When doing PoD cache reclaim as part of a decrease reservation call, avoid
reclaiming special pages for the PoD cache. Otherwise such pages get moved
from the domain ->xenpage_list to the ->page_list, while still being
referenced in ->shared_info domain field.
Prevent PoD cache from reclaiming special pages, as nothing good can come
out of it.
This is XSA-507 / CVE-2026-62434.
Fixes: 41aa0b6269 ("PoD memory 4/9: Decrease reservation")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
(cherry picked from commit 372497ce42)
The hypercall requires at least one input buffer, as both arch-specific
implementations of dm_op() unconditionally assume ->buf[0] to be valid (and
not stack rubble).
Additionally, XEN_DMOP_modified_memory requires two input buffers, yet the
code was assuming the second buffer to always be provided by the user when
checking for the number of extents. In case the caller sets nr_bufs to 1,
the code in modified_memory() will read stack garbage as the size of the
buffer, thus allowing the caller some degree of insight on the contents of
the stack by probing whether the hypercall returns -EINVAL or -EFAULT as a
result of such bogus call.
This is XSA-506 / CVE-2026-62433.
Fixes: e3b93b3c59 ("dmop: add xendevicemodel_modified_memory_bulk()")
Fixes: 85cb15dfe4 ("x86/hvm/dmop: only copy what is needed to/from the guest")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
(cherry picked from commit aef948a300)
evtchn_fifo_expand_array() will check for the domain evtchn_fifo being
populated without holding the event_lock, which can lead to a race with a
concurrent evtchn_reset().
Ensure the checking for evtchn_fifo presence is done while holding the
event_lock.
This is XSA-505 / CVE-2026-62432.
Fixes: 400b3bd642 ("evtchn: make EVTCHNOP_reset suitable for kexec")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
(cherry picked from commit 0e9d647aca)
Otherwise in periodic mode a division by 0 would happen on the second call
to start_stimer() when using periodic mode.
Note that the HyperV specification states: "Writing the value zero to the
Count register will stop the counter, thereby disabling the timer,
independent of the setting of AutoEnable in the configuration register."
so a timer with a 0 count should never be in the enabled state.
This is XSA-504 / CVE-2026-62431.
Fixes: 26fba3c855 ("viridian: add implementation of synthetic timers")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
(cherry picked from commit ff7d9aa77e)
Do the checking for a valid CMOS index while holding the spinlock,
otherwise the value could be changed by the guest after having been
checked.
This is XSA-503 / CVE-2026-62430.
Fixes: 34bef0e6d5 ("hvm: Add locking to platform timers.")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
(cherry picked from commit 52350da926)
Neither current nor curr_vcpu per-CPU fields accurately track which
page-tables are loaded. There are corner cases when dealing with shadow
paging failures that switch to the idle vCPU page-tables without changing
current or curr_vcpu per-CPU fields.
Introduce a new per-CPU field that attempts to track which vCPU page-tables
are loaded. Update such tracking when cr3 is changed, and do so in a
region with interrupts disabled, as to avoid handling interrupts with a
mismatch between the vCPU tracking field and the loaded page-tables.
As a result of this newly more accurate tracking the mapcache override
functionality can be removed: the dom0 PV builder was the only user of it,
and it's updated here to properly signal which vCPU page-tables are loaded
in the calls to switch_cr3_cr4().
Note the EFI page-tables have the Xen owned L4 slots copied from the idle
page-tables, so for the effects of the mapcache the EFI page-tables could
use the idle mapcache if it had one. Pass the idle vCPU in the
switch_cr3_cr4() call that switches to the runtime EFI page-tables.
There are known issues with the use of mapcache in NMI context. This patch
does not alter the behaviour.
This is CVE-2026-42488 / XSA-494.
Fixes: fb0ff49fe9 ("x86/shadow: defer releasing of PV's top-level shadow reference")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
(cherry picked from commit 622c9a5ba9)
In the EPT implementation, the defer flushing logic is used
unconditionally, and that would lead to paging memory being returned to the
paging pool before its references had been flushed.
Issue any pending flushes before freeing the paging memory back to the
pool.
Note AMD (NPT) and Shadow paging are not affected, as they don't implement
the deferred flushing logic.
This is XSA-480 / CVE-2026-23554
Fixes: 4a59e6bb3a ("x86/EPT: squash meaningless TLB flush")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
(cherry picked from commit 4bc2e5333a)
The first parameter passed to align_timer() is the timer expiration, not
the current time. Adjust the calls to align_timer() in the on-demand
governor to pass the expected timer expiration as the first parameter.
Fixes: af74e3a15a ("cpufreq: align dbs timer for better package C state residency")
Fixes: 382b95f627 ("Fix cpufreq HW-ALL coordination handle")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: a0ed5bcfbe
master date: 2026-03-02 17:16:26 +0100
The logic in dbs_check_cpu() resets the sampling window even when the
sampling period is considered too small. This leads to further calls
finding an imbalance between the total window time and the idle time, as
the total window time is possibly shorter than the idle time.
Fix by resetting the sampling window start time in the same block where the
current idle time is stored. While there also prevent a duplicated call to
NOW() and instead re-use the previously fetched value.
Fixes: d6f001cb91 ("x86: Implement cpufreq ondemand policy")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 74867aae25
master date: 2026-03-02 17:16:26 +0100
Currently the -Wa,--noexecstack parameter is only passed to the compiler
when building assembly files, as the assembler doesn't add such note for
stack attributes.
However, the default addition of the .note.GNU-stack section is dependent
on the ELF target used. Adjust the passing of -Wa,--noexecstack so it's
added to CFLAGS instead of AFLAGS, and hence such section is also added to
object files generated from .c sources. This fixes the complains from the
linker about missing .note.GNU-stack section when building x86-64 Xen on
Darwin platforms.
Note we could instead pass `-z noexecstack` to the linker, but that however
would mask any possible errors or unintended mismatches from previous
steps.
Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
master commit: 3dfd77003c
master date: 2026-02-13 16:42:42 +0100
BSD grep only recognizes [[:blank:]] as a valid expression when using
extended (modern) regular expressions. Pass -E to the grep call used in
efi-nr-fixups.
Additionally, the return from `wc -l` is space padded on BSD, and hence
the content of efi-nr-fixups is " 2", not plain "2". Strip the
spaces in the Makefile using $(strip ...).
Fixes: 1be65ec4c8 ('x86/EFI: avoid use of GNU ld's --disable-reloc-section when possible')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
master commit: c13335b5c1
master date: 2026-02-13 16:42:34 +0100
Currently do_kexec_op_internal() will return 0 for unknown hypercalls. Fix
this by returning -EOPNOTSUPP instead.
While there remove the pointless initialization of ret to -EINVAL; it just
makes the code confusing.
Fixes: d046f361dc ("Xen Security Modules: XSM")
Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 1ee8b11c11
master date: 2026-02-06 20:56:16 +0100
The limitation of shared_info being allocated below 4G to fit in the
start_info field only applies to 32bit PV guests. On 64bit PV guests the
start_info field is 64bits wide. HVM guests don't use start_info at all.
Drop the restriction in arch_domain_create() and instead free and
re-allocate the page from memory below 4G if needed in switch_compat(),
when the guest is set to run in 32bit PV mode.
Fixes: 3cadc0469d ("x86_64: shared_info must be allocated below 4GB as it is advertised to 32-bit guests via a 32-bit machine address field in start_info.")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 8a9255405c
master date: 2026-02-05 21:27:40 +0100
Physmap population has the need to use pages as big as possible to reduce
p2m shattering. However that triggers issues when big enough pages are not
yet scrubbed, and so scrubbing must be done at allocation time. On some
scenarios with added contention the watchdog can trigger:
Watchdog timer detects that CPU55 is stuck!
----[ Xen-4.17.5-21 x86_64 debug=n Not tainted ]----
CPU: 55
RIP: e008:[<ffff82d040204c4a>] clear_page_sse2+0x1a/0x30
RFLAGS: 0000000000000202 CONTEXT: hypervisor (d0v12)
[...]
Xen call trace:
[<ffff82d040204c4a>] R clear_page_sse2+0x1a/0x30
[<ffff82d04022a121>] S clear_domain_page+0x11/0x20
[<ffff82d04022c170>] S common/page_alloc.c#alloc_heap_pages+0x400/0x5a0
[<ffff82d04022d4a7>] S alloc_domheap_pages+0x67/0x180
[<ffff82d040226f9f>] S common/memory.c#populate_physmap+0x22f/0x3b0
[<ffff82d040228ec8>] S do_memory_op+0x728/0x1970
Introduce a mechanism to preempt page scrubbing in populate_physmap(). It
relies on stashing the dirty page in the domain struct temporarily to
preempt to guest context, so the scrubbing can resume when the domain
re-enters the hypercall. The added deferral mechanism will only be used for
domain construction, and is designed to be used with a single threaded
domain builder. If the toolstack makes concurrent calls to
XENMEM_populate_physmap for the same target domain it will trash stashed
pages, resulting in slow domain physmap population.
Note a similar issue is present in increase reservation. However that
hypercall is likely to only be used once the domain is already running and
the known implementations use 4K pages. It will be deal with in a separate
patch using a different approach, that will also take care of the
allocation in populate_physmap() once the domain is running.
Fixes: 74d2e11ccf ("mm: Scrub pages in alloc_heap_pages() if needed")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 83a784a15b
master date: 2026-02-05 08:53:07 +0100
Future changes will care about the state of the PGC_need_scrub flag even
when pages have the PGC_allocated set, and hence it's no longer possible to
alias both values. Also introduce PGC_need_scrub to the set of preserved
flags, so it's not dropped by assign_pages().
No functional change intended, albeit the page counter on x86 looses a bit.
Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: cbb484d008
master date: 2026-02-05 08:53:07 +0100
The previous logic attempted to skip an IBPB in the case of vCPU returning to
a CPU on which it was the previous vCPU to run. While safe for Xen's
isolation between vCPUs, this prevents the guest kernel correctly isolation
between tasks. Consider:
1) vCPU runs on CPU A, running task 1.
2) vCPU moves to CPU B, idle gets scheduled on A. Xen skips IBPB.
3) On CPU B, guest kernel switches from task 1 to 2, issuing IBPB.
4) vCPU moves back to CPU A. Xen skips IBPB again.
Now, task 2 is running on CPU A with task 1's training still in the BTB.
Do the flush unconditionally when switching to a vCPU different than the
idle one. Note there's no need to explicitly gate the IBPB to next domain
!= idle, as the context where the IBPB is issued is subject to that
condition already unless the pCPU is going offline, at which point we don't
really care to issue an extra IBPB.
Also add a comment with the reasoning why the IBPB needs to be in
context_switch() rather than __context_switch().
This is XSA-479 / CVE-2026-23553.
Fixes: a2ed643ed7 ("x86/ctxt: Issue a speculation barrier between vcpu contexts")
Reported-by: David Kaplan <david.kaplan@amd.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 7beadd6119
master date: 2026-01-27 13:55:27 +0100
The current logic splits the update of the amount of available memory in
the system (total_avail_pages) and pending claims into two separately
locked regions. This leads to a window between counters adjustments where
the result of total_avail_pages - outstanding_claims doesn't reflect the
real amount of free memory available, and can return a negative value due
to total_avail_pages having been updated ahead of outstanding_claims.
Fix by adjusting outstanding_claims and d->outstanding_pages in the same
place where total_avail_pages is updated. Note that accesses to
d->outstanding_pages is protected by the global heap_lock, just like
total_avail_pages or outstanding_claims. Add a comment to the field
declaration, and also adjust the comment at the top of
domain_set_outstanding_pages() to be clearer in that regard.
Note that failures in assign_pages() causes the claimed amount that has
been allocated to be lost, as the amount is not added back to the domain
quota once pages are freed. Given the intended usage of claims is limited
to initial physmap populate, and the current failure paths in
assign_pages() would lead to the domain being destroyed anyway, don't
add extra logic to recover the claimed amount on failure - it's just adding
complexity for no real benefit.
Fixes: 65c9792df6 ("mmu: Introduce XENMEM_claim_pages (subop of memory ops)")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 75d29d9b5e
master date: 2026-01-08 11:05:30 +0100
XENMAPSPACE_gmfn{_range} allows moving gfn around the guest p2m: the mfn
behind the source gfn is zapped from the origin and mapped at the
requested destination gfn. The destination p2m entries are always created
with type p2m_ram_rw.
With the current checking done in xenmem_add_to_physmap_one() it's possible
to use XENMAPSPACE_gmfn{_range} to change the type of a p2m entry. The
source gfn is only checked to be not shared, and that the underlying page
is owned by the domain.
Make the source checks more strict, by checking that the source gfn is of
type read/write RAM or logdirty. That prevents the operation from
inadvertently changing the type as part of the move.
Fixes: 3e50af3d87 ("New XENMAPSPACE_gmfn parameter for XENMEM_add_to_physmap.")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: c9491ea21a
master date: 2026-01-08 11:05:30 +0100
Add the missing index increase in the M2P clearing loop, otherwise the loop
keeps pointlessly setting the same MFN entry repeatedly. This seems to be
an oversight from the change that introduced support to process high order
pages in one go.
Fixes: 3c352011c0 ("x86/PoD: shorten certain operations on higher order ranges")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 8e4d495cb8
master date: 2025-12-11 09:50:10 +0100
The usage of atomic_dec_and_test() in msixtbl_pt_unregister() is inverted:
the function will return true when the refcount reaches 0. The current
code does the opposite and calls del_msixtbl_entry() when there are still
refcounts held on the object.
However all callers of msixtbl_pt_unregister() are serialized on the domctl
lock, and hence there cannot be parallel calls to msixtbl_pt_unregister()
that could lead to double freeing of the same object.
The incorrect freeing with active msixtlb entries will result in a possible
guest visible malfunction, but no internal Xen state corruption.
While entries are leaked once the last pIRQ is unbound, the same entry
would get re-used if the device has pIRQs bound again. The guest cannot
exploit this incorrect refcount check to leak arbitrary amounts of memory
by repeatedly enabling and disabling (binding and unbinding) MSI-X entries.
Fixes: 34097f0d30 ('hvm: passthrough MSI-X mask bit acceleration')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: ea87662c77
master date: 2025-11-26 09:46:17 +0100
Otherwise it's not possible for device models to map IRQs of devices on
segments different than 0. Keep the same function prototype and pass the
segment in the high 16bits of the bus parameter, like it's done for the
hypercall itself.
Amends: 7620c0cf9a ("PCI multi-seg: add new physdevop-s")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
master commit: b7838d12bd
master date: 2025-10-21 16:56:19 +0100
Commit 53859596c0 depends on the behavioral change introduced in
ee2eb6849d that inverted the initialization of the MSI-X and header vPCI
handlers. 53859596c0 requires the header to be initialized before MSI-X.
Change the order of initialization and adjust init_msix() to carve holes
for the MSI-X tables if necessary, since it will now run after the BARs
have been possibly mapped into the domain p2m.
Fixes: 50b649c742 ("vpci/msix: improve handling of bogus MSI-X capabilities")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
I've had the luck to come across a PCI card that exposes a MSI-X capability
where the BIR of the vector and PBA tables points at a BAR that has 0 size.
This doesn't play nice with the code in vpci_make_msix_hole(), as it would
still use the address of such empty BAR (0) and attempt to carve a hole in
the p2m. This leads to errors like the one below being reported by Xen:
d0v0 0000:22:00.0: existing mapping (mfn: 181c4300 type: 0) at 0 clobbers MSIX MMIO area
And the device left unable to enable memory decoding due to the failure
reported by vpci_make_msix_hole().
Introduce checking in init_msix() to ensure the BARs containing the MSI-X
tables are usable. This requires checking that the BIR points to a
non-empty BAR, and the offset and size of the MSI-X tables can fit in the
target BAR.
This fixes booting PVH dom0 on Supermicro AS -2126HS-TN servers with AMD
EPYC 9965 processors. The broken device is:
22:00.0 SATA controller: Advanced Micro Devices, Inc. [AMD] FCH SATA Controller [AHCI mode] (rev 93)
There are multiple of those integrated controllers in the system, all
broken in the same way.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Stewart Hildebrand <stewart.hildebrand@amd.com>
master commit: 53859596c0
master date: 2025-10-10 11:02:39 +0200
Reading from the E9 port if the emergency console is active should return
0xe9 according to the documentation from Bochs:
https://bochs.sourceforge.io/doc/docbook/user/bochsrc.html
See `port_e9_hack` section description.
Fix Xen so it also returns the port address. OSes can use it to detect
whether the emergency console is available or not.
Fixes: d1bd157fbc ("Big merge the HVM full-virtualisation abstractions.")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: db5a3c31a6
master date: 2025-10-02 18:05:50 +0200
Otherwise the check for the SS feature in
check_memory_type_self_snoop_errata() fails unconditionally, which leads to
X86_FEATURE_XEN_SELFSNOOP never being set.
We could also avoid this by not doing the reset_cpuinfo() for the BSP in
identify_cpu(), because SS detection uses boot_cpu_data. However that
creates an imbalance on the state of the BSP versus the APs in the
identify_cpu() code.
I've opted for the less controversial solution of populating FEATURESET_1d
in generic_identify(), as the value is already there. The same is done for
the AMD faulting probe code. While there also populate the FEATURESET_1c
element, as the value is in context.
Fixes: f2663ca2e5 ("x86/cpu/intel: Clear cache self-snoop capability in CPUs with known errata")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 382dd0d166
master date: 2025-09-25 16:07:05 +0200
The reference TSC page is shared between all vCPUs, and the data stored in
the domain struct. However the handlers to set and clear it are not safe
against concurrent accesses. It's possible for two (or more) vCPUs to call
HV_X64_MSR_REFERENCE_TSC at the same time and cause the in-use reference
TSC page to be freed, while still being on the p2m. This creates a use-
after-free bug, where the page can end up mapped in another domain or used
by Xen while still being part of the original domain's p2m.
It's also possible to underflow the reference counter, as multiple
concurrent writes to HV_X64_MSR_REFERENCE_TSC can create an imbalance on
the number of put_page_and_type() calls.
Introduce a lock to protect the reference TSC domain field, thus
serializing concurrent vCPU accesses.
This is CVE-2025-58143 / part of XSA-472.
Fixes: 386b336522 ('viridian: use viridian_map/unmap_guest_page() for reference tsc page')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 45729a510a
master date: 2025-09-09 14:12:00 +0200
The function is called unconditionally, regardless of whether the SIM page
is mapped. Avoid a NULL pointer dereference in
viridian_synic_deliver_timer_msg() by checking whether the SIM page is
mapped.
This is CVE-2025-58142 / part of XSA-472.
Fixes: 26fba3c855 ('viridian: add implementation of synthetic timers')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 3a7b4f2897
master date: 2025-09-09 14:11:34 +0200
The function is only called when the MSR has the enabled bit set, but even
then the page might not be mapped because the guest provided gfn is not
suitable.
Prevent a NULL pointer dereference in update_reference_tsc() by checking
whether the page is mapped.
This is CVE-2025-27466 / part of XSA-472.
Fixes: 386b336522 ('viridian: use viridian_map/unmap_guest_page() for reference tsc page')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 5776a2e9db
master date: 2025-09-09 14:11:09 +0200
The current code to restore the timer state on resume is incomplete. While
the local APIC Initial Count Register is saved and restored across
suspension (even if possibly no longer accurate since it's not adjusted to
account for the time spent in suspension), the TSC deadline MSR is not
saved and restored, hence hosts using the TSC deadline timer will likely
get stuck when resuming from suspension.
The lack of restoring of the TSC deadline MSR was mitigated by the raising
of a timer softirq in mwait_idle_with_hints() if the timer had expired,
previous to commit 3faf0866a3, which removed
that logic.
This patch fixes the usage of the TSC deadline timer with suspension, by
unconditionally raising a timer softirq on resume, that will take care of
rearming the hardware timer. Given that a timer softirq will be
unconditionally risen, there's no need to save and restore the APIC Initial
Count Register anymore either.
Note that secondary processors don't need this special treatment when
resuming, since they are offlined before suspension and brought back up
during resume, the first timer that gets setup will trigger a timer softirq
unconditionally, for example from sched_migrate_timers() that gets called
for each secondary processor.
Link: https://github.com/QubesOS/qubes-issues/issues/10110
Reported-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Fixes: fd1291a826 ('X86: Prefer TSC-deadline timer in Xen')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Tested-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
master commit: 757abedabb
master date: 2025-08-27 09:33:12 +0200
Otherwise the PCI accesses to segments different than the first one done by
the IOMMU initialization code would silently fail by returning all ones.
Introduce a new helper, called pci_setup(), and move both the creation of
PCI segment 0 internal data structures, plus the parsing of ACPI MMCFG
table to it. This moves acpi_mmcfg_init() slightly earlier from
acpi_boot_init() into pci_setup().
Note that further work will be needed to support systems where access
methods to segments different than 0 is not discoverable by Xen.
Fixes: 3950f2485b ('x86/x2APIC: defer probe until after IOMMU ACPI table parsing')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: c292772b49
master date: 2025-08-21 11:57:25 +0200
The current logic in the HPET interrupt ->ack() hook will perform a local
APIC EOI ahead of enabling interrupts, possibly leading to recursion in the
interrupt handler.
Fix this by doing the local APIC EOI strictly after the window with
interrupt enabled, as that prevents the recursion, and would only allow for
interrupts with higher priority to be serviced.
Use the generic ack_nonmaskable_msi_irq() and end_nonmaskable_irq()
functions, removing the need for hpet_msi_ack().
Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Fixes: 3ba523ff95 ('CPUIDLE: enable MSI capable HPET for timer broadcast')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 1db7829e56
master date: 2025-08-06 10:47:16 +0200
The current p2m type-change loop in ept_vcpu_flush_pml_buffer() relies on
each call to p2m_change_type_one() taking the p2m lock, doing the change
and then dropping the lock and flushing the p2m. Instead take the p2m lock
outside of the loop, so that calls to gfn_{,un}lock() inside
p2m_change_type_one() just take the p2m lock recursively, and more
importantly, the flush is deferred until the p2m is unlocked in
ept_vcpu_flush_pml_buffer().
No functional change intended in the end result of
ept_vcpu_flush_pml_buffer(), however a possibly noticeable performance
improvement is expected.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 060e9ab6e4
master date: 2025-07-18 09:17:25 +0200
For once the message printed when a BAR overlaps with a non-hole regions is
not accurate on x86. While the BAR won't be mapped by the vPCI logic, it
is quite likely overlapping with a reserved region in the memory map, and
already mapped as by default all reserved regions are identity mapped in
the p2m. Avoid printing the warning message in modify_bars(), and instead
print a more lax message in the x86 implementation of pci_check_bar() to
note the current BAR position overlaps with non-hole region(s).
Secondly, when an overlap is detected the BAR 'enabled' field is not set,
hence other vPCI code that depends on it like vPCI MSI-X handling won't
function properly, as it sees the BAR as disabled, even when memory
decoding is enabled for the device and the BAR is likely mapped in the
p2m. Change the handling of BARs that overlap non-hole regions to instead
remove any overlapped regions from the rangeset, so the resulting ranges to
map just contain the hole regions. This requires introducing a new
pci_sanitize_bar_memory() that's implemented per-arch and sanitizes the
address range to add to the p2m.
For x86 pci_sanitize_bar_memory() removes any regions present in the host
memory map, for ARM this is currently left as a dummy handler to not change
existing behavior.
Ultimately the above changes should fix the vPCI MSI-X handlers not working
correctly when the BAR that contains the MSI-X table overlaps with a
non-hole region, as then the 'enabled' BAR bit won't be set and the MSI-X
traps won't handle accesses as expected.
Reported-by: Stefano Stabellini <stefano.stabellini@amd.com>
Fixes: 53d9133638 ('pci: do not disable memory decoding for devices')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Tested-by: Victor M Lira <victorm.lira@amd.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <jgrall@amazon.com>
x86/vpci: fix off-by-one in pci_sanitize_bar_memory()
rangeset_remove_range() uses inclusive ranges, and hence the end of the
range should be calculated using PFN_DOWN(), not PFN_UP().
Fixes: 4acab25a93 ('x86/vpci: fix handling of BAR overlaps with non-hole regions')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 4acab25a93
master date: 2025-05-21 18:29:55 +0200
master commit: 7ab4b392b7
master date: 2025-05-22 16:17:12 +0200
The current emulation of wb{,no}invd is bogus for PV guests: it will only
flush the current pCPU cache, without taking into account pCPUs where the
vCPU had run previously. Resort to flushing the cache on all host pCPUs to
make it correct.
Fixes: 799fed0a7c ("Priv-op emulation in Xen, for RDMSR/WRMSR/WBINVD")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
[backport: no FLUSH_CACHE_WRITEBACK]
master commit: b0f8bf8605
master date: 2025-05-20 16:35:52 +0200
The implementation of MMUEXT_FLUSH_CACHE is bogus, as it doesn't account to
flush the cache of any previous pCPU where the current vCPU might have run,
and hence is likely to not work as expected.
Fix this by resorting to use the same logic as MMUEXT_FLUSH_CACHE_GLOBAL,
which will be correct in all cases.
Adjust MMUEXT_FLUSH_CACHE_GLOBAL return code in case cache flush is not
permitted for the domain to use -EACCES instead of -EINVAL, as that's more
accurate and also matches the error code used by MMUEXT_FLUSH_CACHE.
Fixes: 8e90e37e6d ("Fix WBINVD by adding a new hypercall.")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
master commit: bbaea9878b
master date: 2025-05-13 19:43:48 +0200
It's unclear why -N is being used in the first place. It was added by
commit 4676bbf96d back in 2002 without any justification.
When building a PE image it's actually detrimental to forcefully set the
.text section as writable. The GNU LD man page contains the following
warning regarding the -N option:
> Note: Although a writable text section is allowed for PE-COFF targets, it
> does not conform to the format specification published by Microsoft.
Remove the usage of -N uniformly on all architectures, assuming that the
addition was simply done as a copy and paste of the original x86 linking
rune.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Acked-by: Julien Grall <jgrall@amazon.com>
master commit: d444763f8c
master date: 2025-04-23 16:12:25 +0200
There are several errata on Intel regarding the usage of the MONITOR/MWAIT
instructions, all having in common that stores to the monitored region
might not wake up the CPU.
Fix them by forcing the sending of an IPI for the affected models.
The Ice Lake issue has been reproduced internally on XenServer hardware,
and the fix does seem to prevent it. The symptom was APs getting stuck in
the idle loop immediately after bring up, which in turn prevented the BSP
from making progress. This would happen before the watchdog was
initialized, and hence the whole system would get stuck.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 4aae4452ef
master date: 2025-04-23 16:12:25 +0200
Given the following rangset operation:
{ [0, 1], [4, 5] } - { [3, 4] }
The current rangeset logic will output a rangeset:
{ [0, 2], [5, 5] }
This is incorrect, and also has the undesirable property of being bogus in
a way that the resulting rangeset is expanded.
Fix this by making sure the bounds are correctly checked before modifying
the previous range.
Fixes: 484a058c48 ('Add auto-destructing per-domain rangeset data structure...')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: e118fc98e7
master date: 2025-04-11 12:20:10 +0200
It makes no sense to request for preprocessor only output and also request
object file generation. Fix the _libxl.api-for-check target to only use
-E (preprocessor output).
Also Clang 20.0 reports an error if both options are used.
Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Fixes: 2862bf5b6c ('libxl: enforce prohibitions of internal callers')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Anthony PERARD <anthony.perard@vates.tech>
(cherry picked from commit a235f856e4)
Introduce a basic set of smoke tests using the XTF selftest image, and run
them on QEMU. Use the matrix keyword to create a different task for each
XTF flavor on each FreeBSD build.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Oleksii Kurochko<oleksii.kurochko@gmail.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
(cherry picked from commit 7973cba4db)
In preparation for adding some smoke tests that will consume those outputs.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>
(cherry picked from commit c9de0e2ce4)
In preparation for using the XTF selftests to smoke test the FreeBSD based
Xen builds.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
(cherry picked from commit a92c0a9888)
Move the current logic to use the matrix keyword to generate a task for
each version of FreeBSD we want to build Xen on. The matrix keyword
however cannot be used in YAML aliases, so it needs to be explicitly used
inside of each task, which creates a bit of duplication. At least abstract
the FreeBSD minor version numbers to avoid repetition of image names.
Note that the full build uses matrix over an env variable instead of using
it directly in image_family. This is so that the alias can also be set
based on the FreeBSD version, in preparation for adding further tasks that
will depend on the full build having finished.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
(cherry picked from commit b548a7dc4b)
Instead use env to find the location of expect.
Additionally do not use the -f flag, as it's only meaningful when passing
arguments on the command line, which we never do for console.exp. From the
expect 5.45.4 man page:
> The -f flag prefaces a file from which to read commands from. The flag
> itself is optional as it is only useful when using the #! notation (see
> above), so that other arguments may be supplied on the command line.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
In case the build fails or gets stuck, store the Kconfig file ahead of
starting the build.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>
(cherry picked from commit 421c2bd58f)
Such timestamps can still be disabled from the Web UI using a tick box.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>
(cherry picked from commit 1256159f3c)
Add a new randconfig job for each FreeBSD version. This requires some
rework of the template so common parts can be shared between the full and
the randconfig builds. Such randconfig builds are relevant because FreeBSD
is the only tested system that has a full non-GNU toolchain.
While there replace the usage of the python311 package with python3, which is
already using 3.11, and remove the install of the plain python package for full
builds.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
(cherry picked from commit c3f5d1bb40)
The call to ioremap_wc() in video_init() will always fail, because
video_init() is called ahead of vm_init_type(), and so the underlying
__vmap() call will fail to allocate the linear address space.
Fix by reverting to the previous behavior and use __va() for the VGA text
buffer, as it's below the 1MB boundary, and thus always mapped in the
directmap.
Fixes: 81d195c6c0 ('x86: introduce ioremap_wc()')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 1ca5f69e35
master date: 2025-03-20 17:16:10 +0100
UBSAN complains with:
UBSAN: Undefined behaviour in common/compat/memory.c:90:9
pointer operation overflowed ffff820080000000 to 0000020080000000
[...]
Xen call trace:
[<ffff82d040303782>] R common/ubsan/ubsan.c#ubsan_epilogue+0xa/0xc0
[<ffff82d040304bc3>] F __ubsan_handle_pointer_overflow+0xcb/0x100
[<ffff82d0402a6259>] F compat_memory_op+0xf1/0x4d20
[<ffff82d04041532d>] F hvm_memory_op+0x55/0xe0
[<ffff82d040416150>] F hvm_hypercall+0xae8/0x21b0
[<ffff82d0403b24ca>] F svm_vmexit_handler+0x1252/0x2450
[<ffff82d0402049c0>] F svm_stgi_label+0x5/0x15
Adjust the calculations in COMPAT_ARG_XLAT_VIRT_BASE to subtract from the
per-domain area to obtain the mirrored linear address in the 4th slot,
instead of overflowing the per-domain linear address.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: fc302866f4
master date: 2025-03-20 12:28:30 +0100
UBSAN complains with:
UBSAN: Undefined behaviour in arch/x86/mm/shadow/private.h:515:30
pointer operation overflowed ffff82e000000000 to ffff82dfffffffe0
[...]
Xen call trace:
[<ffff82d040303782>] R common/ubsan/ubsan.c#ubsan_epilogue+0xa/0xc0
[<ffff82d040304bc3>] F __ubsan_handle_pointer_overflow+0xcb/0x100
[<ffff82d040471b2d>] F arch/x86/mm/shadow/guest_2.c#sh_page_fault__guest_2+0x1e350
[<ffff82d0403b206b>] F svm_vmexit_handler+0xdf3/0x2450
[<ffff82d0402049c0>] F svm_stgi_label+0x5/0x15
Fix by moving the call to mfn_to_page() after the check of whether the
passed gmfn is valid. This avoid the call to mfn_to_page() with an
INVALID_MFN parameter.
While there make the page local variable const, it's not modified by the
function.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 45ee73f1b2
master date: 2025-03-20 12:28:30 +0100
mkelf32 attempt to check that the program header defined NOTE segment falls
inside of the LOAD segment, as the build-id should be loaded for Xen at
runtime to check.
However the current code doesn't take into account the LOAD program header
segment offset when calculating overlap with the NOTE segment. This
results in incorrect detection, and the following build error:
arch/x86/boot/mkelf32 --notes xen-syms ./.xen.elf32 0x200000 \
`nm xen-syms | sed -ne 's/^\([^ ]*\) . __2M_rwdata_end$/0x\1/p'`
Expected .note section within .text section!
Offset 4244776 not within 2910364!
When xen-syms has the following program headers:
Program Header:
LOAD off 0x0000000000200000 vaddr 0xffff82d040200000 paddr 0x0000000000200000 align 2**21
filesz 0x00000000002c689c memsz 0x00000000003f7e20 flags rwx
NOTE off 0x000000000040c528 vaddr 0xffff82d04040c528 paddr 0x000000000040c528 align 2**2
filesz 0x0000000000000024 memsz 0x0000000000000024 flags r--
Account for the program header offset of the LOAD segment when checking
whether the NOTE segments is contained within. Also fix the logic to
ensure the NOTE segments is fully contained between the LOAD segment.
Fixes: a353cab905 ('build_id: Provide ld-embedded build-ids')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 6e5fed7cb6
master date: 2025-03-20 12:28:30 +0100
This was reported by clang UBSAN as:
UBSAN: Undefined behaviour in arch/x86/mm.c:6297:40
applying zero offset to null pointer
[...]
Xen call trace:
[<ffff82d040303662>] R common/ubsan/ubsan.c#ubsan_epilogue+0xa/0xc0
[<ffff82d040304aa3>] F __ubsan_handle_pointer_overflow+0xcb/0x100
[<ffff82d0406ebbc0>] F ioremap_wc+0xc8/0xe0
[<ffff82d0406c3728>] F video_init+0xd0/0x180
[<ffff82d0406ab6f5>] F console_init_preirq+0x3d/0x220
[<ffff82d0406f1876>] F __start_xen+0x68e/0x5530
[<ffff82d04020482e>] F __high_start+0x8e/0x90
Fix bt_ioremap() and ioremap{,_wc}() to not add the offset if the returned
pointer from __vmap() is NULL.
Fixes: d0d4635d03 ('implement vmap()')
Fixes: f390941a92 ('x86/DMI: fix table mapping when one lives above 1Mb')
Fixes: 81d195c6c0 ('x86: introduce ioremap_wc()')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 9a6f2c52f7
master date: 2025-03-17 13:33:29 +0100
Consume the return code from hvm_pi_update_irte(), and propagate the error
back to the caller if hvm_pi_update_irte() fails.
Fixes: 35a1caf8b6 ('pass-through: update IRTE according to guest interrupt config changes')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: cb587f620a
master date: 2025-03-12 13:32:31 +0100
The current usage of msi_desc->msg in vmx_pi_update_irte() will make the
field contain a translated MSI message, instead of the expected
untranslated one. This breaks dump_msi(), that use the data in
msi_desc->msg to print the interrupt details.
Fix this by introducing a dummy local msi_msg, and use it with
iommu_update_ire_from_msi(). vmx_pi_update_irte() relies on the MSI
message not changing, so there's no need to propagate the resulting msi_msg
to the hardware, and the contents can be ignored.
Additionally add a comment to clarify that msi_desc->msg must always
contain the untranslated MSI message.
Fixes: a5e25908d1 ('VT-d: introduce new fields in msi_desc to track binding with guest interrupt')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 30f0e55a79
master date: 2025-03-12 13:32:30 +0100
The MMIO_CONF_BASE reports the base of the MCFG range on AMD systems.
Linux pre-6.14 is unconditionally attempting to read the MSR without a
safe MSR accessor, and since Xen doesn't allow access to it Linux reports
the following error:
unchecked MSR access error: RDMSR from 0xc0010058 at rIP: 0xffffffff8101d19f (xen_do_read_msr+0x7f/0xa0)
Call Trace:
xen_read_msr+0x1e/0x30
amd_get_mmconfig_range+0x2b/0x80
quirk_amd_mmconfig_area+0x28/0x100
pnp_fixup_device+0x39/0x50
__pnp_add_device+0xf/0x150
pnp_add_device+0x3d/0x100
pnpacpi_add_device_handler+0x1f9/0x280
acpi_ns_get_device_callback+0x104/0x1c0
acpi_ns_walk_namespace+0x1d0/0x260
acpi_get_devices+0x8a/0xb0
pnpacpi_init+0x50/0x80
do_one_initcall+0x46/0x2e0
kernel_init_freeable+0x1da/0x2f0
kernel_init+0x16/0x1b0
ret_from_fork+0x30/0x50
ret_from_fork_asm+0x1b/0x30
Such access is conditional to the presence of a device with PnP ID
"PNP0c01", which triggers the execution of the quirk_amd_mmconfig_area()
function. Note that prior to commit 3fac3734c43a MSR accesses when running
as a PV guest would always use the safe variant, and thus silently handle
the #GP.
Fix by allowing access to the MSR on AMD systems for the hardware domain.
Write attempts to the MSR will still result in #GP for all domain types.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: b4071d28c5
master date: 2025-03-12 13:32:30 +0100
Xen currently prevents dom0 from creating CPU or IOMMU page-table mappings
into the interrupt address range [0xfee00000, 0xfeefffff]. This range has
two different purposes. For accesses from the CPU is contains the default
position of local APIC page at 0xfee00000. For accesses from devices
it's the MSI address range, so the address field in the MSI entries
(usually) point to an address on that range to trigger an interrupt.
There are reports of Lenovo Thinkpad devices placing what seems to be the
UCSI shared mailbox at address 0xfeec2000 in the interrupt address range.
Attempting to use that device with a Linux PV dom0 leads to an error when
Linux kernel maps 0xfeec2000:
RIP: e030:xen_mc_flush+0x1e8/0x2b0
xen_leave_lazy_mmu+0x15/0x60
vmap_range_noflush+0x408/0x6f0
__ioremap_caller+0x20d/0x350
acpi_os_map_iomem+0x1a3/0x1c0
acpi_ex_system_memory_space_handler+0x229/0x3f0
acpi_ev_address_space_dispatch+0x17e/0x4c0
acpi_ex_access_region+0x28a/0x510
acpi_ex_field_datum_io+0x95/0x5c0
acpi_ex_extract_from_field+0x36b/0x4e0
acpi_ex_read_data_from_field+0xcb/0x430
acpi_ex_resolve_node_to_value+0x2e0/0x530
acpi_ex_resolve_to_value+0x1e7/0x550
acpi_ds_evaluate_name_path+0x107/0x170
acpi_ds_exec_end_op+0x392/0x860
acpi_ps_parse_loop+0x268/0xa30
acpi_ps_parse_aml+0x221/0x5e0
acpi_ps_execute_method+0x171/0x3e0
acpi_ns_evaluate+0x174/0x5d0
acpi_evaluate_object+0x167/0x440
acpi_evaluate_dsm+0xb6/0x130
ucsi_acpi_dsm+0x53/0x80
ucsi_acpi_read+0x2e/0x60
ucsi_register+0x24/0xa0
ucsi_acpi_probe+0x162/0x1e3
platform_probe+0x48/0x90
really_probe+0xde/0x340
__driver_probe_device+0x78/0x110
driver_probe_device+0x1f/0x90
__driver_attach+0xd2/0x1c0
bus_for_each_dev+0x77/0xc0
bus_add_driver+0x112/0x1f0
driver_register+0x72/0xd0
do_one_initcall+0x48/0x300
do_init_module+0x60/0x220
__do_sys_init_module+0x17f/0x1b0
do_syscall_64+0x82/0x170
Remove the restrictions to create mappings in the interrupt address range
for dom0. Note that the restriction to map the local APIC page is enforced
separately, and that continues to be present. Additionally make sure the
emulated local APIC page is also not mapped, in case dom0 is using it.
Note that even if the interrupt address range entries are populated in the
IOMMU page-tables no device access will reach those pages. Device accesses
to the Interrupt Address Range will always be converted into Interrupt
Messages and are not subject to DMA remapping.
There's also the following restriction noted in Intel VT-d:
> Software must not program paging-structure entries to remap any address to
> the interrupt address range. Untranslated requests and translation requests
> that result in an address in the interrupt range will be blocked with
> condition code LGN.4 or SGN.8. Translated requests with an address in the
> interrupt address range are treated as Unsupported Request (UR).
Similarly for AMD-Vi:
> Accesses to the interrupt address range (Table 3) are defined to go through
> the interrupt remapping portion of the IOMMU and not through address
> translation processing. Therefore, when a transaction is being processed as
> an interrupt remapping operation, the transaction attribute of
> pretranslated or untranslated is ignored.
>
> Software Note: The IOMMU should
> not be configured such that an address translation results in a special
> address such as the interrupt address range.
However those restrictions don't apply to the identity mappings possibly
created for dom0, since the interrupt address range is never subject to DMA
remapping, and hence there's no output address after translation that
belongs to the interrupt address range.
Reported-by: Jürgen Groß <jgross@suse.com>
Link: https://lore.kernel.org/xen-devel/baade0a7-e204-4743-bda1-282df74e5f89@suse.com/
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
master commit: 381caa3885
master date: 2025-03-05 10:26:46 +0100
The current code in arch_iommu_hwdom_init() kind of open-codes the same
MMIO permission ranges that are added to the hardware domain ->iomem_caps.
Avoid this duplication and use ->iomem_caps in arch_iommu_hwdom_init() to
filter which memory regions should be added to the dom0 IOMMU page-tables.
Note the IO-APIC and MCFG page(s) must be set as not accessible for a PVH
dom0, otherwise the internal Xen emulation for those ranges won't work.
This requires adjustments in dom0_setup_permissions().
The call to pvh_setup_mmcfg() in dom0_construct_pvh() must now strictly be
done ahead of setting up dom0 permissions, so take the opportunity to also
put it inside the existing is_hardware_domain() region.
Also the special casing of E820_UNUSABLE regions no longer needs to be done
in arch_iommu_hwdom_init(), as those regions are already blocked in
->iomem_caps and thus would be removed from the rangeset as part of
->iomem_caps processing in arch_iommu_hwdom_init(). The E820_UNUSABLE
regions below 1Mb are not removed from ->iomem_caps, that's a slight
difference for the IOMMU created page-tables, but the aim is to allow
access to the same memory either from the CPU or the IOMMU page-tables.
Since ->iomem_caps already takes into account the domain max paddr, there's
no need to remove any regions past the last address addressable by the
domain, as applying ->iomem_caps would have already taken care of that.
Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 62f3fc5296
master date: 2025-03-05 10:26:46 +0100
The logic in dom0_setup_permissions() sets the maximum bound in
->iomem_caps unconditionally using paddr_bits, which is not correct for HVM
based domains. Instead use domain_max_paddr_bits() to get the correct
maximum paddr bits for each possible domain type.
Switch to using PFN_DOWN() instead of PAGE_SHIFT, as that's shorter.
Fixes: 53de839fb4 ('x86: constrain MFN range Dom0 may access')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: a00e08799c
master date: 2025-03-05 10:26:46 +0100
When building a PVH dom0 Xen attempts to map all (relevant) MMIO regions
into the p2m for dom0 access. However the information Xen has about the
host memory map is limited. Xen doesn't have access to any resources
described in ACPI dynamic tables, and hence the p2m mappings provided might
not be complete.
PV doesn't suffer from this issue because a PV dom0 is capable of mapping
into it's page-tables any address not explicitly banned in d->iomem_caps.
Introduce a new command line options that allows Xen to attempt to fixup
the p2m page-faults, by creating p2m identity maps in response to p2m
page-faults.
This is aimed as a workaround to small ACPI regions Xen doesn't know about.
Note that missing large MMIO regions mapped in this way will lead to
slowness due to the VM exit processing, plus the mappings will always use
small pages.
The ultimate aim is to attempt to bring better parity with a classic PV
dom0.
Note such fixup rely on the CPU doing the access to the unpopulated
address. If the access is attempted from a device instead there's no
possible way to fixup, as IOMMU page-fault are asynchronous.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
master commit: 104591f5dd
master date: 2025-03-05 10:26:46 +0100
A PV dom0 can map any host memory as long as it's allowed by the IO
capability range in d->iomem_caps. On the other hand, a PVH dom0 has no
way to populate MMIO region onto it's p2m, so it's limited to what Xen
initially populates on the p2m based on the host memory map and the enabled
device BARs.
Introduce a new debug build only printk that reports attempts by dom0 to
access addresses not populated on the p2m, and not handled by any emulator.
This is for information purposes only, but might allow getting an idea of
what MMIO ranges might be missing on the p2m.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
master commit: 43d8a80a0c
master date: 2025-03-05 10:26:46 +0100
Add a new hook to inhibit interrupt generation by the IOMMU(s). Note the
hook is currently only implemented for x86 IOMMUs. The purpose is to
disable interrupt generation at shutdown so any kexec chained image finds
the IOMMU(s) in a quiesced state.
It would also prevent "Receive accept error" being raised as a result of
non-disabled interrupts targeting offline CPUs.
Note that the iommu_quiesce() call in nmi_shootdown_cpus() is still
required even when there's a preceding iommu_crash_shutdown() call; the
later can become a no-op depending on the setting of the "crash-disable"
command line option.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 819c3cb186
master date: 2025-02-12 15:56:07 +0100
Attempt to disable MSI(-X) capabilities on all PCI devices know by Xen at
shutdown. Doing such disabling should facilitate kexec chained kernel from
booting more reliably, as device MSI(-X) interrupt generation should be
quiesced.
Only attempt to disable MSI(-X) on all devices in the crash context if the
PCI lock is not taken, otherwise the PCI device list could be in an
inconsistent state. This requires introducing a new pcidevs_trylock()
helper to check whether the lock is currently taken.
Disabling MSI(-X) should prevent "Receive accept error" being raised as a
result of non-disabled interrupts targeting offline CPUs.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 7ab6951981
master date: 2025-02-12 15:56:07 +0100
Move the disabling of interrupt sources so it's done ahead of the offlining
of APs. This is to prevent AMD systems triggering "Receive accept error"
when interrupts target CPUs that are no longer online.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: db6daa9bf4
master date: 2025-02-12 15:56:07 +0100
The current shutdown logic in smp_send_stop() will disable the APs while
having interrupts enabled on the BSP or possibly other APs. On AMD systems
this can lead to local APIC errors:
APIC error on CPU0: 00(08), Receive accept error
Such error message can be printed in a loop, thus blocking the system from
rebooting. I assume this loop is created by the error being triggered by
the console interrupt, which is further stirred by the ESR handler
printing to the console.
Intel SDM states:
"Receive Accept Error.
Set when the local APIC detects that the message it received was not
accepted by any APIC on the APIC bus, including itself. Used only on P6
family and Pentium processors."
So the error shouldn't trigger on any Intel CPU supported by Xen.
However AMD doesn't make such claims, and indeed the error is broadcast to
all local APICs when an interrupt targets a CPU that's already offline.
To prevent the error from stalling the shutdown process perform the
disabling of APs and the BSP local APIC with interrupts disabled on all
CPUs in the system, so that by the time interrupts are unmasked on the BSP
the local APIC is already disabled. This can still lead to a spurious:
APIC error on CPU0: 00(00)
As a result of an LVT Error getting injected while interrupts are masked on
the CPU, and the vector only handled after the local APIC is already
disabled. ESR reports 0 because as part of disable_local_APIC() the ESR
register is cleared.
Note the NMI crash path doesn't have such issue, because disabling of APs
and the caller local APIC is already done in the same contiguous region
with interrupts disabled. There's a possible window on the NMI crash path
(nmi_shootdown_cpus()) where some APs might be disabled (and thus
interrupts targeting them raising "Receive accept error") before others APs
have interrupts disabled. However the shutdown NMI will be handled,
regardless of whether the AP is processing a local APIC error, and hence
such interrupts will not cause the shutdown process to get stuck.
Remove the call to fixup_irqs() in smp_send_stop(): it doesn't achieve the
intended goal of moving all interrupts to the BSP anyway. The logic in
fixup_irqs() will move interrupts whose affinity doesn't overlap with the
passed mask, but the movement of interrupts is done to any CPU set in
cpu_online_map. As in the shutdown path fixup_irqs() is called before APs
are cleared from cpu_online_map this leads to interrupts being shuffled
around, but not assigned to the BSP exclusively.
The Fixes tag is more of a guess than a certainty; it's possible the
previous sleep window in fixup_irqs() allowed any in-flight interrupt to be
delivered before APs went offline. However fixup_irqs() was still
incorrectly used, as it didn't (and still doesn't) move all interrupts to
target the provided cpu mask.
Fixes: e2bb28d621 ('x86/irq: forward pending interrupts to new destination in fixup_irqs()')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 1191ce954f
master date: 2025-02-12 15:56:07 +0100
Either when using a 32bit Interrupt Remapping Entry or a 128bit one update
the entry atomically, by using cmpxchg unconditionally as IOMMU depends on
it. No longer disable the entry by setting RemapEn = 0 ahead of updating
it. As a consequence of not toggling RemapEn ahead of the update the
Interrupt Remapping Table needs to be flushed after the entry update.
This avoids a window where the IRTE has RemapEn = 0, which can lead to
IO_PAGE_FAULT if the underlying interrupt source is not masked.
There's no guidance in AMD-Vi specification about how IRTE update should be
performed as opposed to DTE updating which has specific guidance. However
DTE updating claims that reads will always be at least 128bits in size, and
hence for the purposes here assume that reads and caching of the IRTE
entries in either 32 or 128 bit format will be done atomically from
the IOMMU.
Note that as part of introducing a new raw128 field in the IRTE struct, the
current raw field is renamed to raw64 to explicitly contain the size in the
field name.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: b953a99da9
master date: 2025-01-27 13:05:11 +0100
Avoid exiting early from the loop when a pin that could be connected to the
i8259 is found, as such early exit would leave the EOI handler translation
array only partially allocated and/or initialized.
Otherwise on systems with multiple IO-APICs and an unmasked ExtINT pin on
any IO-APIC that's no the last one the following NULL pointer dereference
triggers:
(XEN) Enabling APIC mode. Using 2 I/O APICs
(XEN) ----[ Xen-4.20-unstable x86_64 debug=y Not tainted ]----
(XEN) CPU: 0
(XEN) RIP: e008:[<ffff82d040328046>] __ioapic_write_entry+0x83/0x95
[...]
(XEN) Xen call trace:
(XEN) [<ffff82d040328046>] R __ioapic_write_entry+0x83/0x95
(XEN) [<ffff82d04027464b>] F amd_iommu_ioapic_update_ire+0x1ea/0x273
(XEN) [<ffff82d0402755a1>] F iommu_update_ire_from_apic+0xa/0xc
(XEN) [<ffff82d040328056>] F __ioapic_write_entry+0x93/0x95
(XEN) [<ffff82d0403283c1>] F arch/x86/io_apic.c#clear_IO_APIC_pin+0x7c/0x10e
(XEN) [<ffff82d040328480>] F arch/x86/io_apic.c#clear_IO_APIC+0x2d/0x61
(XEN) [<ffff82d0404448b7>] F enable_IO_APIC+0x2e3/0x34f
(XEN) [<ffff82d04044c9b0>] F smp_prepare_cpus+0x254/0x27a
(XEN) [<ffff82d04044bec2>] F __start_xen+0x1ce1/0x23ae
(XEN) [<ffff82d0402033ae>] F __high_start+0x8e/0x90
(XEN)
(XEN) Pagetable walk from 0000000000000000:
(XEN) L4[0x000] = 000000007dbfd063 ffffffffffffffff
(XEN) L3[0x000] = 000000007dbfa063 ffffffffffffffff
(XEN) L2[0x000] = 000000007dbcc063 ffffffffffffffff
(XEN) L1[0x000] = 0000000000000000 ffffffffffffffff
(XEN)
(XEN) ****************************************
(XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0002]
(XEN) Faulting linear address: 0000000000000000
(XEN) ****************************************
(XEN)
(XEN) Reboot in five seconds...
Reported-by: Sergii Dmytruk <sergii.dmytruk@3mdeb.com>
Fixes: 86001b3970 ('x86/io-apic: fix directed EOI when using AMD-Vi interrupt remapping')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: f38fd27c4c
master date: 2024-12-17 11:15:30 +0100
The tools infrastructure used to build livepatches for Xen
(livepatch-build-tools) consumes some DWARF debug information present in
xen-syms to generate a livepatch (see livepatch-build script usage of readelf
-wi).
The current Kconfig defaults however will enable LIVEPATCH without DEBUG_INFO
on release builds, thus providing a default Kconfig selection that's not
suitable for livepatch-build-tools even when LIVEPATCH support is enabled,
because it's missing the DWARF debug section.
Fix by defaulting DEBUG_INFO to enabled when LIVEPATCH is.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 126b0a6e53
master date: 2024-12-02 15:22:05 +0100
Do not return early in the PVH/HVM case, so that the number of pIRQs is also
printed. While PVH dom0 doesn't have access to the hypercalls to manage pIRQs
itself, nor the knowledge to do so, pIRQs are still used by Xen to map and
bind interrupts to a PVH dom0 behind its back. Hence the pIRQ limit is still
relevant for a PVH dom0.
Fixes: 17f6d398f7 ('cmdline: document and enforce "extra_guest_irqs" upper bounds')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 722d93bc91
master date: 2024-11-25 11:33:38 +0100
The current calculation of PV dom0 pIRQs uses:
n = min(fls(num_present_cpus()), dom0_max_vcpus());
The usage of fls() is wrong, as num_present_cpus() already returns the number
of present CPUs, not the bitmap mask of CPUs.
Fix by removing the usage of fls().
Fixes: 7e73a6e7f1 ('have architectures specify the number of PIRQs a hardware domain gets')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 5c56361c61
master date: 2024-11-25 11:33:06 +0100
The current logic in modify_xen_mappings() allows for fully empty L2 tables to
not be freed and unhooked from the parent L3 if the last L2 slot is not
populated.
Ensure that even when an L2 slot is empty the logic to check whether the whole
L2 can be removed is not skipped.
Fixes: 4376c05c31 ('x86-64: use 1GB pages in 1:1 mapping if available')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 41c8049608
master date: 2024-11-14 16:13:10 +0100
INVALID_MFN is ~0, so by it having all bits as 1s it doesn't fulfill the
super-page address alignment checks for L3 and L2 entries. Skip the alignment
checks if the new entry is a non-present one.
This fixes a regression introduced by 0b6b51a69f, where the switch from 0 to
INVALID_MFN caused all super-pages to be shattered when attempting to remove
mappings by passing INVALID_MFN instead of 0.
Fixes: 0b6b51a69f ('xen/mm: Switch map_pages_to_xen to use MFN typesafe')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
x86/mm: fix alignment check for non-present entries
While the alignment of the mfn is not relevant for non-present entries, the
alignment of the linear address is. Commit 5b52e1b043 introduced a
regression by not checking the alignment of the linear address when the new
entry was a non-present one.
Fix by always checking the alignment of the linear address, non-present entries
must just skip the alignment check of the physical address.
Fixes: 5b52e1b043 ('x86/mm: skip super-page alignment checks for non-present entries')
Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 5b52e1b043
master date: 2024-11-14 16:12:35 +0100
master commit: b1ebb6461a
master date: 2024-11-15 14:14:12 +0100
Split the code that detects whether the physical and linear address of a
mapping request are suitable to be used in an L3 or L2 slot.
No functional change intended.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 97fb6fcf26
master date: 2024-11-14 16:12:12 +0100
alternatives is used both at boot time, and when loading livepatch payloads.
While for the former it makes sense to panic, it's not useful for the later, as
for livepatches it's possible to fail to load the livepatch if alternatives
cannot be resolved and continue operating normally.
Relax the BUGs in _apply_alternatives() to instead return an error code. The
caller will figure out whether the failures are fatal and panic.
Print an error message to provide some user-readable information about what
went wrong.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: aa5a06d5d6
master date: 2024-09-26 14:18:03 +0100
The check against the expected Xen build ID should be done ahead of attempting
to apply the alternatives contained in the livepatch.
If the CPUID in the alternatives patching data is out of the scope of the
running Xen featureset the BUG() in _apply_alternatives() will trigger thus
bringing the system down. Note the layout of struct alt_instr could also
change between versions. It's also possible for struct exception_table_entry
to have changed format, hence leading to other kind of errors if parsing of the
payload is done ahead of checking if the Xen build-id matches.
Move the Xen build ID check as early as possible. To do so introduce a new
check_xen_buildid() function that parses and checks the Xen build-id before
moving the payload. Since the expected Xen build-id is used early to
detect whether the livepatch payload could be loaded, there's no reason to
store it in the payload struct, as a non-matching Xen build-id won't get the
payload populated in the first place.
Note printing the expected Xen build ID has part of dumping the payload
information is no longer done: all loaded payloads would have Xen build IDs
matching the running Xen, otherwise they would have failed to load.
Fixes: 879615f5db ('livepatch: Always check hypervisor build ID upon livepatch upload')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: fa49f4be41
master date: 2024-09-26 14:18:03 +0100
The following sections: .note.gnu.build-id, .livepatch.xen_depends and
.livepatch.depends are mandatory and ensured to be present by
check_special_sections() before prepare_payload() is called.
Simplify the logic in prepare_payload() by introducing a generic function to
parse the sections that contain a buildid. Note the function assumes the
buildid related section to always be present.
No functional change intended.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 86d09d16dd
master date: 2024-09-26 14:18:03 +0100
The Elf loading logic will initially use the `data` section field to stash a
pointer to the temporary loaded data (from the buffer allocated in
livepatch_upload(), which is later relocated and the new pointer stashed in
`load_addr`.
Remove this dual field usage and use an `addr` uniformly. Initially data will
point to the temporary buffer, until relocation happens, at which point the
pointer will be updated to the relocated address.
This avoids leaving a dangling pointer in the `data` field once the temporary
buffer is freed by livepatch_upload().
Note the `addr` field cannot retain the const attribute from the previous
`data`field, as there's logic that performs manipulations against the loaded
sections, like applying relocations or sorting the exception table.
No functional change intended.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Michal Orzel <michal.orzel@amd.com>
master commit: 8c81423038
master date: 2024-09-26 14:18:03 +0100
The current logic to chose the preferred reboot method is based on the mode Xen
has been booted into, so if the box is booted from UEFI, the preferred reboot
method will be to use the ResetSystem() run time service call.
However, that method seems to be widely untested, and quite often leads to a
result similar to:
Hardware Dom0 shutdown: rebooting machine
----[ Xen-4.18-unstable x86_64 debug=y Tainted: C ]----
CPU: 0
RIP: e008:[<0000000000000017>] 0000000000000017
RFLAGS: 0000000000010202 CONTEXT: hypervisor
[...]
Xen call trace:
[<0000000000000017>] R 0000000000000017
[<ffff83207eff7b50>] S ffff83207eff7b50
[<ffff82d0403525aa>] F machine_restart+0x1da/0x261
[<ffff82d04035263c>] F apic_wait_icr_idle+0/0x37
[<ffff82d040233689>] F smp_call_function_interrupt+0xc7/0xcb
[<ffff82d040352f05>] F call_function_interrupt+0x20/0x34
[<ffff82d04033b0d5>] F do_IRQ+0x150/0x6f3
[<ffff82d0402018c2>] F common_interrupt+0x132/0x140
[<ffff82d040283d33>] F arch/x86/acpi/cpu_idle.c#acpi_idle_do_entry+0x113/0x129
[<ffff82d04028436c>] F arch/x86/acpi/cpu_idle.c#acpi_processor_idle+0x3eb/0x5f7
[<ffff82d04032a549>] F arch/x86/domain.c#idle_loop+0xec/0xee
****************************************
Panic on CPU 0:
FATAL TRAP: vector = 6 (invalid opcode)
****************************************
Which in most cases does lead to a reboot, however that's unreliable.
Change the default reboot preference to prefer ACPI over UEFI if available and
not in reduced hardware mode.
This is in line to what Linux does, so it's unlikely to cause issues on current
and future hardware, since there's a much higher chance of vendors testing
hardware with Linux rather than Xen.
Add a special case for one Acer model that does require being rebooted using
ResetSystem(). See Linux commit 0082517fa4bce for rationale.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
# Commit 44a37c7832
# Date 2024-08-06 08:37:10 +0200
# Author Jan Beulich <jbeulich@suse.com>
# Committer Jan Beulich <jbeulich@suse.com>
x86/shutdown: use DMI_MATCH2()
... to please Misra C:2012 Rule 9.3 (Arrays shall not be partially
initialized).
Fixes: d81dd31303 ("x86/shutdown: change default reboot method preference")
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
master commit: d81dd31303
master date: 2024-08-05 10:16:54 +0200
master commit: 44a37c7832
master date: 2024-08-06 08:37:10 +0200
GNU assembly that supports such feature will unconditionally add a
.note.gnu.property section to object files. The content of that section can
change depending on the generated instructions. The current logic in
livepatch-build-tools doesn't know how to deal with such section changing
as a result of applying a patch and rebuilding.
Since .note.gnu.property is not consumed by the Xen build, suppress its
addition when livepatch support is enabled.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 718400a54d
master date: 2024-11-11 13:19:45 +0100
Clang 19 complains with the following error when building libxl:
libxl_utils.c:48:15: error: variable length array folded to constant array as an extension [-Werror,-Wgnu-folding-constant]
48 | char path[strlen("/local/domain") + 12];
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
Replace the usage of strlen() with sizeof, which allows the literal
string length to be known at build time. Note sizeof accounts for the
NUL terminator while strlen() didn't, hence subtract 1 from the total
size calculation.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Frediano Ziglio <frediano.ziglio@cloud.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Anthony PERARD <anthony.perard@vates.tech>
master commit: a7c7c3f642
master date: 2024-11-05 19:59:50 +0000
When using AMD-Vi interrupt remapping the vector field in the IO-APIC RTE is
repurposed to contain part of the offset into the remapping table. Previous to
2ca9fbd739 Xen had logic so that the offset into the interrupt remapping
table would match the vector. Such logic was mandatory for end of interrupt to
work, since the vector field (even when not containing a vector) is used by the
IO-APIC to find for which pin the EOI must be performed.
A simple solution wold be to read the IO-APIC RTE each time an EOI is to be
performed, so the raw value of the vector field can be obtained. However
that's likely to perform poorly. Instead introduce a cache to store the
EOI handles when using interrupt remapping, so that the IO-APIC driver can
translate pins into EOI handles without having to read the IO-APIC RTE entry.
Note that to simplify the logic such cache is used unconditionally when
interrupt remapping is enabled, even if strictly it would only be required
for AMD-Vi.
Reported-by: Willi Junga <xenproject@ymy.be>
Suggested-by: David Woodhouse <dwmw@amazon.co.uk>
Fixes: 2ca9fbd739 ('AMD IOMMU: allocate IRTE entries instead of using a static mapping')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Tested-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 86001b3970
master date: 2024-11-05 10:36:53 +0000
IVMD table contains restrictions about memory which must be mandatory assigned
to devices (and which permissions it should use), or memory that should be
never accessible to devices.
Some hardware however contains ranges in IVMD that reference devices outside of
the IVHD tables (in other words, devices not behind any IOMMU). Such mismatch
will cause Xen to fail in register_range_for_device(), ultimately leading to
the IOMMU being disabled, and Xen crashing as x2APIC support might be already
enabled and relying on the IOMMU functionality.
Relax IVMD parsing: allow IVMD blocks to reference devices not assigned to any
IOMMU. It's impossible for Xen to fulfill the requirement in the IVMD block if
the device is not behind any IOMMU, but it's no worse than booting without
IOMMU support, and thus not parsing ACPI IVRS in the first place.
Reported-by: Willi Junga <xenproject@ymy.be>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
master commit: 2defb54490
master date: 2024-10-15 14:23:59 +0200
Since the addition of the MSR_AMD64_DR{1-4}_ADDRESS_MASK MSRs to the
msrs_to_send array, the calculations for the maximum number of MSRs that
the hypercall can handle is off by 4.
Remove the addition of 4 to the maximum number of MSRs that
XEN_DOMCTL_{set,get}_vcpu_msrs supports, as those are already part of the
array.
A further adjustment could be to subtract 4 from the maximum size if the DBEXT
CPUID feature is not exposed to the guest, but guest_{rd,wr}msr() will already
perform that check when fetching or loading the MSRs. The maximum array is
used to indicate the caller of the buffer it needs to allocate in the get case,
and as an early input sanitation in the set case, using a buffer size slightly
lager than required is not an issue.
Fixes: 86d47adcd3 ('x86/msr: Handle MSR_AMD64_DR{0-3}_ADDRESS_MASK in the new MSR infrastructure')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: c95cd5f9c5
master date: 2024-10-08 14:37:53 +0200
The current dpci logic relies on a softirq being executed as a side effect of
the cpu_notifier_call_chain() call in the code path that offlines the target
CPU. However the call to cpu_notifier_call_chain() won't trigger any softirq
processing, and even if it did, such processing should be done after all
interrupts have been migrated off the current CPU, otherwise new pending dpci
interrupts could still appear.
Currently the ASSERT() in the cpu callback notifier is fairly easy to trigger
by doing CPU offline from a PVH dom0.
Solve this by instead moving out any dpci interrupts pending processing once
the CPU is dead. This might introduce more latency than attempting to drain
before the CPU is put offline, but it's less complex, and CPU online/offline is
not a common action. Any extra introduced latency should be tolerable.
Fixes: f6dd295381 ('dpci: replace tasklet with softirq')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 29555668b5
master date: 2024-10-07 11:10:21 +0200
Current blkif implementations (both backends and frontends) have all slight
differences about how they handle the 'sector-size' xenstore node, and how
other fields are derived from this value or hardcoded to be expressed in units
of 512 bytes.
To give some context, this is an excerpt of how different implementations use
the value in 'sector-size' as the base unit for to other fields rather than
just to set the logical sector size of the block device:
│ sectors xenbus node │ requests sector_number │ requests {first,last}_sect
────────────────────────┼─────────────────────┼────────────────────────┼───────────────────────────
FreeBSD blk{front,back} │ sector-size │ sector-size │ 512
────────────────────────┼─────────────────────┼────────────────────────┼───────────────────────────
Linux blk{front,back} │ 512 │ 512 │ 512
────────────────────────┼─────────────────────┼────────────────────────┼───────────────────────────
QEMU blkback │ sector-size │ sector-size │ sector-size
────────────────────────┼─────────────────────┼────────────────────────┼───────────────────────────
Windows blkfront │ sector-size │ sector-size │ sector-size
────────────────────────┼─────────────────────┼────────────────────────┼───────────────────────────
MiniOS │ sector-size │ 512 │ 512
An attempt was made by 67e1c050e3 in order to change the base units of the
request fields and the xenstore 'sectors' node. That however only lead to more
confusion, as the specification now clearly diverged from the reference
implementation in Linux. Such change was only implemented for QEMU Qdisk
and Windows PV blkfront.
Partially revert to the state before 67e1c050e3 while adjusting the
documentation for 'sectors' to match what it used to be previous to
2fa701e534:
* Declare 'feature-large-sector-size' deprecated. Frontends should not expose
the node, backends should not make decisions based on its presence.
* Clarify that 'sectors' xenstore node and the requests fields are always in
512-byte units, like it was previous to 2fa701e534 and 67e1c050e3.
All base units for the fields used in the protocol are 512-byte based, the
xenbus 'sector-size' field is only used to signal the logic block size. When
'sector-size' is greater than 512, blkfront implementations must make sure that
the offsets and sizes (despite being expressed in 512-byte units) are aligned
to the logical block size specified in 'sector-size', otherwise the backend
will fail to process the requests.
This will require changes to some of the frontends and backends in order to
properly support 'sector-size' nodes greater than 512.
Fixes: 2fa701e534 ('blkif.h: Provide more complete documentation of the blkif interface')
Fixes: 67e1c050e3 ('public/io/blkif.h: try to fix the semantics of sector based quantities')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Anthony PERARD <anthony.perard@vates.tech>
master commit: 221f2748e8
master date: 2024-09-12 14:04:56 +0200
Move the logic that disables SMAP so it's only performed when building a PV
dom0, PVH dom0 builder doesn't require disabling SMAP.
The fixes tag is to account for the wrong usage of cpu_has_smap in
create_dom0(), it should instead have used
boot_cpu_has(X86_FEATURE_XEN_SMAP). Fix while moving the logic to apply to PV
only.
While there also make cr4_pv32_mask __ro_after_init.
Fixes: 493ab190e5 ('xen/sm{e, a}p: allow disabling sm{e, a}p for Xen itself')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: fb1658221a
master date: 2024-08-28 19:59:07 +0100
The current code in ALT_CALL_ARG() won't successfully workaround the clang
code-generation issue if the arg parameter has a size that's not a power of 2.
While there are no such sized parameters at the moment, improve the workaround
to also be effective when such sizes are used.
Instead of using a union with a long use an unsigned long that's first
initialized to 0 and afterwards set to the argument value.
Reported-by: Alejandro Vallejo <alejandro.vallejo@cloud.com>
Suggested-by: Alejandro Vallejo <alejandro.vallejo@cloud.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 561cba38ff
master date: 2024-07-31 12:41:22 +0200
One of the error paths in the PV dom0 builder section that runs on the guest
page-tables wasn't restoring the Xen value of %cr3, neither removing the
mapcache override.
Fixes: 079ff2d32c ('libelf-loader: introduce elf_load_image')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
master commit: 1fc3f77113
master date: 2024-07-31 12:41:02 +0200
Yet another clang code generation issue when using altcalls.
The issue this time is with using loop constructs around alternative_{,v}call
instances using parameter types smaller than the register size.
Given the following example code:
static void bar(bool b)
{
unsigned int i;
for ( i = 0; i < 10; i++ )
{
int ret_;
register union {
bool e;
unsigned long r;
} di asm("rdi") = { .e = b };
register unsigned long si asm("rsi");
register unsigned long dx asm("rdx");
register unsigned long cx asm("rcx");
register unsigned long r8 asm("r8");
register unsigned long r9 asm("r9");
register unsigned long r10 asm("r10");
register unsigned long r11 asm("r11");
asm volatile ( "call %c[addr]"
: "+r" (di), "=r" (si), "=r" (dx),
"=r" (cx), "=r" (r8), "=r" (r9),
"=r" (r10), "=r" (r11), "=a" (ret_)
: [addr] "i" (&(func)), "g" (func)
: "memory" );
}
}
See: https://godbolt.org/z/qvxMGd84q
Clang will generate machine code that only resets the low 8 bits of %rdi
between loop calls, leaving the rest of the register possibly containing
garbage from the use of %rdi inside the called function. Note also that clang
doesn't truncate the input parameters at the callee, thus breaking the psABI.
Fix this by turning the `e` element in the anonymous union into an array that
consumes the same space as an unsigned long, as this forces clang to reset the
whole %rdi register instead of just the low 8 bits.
Fixes: 2ce562b2a4 ('x86/altcall: use a union as register type for function parameters on clang')
Suggested-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
master commit: d51b2f5ea1
master date: 2024-07-23 13:59:30 +0200
fixup_irqs() is used to evacuate interrupts from to be offlined CPUs. Given
the CPU is to become offline, the normal migration logic used by Xen where the
vector in the previous target(s) is left configured until the interrupt is
received on the new destination is not suitable.
Instead attempt to do as much as possible in order to prevent loosing
interrupts. If fixup_irqs() is called from the CPU to be offlined (as is
currently the case for CPU hot unplug) attempt to forward pending vectors when
interrupts that target the current CPU are migrated to a different destination.
Additionally, for interrupts that have already been moved from the current CPU
prior to the call to fixup_irqs() but that haven't been delivered to the new
destination (iow: interrupts with move_in_progress set and the current CPU set
in ->arch.old_cpu_mask) also check whether the previous vector is pending and
forward it to the new destination.
This allows us to remove the window with interrupts enabled at the bottom of
fixup_irqs(). Such window wasn't safe anyway: references to the CPU to become
offline are removed from interrupts masks, but the per-CPU vector_irq[] array
is not updated to reflect those changes (as the CPU is going offline anyway).
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Currently there's logic in fixup_irqs() that attempts to prevent
_assign_irq_vector() from failing, as fixup_irqs() is required to evacuate all
interrupts from the CPUs not present in the input mask. The current logic in
fixup_irqs() is incomplete, as it doesn't deal with interrupts that have
move_cleanup_count > 0 and a non-empty ->arch.old_cpu_mask field.
Instead of attempting to fixup the interrupt descriptor in fixup_irqs() so that
_assign_irq_vector() cannot fail, introduce logic in _assign_irq_vector()
to deal with interrupts that have either move_{in_progress,cleanup_count} set
and no remaining online CPUs in ->arch.cpu_mask.
If _assign_irq_vector() is requested to move an interrupt in the state
described above, first attempt to see if ->arch.old_cpu_mask contains any valid
CPUs that could be used as fallback, and if that's the case do move the
interrupt back to the previous destination. Note this is easier because the
vector hasn't been released yet, so there's no need to allocate and setup a new
vector on the destination.
Due to the logic in fixup_irqs() that clears offline CPUs from
->arch.old_cpu_mask (and releases the old vector if the mask becomes empty) it
shouldn't be possible to get into _assign_irq_vector() with
->arch.move_{in_progress,cleanup_count} set but no online CPUs in
->arch.old_cpu_mask.
However if ->arch.move_{in_progress,cleanup_count} is set and the interrupt has
also changed affinity, it's possible the members of ->arch.old_cpu_mask are no
longer part of the affinity set, move the interrupt to a different CPU part of
the provided mask and keep the current ->arch.old_{cpu_mask,vector} for the
pending interrupt movement to be completed.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Given the current logic it's possible for ->arch.old_cpu_mask to get out of
sync: if a CPU set in old_cpu_mask is offlined and then onlined
again without old_cpu_mask having been updated the data in the mask will no
longer be accurate, as when brought back online the CPU will no longer have
old_vector configured to handle the old interrupt source.
If there's an interrupt movement in progress, and the to be offlined CPU (which
is the call context) is in the old_cpu_mask, clear it and update the mask, so
it doesn't contain stale data.
Note that when the system is going down fixup_irqs() will be called by
smp_send_stop() from CPU 0 with a mask with only CPU 0 on it, effectively
asking to move all interrupts to the current caller (CPU 0) which is the only
CPU to remain online. In that case we don't care to migrate interrupts that
are in the process of being moved, as it's likely we won't be able to move all
interrupts to CPU 0 due to vector shortage anyway.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
The current check used in fixup_irqs() to decide whether to move around
interrupts is based on the affinity mask, but such mask can have all bits set,
and hence is unlikely to be a subset of the input mask. For example if an
interrupt has an affinity mask of all 1s, any input to fixup_irqs() that's not
an all set CPU mask would cause that interrupt to be shuffled around
unconditionally.
What fixup_irqs() care about is evacuating interrupts from CPUs not set on the
input CPU mask, and for that purpose it should check whether the interrupt is
assigned to a CPU not present in the input mask. Assume that ->arch.cpu_mask
is a subset of the ->affinity mask, and keep the current logic that resets the
->affinity mask if the interrupt has to be shuffled around.
Doing the affinity movement based on ->arch.cpu_mask requires removing the
special handling to ->arch.cpu_mask done for high priority vectors, otherwise
the adjustment done to cpu_mask makes them always skip the CPU interrupt
movement.
While there also adjust the comment as to the purpose of fixup_irqs().
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-By: Oleksii Kurochko <oleksii.kurochko@gmail.com>
The logic to move interrupts across CPUs is complex, attempt to provide a
comment that describes the expected behavior so users of the interrupt system
have more context about the usage of the arch_irq_desc structure fields.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
Due to the current rwlock logic, if the CPU calling get_cpu_maps() does
so from a cpu_hotplug_{begin,done}() region the function will still
return success, because a CPU taking the rwlock in read mode after
having taken it in write mode is allowed. Such corner case makes using
get_cpu_maps() alone not enough to prevent using the shorthand in CPU
hotplug regions.
Introduce a new helper to detect whether the current caller is between a
cpu_hotplug_{begin,done}() region and use it in send_IPI_mask() to restrict
shorthand usage.
Fixes: 5500d265a2 ('x86/smp: use APIC ALLBUT destination shorthand when possible')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
PVH dom0 is functionally very similar to PVH domU except for the domain
builder and the added set of hypercalls available to it.
The main concern with declaring it "Supported" is the lack of some features
when compared to classic PV dom0, hence switch it's status to supported with
caveats. List the known missing features, there might be more features missing
or not working as expected apart from the ones listed.
Note there's some (limited) PVH dom0 testing on both osstest and gitlab.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Release-Acked-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
When adjusting move_cleanup_count to account for CPUs that are offline also
adjust old_cpu_mask, otherwise further calls to fixup_irqs() could subtract
those again and create an imbalance in move_cleanup_count.
Fixes: 472e0b74c5 ('x86/IRQ: deal with move cleanup count state in fixup_irqs()')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Roger Pau Monne