Scrubbing pages on initial balloon down can take some time, especially
in nested virtualization case (nested EPT is slow). When HVM/PVH guest is
started with memory= significantly lower than maxmem=, all the extra
pages will be scrubbed before returning to Xen. But since most of them
weren't used at all at that point, Xen needs to populate them first
(from populate-on-demand pool). In nested virt case (Xen inside KVM)
this slows down the guest boot by 15-30s with just 1.5GB needed to be
returned to Xen.
Add runtime parameter to enable/disable it, to allow initially disabling
scrubbing, then enable it back during boot (for example in initramfs).
Such usage relies on assumption that a) most pages ballooned out during
initial boot weren't used at all, and b) even if they were, very few
secrets are in the guest at that time (before any serious userspace
kicks in).
Convert CONFIG_XEN_SCRUB_PAGES to CONFIG_XEN_SCRUB_PAGES_DEFAULT (also
enabled by default), controlling default value for the new runtime
switch.
Signed-off-by: Marek Marczykowski-Górecki <[email protected]>
---
Changes in v2:
- move sysfs control to /sys/devices/system/xen_memory
- use core_param() to avoid confusing prefix for the option name
- document option
- change CONFIG_XEN_SCRUB_PAGES to CONFIG_XEN_SCRUB_PAGES_DEFAULT,
controlling only the default value for the runtime option
---
.../ABI/stable/sysfs-devices-system-xen_memory | 9 +++++++++
Documentation/admin-guide/kernel-parameters.txt | 6 ++++++
drivers/xen/Kconfig | 10 +++++++---
drivers/xen/mem-reservation.c | 8 ++++++++
drivers/xen/xen-balloon.c | 3 +++
include/xen/mem-reservation.h | 7 ++++---
6 files changed, 37 insertions(+), 6 deletions(-)
diff --git a/Documentation/ABI/stable/sysfs-devices-system-xen_memory b/Documentation/ABI/stable/sysfs-devices-system-xen_memory
index caa311d59ac1..6d83f95a8a8e 100644
--- a/Documentation/ABI/stable/sysfs-devices-system-xen_memory
+++ b/Documentation/ABI/stable/sysfs-devices-system-xen_memory
@@ -75,3 +75,12 @@ Contact: Konrad Rzeszutek Wilk <[email protected]>
Description:
Amount (in KiB) of low (or normal) memory in the
balloon.
+
+What: /sys/devices/system/xen_memory/xen_memory0/scrub_pages
+Date: September 2018
+KernelVersion: 4.20
+Contact: [email protected]
+Description:
+ Control scrubbing pages before returning them to Xen for others domains
+ use. Can be set with xen_scrub_pages cmdline
+ parameter. Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 9871e649ffef..0f20282629de 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -4994,6 +4994,12 @@
Disables the PV optimizations forcing the HVM guest to
run as generic HVM guest with no PV drivers.
+ xen_scrub_pages= [XEN]
+ Boolean option to control scrubbing pages before giving them back
+ to Xen, for use by other domains. Can be also changed at runtime
+ with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
+ Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
+
xirc2ps_cs= [NET,PCMCIA]
Format:
<irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
diff --git a/drivers/xen/Kconfig b/drivers/xen/Kconfig
index b459edfacff3..90d387b50ab7 100644
--- a/drivers/xen/Kconfig
+++ b/drivers/xen/Kconfig
@@ -79,15 +79,19 @@ config XEN_BALLOON_MEMORY_HOTPLUG_LIMIT
This value is used to allocate enough space in internal
tables needed for physical memory administration.
-config XEN_SCRUB_PAGES
- bool "Scrub pages before returning them to system"
+config XEN_SCRUB_PAGES_DEFAULT
+ bool "Scrub pages before returning them to system by default"
depends on XEN_BALLOON
default y
help
Scrub pages before returning them to the system for reuse by
other domains. This makes sure that any confidential data
is not accidentally visible to other domains. Is it more
- secure, but slightly less efficient.
+ secure, but slightly less efficient. This can be controlled with
+ xen_scrub_pages=0 parameter and
+ /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
+ This option only sets the default value.
+
If in doubt, say yes.
config XEN_DEV_EVTCHN
diff --git a/drivers/xen/mem-reservation.c b/drivers/xen/mem-reservation.c
index 084799c6180e..14276e2d76fe 100644
--- a/drivers/xen/mem-reservation.c
+++ b/drivers/xen/mem-reservation.c
@@ -14,6 +14,14 @@
#include <xen/interface/memory.h>
#include <xen/mem-reservation.h>
+#include <linux/moduleparam.h>
+
+#ifdef CONFIG_XEN_SCRUB_PAGES_DEFAULT
+bool __read_mostly xen_scrub_pages = true;
+#else
+bool __read_mostly xen_scrub_pages = false;
+#endif
+core_param(xen_scrub_pages, xen_scrub_pages, bool, 0);
/*
* Use one extent per PAGE_SIZE to avoid to break down the page into
diff --git a/drivers/xen/xen-balloon.c b/drivers/xen/xen-balloon.c
index 294f35ce9e46..63c1494a8d73 100644
--- a/drivers/xen/xen-balloon.c
+++ b/drivers/xen/xen-balloon.c
@@ -44,6 +44,7 @@
#include <xen/xenbus.h>
#include <xen/features.h>
#include <xen/page.h>
+#include <xen/mem-reservation.h>
#define PAGES2KB(_p) ((_p)<<(PAGE_SHIFT-10))
@@ -137,6 +138,7 @@ static DEVICE_ULONG_ATTR(schedule_delay, 0444, balloon_stats.schedule_delay);
static DEVICE_ULONG_ATTR(max_schedule_delay, 0644, balloon_stats.max_schedule_delay);
static DEVICE_ULONG_ATTR(retry_count, 0444, balloon_stats.retry_count);
static DEVICE_ULONG_ATTR(max_retry_count, 0644, balloon_stats.max_retry_count);
+static DEVICE_BOOL_ATTR(scrub_pages, 0644, xen_scrub_pages);
static ssize_t show_target_kb(struct device *dev, struct device_attribute *attr,
char *buf)
@@ -203,6 +205,7 @@ static struct attribute *balloon_attrs[] = {
&dev_attr_max_schedule_delay.attr.attr,
&dev_attr_retry_count.attr.attr,
&dev_attr_max_retry_count.attr.attr,
+ &dev_attr_scrub_pages.attr.attr,
NULL
};
diff --git a/include/xen/mem-reservation.h b/include/xen/mem-reservation.h
index 80b52b4945e9..a2ab516fcd2c 100644
--- a/include/xen/mem-reservation.h
+++ b/include/xen/mem-reservation.h
@@ -17,11 +17,12 @@
#include <xen/page.h>
+extern bool xen_scrub_pages;
+
static inline void xenmem_reservation_scrub_page(struct page *page)
{
-#ifdef CONFIG_XEN_SCRUB_PAGES
- clear_highpage(page);
-#endif
+ if (xen_scrub_pages)
+ clear_highpage(page);
}
#ifdef CONFIG_XEN_HAVE_PVMMU
--
2.17.1
>>> On 07.09.18 at 17:31, <[email protected]> wrote:
> --- a/drivers/xen/mem-reservation.c
> +++ b/drivers/xen/mem-reservation.c
> @@ -14,6 +14,14 @@
>
> #include <xen/interface/memory.h>
> #include <xen/mem-reservation.h>
> +#include <linux/moduleparam.h>
> +
> +#ifdef CONFIG_XEN_SCRUB_PAGES_DEFAULT
> +bool __read_mostly xen_scrub_pages = true;
> +#else
> +bool __read_mostly xen_scrub_pages = false;
> +#endif
Use IS_ENABLED() here instead of the #ifdef?
Jan
Scrubbing pages on initial balloon down can take some time, especially
in nested virtualization case (nested EPT is slow). When HVM/PVH guest is
started with memory= significantly lower than maxmem=, all the extra
pages will be scrubbed before returning to Xen. But since most of them
weren't used at all at that point, Xen needs to populate them first
(from populate-on-demand pool). In nested virt case (Xen inside KVM)
this slows down the guest boot by 15-30s with just 1.5GB needed to be
returned to Xen.
Add runtime parameter to enable/disable it, to allow initially disabling
scrubbing, then enable it back during boot (for example in initramfs).
Such usage relies on assumption that a) most pages ballooned out during
initial boot weren't used at all, and b) even if they were, very few
secrets are in the guest at that time (before any serious userspace
kicks in).
Convert CONFIG_XEN_SCRUB_PAGES to CONFIG_XEN_SCRUB_PAGES_DEFAULT (also
enabled by default), controlling default value for the new runtime
switch.
Signed-off-by: Marek Marczykowski-Górecki <[email protected]>
---
Changes in v2:
- move sysfs control to /sys/devices/system/xen_memory
- use core_param() to avoid confusing prefix for the option name
- document option
- change CONFIG_XEN_SCRUB_PAGES to CONFIG_XEN_SCRUB_PAGES_DEFAULT,
controlling only the default value for the runtime option
Changes in v3:
- use IS_ENABLED()
---
.../ABI/stable/sysfs-devices-system-xen_memory | 9 +++++++++
Documentation/admin-guide/kernel-parameters.txt | 6 ++++++
drivers/xen/Kconfig | 10 +++++++---
drivers/xen/mem-reservation.c | 4 ++++
drivers/xen/xen-balloon.c | 3 +++
include/xen/mem-reservation.h | 7 ++++---
6 files changed, 33 insertions(+), 6 deletions(-)
diff --git a/Documentation/ABI/stable/sysfs-devices-system-xen_memory b/Documentation/ABI/stable/sysfs-devices-system-xen_memory
index caa311d59ac1..6d83f95a8a8e 100644
--- a/Documentation/ABI/stable/sysfs-devices-system-xen_memory
+++ b/Documentation/ABI/stable/sysfs-devices-system-xen_memory
@@ -75,3 +75,12 @@ Contact: Konrad Rzeszutek Wilk <[email protected]>
Description:
Amount (in KiB) of low (or normal) memory in the
balloon.
+
+What: /sys/devices/system/xen_memory/xen_memory0/scrub_pages
+Date: September 2018
+KernelVersion: 4.20
+Contact: [email protected]
+Description:
+ Control scrubbing pages before returning them to Xen for others domains
+ use. Can be set with xen_scrub_pages cmdline
+ parameter. Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 9871e649ffef..0f20282629de 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -4994,6 +4994,12 @@
Disables the PV optimizations forcing the HVM guest to
run as generic HVM guest with no PV drivers.
+ xen_scrub_pages= [XEN]
+ Boolean option to control scrubbing pages before giving them back
+ to Xen, for use by other domains. Can be also changed at runtime
+ with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
+ Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
+
xirc2ps_cs= [NET,PCMCIA]
Format:
<irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
diff --git a/drivers/xen/Kconfig b/drivers/xen/Kconfig
index b459edfacff3..90d387b50ab7 100644
--- a/drivers/xen/Kconfig
+++ b/drivers/xen/Kconfig
@@ -79,15 +79,19 @@ config XEN_BALLOON_MEMORY_HOTPLUG_LIMIT
This value is used to allocate enough space in internal
tables needed for physical memory administration.
-config XEN_SCRUB_PAGES
- bool "Scrub pages before returning them to system"
+config XEN_SCRUB_PAGES_DEFAULT
+ bool "Scrub pages before returning them to system by default"
depends on XEN_BALLOON
default y
help
Scrub pages before returning them to the system for reuse by
other domains. This makes sure that any confidential data
is not accidentally visible to other domains. Is it more
- secure, but slightly less efficient.
+ secure, but slightly less efficient. This can be controlled with
+ xen_scrub_pages=0 parameter and
+ /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
+ This option only sets the default value.
+
If in doubt, say yes.
config XEN_DEV_EVTCHN
diff --git a/drivers/xen/mem-reservation.c b/drivers/xen/mem-reservation.c
index 084799c6180e..3782cf070338 100644
--- a/drivers/xen/mem-reservation.c
+++ b/drivers/xen/mem-reservation.c
@@ -14,6 +14,10 @@
#include <xen/interface/memory.h>
#include <xen/mem-reservation.h>
+#include <linux/moduleparam.h>
+
+bool __read_mostly xen_scrub_pages = IS_ENABLED(CONFIG_XEN_SCRUB_PAGES_DEFAULT);
+core_param(xen_scrub_pages, xen_scrub_pages, bool, 0);
/*
* Use one extent per PAGE_SIZE to avoid to break down the page into
diff --git a/drivers/xen/xen-balloon.c b/drivers/xen/xen-balloon.c
index 294f35ce9e46..63c1494a8d73 100644
--- a/drivers/xen/xen-balloon.c
+++ b/drivers/xen/xen-balloon.c
@@ -44,6 +44,7 @@
#include <xen/xenbus.h>
#include <xen/features.h>
#include <xen/page.h>
+#include <xen/mem-reservation.h>
#define PAGES2KB(_p) ((_p)<<(PAGE_SHIFT-10))
@@ -137,6 +138,7 @@ static DEVICE_ULONG_ATTR(schedule_delay, 0444, balloon_stats.schedule_delay);
static DEVICE_ULONG_ATTR(max_schedule_delay, 0644, balloon_stats.max_schedule_delay);
static DEVICE_ULONG_ATTR(retry_count, 0444, balloon_stats.retry_count);
static DEVICE_ULONG_ATTR(max_retry_count, 0644, balloon_stats.max_retry_count);
+static DEVICE_BOOL_ATTR(scrub_pages, 0644, xen_scrub_pages);
static ssize_t show_target_kb(struct device *dev, struct device_attribute *attr,
char *buf)
@@ -203,6 +205,7 @@ static struct attribute *balloon_attrs[] = {
&dev_attr_max_schedule_delay.attr.attr,
&dev_attr_retry_count.attr.attr,
&dev_attr_max_retry_count.attr.attr,
+ &dev_attr_scrub_pages.attr.attr,
NULL
};
diff --git a/include/xen/mem-reservation.h b/include/xen/mem-reservation.h
index 80b52b4945e9..a2ab516fcd2c 100644
--- a/include/xen/mem-reservation.h
+++ b/include/xen/mem-reservation.h
@@ -17,11 +17,12 @@
#include <xen/page.h>
+extern bool xen_scrub_pages;
+
static inline void xenmem_reservation_scrub_page(struct page *page)
{
-#ifdef CONFIG_XEN_SCRUB_PAGES
- clear_highpage(page);
-#endif
+ if (xen_scrub_pages)
+ clear_highpage(page);
}
#ifdef CONFIG_XEN_HAVE_PVMMU
--
2.17.1
On 07/09/18 18:49, Marek Marczykowski-Górecki wrote:
> Scrubbing pages on initial balloon down can take some time, especially
> in nested virtualization case (nested EPT is slow). When HVM/PVH guest is
> started with memory= significantly lower than maxmem=, all the extra
> pages will be scrubbed before returning to Xen. But since most of them
> weren't used at all at that point, Xen needs to populate them first
> (from populate-on-demand pool). In nested virt case (Xen inside KVM)
> this slows down the guest boot by 15-30s with just 1.5GB needed to be
> returned to Xen.
>
> Add runtime parameter to enable/disable it, to allow initially disabling
> scrubbing, then enable it back during boot (for example in initramfs).
> Such usage relies on assumption that a) most pages ballooned out during
> initial boot weren't used at all, and b) even if they were, very few
> secrets are in the guest at that time (before any serious userspace
> kicks in).
> Convert CONFIG_XEN_SCRUB_PAGES to CONFIG_XEN_SCRUB_PAGES_DEFAULT (also
> enabled by default), controlling default value for the new runtime
> switch.
>
> Signed-off-by: Marek Marczykowski-Górecki <[email protected]>
Reviewed-by: Juergen Gross <[email protected]>
Juergen
On 9/7/18 12:49 PM, Marek Marczykowski-Górecki wrote:
> Scrubbing pages on initial balloon down can take some time, especially
> in nested virtualization case (nested EPT is slow). When HVM/PVH guest is
> started with memory= significantly lower than maxmem=, all the extra
> pages will be scrubbed before returning to Xen. But since most of them
> weren't used at all at that point, Xen needs to populate them first
> (from populate-on-demand pool). In nested virt case (Xen inside KVM)
> this slows down the guest boot by 15-30s with just 1.5GB needed to be
> returned to Xen.
>
> Add runtime parameter to enable/disable it, to allow initially disabling
> scrubbing, then enable it back during boot (for example in initramfs).
> Such usage relies on assumption that a) most pages ballooned out during
> initial boot weren't used at all, and b) even if they were, very few
> secrets are in the guest at that time (before any serious userspace
> kicks in).
> Convert CONFIG_XEN_SCRUB_PAGES to CONFIG_XEN_SCRUB_PAGES_DEFAULT (also
> enabled by default), controlling default value for the new runtime
> switch.
>
> Signed-off-by: Marek Marczykowski-Górecki <[email protected]>
Applied to for-linux-4.19b.
-boris