page), mattr, t, a);
if ( level < 3 )
pte.p2m.table = 0;
+
+ if ( p2m_mem_access_radix_set(p2m, paddr_to_pfn(*addr), a) < 0 )
+ {
+ /* If we fail to save the setting in the Radix tree, we
+ * need to clear the PTE mem_access permissions. */
+ p2m_set_permission(&pte, pte.p2m.type, p2m_access_rwx);
+ }
+
p2m_write_pte(entry, pte, flush_cache);
p2m->stats.mappings[level]++;
@@ -538,6 +612,13 @@ static int apply_one_level(struct domain *d,
if ( level < 3 )
pte.p2m.table = 0; /* Superpage entry */
+ if ( p2m_mem_access_radix_set(p2m, paddr_to_pfn(*addr), a) < 0 )
+ {
+ /* If we fail to save the setting in the Radix tree, we
+ * need to clear the PTE mem_access permissions. */
+ p2m_set_permission(&pte, pte.p2m.type, p2m_access_rwx);
+ }
+
p2m_write_pte(entry, pte, flush_cache);
*flush |= p2m_valid(orig_pte);
@@ -641,6 +722,7 @@ static int apply_one_level(struct domain *d,
memset(&pte, 0x00, sizeof(pte));
p2m_write_pte(entry, pte, flush_cache);
+ radix_tree_delete(&p2m->mem_access_settings, paddr_to_pfn(*addr));
*addr += level_size;
@@ -1081,6 +1163,305 @@ err:
return page;
}
+static int p2m_set_entry(struct domain *d, paddr_t paddr, p2m_access_t p2ma)
+{
+ int rc = 0;
+ struct p2m_domain *p2m = &d->arch.p2m;
+ lpae_t *pte, *first = NULL, *second = NULL, *third = NULL;
+
+ first = p2m_map_first(p2m, paddr);
+ if ( !first )
+ goto out;
+
+ pte = &first[first_table_offset(paddr)];
+ if( !p2m_valid(*pte) )
+ goto out;
+
+ if ( !p2m_table(*pte) )
+ goto done;
+
+ second = map_domain_page(pte->p2m.base);
+ if ( !second )
+ goto out;
+
+ pte = &second[second_table_offset(paddr)];
+ if( !p2m_valid(*pte) )
+ goto out;
+
+ if ( !p2m_table(*pte) )
+ goto done;
+
+ third = map_domain_page(pte->p2m.base);
+ if ( !third )
+ goto out;
+
+ pte = &third[third_table_offset(paddr)];
+
+ BUILD_BUG_ON(THIRD_MASK != PAGE_MASK);
+
+ /* This bit must be one in the level 3 entry */
+ if ( !p2m_table(*pte) )
+ pte->bits = 0;
+
+done:
+ if ( p2m_valid(*pte) )
+ {
+ ASSERT(pte->p2m.type != p2m_invalid);
+
+ p2m_set_permission(pte, pte->p2m.type, p2ma);
+ p2m_write_pte(pte, *pte, 1);
+ rc = 1;
+ }
+
+out:
+ if (third) unmap_domain_page(third);
+ if (second) unmap_domain_page(second);
+ if (first) unmap_domain_page(first);
+
+ return rc;
+}
+
+bool_t p2m_mem_access_check(paddr_t gpa, vaddr_t gla, struct npfec npfec)
+{
+ struct vcpu *v = current;
+ struct p2m_domain *p2m = p2m_get_hostp2m(v->domain);
+ mem_event_request_t *req = NULL;
+ xenmem_access_t xma;
+ bool_t violation;
+ int rc;
+
+ rc = p2m_get_mem_access(v->domain, paddr_to_pfn(gpa), &xma);
+ if ( rc )
+ {
+ /* No setting was found, reinject */
+ return 1;
+ }
+ else
+ {
+ /* First, handle rx2rw and n2rwx conversion automatically. */
+ if ( npfec.write_access && xma == XENMEM_access_rx2rw )
+ {
+ rc = p2m_set_mem_access(v->domain, paddr_to_pfn(gpa), 1,
+ 0, ~0, XENMEM_access_rw);
+ return 0;
+ }
+ else if ( xma == XENMEM_access_n2rwx )
+ {
+ rc = p2m_set_mem_access(v->domain, paddr_to_pfn(gpa), 1,
+ 0, ~0, XENMEM_access_rwx);
+ }
+ }
+
+ /* Otherwise, check if there is a memory event listener, and send the message along */
+ if ( !mem_event_check_ring( &v->domain->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, v->domain->domain_id);
+ domain_crash(v->domain);
+ }
+ else
+ {
+ /* n2rwx was already handled */
+ if ( xma != XENMEM_access_n2rwx)
+ {
+ /* A listener is not required, so clear the access
+ * restrictions. */
+ rc = p2m_set_mem_access(v->domain, paddr_to_pfn(gpa), 1,
+ 0, ~0, XENMEM_access_rwx);
+ }
+ }
+
+ /* No need to reinject */
+ return 0;
+ }
+
+ switch ( xma )
+ {
+ default:
+ case XENMEM_access_n:
+ violation = npfec.read_access || npfec.write_access || npfec.insn_fetch;
+ break;
+ case XENMEM_access_r:
+ violation = npfec.write_access || npfec.insn_fetch;
+ break;
+ case XENMEM_access_w:
+ violation = npfec.read_access || npfec.insn_fetch;
+ break;
+ case XENMEM_access_x:
+ violation = npfec.read_access || npfec.write_access;
+ break;
+ case XENMEM_access_rx:
+ violation = npfec.write_access;
+ break;
+ case XENMEM_access_wx:
+ violation = npfec.read_access;
+ break;
+ case XENMEM_access_rw:
+ violation = npfec.insn_fetch;
+ break;
+ case XENMEM_access_rwx:
+ violation = 0;
+ break;
+ }
+
+ if ( !violation )
+ return 1;
+
+ req = xzalloc(mem_event_request_t);
+ if ( req )
+ {
+ req->reason = MEM_EVENT_REASON_VIOLATION;
+ if ( xma != XENMEM_access_n2rwx )
+ req->flags |= MEM_EVENT_FLAG_VCPU_PAUSED;
+ req->gfn = gpa >> PAGE_SHIFT;
+ req->offset = gpa & ((1 << PAGE_SHIFT) - 1);
+ req->gla = gla;
+ req->gla_valid = npfec.gla_valid;
+ req->access_r = npfec.read_access;
+ req->access_w = npfec.write_access;
+ req->access_x = npfec.insn_fetch;
+ if ( npfec_kind_in_gpt == npfec.kind )
+ req->fault_in_gpt = 1;
+ if ( npfec_kind_with_gla == npfec.kind )
+ req->fault_with_gla = 1;
+ req->vcpu_id = v->vcpu_id;
+
+ mem_access_send_req(v->domain, req);
+ xfree(req);
+ }
+
+ /* Pause the current VCPU */
+ if ( xma != XENMEM_access_n2rwx )
+ mem_event_vcpu_pause(v);
+
+ return 0;
+}
+
+/* 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),
+#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;
+ }
+
+ spin_lock(&p2m->lock);
+ for ( pfn += start; nr > start; ++pfn )
+ {
+
+ rc = p2m_set_entry(d, pfn_to_paddr(pfn), a);
+ if ( rc < 0 )
+ break;
+
+ if ( rc )
+ {
+ rc = p2m_mem_access_radix_set(p2m, pfn, a);
+ if ( rc < 0 )
+ {
+ /* If we fail to save the setting in the Radix tree, we
+ * need to clear the PTE mem_access permissions. */
+ p2m_set_entry(d, pfn_to_paddr(pfn), p2m_access_rwx);
+ break;
+ }
+ }
+
+ /* Check for continuation if it's not the last iteration. */
+ if ( nr > ++start && !(start & mask) && hypercall_preempt_check() )
+ {
+ rc = start;
+ break;
+ }
+ }
+
+ /* Flush the TLB of the domain to ensure consistency */
+ flush_tlb_domain(d);
+
+ spin_unlock(&p2m->lock);
+ return rc;
+}
+
+int p2m_get_mem_access(struct domain *d, unsigned long gpfn,
+ xenmem_access_t *access)
+{
+ struct p2m_domain *p2m = p2m_get_hostp2m(d);
+ void *i;
+ unsigned int index;
+
+ static const xenmem_access_t memaccess[] = {
+#define ACCESS(ac) [XENMEM_access_##ac] = XENMEM_access_##ac
+ ACCESS(n),
+ ACCESS(r),
+ ACCESS(w),
+ ACCESS(rw),
+ ACCESS(x),
+ ACCESS(rx),
+ ACCESS(wx),
+ ACCESS(rwx),
+#undef ACCESS
+ };
+
+ /* If request to get default access */
+ if ( gpfn == ~0ull )
+ {
+ *access = memaccess[p2m->default_access];
+ return 0;
+ }
+
+ spin_lock(&p2m->lock);
+
+ i = radix_tree_lookup(&p2m->mem_access_settings, gpfn);
+
+ spin_unlock(&p2m->lock);
+
+ if ( !i )
+ return -ESRCH;
+
+ index = radix_tree_ptr_to_int(i);
+
+ if ( index >= ARRAY_SIZE(memaccess) )
+ return -ERANGE;
+
+ *access = memaccess[index];
+ return 0;
+}
+
/*
* Local variables:
* mode: C
diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
index 019991f..7eb875a 100644
--- a/xen/arch/arm/traps.c
+++ b/xen/arch/arm/traps.c
@@ -1852,13 +1852,38 @@ static void do_trap_data_abort_guest(struct cpu_user_regs *regs,
info.gva = READ_SYSREG64(FAR_EL2);
#endif
- if (dabt.s1ptw)
- goto bad_data_abort;
-
rc = gva_to_ipa(info.gva, &info.gpa);
if ( rc == -EFAULT )
goto bad_data_abort;
+ switch ( dabt.dfsc )
+ {
+ case DABT_DFSC_PERMISSION_1:
+ case DABT_DFSC_PERMISSION_2:
+ case DABT_DFSC_PERMISSION_3:
+ {
+ 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 = p2m_mem_access_check(info.gpa, info.gva, npfec);
+
+ /* Trap was triggered by mem_access, work here is done */
+ if ( !rc )
+ return;
+ }
+ break;
+
+ default:
+ break;
+ }
+
+ if ( dabt.s1ptw )
+ goto bad_data_abort;
+
/* XXX: Decode the instruction if ISS is not valid */
if ( !dabt.valid )
goto bad_data_abort;
diff --git a/xen/include/asm-arm/p2m.h b/xen/include/asm-arm/p2m.h
index b2009ee..35466dd 100644
--- a/xen/include/asm-arm/p2m.h
+++ b/xen/include/asm-arm/p2m.h
@@ -247,6 +247,23 @@ static inline bool_t p2m_mem_event_sanity_check(struct domain *d)
return 1;
}
+/* Send mem event based on the access (gla is -1ull if not available). Boolean
+ * return value indicates if trap needs to be injected into guest. */
+bool_t p2m_mem_access_check(paddr_t gpa, vaddr_t gla, struct npfec npfec);
+
+/* 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 */
+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 0cc5b6d..b844f1d 100644
--- a/xen/include/asm-arm/processor.h
+++ b/xen/include/asm-arm/processor.h
@@ -262,6 +262,36 @@ 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] 45+ messages in thread
* [PATCH v4 12/16] xen/arm: Instruction prefetch abort (X) mem_event handling
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (10 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 11/16] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 13/16] xen/arm: Shatter large pages when using mem_acces Tamas K Lengyel
` (3 subsequent siblings)
15 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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>
---
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 | 42 ++++++++++++++++++++++++++++++++++++++++-
xen/include/asm-arm/processor.h | 40 ++++++++++++++++++++++++++++++++++++++-
2 files changed, 80 insertions(+), 2 deletions(-)
diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
index 7eb875a..6f851cc 100644
--- a/xen/arch/arm/traps.c
+++ b/xen/arch/arm/traps.c
@@ -1828,7 +1828,47 @@ 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 ( rc == -EFAULT )
+ return;
+
+ switch ( iabt.ifsc )
+ {
+ case IABT_IFSC_PERMISSION_1:
+ case IABT_IFSC_PERMISSION_2:
+ case IABT_IFSC_PERMISSION_3:
+ {
+ struct npfec npfec = {
+ .insn_fetch = 1,
+ .gla_valid = 1,
+ .kind = iabt.s1ptw ? npfec_kind_in_gpt : npfec_kind_with_gla
+ };
+
+ rc = p2m_mem_access_check(gpa, gva, npfec);
+
+ /* Trap was triggered by mem_access, work here is done */
+ if ( !rc )
+ return;
+ }
+ break;
+
+ default:
+ 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 b844f1d..044de12 100644
--- a/xen/include/asm-arm/processor.h
+++ b/xen/include/asm-arm/processor.h
@@ -292,6 +292,36 @@ 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 {
@@ -371,10 +401,18 @@ 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] 45+ messages in thread
* [PATCH v4 13/16] xen/arm: Shatter large pages when using mem_acces
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (11 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 12/16] xen/arm: Instruction prefetch abort (X) mem_event handling Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 14/16] xen/arm: Enable the compilation of mem_access and mem_event on ARM Tamas K Lengyel
` (2 subsequent siblings)
15 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
When using mem_acccess+mem_event, the coarsest granularity is required
in the pagetables, therefore we shatter each large page (1GB/2MB) as we
apply the permission changes.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
xen/arch/arm/p2m.c | 23 ++++++++++++++++++++---
1 file changed, 20 insertions(+), 3 deletions(-)
diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index c723f9a..35ca6c1 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -1178,7 +1178,16 @@ static int p2m_set_entry(struct domain *d, paddr_t paddr, p2m_access_t p2ma)
goto out;
if ( !p2m_table(*pte) )
- goto done;
+ {
+ /* This is a mapping of a 1GB page, shatter it */
+ rc = p2m_create_table(d, pte, FIRST_SHIFT - PAGE_SHIFT, 1);
+ if ( rc < 0 )
+ goto out;
+
+ p2m->stats.shattered[1]++;
+ p2m->stats.mappings[1]--;
+ p2m->stats.mappings[2] += LPAE_ENTRIES;
+ }
second = map_domain_page(pte->p2m.base);
if ( !second )
@@ -1189,7 +1198,16 @@ static int p2m_set_entry(struct domain *d, paddr_t paddr, p2m_access_t p2ma)
goto out;
if ( !p2m_table(*pte) )
- goto done;
+ {
+ /* This is a mapping of 2MB page, shatter it */
+ rc = p2m_create_table(d, pte, SECOND_SHIFT - PAGE_SHIFT, 1);
+ if ( rc < 0 )
+ goto out;
+
+ p2m->stats.shattered[2]++;
+ p2m->stats.mappings[2]--;
+ p2m->stats.mappings[3] += LPAE_ENTRIES;
+ }
third = map_domain_page(pte->p2m.base);
if ( !third )
@@ -1203,7 +1221,6 @@ static int p2m_set_entry(struct domain *d, paddr_t paddr, p2m_access_t p2ma)
if ( !p2m_table(*pte) )
pte->bits = 0;
-done:
if ( p2m_valid(*pte) )
{
ASSERT(pte->p2m.type != p2m_invalid);
--
2.1.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v4 14/16] xen/arm: Enable the compilation of mem_access and mem_event on ARM.
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (12 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 13/16] xen/arm: Shatter large pages when using mem_acces Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 15/16] tools/libxc: Allocate magic page for mem access " Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 16/16] tools/tests: Enable xen-access " Tamas K Lengyel
15 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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 fbbed43..096cbff 100644
--- a/xen/common/mem_event.c
+++ b/xen/common/mem_event.c
@@ -27,8 +27,15 @@
#include <asm/p2m.h>
#include <xen/mem_event.h>
#include <xen/mem_access.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 */
@@ -424,12 +431,14 @@ 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)
@@ -438,15 +447,20 @@ 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;
@@ -473,6 +487,7 @@ 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 */
@@ -533,6 +548,7 @@ 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;
@@ -583,6 +599,7 @@ int mem_event_domctl(struct domain *d, xen_domctl_mem_event_op_t *mec,
}
}
break;
+#endif
case XEN_DOMCTL_MEM_EVENT_OP_ACCESS:
{
@@ -617,6 +634,7 @@ 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;
@@ -655,6 +673,7 @@ 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 c5aa316..cea2e63 100644
--- a/xen/include/xsm/dummy.h
+++ b/xen/include/xsm/dummy.h
@@ -507,6 +507,20 @@ 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)
{
@@ -550,18 +564,6 @@ 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 a85045d..6c3032c 100644
--- a/xen/include/xsm/xsm.h
+++ b/xen/include/xsm/xsm.h
@@ -140,6 +140,11 @@ struct xsm_operations {
int (*hvm_control) (struct domain *d, unsigned long op);
int (*hvm_param_nested) (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);
@@ -148,8 +153,6 @@ 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);
@@ -534,6 +537,18 @@ static inline int xsm_hvm_param_nested (xsm_default_t def, struct domain *d)
return xsm_ops->hvm_param_nested(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)
{
@@ -570,16 +585,6 @@ 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 c95c803..e9cdc01 100644
--- a/xen/xsm/dummy.c
+++ b/xen/xsm/dummy.c
@@ -117,6 +117,11 @@ 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);
@@ -125,8 +130,6 @@ 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 f2f59ea..96efd0b 100644
--- a/xen/xsm/flask/hooks.c
+++ b/xen/xsm/flask/hooks.c
@@ -571,6 +571,9 @@ static int flask_domctl(struct domain *d, int cmd)
case XEN_DOMCTL_irq_permission:
case XEN_DOMCTL_iomem_permission:
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:
@@ -579,7 +582,6 @@ static int flask_domctl(struct domain *d, int cmd)
case XEN_DOMCTL_unbind_pt_irq:
case XEN_DOMCTL_memory_mapping:
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:
@@ -1181,6 +1183,18 @@ 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)
{
@@ -1291,16 +1305,6 @@ 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);
@@ -1567,6 +1571,11 @@ 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,
@@ -1575,8 +1584,6 @@ 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] 45+ messages in thread
* [PATCH v4 15/16] tools/libxc: Allocate magic page for mem access on ARM
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (13 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 14/16] xen/arm: Enable the compilation of mem_access and mem_event on ARM Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 16/16] tools/tests: Enable xen-access " Tamas K Lengyel
15 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 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] 45+ messages in thread
* [PATCH v4 16/16] tools/tests: Enable xen-access on ARM
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
` (14 preceding siblings ...)
2014-09-05 8:58 ` [PATCH v4 15/16] tools/libxc: Allocate magic page for mem access " Tamas K Lengyel
@ 2014-09-05 8:58 ` Tamas K Lengyel
2014-09-09 0:29 ` Julien Grall
15 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 8:58 UTC (permalink / raw)
To: xen-devel
Cc: ian.campbell, tim, julien.grall, ian.jackson, stefano.stabellini,
andres, jbeulich, dgdegra, Tamas K Lengyel
On ARM the guest memory doesn't start from 0 and has multiple banks
of memory, thus we adjust the setting/unsetting routines accordingly.
We also define the ARM specific test_and_set_bit function.
Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
---
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 | 94 +++++++++++++++++++++++++++++++++----
xen/arch/arm/Rules.mk | 1 -
xen/arch/x86/Rules.mk | 1 -
8 files changed, 97 insertions(+), 14 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 090df5f..e122ef8 100644
--- a/tools/tests/xen-access/xen-access.c
+++ b/tools/tests/xen-access/xen-access.c
@@ -41,22 +41,15 @@
#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 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 +62,44 @@ 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 GUEST_RAM0_4K_PAGES GUEST_RAM0_SIZE / (1ull<<PAGE_SHIFT)
+#define GUEST_RAM1_4K_PAGES GUEST_RAM1_SIZE / (1ull<<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)
@@ -403,7 +434,7 @@ int main(int argc, char *argv[])
mem_event_request_t req;
mem_event_response_t rsp;
int rc = -1;
- int rc1;
+ int rc1, rc2;
xc_interface *xch;
xenmem_access_t default_access = XENMEM_access_rwx;
xenmem_access_t after_first_access = XENMEM_access_rwx;
@@ -492,8 +523,30 @@ int main(int argc, char *argv[])
goto exit;
}
+#ifdef CONFIG_X86
rc = xc_set_mem_access(xch, domain_id, default_access, 0,
xenaccess->domain_info->max_pages);
+#elif CONFIG_ARM
+ /* On ARM the guest memory is mapped with a gap after the first 3GB. */
+ if ( xenaccess->domain_info->max_pages <= GUEST_RAM0_4K_PAGES )
+ {
+ rc = xc_set_mem_access(xch, domain_id, default_access,
+ GUEST_RAM0_BASE >> PAGE_SHIFT,
+ xenaccess->domain_info->max_pages);
+ }
+ else
+ {
+ rc1 = xc_set_mem_access(xch, domain_id, default_access,
+ GUEST_RAM0_BASE >> PAGE_SHIFT,
+ GUEST_RAM0_4K_PAGES);
+ rc2 = xc_set_mem_access(xch, domain_id, default_access,
+ GUEST_RAM1_BASE >> PAGE_SHIFT,
+ xenaccess->domain_info->max_pages
+ - GUEST_RAM0_4K_PAGES);
+ rc = rc1 < 0 ? rc1 : (rc2 < 0 ? rc2 : 0);
+ }
+#endif
+
if ( rc < 0 )
{
ERROR("Error %d setting all memory to access type %d\n", rc,
@@ -520,8 +573,29 @@ int main(int argc, char *argv[])
/* Unregister for every event */
rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx, ~0ull, 0);
+#ifdef CONFIG_X86
rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx, 0,
xenaccess->domain_info->max_pages);
+#elif CONFIG_ARM
+ /* On ARM the guest memory is mapped with a gap after the first 3GB. */
+ if ( xenaccess->domain_info->max_pages <= GUEST_RAM0_4K_PAGES )
+ {
+ rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx,
+ GUEST_RAM0_BASE >> PAGE_SHIFT,
+ xenaccess->domain_info->max_pages);
+ }
+ else
+ {
+ rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx,
+ GUEST_RAM0_BASE >> PAGE_SHIFT,
+ GUEST_RAM0_4K_PAGES);
+ rc = xc_set_mem_access(xch, domain_id, XENMEM_access_rwx,
+ GUEST_RAM1_BASE >> PAGE_SHIFT,
+ xenaccess->domain_info->max_pages
+ - GUEST_RAM0_4K_PAGES);
+ }
+#endif
+
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] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 8:58 ` [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
@ 2014-09-05 9:14 ` Jan Beulich
2014-09-05 9:45 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:14 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> --- a/xen/arch/x86/x86_64/mm.c
> +++ b/xen/arch/x86/x86_64/mm.c
> @@ -35,9 +35,9 @@
> #include <asm/msr.h>
> #include <asm/setup.h>
> #include <asm/numa.h>
> -#include <asm/mem_event.h>
> +#include <xen/mem_event.h>
> #include <asm/mem_sharing.h>
> -#include <asm/mem_access.h>
> +#include <xen/mem_access.h>
> #include <public/memory.h>
This is not the only place, but a specifically bad example: I'm pretty
sure I asked you before to not make a mess by mixing asm/ and
xen/ included - move the now xen/ ones to the other ones already
coming from that directory. And do so consistently throughout the
patch.
> --- 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
The addition should go alongside the other CONFIG_* ones instead
of at an arbitrary place. Looking at this again I wonder though
whether we really need these - the use sites could easily check
whether p2m_is_{paging,shared} is defined instead.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common
2014-09-05 8:58 ` [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
@ 2014-09-05 9:16 ` Jan Beulich
2014-09-05 9:25 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:16 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> --- a/xen/common/mem_access.c
> +++ b/xen/common/mem_access.c
> @@ -29,6 +29,30 @@
> #include <xen/mem_event.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) )
Please take the opportunity to fix the coding style here.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop into common.
2014-09-05 8:58 ` [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
@ 2014-09-05 9:23 ` Jan Beulich
2014-09-05 9:29 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:23 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> ---
> v4: Don't remove memop handling from x86_64/compat and style fixes.
Here you acted a little too lax one the previous version's comments
I gave: If the handling of the native memop gets moved from
arch/x86/ to common/, so should the compat one do.
> @@ -967,6 +968,12 @@ 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;
> +
> default:
> ret = arch_do_domctl(op, d, u_domctl);
> break;
Additions of new domctls are fine to go here, but moving existing
ones here from arch code should try to put it in a place at least
roughly expectable by the actual domctl number (I'm not sure the
switch statement is fully ordered, but at least I'm pretty sure you'll
find it to at least be approximately).
> @@ -969,6 +970,10 @@ 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;
> +
> default:
> rc = arch_memory_op(cmd, arg);
> break;
Same here then, obviously.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common
2014-09-05 9:16 ` Jan Beulich
@ 2014-09-05 9:25 ` Tamas K Lengyel
2014-09-05 9:43 ` Jan Beulich
0 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:25 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: 704 bytes --]
On Fri, Sep 5, 2014 at 11:16 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> > --- a/xen/common/mem_access.c
> > +++ b/xen/common/mem_access.c
> > @@ -29,6 +29,30 @@
> > #include <xen/mem_event.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) )
>
> Please take the opportunity to fix the coding style here.
>
> Jan
>
Surely, but what would be preferred coding style here? A separate int rc
for the value returned by mem_event_get_response?
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1322 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] 45+ messages in thread
* Re: [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds
2014-09-05 8:58 ` [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
@ 2014-09-05 9:25 ` Jan Beulich
2014-09-05 9:26 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:25 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> --- a/xen/common/mem_event.c
> +++ b/xen/common/mem_event.c
> @@ -278,7 +278,12 @@ 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,
> + "Dom%"PRIu32" VCPU %"PRIu32" was not paused.\n",
> + d->domain_id, req->vcpu_id);
> +#endif
And now please use the conventional d%dv%d, to make domain/vcpu
specific log entries greppable.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds
2014-09-05 9:25 ` Jan Beulich
@ 2014-09-05 9:26 ` Tamas K Lengyel
0 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:26 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: 823 bytes --]
On Fri, Sep 5, 2014 at 11:25 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> > --- a/xen/common/mem_event.c
> > +++ b/xen/common/mem_event.c
> > @@ -278,7 +278,12 @@ 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,
> > + "Dom%"PRIu32" VCPU %"PRIu32" was not paused.\n",
> > + d->domain_id, req->vcpu_id);
> > +#endif
>
> And now please use the conventional d%dv%d, to make domain/vcpu
> specific log entries greppable.
>
> Jan
>
Ack.
[-- Attachment #1.2: Type: text/html, Size: 1438 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] 45+ messages in thread
* Re: [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks
2014-09-05 8:58 ` [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
@ 2014-09-05 9:28 ` Jan Beulich
2014-09-05 9:35 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:28 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> --- a/xen/include/asm-x86/p2m.h
> +++ b/xen/include/asm-x86/p2m.h
> @@ -611,6 +611,20 @@ 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)
> +{
> + /* Only HAP is supported */
> + if ( !hap_enabled(d) )
> + return 0;
> +
> + /* Currently only EPT is supported */
> + if ( !cpu_has_vmx )
> + return 0;
> +
> + return 1;
> +}
For simple inline functions like this, may I suggest have them a use
simple return statement:
return hap_enabled(d) && cpu_has_vmx;
This is the more that the first comment is redundant with the
second, i.e. the single statement would then need to be
accompanied only by the second of them.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check
2014-09-05 8:58 ` [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
@ 2014-09-05 9:29 ` Jan Beulich
2014-09-05 9:51 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:29 UTC (permalink / raw)
To: Tamas K Lengyel
Cc: ian.campbell, tim, julien.grall, ian.jackson, xen-devel,
stefano.stabellini, andres, dgdegra
>>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> --- a/xen/include/asm-x86/p2m.h
> +++ b/xen/include/asm-x86/p2m.h
> @@ -625,6 +625,14 @@ static inline bool_t p2m_mem_event_sanity_check(struct domain *d)
> return 1;
> }
>
> +/* Sanity check for mem_access hardware support */
> +static inline bool_t p2m_mem_access_sanity_check(struct domain *d)
> +{
> + if ( !is_hvm_domain(d) )
> + return 0;
> + return 1;
> +}
In line with the previous one - just "return is_hvm_domain(d);" or
even make this a #define.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop into common.
2014-09-05 9:23 ` Jan Beulich
@ 2014-09-05 9:29 ` Tamas K Lengyel
0 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:29 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: 1796 bytes --]
On Fri, Sep 5, 2014 at 11:23 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> > Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> > ---
> > v4: Don't remove memop handling from x86_64/compat and style fixes.
>
> Here you acted a little too lax one the previous version's comments
> I gave: If the handling of the native memop gets moved from
> arch/x86/ to common/, so should the compat one do.
>
>
Ah yes, that would make sense.
> > @@ -967,6 +968,12 @@ 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;
> > +
> > default:
> > ret = arch_do_domctl(op, d, u_domctl);
> > break;
>
> Additions of new domctls are fine to go here, but moving existing
> ones here from arch code should try to put it in a place at least
> roughly expectable by the actual domctl number (I'm not sure the
> switch statement is fully ordered, but at least I'm pretty sure you'll
> find it to at least be approximately).
>
>
Ack.
> > @@ -969,6 +970,10 @@ 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;
> > +
> > default:
> > rc = arch_memory_op(cmd, arg);
> > break;
>
> Same here then, obviously.
>
> Jan
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
>
[-- Attachment #1.2: Type: text/html, Size: 3092 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] 45+ messages in thread
* Re: [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks
2014-09-05 9:28 ` Jan Beulich
@ 2014-09-05 9:35 ` Tamas K Lengyel
2014-09-05 9:44 ` Jan Beulich
0 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:35 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: 1525 bytes --]
On Fri, Sep 5, 2014 at 11:28 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> > --- a/xen/include/asm-x86/p2m.h
> > +++ b/xen/include/asm-x86/p2m.h
> > @@ -611,6 +611,20 @@ 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)
> > +{
> > + /* Only HAP is supported */
> > + if ( !hap_enabled(d) )
> > + return 0;
> > +
> > + /* Currently only EPT is supported */
> > + if ( !cpu_has_vmx )
> > + return 0;
> > +
> > + return 1;
> > +}
>
> For simple inline functions like this, may I suggest have them a use
> simple return statement:
>
> return hap_enabled(d) && cpu_has_vmx;
>
>
Surely, my aim here really was just to minimize look&feel changes compared
to existing code. The current approach might be a tiny bit faster on
domains without hap, but not like it really would count for much, so I'm OK
with having the compact one-liner version.
> This is the more that the first comment is redundant with the
> second, i.e. the single statement would then need to be
> accompanied only by the second of them.
>
> Jan
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
>
[-- Attachment #1.2: Type: text/html, Size: 2458 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] 45+ messages in thread
* Re: [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common
2014-09-05 9:25 ` Tamas K Lengyel
@ 2014-09-05 9:43 ` Jan Beulich
2014-09-05 9:48 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:43 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 05.09.14 at 11:25, <tamas.lengyel@zentific.com> wrote:
> On Fri, Sep 5, 2014 at 11:16 AM, Jan Beulich <JBeulich@suse.com> wrote:
>
>> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
>> > --- a/xen/common/mem_access.c
>> > +++ b/xen/common/mem_access.c
>> > @@ -29,6 +29,30 @@
>> > #include <xen/mem_event.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) )
>>
>> Please take the opportunity to fix the coding style here.
>
> Surely, but what would be preferred coding style here? A separate int rc
> for the value returned by mem_event_get_response?
No, I really just mean formatting: There is a stop missing in the
comment (and there was another such comment further down
iirc), and a blank missing after "while".
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks
2014-09-05 9:35 ` Tamas K Lengyel
@ 2014-09-05 9:44 ` Jan Beulich
2014-09-05 9:47 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 9:44 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 05.09.14 at 11:35, <tamas.lengyel@zentific.com> wrote:
> On Fri, Sep 5, 2014 at 11:28 AM, Jan Beulich <JBeulich@suse.com> wrote:
>
>> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
>> > --- a/xen/include/asm-x86/p2m.h
>> > +++ b/xen/include/asm-x86/p2m.h
>> > @@ -611,6 +611,20 @@ 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)
>> > +{
>> > + /* Only HAP is supported */
>> > + if ( !hap_enabled(d) )
>> > + return 0;
>> > +
>> > + /* Currently only EPT is supported */
>> > + if ( !cpu_has_vmx )
>> > + return 0;
>> > +
>> > + return 1;
>> > +}
>>
>> For simple inline functions like this, may I suggest have them a use
>> simple return statement:
>>
>> return hap_enabled(d) && cpu_has_vmx;
>>
>>
> Surely, my aim here really was just to minimize look&feel changes compared
> to existing code. The current approach might be a tiny bit faster on
> domains without hap,
Why would that be?
Jan
> but not like it really would count for much, so I'm OK
> with having the compact one-liner version.
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 9:14 ` Jan Beulich
@ 2014-09-05 9:45 ` Tamas K Lengyel
2014-09-05 10:07 ` Jan Beulich
0 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:45 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: 1593 bytes --]
On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> > --- a/xen/arch/x86/x86_64/mm.c
> > +++ b/xen/arch/x86/x86_64/mm.c
> > @@ -35,9 +35,9 @@
> > #include <asm/msr.h>
> > #include <asm/setup.h>
> > #include <asm/numa.h>
> > -#include <asm/mem_event.h>
> > +#include <xen/mem_event.h>
> > #include <asm/mem_sharing.h>
> > -#include <asm/mem_access.h>
> > +#include <xen/mem_access.h>
> > #include <public/memory.h>
>
> This is not the only place, but a specifically bad example: I'm pretty
> sure I asked you before to not make a mess by mixing asm/ and
> xen/ included - move the now xen/ ones to the other ones already
> coming from that directory. And do so consistently throughout the
> patch.
>
>
Sure. Is this a style-thing or does it have some other effect on the code?
> > --- 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
>
> The addition should go alongside the other CONFIG_* ones instead
> of at an arbitrary place.
Ack.
> Looking at this again I wonder though
> whether we really need these - the use sites could easily check
> whether p2m_is_{paging,shared} is defined instead.
>
> Jan
>
I would still need to wrap them in CONFIG_X86 blocks as common/memory.c
does, so ultimately I'm not sure there is much difference.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 2729 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] 45+ messages in thread
* Re: [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks
2014-09-05 9:44 ` Jan Beulich
@ 2014-09-05 9:47 ` Tamas K Lengyel
0 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:47 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: 1543 bytes --]
On Fri, Sep 5, 2014 at 11:44 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 11:35, <tamas.lengyel@zentific.com> wrote:
> > On Fri, Sep 5, 2014 at 11:28 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >
> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> >> > --- a/xen/include/asm-x86/p2m.h
> >> > +++ b/xen/include/asm-x86/p2m.h
> >> > @@ -611,6 +611,20 @@ 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)
> >> > +{
> >> > + /* Only HAP is supported */
> >> > + if ( !hap_enabled(d) )
> >> > + return 0;
> >> > +
> >> > + /* Currently only EPT is supported */
> >> > + if ( !cpu_has_vmx )
> >> > + return 0;
> >> > +
> >> > + return 1;
> >> > +}
> >>
> >> For simple inline functions like this, may I suggest have them a use
> >> simple return statement:
> >>
> >> return hap_enabled(d) && cpu_has_vmx;
> >>
> >>
> > Surely, my aim here really was just to minimize look&feel changes
> compared
> > to existing code. The current approach might be a tiny bit faster on
> > domains without hap,
>
> Why would that be?
>
> Jan
>
>
Never mind, rethinking it the compact one-liner version doesn't evaluate
the part after && if the first half already failed, same as it is now.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 2604 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] 45+ messages in thread
* Re: [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common
2014-09-05 9:43 ` Jan Beulich
@ 2014-09-05 9:48 ` Tamas K Lengyel
0 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:48 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: 1089 bytes --]
On Fri, Sep 5, 2014 at 11:43 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 11:25, <tamas.lengyel@zentific.com> wrote:
> > On Fri, Sep 5, 2014 at 11:16 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >
> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> >> > --- a/xen/common/mem_access.c
> >> > +++ b/xen/common/mem_access.c
> >> > @@ -29,6 +29,30 @@
> >> > #include <xen/mem_event.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) )
> >>
> >> Please take the opportunity to fix the coding style here.
> >
> > Surely, but what would be preferred coding style here? A separate int rc
> > for the value returned by mem_event_get_response?
>
> No, I really just mean formatting: There is a stop missing in the
> comment (and there was another such comment further down
> iirc), and a blank missing after "while".
>
> Jan
>
I see, OK.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1951 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] 45+ messages in thread
* Re: [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check
2014-09-05 9:29 ` Jan Beulich
@ 2014-09-05 9:51 ` Tamas K Lengyel
0 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 9:51 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: 769 bytes --]
On Fri, Sep 5, 2014 at 11:29 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> > --- a/xen/include/asm-x86/p2m.h
> > +++ b/xen/include/asm-x86/p2m.h
> > @@ -625,6 +625,14 @@ static inline bool_t
> p2m_mem_event_sanity_check(struct domain *d)
> > return 1;
> > }
> >
> > +/* Sanity check for mem_access hardware support */
> > +static inline bool_t p2m_mem_access_sanity_check(struct domain *d)
> > +{
> > + if ( !is_hvm_domain(d) )
> > + return 0;
> > + return 1;
> > +}
>
> In line with the previous one - just "return is_hvm_domain(d);" or
> even make this a #define.
>
> Jan
>
Ack. I prefer static inlines over #defines, it's easier to see what input
the function expects that way.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1350 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] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 9:45 ` Tamas K Lengyel
@ 2014-09-05 10:07 ` Jan Beulich
2014-09-05 10:36 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 10:07 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 05.09.14 at 11:45, <tamas.lengyel@zentific.com> wrote:
> On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com> wrote:
>
>> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
>> > --- a/xen/arch/x86/x86_64/mm.c
>> > +++ b/xen/arch/x86/x86_64/mm.c
>> > @@ -35,9 +35,9 @@
>> > #include <asm/msr.h>
>> > #include <asm/setup.h>
>> > #include <asm/numa.h>
>> > -#include <asm/mem_event.h>
>> > +#include <xen/mem_event.h>
>> > #include <asm/mem_sharing.h>
>> > -#include <asm/mem_access.h>
>> > +#include <xen/mem_access.h>
>> > #include <public/memory.h>
>>
>> This is not the only place, but a specifically bad example: I'm pretty
>> sure I asked you before to not make a mess by mixing asm/ and
>> xen/ included - move the now xen/ ones to the other ones already
>> coming from that directory. And do so consistently throughout the
>> patch.
>>
>>
> Sure. Is this a style-thing or does it have some other effect on the code?
A style thing that is not supposed to have an effect (we should
strive for headers to include their dependencies rather than relying
on their users doing so up front).
>> Looking at this again I wonder though
>> whether we really need these - the use sites could easily check
>> whether p2m_is_{paging,shared} is defined instead.
>
> I would still need to wrap them in CONFIG_X86 blocks as common/memory.c
> does, so ultimately I'm not sure there is much difference.
I don't understand why - ARM doesn't define p2m_is_paging()
or p2m_is_shared().
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 10:07 ` Jan Beulich
@ 2014-09-05 10:36 ` Tamas K Lengyel
2014-09-05 10:48 ` Jan Beulich
0 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 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: 1790 bytes --]
On Fri, Sep 5, 2014 at 12:07 PM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 11:45, <tamas.lengyel@zentific.com> wrote:
> > On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com> wrote:
> >
> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> >> > --- a/xen/arch/x86/x86_64/mm.c
> >> > +++ b/xen/arch/x86/x86_64/mm.c
> >> > @@ -35,9 +35,9 @@
> >> > #include <asm/msr.h>
> >> > #include <asm/setup.h>
> >> > #include <asm/numa.h>
> >> > -#include <asm/mem_event.h>
> >> > +#include <xen/mem_event.h>
> >> > #include <asm/mem_sharing.h>
> >> > -#include <asm/mem_access.h>
> >> > +#include <xen/mem_access.h>
> >> > #include <public/memory.h>
> >>
> >> This is not the only place, but a specifically bad example: I'm pretty
> >> sure I asked you before to not make a mess by mixing asm/ and
> >> xen/ included - move the now xen/ ones to the other ones already
> >> coming from that directory. And do so consistently throughout the
> >> patch.
> >>
> >>
> > Sure. Is this a style-thing or does it have some other effect on the
> code?
>
> A style thing that is not supposed to have an effect (we should
> strive for headers to include their dependencies rather than relying
> on their users doing so up front).
>
> >> Looking at this again I wonder though
> >> whether we really need these - the use sites could easily check
> >> whether p2m_is_{paging,shared} is defined instead.
> >
> > I would still need to wrap them in CONFIG_X86 blocks as common/memory.c
> > does, so ultimately I'm not sure there is much difference.
>
> I don't understand why - ARM doesn't define p2m_is_paging()
> or p2m_is_shared().
>
> Jan
>
Hm, I guess I could do that, just define them to return 0 and hope the
compiler just drops the always-dead if blocks.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 2807 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] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 10:36 ` Tamas K Lengyel
@ 2014-09-05 10:48 ` Jan Beulich
2014-09-05 10:52 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 10:48 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 05.09.14 at 12:36, <tamas.lengyel@zentific.com> wrote:
> On Fri, Sep 5, 2014 at 12:07 PM, Jan Beulich <JBeulich@suse.com> wrote:
>
>> >>> On 05.09.14 at 11:45, <tamas.lengyel@zentific.com> wrote:
>> > On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com> wrote:
>> >
>> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
>> >> > --- a/xen/arch/x86/x86_64/mm.c
>> >> > +++ b/xen/arch/x86/x86_64/mm.c
>> >> > @@ -35,9 +35,9 @@
>> >> > #include <asm/msr.h>
>> >> > #include <asm/setup.h>
>> >> > #include <asm/numa.h>
>> >> > -#include <asm/mem_event.h>
>> >> > +#include <xen/mem_event.h>
>> >> > #include <asm/mem_sharing.h>
>> >> > -#include <asm/mem_access.h>
>> >> > +#include <xen/mem_access.h>
>> >> > #include <public/memory.h>
>> >>
>> >> This is not the only place, but a specifically bad example: I'm pretty
>> >> sure I asked you before to not make a mess by mixing asm/ and
>> >> xen/ included - move the now xen/ ones to the other ones already
>> >> coming from that directory. And do so consistently throughout the
>> >> patch.
>> >>
>> >>
>> > Sure. Is this a style-thing or does it have some other effect on the
>> code?
>>
>> A style thing that is not supposed to have an effect (we should
>> strive for headers to include their dependencies rather than relying
>> on their users doing so up front).
>>
>> >> Looking at this again I wonder though
>> >> whether we really need these - the use sites could easily check
>> >> whether p2m_is_{paging,shared} is defined instead.
>> >
>> > I would still need to wrap them in CONFIG_X86 blocks as common/memory.c
>> > does, so ultimately I'm not sure there is much difference.
>>
>> I don't understand why - ARM doesn't define p2m_is_paging()
>> or p2m_is_shared().
>
> Hm, I guess I could do that, just define them to return 0 and hope the
> compiler just drops the always-dead if blocks.
No, that's not what I was asking you to do, as that may require you
to defined further stubs used inside the conditionals. Instead I asked
that you replace "#ifdef CONFIG_..." with "#ifdef p2m_is_...".
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 10:48 ` Jan Beulich
@ 2014-09-05 10:52 ` Tamas K Lengyel
2014-09-05 11:26 ` Jan Beulich
0 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 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: 2333 bytes --]
On Fri, Sep 5, 2014 at 12:48 PM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 12:36, <tamas.lengyel@zentific.com> wrote:
> > On Fri, Sep 5, 2014 at 12:07 PM, Jan Beulich <JBeulich@suse.com> wrote:
> >
> >> >>> On 05.09.14 at 11:45, <tamas.lengyel@zentific.com> wrote:
> >> > On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com>
> wrote:
> >> >
> >> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> >> >> > --- a/xen/arch/x86/x86_64/mm.c
> >> >> > +++ b/xen/arch/x86/x86_64/mm.c
> >> >> > @@ -35,9 +35,9 @@
> >> >> > #include <asm/msr.h>
> >> >> > #include <asm/setup.h>
> >> >> > #include <asm/numa.h>
> >> >> > -#include <asm/mem_event.h>
> >> >> > +#include <xen/mem_event.h>
> >> >> > #include <asm/mem_sharing.h>
> >> >> > -#include <asm/mem_access.h>
> >> >> > +#include <xen/mem_access.h>
> >> >> > #include <public/memory.h>
> >> >>
> >> >> This is not the only place, but a specifically bad example: I'm
> pretty
> >> >> sure I asked you before to not make a mess by mixing asm/ and
> >> >> xen/ included - move the now xen/ ones to the other ones already
> >> >> coming from that directory. And do so consistently throughout the
> >> >> patch.
> >> >>
> >> >>
> >> > Sure. Is this a style-thing or does it have some other effect on the
> >> code?
> >>
> >> A style thing that is not supposed to have an effect (we should
> >> strive for headers to include their dependencies rather than relying
> >> on their users doing so up front).
> >>
> >> >> Looking at this again I wonder though
> >> >> whether we really need these - the use sites could easily check
> >> >> whether p2m_is_{paging,shared} is defined instead.
> >> >
> >> > I would still need to wrap them in CONFIG_X86 blocks as
> common/memory.c
> >> > does, so ultimately I'm not sure there is much difference.
> >>
> >> I don't understand why - ARM doesn't define p2m_is_paging()
> >> or p2m_is_shared().
> >
> > Hm, I guess I could do that, just define them to return 0 and hope the
> > compiler just drops the always-dead if blocks.
>
> No, that's not what I was asking you to do, as that may require you
> to defined further stubs used inside the conditionals. Instead I asked
> that you replace "#ifdef CONFIG_..." with "#ifdef p2m_is_...".
>
> Jan
>
>
Oh, OK. Isn't that a bit hack-ish looking though?
Tamas
[-- Attachment #1.2: Type: text/html, Size: 3777 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] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 10:52 ` Tamas K Lengyel
@ 2014-09-05 11:26 ` Jan Beulich
2014-09-05 11:30 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 11:26 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 05.09.14 at 12:52, <tamas.lengyel@zentific.com> wrote:
> On Fri, Sep 5, 2014 at 12:48 PM, Jan Beulich <JBeulich@suse.com> wrote:
>
>> >>> On 05.09.14 at 12:36, <tamas.lengyel@zentific.com> wrote:
>> > On Fri, Sep 5, 2014 at 12:07 PM, Jan Beulich <JBeulich@suse.com> wrote:
>> >
>> >> >>> On 05.09.14 at 11:45, <tamas.lengyel@zentific.com> wrote:
>> >> > On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com>
>> wrote:
>> >> >
>> >> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
>> >> >> > --- a/xen/arch/x86/x86_64/mm.c
>> >> >> > +++ b/xen/arch/x86/x86_64/mm.c
>> >> >> > @@ -35,9 +35,9 @@
>> >> >> > #include <asm/msr.h>
>> >> >> > #include <asm/setup.h>
>> >> >> > #include <asm/numa.h>
>> >> >> > -#include <asm/mem_event.h>
>> >> >> > +#include <xen/mem_event.h>
>> >> >> > #include <asm/mem_sharing.h>
>> >> >> > -#include <asm/mem_access.h>
>> >> >> > +#include <xen/mem_access.h>
>> >> >> > #include <public/memory.h>
>> >> >>
>> >> >> This is not the only place, but a specifically bad example: I'm
>> pretty
>> >> >> sure I asked you before to not make a mess by mixing asm/ and
>> >> >> xen/ included - move the now xen/ ones to the other ones already
>> >> >> coming from that directory. And do so consistently throughout the
>> >> >> patch.
>> >> >>
>> >> >>
>> >> > Sure. Is this a style-thing or does it have some other effect on the
>> >> code?
>> >>
>> >> A style thing that is not supposed to have an effect (we should
>> >> strive for headers to include their dependencies rather than relying
>> >> on their users doing so up front).
>> >>
>> >> >> Looking at this again I wonder though
>> >> >> whether we really need these - the use sites could easily check
>> >> >> whether p2m_is_{paging,shared} is defined instead.
>> >> >
>> >> > I would still need to wrap them in CONFIG_X86 blocks as
>> common/memory.c
>> >> > does, so ultimately I'm not sure there is much difference.
>> >>
>> >> I don't understand why - ARM doesn't define p2m_is_paging()
>> >> or p2m_is_shared().
>> >
>> > Hm, I guess I could do that, just define them to return 0 and hope the
>> > compiler just drops the always-dead if blocks.
>>
>> No, that's not what I was asking you to do, as that may require you
>> to defined further stubs used inside the conditionals. Instead I asked
>> that you replace "#ifdef CONFIG_..." with "#ifdef p2m_is_...".
>>
> Oh, OK. Isn't that a bit hack-ish looking though?
Some may say it is, but I personally don't think so. After all there's
a clear connection between the #ifdef and the subsequent code.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 11:26 ` Jan Beulich
@ 2014-09-05 11:30 ` Tamas K Lengyel
2014-09-05 11:46 ` Jan Beulich
0 siblings, 1 reply; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-05 11:30 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: 2914 bytes --]
On Fri, Sep 5, 2014 at 1:26 PM, Jan Beulich <JBeulich@suse.com> wrote:
> >>> On 05.09.14 at 12:52, <tamas.lengyel@zentific.com> wrote:
> > On Fri, Sep 5, 2014 at 12:48 PM, Jan Beulich <JBeulich@suse.com> wrote:
> >
> >> >>> On 05.09.14 at 12:36, <tamas.lengyel@zentific.com> wrote:
> >> > On Fri, Sep 5, 2014 at 12:07 PM, Jan Beulich <JBeulich@suse.com>
> wrote:
> >> >
> >> >> >>> On 05.09.14 at 11:45, <tamas.lengyel@zentific.com> wrote:
> >> >> > On Fri, Sep 5, 2014 at 11:14 AM, Jan Beulich <JBeulich@suse.com>
> >> wrote:
> >> >> >
> >> >> >> >>> On 05.09.14 at 10:58, <tklengyel@sec.in.tum.de> wrote:
> >> >> >> > --- a/xen/arch/x86/x86_64/mm.c
> >> >> >> > +++ b/xen/arch/x86/x86_64/mm.c
> >> >> >> > @@ -35,9 +35,9 @@
> >> >> >> > #include <asm/msr.h>
> >> >> >> > #include <asm/setup.h>
> >> >> >> > #include <asm/numa.h>
> >> >> >> > -#include <asm/mem_event.h>
> >> >> >> > +#include <xen/mem_event.h>
> >> >> >> > #include <asm/mem_sharing.h>
> >> >> >> > -#include <asm/mem_access.h>
> >> >> >> > +#include <xen/mem_access.h>
> >> >> >> > #include <public/memory.h>
> >> >> >>
> >> >> >> This is not the only place, but a specifically bad example: I'm
> >> pretty
> >> >> >> sure I asked you before to not make a mess by mixing asm/ and
> >> >> >> xen/ included - move the now xen/ ones to the other ones already
> >> >> >> coming from that directory. And do so consistently throughout the
> >> >> >> patch.
> >> >> >>
> >> >> >>
> >> >> > Sure. Is this a style-thing or does it have some other effect on
> the
> >> >> code?
> >> >>
> >> >> A style thing that is not supposed to have an effect (we should
> >> >> strive for headers to include their dependencies rather than relying
> >> >> on their users doing so up front).
> >> >>
> >> >> >> Looking at this again I wonder though
> >> >> >> whether we really need these - the use sites could easily check
> >> >> >> whether p2m_is_{paging,shared} is defined instead.
> >> >> >
> >> >> > I would still need to wrap them in CONFIG_X86 blocks as
> >> common/memory.c
> >> >> > does, so ultimately I'm not sure there is much difference.
> >> >>
> >> >> I don't understand why - ARM doesn't define p2m_is_paging()
> >> >> or p2m_is_shared().
> >> >
> >> > Hm, I guess I could do that, just define them to return 0 and hope the
> >> > compiler just drops the always-dead if blocks.
> >>
> >> No, that's not what I was asking you to do, as that may require you
> >> to defined further stubs used inside the conditionals. Instead I asked
> >> that you replace "#ifdef CONFIG_..." with "#ifdef p2m_is_...".
> >>
> > Oh, OK. Isn't that a bit hack-ish looking though?
>
> Some may say it is, but I personally don't think so. After all there's
> a clear connection between the #ifdef and the subsequent code.
>
> Jan
>
>
I agree I just don't see such an approach used anywhere else. Let's wait to
hear from others what their thoughts are on this.
Tamas
[-- Attachment #1.2: Type: text/html, Size: 4850 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] 45+ messages in thread
* Re: [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common.
2014-09-05 11:30 ` Tamas K Lengyel
@ 2014-09-05 11:46 ` Jan Beulich
0 siblings, 0 replies; 45+ messages in thread
From: Jan Beulich @ 2014-09-05 11:46 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 05.09.14 at 13:30, <tamas.lengyel@zentific.com> wrote:
> I agree I just don't see such an approach used anywhere else. Let's wait to
> hear from others what their thoughts are on this.
Just grep for "run_in_exception_handler" for example.
Jan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 09/16] xen/arm: p2m type definitions and changes
2014-09-05 8:58 ` [PATCH v4 09/16] xen/arm: p2m type definitions and changes Tamas K Lengyel
@ 2014-09-09 0:06 ` Julien Grall
0 siblings, 0 replies; 45+ messages in thread
From: Julien Grall @ 2014-09-09 0:06 UTC (permalink / raw)
To: Tamas K Lengyel, xen-devel
Cc: ian.campbell, tim, ian.jackson, stefano.stabellini, andres,
jbeulich, dgdegra
Hi Tamas,
On 05/09/14 01:58, Tamas K Lengyel wrote:
> diff --git a/xen/include/asm-arm/p2m.h b/xen/include/asm-arm/p2m.h
> index 06c93a0..b2009ee 100644
> --- a/xen/include/asm-arm/p2m.h
> +++ b/xen/include/asm-arm/p2m.h
> @@ -2,9 +2,54 @@
[..]
> +#include <public/memory.h>
> +#include <public/mem_event.h>
Why do you need this 2 includes?
>
> 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;
> +
Any reason to move the enum earlier? If not, I would prefer to keep at
the same place. It will be easier with git-blame to find when a new type
has been added.
> +/*
> + * Additional access types, which are used to further restrict
> + * the permissions given by 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
ambiguous
[..]
> + /* Default P2M access type for each page in the the domain: new pages,
> + * swapped in pages, cleared pages, and pages that are ambiquously
Did you intend to mean ambiguously rather than ambiquously?
[..]
> +/* get host p2m table */
> +#define p2m_get_hostp2m(d) (&((d)->arch.p2m))
> +
> +/* mem_event and mem_access are supported on all ARM */
I don't find "all ARM" clear. I would replace by "any ARM guest"
Regards,
--
Julien Grall
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 10/16] xen/arm: Add set access required domctl
2014-09-05 8:58 ` [PATCH v4 10/16] xen/arm: Add set access required domctl Tamas K Lengyel
@ 2014-09-09 0:07 ` Julien Grall
2014-09-12 15:54 ` Tamas K Lengyel
0 siblings, 1 reply; 45+ messages in thread
From: Julien Grall @ 2014-09-09 0:07 UTC (permalink / raw)
To: Tamas K Lengyel, xen-devel
Cc: ian.campbell, tim, ian.jackson, stefano.stabellini, andres,
jbeulich, dgdegra
Hi Tamas,
On 05/09/14 01:58, Tamas K Lengyel wrote:
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
Reviewed-by: Julien Grall <julien.grall@linaro.org>
Regards,
--
Julien Grall
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 11/16] xen/arm: Data abort exception (R/W) mem_events.
2014-09-05 8:58 ` [PATCH v4 11/16] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
@ 2014-09-09 0:25 ` Julien Grall
0 siblings, 0 replies; 45+ messages in thread
From: Julien Grall @ 2014-09-09 0:25 UTC (permalink / raw)
To: Tamas K Lengyel, xen-devel
Cc: ian.campbell, tim, ian.jackson, stefano.stabellini, andres,
jbeulich, dgdegra
Hi Tamas,
On 05/09/14 01:58, Tamas K Lengyel wrote:
> #define P2M_ONE_DESCEND 0
> #define P2M_ONE_PROGRESS_NOP 0x1
> #define P2M_ONE_PROGRESS 0x10
> @@ -502,6 +568,14 @@ static int apply_one_level(struct domain *d,
> pte = mfn_to_p2m_entry(page_to_mfn(page), mattr, t, a);
> if ( level < 3 )
> pte.p2m.table = 0;
> +
> + if ( p2m_mem_access_radix_set(p2m, paddr_to_pfn(*addr), a) < 0 )
> + {
> + /* If we fail to save the setting in the Radix tree, we
> + * need to clear the PTE mem_access permissions. */
> + p2m_set_permission(&pte, pte.p2m.type, p2m_access_rwx);
> + }
> +
[..]
> @@ -538,6 +612,13 @@ static int apply_one_level(struct domain *d,
> if ( level < 3 )
> pte.p2m.table = 0; /* Superpage entry */
>
> + if ( p2m_mem_access_radix_set(p2m, paddr_to_pfn(*addr), a) < 0 )
> + {
> + /* If we fail to save the setting in the Radix tree, we
> + * need to clear the PTE mem_access permissions. */
> + p2m_set_permission(&pte, pte.p2m.type, p2m_access_rwx);
> + }
> +
Can't this be abstract to another helper?
Also is it acceptable from mem access POV to unconditionally set the
access to rwx when an error occured? Shouldn't you report an error?
> +static int p2m_set_entry(struct domain *d, paddr_t paddr, p2m_access_t p2ma)
> +{
As I said on the previous version, I still think this can be merge in
apply_p2m_changes without any difficulties.
The main reason is to avoid increasing the number of function that have
a common pattern to use the p2m. Also apply_p2m_changes have some
interesting performance improvement (such as avoid map/unmap every time
page table).
Finally, it will make life simpler when 4-Level page table will be
supported (see Ian's series [1])
[..]
> +bool_t p2m_mem_access_check(paddr_t gpa, vaddr_t gla, struct npfec npfec)
const struct npfect
[..]
> diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
> index 019991f..7eb875a 100644
> --- a/xen/arch/arm/traps.c
> +++ b/xen/arch/arm/traps.c
> @@ -1852,13 +1852,38 @@ static void do_trap_data_abort_guest(struct cpu_user_regs *regs,
[..]
> + switch ( dabt.dfsc )
> + {
> + case DABT_DFSC_PERMISSION_1:
> + case DABT_DFSC_PERMISSION_2:
> + case DABT_DFSC_PERMISSION_3:
> + {
> + struct npfec npfec = {
const as you won't modify it later.
Regards,
[1] http://lists.xen.org/archives/html/xen-devel/2014-09/msg00596.html
--
Julien Grall
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 16/16] tools/tests: Enable xen-access on ARM
2014-09-05 8:58 ` [PATCH v4 16/16] tools/tests: Enable xen-access " Tamas K Lengyel
@ 2014-09-09 0:29 ` Julien Grall
0 siblings, 0 replies; 45+ messages in thread
From: Julien Grall @ 2014-09-09 0:29 UTC (permalink / raw)
To: Tamas K Lengyel, xen-devel
Cc: ian.campbell, tim, ian.jackson, stefano.stabellini, andres,
jbeulich, dgdegra
Hi Tamas,
On 05/09/14 01:58, Tamas K Lengyel wrote:
> On ARM the guest memory doesn't start from 0 and has multiple banks
> of memory, thus we adjust the setting/unsetting routines accordingly.
> We also define the ARM specific test_and_set_bit function.
>
> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
> ---
> 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 | 94 +++++++++++++++++++++++++++++++++----
> xen/arch/arm/Rules.mk | 1 -
> xen/arch/x86/Rules.mk | 1 -
> 8 files changed, 97 insertions(+), 14 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
Why can't you move HAS_MEM_ACCESS directly here in patch #1?
It would make this patch simpler.
--
Julien Grall
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v4 10/16] xen/arm: Add set access required domctl
2014-09-09 0:07 ` Julien Grall
@ 2014-09-12 15:54 ` Tamas K Lengyel
0 siblings, 0 replies; 45+ messages in thread
From: Tamas K Lengyel @ 2014-09-12 15:54 UTC (permalink / raw)
To: Julien Grall
Cc: Ian Campbell, Tim Deegan, Ian Jackson, xen-devel,
Stefano Stabellini, Andres Lagar-Cavilla, Jan Beulich,
Daniel De Graaf, Tamas K Lengyel
[-- Attachment #1.1: Type: text/plain, Size: 392 bytes --]
On Tue, Sep 9, 2014 at 2:07 AM, Julien Grall <julien.grall@linaro.org>
wrote:
> Hi Tamas,
>
> On 05/09/14 01:58, Tamas K Lengyel wrote:
>
>> Signed-off-by: Tamas K Lengyel <tklengyel@sec.in.tum.de>
>>
> Reviewed-by: Julien Grall <julien.grall@linaro.org>
>
> Regards,
>
> --
> Julien Grall
>
I seem to have missed some of your reviews on v4 so I'm just digging them
up now..
Thanks!
Tamas
[-- Attachment #1.2: Type: text/html, Size: 1100 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] 45+ messages in thread
end of thread, other threads:[~2014-09-12 15:54 UTC | newest]
Thread overview: 45+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-09-05 8:58 [PATCH v4 00/16] Mem_event and mem_access for ARM Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 01/16] xen: Relocate mem_access and mem_event into common Tamas K Lengyel
2014-09-05 9:14 ` Jan Beulich
2014-09-05 9:45 ` Tamas K Lengyel
2014-09-05 10:07 ` Jan Beulich
2014-09-05 10:36 ` Tamas K Lengyel
2014-09-05 10:48 ` Jan Beulich
2014-09-05 10:52 ` Tamas K Lengyel
2014-09-05 11:26 ` Jan Beulich
2014-09-05 11:30 ` Tamas K Lengyel
2014-09-05 11:46 ` Jan Beulich
2014-09-05 8:58 ` [PATCH v4 02/16] xen: Relocate p2m_mem_access_resume to mem_access common Tamas K Lengyel
2014-09-05 9:16 ` Jan Beulich
2014-09-05 9:25 ` Tamas K Lengyel
2014-09-05 9:43 ` Jan Beulich
2014-09-05 9:48 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 03/16] xen: Relocate struct npfec definition into common Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 04/16] xen: Relocate mem_event_op domctl and access_op memop " Tamas K Lengyel
2014-09-05 9:23 ` Jan Beulich
2014-09-05 9:29 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 05/16] xen/mem_event: Clean out superfluous white-spaces Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 06/16] xen/mem_event: Relax error condition on debug builds Tamas K Lengyel
2014-09-05 9:25 ` Jan Beulich
2014-09-05 9:26 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 07/16] xen/mem_event: Abstract architecture specific sanity checks Tamas K Lengyel
2014-09-05 9:28 ` Jan Beulich
2014-09-05 9:35 ` Tamas K Lengyel
2014-09-05 9:44 ` Jan Beulich
2014-09-05 9:47 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 08/16] xen/mem_access: Abstract architecture specific sanity check Tamas K Lengyel
2014-09-05 9:29 ` Jan Beulich
2014-09-05 9:51 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 09/16] xen/arm: p2m type definitions and changes Tamas K Lengyel
2014-09-09 0:06 ` Julien Grall
2014-09-05 8:58 ` [PATCH v4 10/16] xen/arm: Add set access required domctl Tamas K Lengyel
2014-09-09 0:07 ` Julien Grall
2014-09-12 15:54 ` Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 11/16] xen/arm: Data abort exception (R/W) mem_events Tamas K Lengyel
2014-09-09 0:25 ` Julien Grall
2014-09-05 8:58 ` [PATCH v4 12/16] xen/arm: Instruction prefetch abort (X) mem_event handling Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 13/16] xen/arm: Shatter large pages when using mem_acces Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 14/16] xen/arm: Enable the compilation of mem_access and mem_event on ARM Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 15/16] tools/libxc: Allocate magic page for mem access " Tamas K Lengyel
2014-09-05 8:58 ` [PATCH v4 16/16] tools/tests: Enable xen-access " Tamas K Lengyel
2014-09-09 0:29 ` Julien Grall
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.