Most of this patch is an analysis of the safety of the trap handlers.
Traps 0, 4, 5, 9-12, 16, 17 and 19 all end up in do_trap(). do_trap() is
mostly safe, performing an exception table search and possibly panic()s.
There is one complication with traps 16 and 19 which will see about calling
the fpu_exception_callback. This involves following current which is not
valid early on boot. The has_hvm_container_vcpu(curr) check is preceded with
a system_state check, so in the exceedingly unlikely case that Xen takes an
x87/SIMD trap while booting, it will panic() instead of following a bogus
current vcpu.
Traps 1, 3, 6-8, 13 and 15 are completely safe with respect to running during
early boot. They all have well formed and obvious differences between faults
in Xen and faults in guests, with the Xen faults doing little more than
exception table walks or panic()s.
Trap 2 is a complicated codepath, but appears safe. For the possible
injection of NMIs into dom0 there is a NULL domain pointer check. The
possible softirq raised for PCI SERR will be delivered until we start the idle
vcpu, but is safe.
Trap 14 is very complicated. The code is certainly unsafe for boot as
fixup_page_fault() will dereference current to find the running domain. There
exists an explicit do_early_page_fault() handler which shall continue to be
used.
Trap 18 has a default handler before the MCE infrastructure is set up, which
has always been unsafe and liable to deadlock itself with the console lock.
As it is expected never to trigger, and if it did we would be in serious
problems, the simple printk() is replaced with a fatal error path.
Trap 20 (Virtualisation Exception) is currently not implemented. It is fatal
one way or another, and will become more explicitly so with later changes.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reverse two conditions in show_registers(). For an early crash, it is not
safe to dereference 'current' for its HVM status before knowing that it is a
guest vcpu.
Introduce SYS_STATE_smp_boot to distinguish the point at which APs need
considering before boot is complete. There is one code change required as a
result; .init.text symbols are still in use before Xen is active, so alter its
predicate in is_active_kernel_text().
Make use of SYS_STATE_smp_boot in machine_{halt,restart}(). Before Xen starts
booting the APs, any execution here is certainly the BSP.
When halting or rebooting particularly early, this avoids the risks of a #PF
or #GP when accessing the LAPIC before generic_apic_probe(), as well as trying
to enable interrupts before init_IRQ() is complete.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
This field has not been used since a while, last use was before the
commit 4df76b3 "xen/arm: disable the event optimization in the gic" back
in July 2012.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Currently we set the stdvga video memory with qemu-traditional only, add the
necessary settings for qemu upstream too.
Signed-off-by: Fabio Fantoni <fabio.fantoni@m2r.biz>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The Cirrus VGA videoram setting used with upstream qemu is wrong. Qemu
silently ignores the incorrect setting.
Switch to the correct vgamem_mb property which was added in qemu 1.3.
Signed-off-by: Fabio Fantoni <fabio.fantoni@m2r.biz>
Reviewed-by: Don Slutz <dslutz@verizon.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- updated changelog. ]
There's no need to print all the CPUs that are ok, only the ones that
got stuck.
The resulting output is either:
Testing NMI watchdog on all CPUs: 1 4 6 stuck
or
Testing NMI watchdog on all CPUs: ok
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Tim Deegan <tim@xen.org>
All callers of check_nmi_watchdog() already have local irqs enabled so
remove the unpaired local_irq_enable().
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Tim Deegan <tim@xen.org>
This should only cover bits 12-19, in line with MSI_ADDR_DEST_ID_SHIFT.
Also replace a couple of open-coded uses of this shift and mask.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
-EAGAIN being a return value we want to return to the actual caller in
a couple of cases makes this unsuitable for restart indication, and x86
already developed two cases where -EAGAIN could not be returned as
intended due to this (which is being fixed here at once).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com
Acked-by: Aravind Gopalakrishnan<Aravind.Gopalakrishnan@amd.com>
Reviewed-by: Tim Deegan <tim@xen.org>
The code to check for an appended DTB was confusing and unnecessary. Since we
know the size of the kernel binary passed to us we should just load the entire
thing into guest RAM (subject to the limits checks). Removing this code avoids
a whole raft of overflow and alignment issues.
We also need to validate the limits of the segment where we intend to load the
kernel to avoid overflow issues.
For ARM32 we control the load address, but we need to validate the size. The
entry point is only relevant within the guest so we don't need to worry about
that.
For ARM64 we need to validate both the load address (which is the same as the
entry point) and the size.
This is XSA-95.
Reported-by: Thomas Leonard <talex5@gmail.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
When iommu={disable,off,no,false} is given to Xen command line, the IOMMU
framework won't specify that the device shouldn't be passthrough to DOM0.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This patch contains the architecture to use IOMMUs on ARM. There is no
IOMMU drivers on this patch.
In this implementation, IOMMU page table will be shared with the P2M.
The code will run through the device tree and will initialize every IOMMU.
It's possible to have multiple IOMMUs on the same platform, but they must
be handled with the same driver. For now, there is no support for using
multiple iommu drivers at runtime.
Each new IOMMU drivers should contain:
static const char * const myiommu_dt_compat[] __initconst =
{
/* list of device compatible with the drivers. Will be matched with
* the "compatible" property on the device tree
*/
NULL,
};
DT_DEVICE_START(myiommu, "MY IOMMU", DEVICE_IOMMU)
.compatible = myiommu_compatible,
.init = myiommu_init,
DT_DEVICE_END
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Xiantao Zhang <xiantao.zhang@intel.com>
Cc: Jan Beulich <jbeulich@suse.com>
Add IOMMU helpers to support device tree assignment/deassignment. This patch
introduces 2 new fields in the dt_device_node:
- is_protected: Does the device is protected by an IOMMU
- domain_list: Pointer to the next device assigned to the same
domain
This commit contains only support to protected a device with DOM0.
Device passthrough to another guest won't work out-of-box.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Xiantao Zhang <xiantao.zhang@intel.com>
The pattern p2m_load_VTTBR(d) -> flush_tlb -> p2m_load_VTTBR(current->domain)
is used in few places.
Replace this usage by flush_tlb_domain which will take care of this pattern.
This will help to the readability of apply_p2m_changes which begin to be big.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- s/lisibility/readability/, s/speficied/specified/ ]
Dom0 may bring up less number of vCPUs than xen hypervisor actually created for
it, and in this case, on Intel platform, vMCE injection to dom0 will fail due to
injecting vMCE to uninitialized vcpu, and cause dom0 crash.
Signed-off-by: Kai Huang <kai.huang@linux.intel.com>
Acked-by: Christoph Egger <chegger@amazon.de>
Currently the structure hvm_iommu (xen/include/xen/hvm/iommu.h) contains
x86 specific fields.
This patch creates:
- arch_hvm_iommu structure which will contain architecture depend
fields
- arch_iommu_domain_{init,destroy} function to execute arch
specific during domain creation/destruction
Also move iommu_use_hap_pt and domain_hvm_iommu in asm-x86/iommu.h.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@amd.com>
Tested-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@amd.com>
The generic IOMMU framework code (xen/drivers/passthrough/iommu.c) contains
functions specific to x86 and PCI.
Split the framework in 3 distincts files:
- iommu.c: contains generic functions shared between x86 and ARM
(when it will be supported)
- pci.c: contains specific functions for PCI passthrough
- x86/iommu.c: contains specific functions for x86
io.c contains x86 HVM specific code. Only compile for x86.
This patch is mostly code movement in new files.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
Hardware domain on ARM will have the same requirements as hwdom PVH when iommu
is enabled. Both PVH and ARM guest has paging mode translate enabled, so Xen
can use it to know if it needs to check the requirements.
Rename the function and remove "pvh" word in the panic message.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This function was exported in common header. Rename it and move the declaration
in drivers/passtrough/vtd/extern.h
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
Furthermore, trapnr can be pulled from regs->entry_vector to avoid a risk of
mismatching the trap number with the underlying exception state.
This nets 2K of code reduction.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Intel new CPU supports SMAP (Supervisor Mode Access Prevention).
SMAP prevents supervisor-mode accesses to any linear address with
a valid translation for which the U/S flag (bit 2) is 1 in every
paging-structure entry controlling the translation for the linear
address.
Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Tim Deegan <tim@xen.org>
Supervisor Mode Access Prevention (SMAP) is a new security
feature disclosed by Intel, please refer to the following
document:
http://software.intel.com/sites/default/files/319433-014.pdf
If CR4.SMAP = 1, supervisor-mode data accesses are not allowed
to linear addresses that are accessible in user mode. If CPL < 3,
SMAP protections are disabled if EFLAGS.AC = 1. If CPL = 3, SMAP
applies to all supervisor-mode data accesses (these are implicit
supervisor accesses) regardless of the value of EFLAGS.AC.
This patch enables SMAP in Xen to prevent Xen hypervisor from
accessing pv guest data, whose translation paging-structure
entries' U/S flags are all set.
Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
SMAP is disabled if CPU is in non-paging mode in hardware.
However Xen always uses paging mode to emulate guest non-paging
mode with HAP. To emulate this behavior, SMAP needs to be manually
disabled when guest switches to non-paging mode.
This logic is similiar with SMEP.
Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Use STAC/CLAC to temporarily disable SMAP to allow legal accesses to
user pages in kernel mode
STAC/CLAC is not needed for compat_create_bounce_frame, since in this
chunk of code, it only accesses the pv guest's kernel stack, which is
in ring 1 for 32-bit pv guests.
Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Clear AC bit in RFLAGS at the beginning of exception, interrupt, hypercall,
so Xen itself can be protected by SMAP mechanism. This patch also sets AC
bit at the beginning of double_fault and fatal_trap() to reduce the likelihood
of taking a further fault while trying to dump state.
Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
The STAC/CLAC instructions are only available when SMAP feature is
available, but on the other hand they aren't needed if SMAP is not
enabled, or before we start to run userspace, in that case, the
functions and macros do nothing.
Signed-off-by: Feng Wu <feng.wu@intel.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
If no partitions at all are found, pygrub never creates the name 'fs',
resulting in a NameError indicating the lack of fs, rather than a
RuntimeError explaining that no partitions were found.
Set fs to None right at the start, and use the pythonic idiom "if fs is None:"
to protect against otherwise valid values for fs which compare equal to
0/False.
Reported-by: Sven Köhler <sven.koehler@gmail.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Clarify the behaviour of SCHEDOP_shutdown: PV x86 guests need to pass a
third argument, that is unused on HVM and ARM guests.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Xen bitmaps can be 4 rather than 8 byte aligned, so use the appropriate type.
Otherwise the compiler can generate unaligned 8 byte accesses and cause traps.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
In this patch, we add some checks and restrictions in the relevant
p2m paths for p2m_is_foreign.
Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
Acked-by: Tim Deegan <tim@xen.org>
A subsequent patch exposes rangesets to secondary emulators, so to allow a
limit to be placed on the amount of xenheap that an emulator can cause to be
consumed, the function rangeset_limit() has been created to set the allowed
number of ranges in a rangeset. By default, there is no limit.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Also needed to fix vsnprintf() et al so it can be called with a NULL buf
(and zero size, of course).
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
This patch only creates the ioreq server when the legacy HVM parameters
are read (by an emulator).
A lock is introduced to protect access to the ioreq server by multiple
emulator/tool invocations should such an eventuality arise. The guest is
protected by creation of the ioreq server only being done whilst the
domain is paused.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Collect together data structures concerning device emulation together into
a new struct hvm_ioreq_server.
Code that deals with the shared and buffered ioreq pages is extracted from
functions such as hvm_domain_initialise, hvm_vcpu_initialise and do_hvm_op
and consolidated into a set of hvm_ioreq_server manipulation functions. The
lock in the hvm_ioreq_page served two different purposes and has been
replaced by separate locks in the hvm_ioreq_server structure.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
To simplify creation of the ioreq server abstraction in a subsequent patch,
this patch centralizes all use of the shared ioreq structure and the
buffered ioreq ring to the source module xen/arch/x86/hvm/hvm.c.
The patch moves an rmb() from inside hvm_io_assist() to hvm_do_resume()
because the former may now be passed a data structure on stack, in which
case the barrier is unnecessary.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
This patch tidies up various parts of the code that following patches move
around. If these modifications were combined with the code motion it would
be easy to miss them.
There's also some function renaming to reflect purpose and a single
whitespace fix.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
This patch changes construct_dom0() to boot in pvh mode:
- Make sure dom0 elf supports pvh mode.
- Call guest_physmap_add_page for pvh rather than simple p2m setting
- Map all non-RAM regions 1:1 upto the end region in e820 or 4GB which
ever is higher.
- Allocate p2m, copying calculation from toolstack.
- Allocate shared info page from the virtual space so that dom0 PT
can be updated. Then update p2m for it with the actual mfn.
- Since we build the page tables for pvh same as for pv, in
pvh_fixup_page_tables_for_hap we replace the mfns with pfns.
Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Tim Deegan <tim@xen.org>
After commit 44e24f8567 (x86: don't call generic_identify() redundantly)
the struct cpu_dev don't need the c_identify.
Signed-off-by: Yi Li <peteryili@tencent.com>
For auto-translated guests the p2m code will do the necessary TLB
flushes, so there's no need to perform any TLB flushes in generic
grant table code.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Modify the function's name to help eventual backports involving this
function, and in one case where this is trivially possible also stop
ignoring its return value.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Tim Deegan <tim@xen.org>
Andrew Cooper
Julien Grall
Zhigang Wang
Fabio Fantoni
David Vrabel
Jan Beulich
Ian Campbell
Kai Huang
Feng Wu
Stefano Stabellini
Mukesh Rathor
Paul Durrant
Edmund H White
Yi Li
Roger Pau Monné