The bug occurs when the guest is entering protected mode and reaches
the far jump to set up the 32-bit segment registers for the virtual
VMENTER we're about to perform. vmxassist, in protected_mode(), looks
at the segment registers, which might be 16-bit segments or might be
32-bit segment selectors at this stage (depending on whether the OS
has reloaded them since entering protected mode); and it tries to load
the segments from the GDT. Unconditionally. Even if the segment
register still actually contains a real mode 16-bit segment. Whoops.
vmxassist already has code to detect 16-bit segments that survived
unmodified from a transition into and out of protected mode, and to
save and restore those appropriately. It does this using
"saved_rm_regs", which get cleared on entry to protected mode, and
then set to the old segment value if we fail to set a given 32-bit
segment correctly.
The fix is to save the 16-bit segments *always*, on entry to protected
mode when %CR0(PE) is first set; and to clear the saved 16-bit segment
and set the 32-bit variant in oldctx whenever a 32-bit segment
descriptor is set during the transition to 32-bit CS. Then, when we
finally do the VMENTER, we will set up the VMCS from only the 32-bit
segments, clearing the VMCS entries for segments that have not been
assigned valid 32-bit segments yet.
Tested on various RHEL-5 boot.isos, including ones which worked before
and ones which triggered the bug; all now boot correctly.
Signed-off-by: Stephen Tweedie <sct@redhat.com>
and extra code in hvm_vcpu_initialise(). Remove 'apic' config
option -- HVM CPUs will always have an APIC (which should be
set up in virtual wire mode for backward compatibility, just as
in a real system).
Signed-off-by: Keir Fraser <keir@xensource.com>
parse HVM e820 tables in Xen (add some extra HVM parameters as a
cleaner alternative). Lots of code removal.
Signed-off-by: Keir Fraser <keir@xensource.com>
directory pending tighter integration of table generation
into hvmloader. This patch also includes the TPM TIS SSDT
generator from Stefan Berger at IBM.
Signed-off-by: Keir Fraser <keir@xensource.com>
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>
Address space access limit in VMXAssist is 4G, because IA-32 only has
4GB virtual address space which VMXassist can use to map physical
memory. The issue is, win2k3 server with more than 4G memory will put
AP GDT above 4G, so when AP changes its mode from real mode to PAE
paging mode, the long jump instrction it uses need access AP GDT
entries which resides above 4G, but because of this constraint, it can
not access GDT and so fails boot.
Signed-off-by: Xin Li <xin.b.li@intel.com>
1. GNU ld does not understand -m{32,64}. It must be cooked for it by
the GCC driver program.
2. Where GNU ld is directly called we must use -melf_{i386,x86_64}.
3. We cannot avoid calling GNU ld directly in some cases (e.g., when
specifying GNU-specific linker scripts) as on some host
architectures the GCC driver is configured to call the host
linker.
4. We cannot add -melf_{i386,x86_64} to LDFLAGS as the option is
not recognised by GCC.
Hence we define new LDFLAGS_DIRECT, to be added to the command line
only when invoking GNU ld directly.
Signed-off-by: Keir Fraser <keir@xensource.com>
Currently, guest_linear_to_real to_real() in vmxassist assumes guest HVM
use 2 level page table when entering protect mode with PG enabled. Vista
uses 3 level (PAE enabled) page table, so memory addressing is wrong.
This patch fixes it, by considering all 4 possible cases.
Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
Windows zeroes out the UUID in the SMBIOS tables, but the identifier
is often needed by systems management code. Duplicating the UUID
as the virtual system's serial number should help.
Signed-off-by: Andrew D. Ball <aball@us.ibm.com>
1) Fixed HCT test "Unreported Memory and I/O Port test" for memory
range 0xC0000~0xFFFFF and IO 0x378
2) Make consistent with non-acpi OS by enabling LPT1, remove com2
which conflicts with usb mouse
3) Fixed the USB HID-Compliant mouse when set
usbdevice='tablet'/'mouse'
4) Fixed Vista boot by removing old NIC "resourceConsumer"
Signed-off-by: Winston Wang <winston.l.wang@intel.com>
executes an idling HLT. This allow us to re-enable the 'shutdown on
HLT-with-interrupts-disabled' heuristic.
Signed-off-by: Keir Fraser <keir@xensource.com>
This patch fixes 2 vmxassist bugs to boot Vista:
1) When switching from VM86_REAL_TO_PROTECTED to VM86_PROTECTED,
protected_mode() sets return eip to switch_to_protected_mode and
segments to vmxassist's. But in later set_mode(), it still consides
these segments are guest's (Bios at this time) and use its gdt table
to calculate address(), which results in segment limitation fault. 2) In
current vmxassist, when entering VM86_PROTECTED_TO_REAL, we need to
decode every instruction manually until we catch instructions like
jmpl to reset cs to really return to real mode. Vista boot manager uses
more instructions.
Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
in the usual idiomatic way. This avoids some BSS symbols being
defined oustide of range _sbss/_ebss.
Issue raised by Xiaowei Yang from Intel.
Signed-off-by: Keir Fraser <keir@xensource.com>
it to interrogate Xen version information. Also add support
for HVM hypercall execution on 64-bit host.
Signed-off-by: Steven Smith <ssmith@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
1)Code style clean up
2)Mininor fix for pci bus 0, dev 4 apic IRQ number
3)Add support for HCT "unreported IO and Memory" test by reporting low
640 MB mem as motherboard resource
Signed-off-by: Winston Wang <winston.l.wang@intel.com>
The gdtr information in oldctx is an address for guest, not for
vmxassist. When access descriptor on guest gdt, we need to go through
guest page table if guest enable paging. This error may happen if
guest enable PE/PG in one instruction.
Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>=20
Signed-off-by: Xin Li <xin.b.li@intel.com>
Currently the ROMBIOS includes static MP tables for only four
configurations: 1-way, 2-way, 4-way, and 8-way.
This patch makes the creation of MP tables more dynamic by creating them
from within HVMLOADER based upon the VCPU number provided by the HVM
information table.
Note: Some of the information included used to create the MP tables is
still hard coded and based upon values used in the original ROMBIOS MP
tables. Data such as the CPU family/model/stepping information should
probably be made to match the information provided to the guest when
executing a CPUID.
Signed-off-by: Travis Betak <travis.betak@amd.com>
Adds ACPI _PRT table for PCI IRQ routing table in PIC mode
and fixes the _PRT table in APIC mode
Signed-off-by: Tang Liang <tangliang@lenovo.com>
Signed-off-by: Winston Wang <winston.l.wang@intel.com>
The patch support ACPI UP Guest Windows 2000, XP and 2003, based on
this we are going to add more support on SMP and 64bit Windows( which
are only working in ACPI mode) later. We tested: in ia32 guest:
a. Installation and boot of Windows XP sp2 and 2003 server in ACPI
mode
b when boot to both window, no yellow mark in device manger and USB,
network card and IDE DMA are working.
c. clean boot (without ACPI related errors) on UP and 4 vcpu rhel4u1
SMP guest and got ioapic Interrupt assigned according to ACPI DSDT
table's _PRT table's GSI.
To minimize the implementation efforts to satisfy the ACPI OS
installation and boot, we implemented ACPI only event logical model by
emulating ACPI hardware based on PIIX4 (Intel 82371ab) by: Populating
the PIIX4 ACPI bridge PCI configuration space; Adding ACPI timer
(required during windows installation/boot), PM1 event register block
and PM1 control register block (both are required during
installation/boot, shutdown, entering and leaving PM S state if adding
those features later), we provided full ACPI hardware registers above
using byte/word/dword access per ACPI spec requirement.
The patch also fixed many existing hvm ACPI related bugs, such as
warnings when boot to ACPI UP and SMP guest Linux OS; adding new ACPI
_PRT table reporting PCI IRQ routing table in PIC mode, adding new _PRT
table in APIC mode, that fixed no IRQ GSI cause it is forced to shutdown
when OS find new device to assign IRQ without ACPI event service; The
patch also fixed the ACPI DSDT table's PCI resource conflict with guest
e820 table when booting and installing ACPI Windows; we also added ACPI
shutdown service by incorporating the code in ACPI "hardware enable
register writer" from Ben's Virtual Iron's team's patch in changeset
9989:f8d20c3e4225.
We only provide minimum power management requirement support per spec-c1
and S5 for shutdown, may be S4 (hibernating which is meaningful for user
to save their working section) in the future.
Please note, current patch needs to have following switches in guest
configuration to enable proper ACPI support
a. need to add ACPI=1 (apic=1 if boot SMP Linux guest plus vcpu=
number want to test) in xmexample.hvm
b. need to add USB support: USB=1 and usbdevice='device' (or
='tablet') as the pm register base is static..., I will adding dynamic
ACPI FADT table for pm register base pointer later.
When installing guest OS in ACPI mode, need to manually tell Windows
setup program to install single core ACPI HAL:
After passing the 1st screen of windows setup of install Windows,
the first thing you'll see at the bottom of the screen is the
option to press F6 if you need to install a SCSI or RAID
controller. Don't press F6. Press F5 instead. This will take
you to a separate menu of Hardware Abstraction Layer's where
you can choose an appropriate HAL .The choices will be:
ACPI Multiprocessor PC
ACPI Uniprocessor PC
Advanced Configuration and Power Interface (ACPI) PC
...
Standard PC
Standard PC with C-Step i486
Please select "Advanced Configuration and Power Interface (ACPI) PC"
using current patch.
Signed-off-by: Winston Wang <winston.l.wang@intel.com>
Fabrice Bellard on the QEMU project). This patch clears the
memory in the BIOS call rather than when the Cirrus Logic register changes,
which more closely matches what happens on the real hardware.
Signed-off-by: Don Dugger <donald.d.dugger@intel.com>
Advertising more CPUs than the guest is allocated hangs boot
of Windows XP (selects wrong HAL). The correct fix here is
to dynamically generate the MPS table (at least the CPU
entries) from hvmloader binary.
Signed-off-by: Keir Fraser <keir@xensource.com>
