We found Vista polls ACPI timer very frequently (about 15 times
averagely) when handling timer (RTC @ 64HZ) interrupt routine. Though
the exact reason is known, it should be related to system time
adjustment. When it's in Qemu, the overhead is big. After moving,
Vista's idle overhead decreases dramatically from ~10% to 0.9%.=20
Another benefit is that Vista can only pass Performance rating with
this patch. The root cause is that ACPI timer in Qemu isn't
synchronous with other platform timer in HV, which results in Vista
complains "can't measure TSC frequency".
This patch changes vpit.h to vpt.h, for it not only has pit structure
in it, but other platform timer's structure. Another change is moving
ACPI timer and related address from acpi.h to ioreq.h, which can be shared
by HV and ACPI firmware.
Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
As it stands grant entries for the actual data are mapped in bulk.
If one of the earlier entries fail to be mapped, it can cause all
subsequent entries to be leaked if they were successfully mapped.
This patch changes it so that we will continue to reap grant entries
even when an error is detected.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
- it turns off verbose mode of the ACM module
- copies the test policy to the policies directory only if the directory
has been created
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
PowerPC needs console information from the command line. Resetting the whole
command line causes false failures.
Signed-off-by: Tony Breeds <tony@bakeyournoodle.com>
Information is located under /sys/devices/system/memory.
Original patch from Satoshi UCHIDA <s-uchida@ap.jp.nec.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
interface is blocked for longer than 500ms. This avoids deadlock
situations where interfaces cannot be destroyed because some other
dormant interface is holding resources.
Signed-off-by: Christopher Clark <christopher.clark@cl.cam.ac.uk>
Signed-off-by: Keir Fraser <keir@xensource.com>
The blkback code checks first_sect by computing last_sect - first_sect
+ 1. If this is not positive then the request is
rejected. Unfortunately, the calculation is stored in an unsigned
integer before the check is made so negative values will not be
rejected.
This patch fixes this by explicitly comparing first_sect with
last_sect.
The blktap code is similar but it stores the difference in a signed
integer so it's okay.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
As it is blkback verifies the metadata from the frontend in place.
This means we run the risk of the frontend changing the data after
we've verified it. This patch copies the data onto the stack before
verifying and using it to ensure we see a consistent snapshot.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
On solaris mlock requires a page aligned address and mlock doesn't
ensure the pages won't minor page fault.
Signed-off-by: Mark Johnson <mark.johnson@sun.com>
max_pages is the largest number of pages the domain can have
allocated, not the gpfn of the highest one.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
It's OK for the guest to see old entries in the TLB, but not for the
shadow fault handler to see them in its linear mappings.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
with timer interrupts being delivered to VCPUs other than 0, so spit a
warning and then subsequently ignore the target. This fixes the recent
issues Intel reported when booting SMP FC5 and FC6 kernels.
Needs a better fix at some point.
Signed-off-by: Steven Hand <steven@xensource.com>
This patch eliminates redundant security tools information that was
integrated into the Xen user guide and the xm man page.
Signed-off by: Reiner Sailer <sailer@us.ibm.com>
I noticed that with default TCP window sizes TSO + Xen would slow to a
crawl on certain machines. It turned out that there is a bug in the
TCP stack when it comes to receiving LRO (counter part of TSO for rx)
packets.
The following fix has been applied upstream.
[TCP]: Fix rcv mss estimate for LRO
By passing a Linux-generated TSO packet straight back into
Linux, Xen
becomes our first LRO user :) Unfortunately, there is at least
one spot
in our stack that needs to be changed to cope with this.
The receive MSS estimate is computed from the raw packet
size. This is
broken if the packet is GSO/LRO. Fortunately the real MSS can
be found
in gso_size so we simply need to use that if it is non-zero.
Real LRO NICs should of course set the gso_size field in
future.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Remove the do_nmi() call from the vmexit logic, as the NMI is held
pending and will be handled in the context of the host when the STGI
instruction is executed.
Signed-off-by: Tom Woller <thomas.woller@amd.com>
Add flag to indicate that an exception event needs injecting, and to
delay the ext interrupt injection.
Remove unnecessary check of RFLAGS.IF for ExtInt injection.
Signed-off-by: Travis Betak <travis.betak@amd.com>
Signed-off-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Tom Woller <thomas.woller@amd.com>
This patch uses the VINTR intercept to signal the hypervisor when a
guest can take an interrupt. When guest's interrupts are masked by
EFLAGS.IF or the guests are in an interrupt shadow, we create a 'fake'
virtual interrupt to inject while also enabling the VINTR intercept.
When the guest _can_ take interrupts, the hypervisor will #VMEXIT
on VINTR. The VINTR exit handler then clears the VINTR intercept bit
and clears the V_IRQ bit so that svm_intr_assist() can inject a
legitimate interrupt.
Signed-off-by: Travis Betak <travis.betak@amd.com>
Signed-off-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Tom Woller <thomas.woller@amd.com>
