Pull kvm fixes from Paolo Bonzini:
"ARM64:
- Pass the correct address to mte_clear_page_tags() on initialising a
tagged page
- Plug a race against a GICv4.1 doorbell interrupt while saving the
vgic-v3 pending state.
x86:
- A command line parsing fix and a clang compilation fix for
selftests
- A fix for a longstanding VMX issue, that surprisingly was only
found now to affect real world guests"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: selftests: Make reclaim_period_ms input always be positive
KVM: x86/vmx: Do not skip segment attributes if unusable bit is set
selftests: kvm: move declaration at the beginning of main()
KVM: arm64: GICv4.1: Fix race with doorbell on VPE activation/deactivation
KVM: arm64: Pass the actual page address to mte_clear_page_tags()
Pull EFI fixes from Ard Biesheuvel:
"Another couple of EFI fixes, of which the first two were already in
-next when I sent out the previous PR, but they caused some issues on
non-EFI boots so I let them simmer for a bit longer.
- ensure the EFI ResetSystem and ACPI PRM calls are recognized as
users of the EFI runtime, and therefore protected against
exceptions
- account for the EFI runtime stack in the stacktrace code
- remove Matthew Garrett's MAINTAINERS entry for efivarfs"
* tag 'efi-fixes-for-v6.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
efi: Remove Matthew Garrett as efivarfs maintainer
arm64: efi: Account for the EFI runtime stack in stack unwinder
arm64: efi: Avoid workqueue to check whether EFI runtime is live
Pull scheduler fixes from Borislav Petkov:
- Make sure the scheduler doesn't use stale frequency scaling values
when latter get disabled due to a value error
- Fix a NULL pointer access on UP configs
- Use the proper locking when updating CPU capacity
* tag 'sched_urgent_for_v6.2_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/aperfmperf: Erase stale arch_freq_scale values when disabling frequency invariance readings
sched/core: Fix NULL pointer access fault in sched_setaffinity() with non-SMP configs
sched/fair: Fixes for capacity inversion detection
sched/uclamp: Fix a uninitialized variable warnings
Pull perf fix from Borislav Petkov:
- Add Emerald Rapids model support to more perf machinery
* tag 'perf_urgent_for_v6.2_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/x86/intel/cstate: Add Emerald Rapids
perf/x86/intel: Add Emerald Rapids
When serializing and deserializing kvm_sregs, attributes of the segment
descriptors are stored by user space. For unusable segments,
vmx_segment_access_rights skips all attributes and sets them to 0.
This means we zero out the DPL (Descriptor Privilege Level) for unusable
entries.
Unusable segments are - contrary to their name - usable in 64bit mode and
are used by guests to for example create a linear map through the
NULL selector.
VMENTER checks if SS.DPL is correct depending on the CS segment type.
For types 9 (Execute Only) and 11 (Execute Read), CS.DPL must be equal to
SS.DPL [1].
We have seen real world guests setting CS to a usable segment with DPL=3
and SS to an unusable segment with DPL=3. Once we go through an sregs
get/set cycle, SS.DPL turns to 0. This causes the virtual machine to crash
reproducibly.
This commit changes the attribute logic to always preserve attributes for
unusable segments. According to [2] SS.DPL is always saved on VM exits,
regardless of the unusable bit so user space applications should have saved
the information on serialization correctly.
[3] specifies that besides SS.DPL the rest of the attributes of the
descriptors are undefined after VM entry if unusable bit is set. So, there
should be no harm in setting them all to the previous state.
[1] Intel SDM Vol 3C 26.3.1.2 Checks on Guest Segment Registers
[2] Intel SDM Vol 3C 27.3.2 Saving Segment Registers and Descriptor-Table
Registers
[3] Intel SDM Vol 3C 26.3.2.2 Loading Guest Segment Registers and
Descriptor-Table Registers
Cc: Alexander Graf <graf@amazon.de>
Cc: stable@vger.kernel.org
Signed-off-by: Hendrik Borghorst <hborghor@amazon.de>
Reviewed-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Alexander Graf <graf@amazon.com>
Message-Id: <20221114164823.69555-1-hborghor@amazon.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM/arm64 fixes for 6.2, take #2
- Pass the correct address to mte_clear_page_tags() on initialising
a tagged page
- Plug a race against a GICv4.1 doorbell interrupt while saving
the vgic-v3 pending state.
To save the vgic LPI pending state with GICv4.1, the VPEs must all be
unmapped from the ITSs so that the sGIC caches can be flushed.
The opposite is done once the state is saved.
This is all done by using the activate/deactivate irqdomain callbacks
directly from the vgic code. Crutially, this is done without holding
the irqdesc lock for the interrupts that represent the VPE. And these
callbacks are changing the state of the irqdesc. What could possibly
go wrong?
If a doorbell fires while we are messing with the irqdesc state,
it will acquire the lock and change the interrupt state concurrently.
Since we don't hole the lock, curruption occurs in on the interrupt
state. Oh well.
While acquiring the lock would fix this (and this was Shanker's
initial approach), this is still a layering violation we could do
without. A better approach is actually to free the VPE interrupt,
do what we have to do, and re-request it.
It is more work, but this usually happens only once in the lifetime
of the VM and we don't really care about this sort of overhead.
Fixes: f66b7b151e ("KVM: arm64: GICv4.1: Try to save VLPI state in save_pending_tables")
Reported-by: Shanker Donthineni <sdonthineni@nvidia.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230118022348.4137094-1-sdonthineni@nvidia.com
Commit d77e59a8fc ("arm64: mte: Lock a page for MTE tag
initialisation") added a call to mte_clear_page_tags() in case a
prior mte_copy_tags_from_user() failed in order to avoid stale tags in
the guest page (it should have really been a separate commit).
Unfortunately, the argument passed to this function was the address of
the struct page rather than the actual page address. Fix this function
call.
Fixes: d77e59a8fc ("arm64: mte: Lock a page for MTE tag initialisation")
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Marc Zyngier <maz@kernel.org>
Reviewed-by: Oliver Upton <oliver.upton@linux.dev>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20230119170902.1574756-1-catalin.marinas@arm.com
Pull ARM SoC DT and driver fixes from Arnd Bergmann:
"Lots of dts fixes for Qualcomm Snapdragon and NXP i.MX platforms,
including:
- A regression fix for SDHCI controllers on Inforce 6540, and another
SDHCI fix on SM8350
- Reenable cluster idle on sm8250 after the the code fix is upstream
- multiple fixes for the QMP PHY binding, needing an incompatible dt
change
- The reserved memory map is updated on Xiaomi Mi 4C and Huawei Nexus
6P, to avoid instabilities caused by use of protected memory
regions
- Fix i.MX8MP DT for missing GPC Interrupt, power-domain typo and USB
clock error
- A couple of verdin-imx8mm DT fixes for audio playback support
- Fix pca9547 i2c-mux node name for i.MX and Vybrid device trees
- Fix an imx93-11x11-evk uSDHC pad setting problem that causes Micron
eMMC CMD8 CRC error in HS400ES/HS400 mode
The remaining ARM and RISC-V platforms only have very few smaller dts
bugfixes this time:
- A fix for the SiFive unmatched board's PCI memory space
- A revert to fix a regression with GPIO on Marvell Armada
- A fix for the UART address on Marvell AC5
- Missing chip-select phandles for stm32 boards
- Selecting the correct clock for the sam9x60 memory controller
- Amlogic based Odroid-HC4 needs a revert to restore USB
functionality.
And finally, there are some minor code fixes:
- Build fixes for OMAP1, pxa, riscpc, raspberry pi firmware, and zynq
firmware
- memory controller driver fixes for an OMAP regression and older
bugs on tegra, atmel and mvebu
- reset controller fixes for ti-sci and uniphier platforms
- ARM SCMI firmware fixes for a couple of rare corner cases
- Qualcomm platform driver fixes for incorrect error handling and a
backwards compatibility fix for the apr driver using older dtb
- NXP i.MX SoC driver fixes for HDMI output, error handling in the
imx8 soc-id and missing reference counting on older cpuid code"
* tag 'soc-fixes-6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (60 commits)
firmware: zynqmp: fix declarations for gcc-13
ARM: dts: stm32: Fix qspi pinctrl phandle for stm32mp151a-prtt1l
ARM: dts: stm32: Fix qspi pinctrl phandle for stm32mp157c-emstamp-argon
ARM: dts: stm32: Fix qspi pinctrl phandle for stm32mp15xx-dhcom-som
ARM: dts: stm32: Fix qspi pinctrl phandle for stm32mp15xx-dhcor-som
ARM: dts: at91: sam9x60: fix the ddr clock for sam9x60
ARM: omap1: fix building gpio15xx
ARM: omap1: fix !ARCH_OMAP1_ANY link failures
firmware: raspberrypi: Fix type assignment
arm64: dts: qcom: msm8992-libra: Fix the memory map
arm64: dts: qcom: msm8992: Don't use sfpb mutex
PM: AVS: qcom-cpr: Fix an error handling path in cpr_probe()
arm64: dts: msm8994-angler: fix the memory map
arm64: dts: marvell: AC5/AC5X: Fix address for UART1
ARM: footbridge: drop unnecessary inclusion
Revert "ARM: dts: armada-39x: Fix compatible string for gpios"
Revert "ARM: dts: armada-38x: Fix compatible string for gpios"
ARM: pxa: enable PXA310/PXA320 for DT-only build
riscv: dts: sifive: fu740: fix size of pcie 32bit memory
soc: qcom: apr: Make qcom,protection-domain optional again
...
Pull s390 build fix from Heiko Carstens:
- Workaround invalid gcc-11 out of bounds read warning caused by s390's
S390_lowcore definition. This happens only with gcc 11.1.0 and
11.2.0.
The code which causes this warning will be gone with the next merge
window. Therefore just replace the memcpy() with a for loop to get
rid of the warning.
* tag 's390-6.2-3' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
s390: workaround invalid gcc-11 out of bounds read warning
GCC 11.1.0 and 11.2.0 generate a wrong warning when compiling the
kernel e.g. with allmodconfig:
arch/s390/kernel/setup.c: In function ‘setup_lowcore_dat_on’:
./include/linux/fortify-string.h:57:33: error: ‘__builtin_memcpy’ reading 128 bytes from a region of size 0 [-Werror=stringop-overread]
...
arch/s390/kernel/setup.c:526:9: note: in expansion of macro ‘memcpy’
526 | memcpy(abs_lc->cregs_save_area, S390_lowcore.cregs_save_area,
| ^~~~~~
This could be addressed by using absolute_pointer() with the
S390_lowcore macro, but this is not a good idea since this generates
worse code for performance critical paths.
Therefore simply use a for loop to copy the array in question and get
rid of the warning.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
When exception is triggered, code flow go handle_\exception in some
cases. One of stackframe in this case as follows,
high -> +-------+
| REGS | <- a pt_regs
| |
| | <- ex trigger
| REGS | <- ex pt_regs <-+
| | |
| | |
low -> +-------+ ->unwind-+
When unwinder unwinds to handler_\exception it cannot go on prologue
analysis. Because it is an asynchronous code flow, we should get the
next frame PC from regs->csr_era rather than regs->regs[1]. At init time
we copy the handlers to eentry and also copy them to NUMA-affine memory
named pcpu_handlers if NUMA is enabled. Thus, unwinder cannot unwind
normally. To solve this, we try to give some hints in handler_\exception
and fixup unwinders in unwind_next_frame().
Reported-by: Qing Zhang <zhangqing@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
The prolugue unwinder rely on symbol info. When PC is not in kernel text
address, it cannot find relative symbol info and it will be broken. The
guess unwinder will be used in this case. And the guess unwinder code in
prolugue unwinder is redundant. Strip it out and set the unwinder type
in unwind_state. Make guess_unwinder::unwind_next_frame() as default way
when other unwinders cannot unwind in some extreme case.
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
The stack frame when function_graph enable like follows,
--------- <- function sp_on_entry
|
|
|
FAKE_RA <- sp_on_entry - sizeof(pt_regs) + PT_R1
|
--------- <- sp_on_entry - sizeof(pt_regs)
So if we want to get the &FAKE_RA we should get sp_on_entry first. In
the unwinder_prologue case, we can get the sp_on_entry as state->sp,
because we try to calculate each CFA and the ra saved address. But in
the unwinder_guess case, we cannot get it because we do not try to
calculate the CFA. Although LoongArch have not fixed frame, the $ra is
saved at CFA - 8 in most cases, we can try guess, too. As we store the
pc in state, we not need to dereference state->sp, too.
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
At unwind_start(), it is better to get its frame info here rather than
get them outside, even we don't have 'regs'. In this way we can simply
use unwind_{start, next_frame, done} outside.
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
When state->first is not set, the PC is a return address in the previous
frame. We need to adjust its value in case overflow to the next symbol.
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
There exists a common function sign_extend64() to sign extend a 64-bit
value using specified bit as sign-bit in include/linux/bitops.h, it is
more efficient, let us use it and remove the arch-specific sign_extend()
under arch/loongarch.
Suggested-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
The EFI runtime services run from a dedicated stack now, and so the
stack unwinder needs to be informed about this.
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Comparing current_work() against efi_rts_work.work is sufficient to
decide whether current is currently running EFI runtime services code at
any level in its call stack.
However, there are other potential users of the EFI runtime stack, such
as the ACPI subsystem, which may invoke efi_call_virt_pointer()
directly, and so any sync exceptions occurring in firmware during those
calls are currently misidentified.
So instead, let's check whether the stashed value of the thread stack
pointer points into current's thread stack. This can only be the case if
current was interrupted while running EFI runtime code. Note that this
implies that we should clear the stashed value after switching back, to
avoid false positives.
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
mvebu fixes for 6.2 (part 1)
Fix regression for gpio support on Armada 38x and Armada 38x
Fix address for UART1 on AC5/AC5X
* tag 'mvebu-fixes-6.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/gclement/mvebu:
arm64: dts: marvell: AC5/AC5X: Fix address for UART1
Revert "ARM: dts: armada-39x: Fix compatible string for gpios"
Revert "ARM: dts: armada-38x: Fix compatible string for gpios"
Link: https://lore.kernel.org/r/87mt6mg08k.fsf@BL-laptop
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Once disable_freq_invariance_work is called the scale_freq_tick function
will not compute or update the arch_freq_scale values.
However the scheduler will still read these values and use them.
The result is that the scheduler might perform unfair decisions based on stale
values.
This patch adds the step of setting the arch_freq_scale values for all
cpus to the default (max) value SCHED_CAPACITY_SCALE, Once all cpus
have the same arch_freq_scale value the scaling is meaningless.
Signed-off-by: Yair Podemsky <ypodemsk@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230110160206.75912-1-ypodemsk@redhat.com
Pull x86 fixes from Borislav Petkov:
- Make sure the poking PGD is pinned for Xen PV as it requires it this
way
- Fixes for two resctrl races when moving a task or creating a new
monitoring group
- Fix SEV-SNP guests running under HyperV where MTRRs are disabled to
not return a UC- type mapping type on memremap() and thus cause a
serious slowdown
- Fix insn mnemonics in bioscall.S now that binutils is starting to fix
confusing insn suffixes
* tag 'x86_urgent_for_v6.2_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/mm: fix poking_init() for Xen PV guests
x86/resctrl: Fix event counts regression in reused RMIDs
x86/resctrl: Fix task CLOSID/RMID update race
x86/pat: Fix pat_x_mtrr_type() for MTRR disabled case
x86/boot: Avoid using Intel mnemonics in AT&T syntax asm
Pull powerpc fixes from Michael Ellerman:
- Fix a build failure with some versions of ld that have an odd version
string
- Fix incorrect use of mutex in the IMC PMU driver
Thanks to Kajol Jain, Michael Petlan, Ojaswin Mujoo, Peter Zijlstra, and
Yang Yingliang.
* tag 'powerpc-6.2-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
powerpc/64s/hash: Make stress_hpt_timer_fn() static
powerpc/imc-pmu: Fix use of mutex in IRQs disabled section
powerpc/boot: Fix incorrect version calculation issue in ld_version
Pull pci fixes from Bjorn Helgaas:
- Work around apparent firmware issue that made Linux reject MMCONFIG
space, which broke PCI extended config space (Bjorn Helgaas)
- Fix CONFIG_PCIE_BT1 dependency due to mid-air collision between a
PCI_MSI_IRQ_DOMAIN -> PCI_MSI change and addition of PCIE_BT1 (Lukas
Bulwahn)
* tag 'pci-v6.2-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
x86/pci: Treat EfiMemoryMappedIO as reservation of ECAM space
x86/pci: Simplify is_mmconf_reserved() messages
PCI: dwc: Adjust to recent removal of PCI_MSI_IRQ_DOMAIN
Pull kvm fixes from Paolo Bonzini:
"ARM:
- Fix the PMCR_EL0 reset value after the PMU rework
- Correctly handle S2 fault triggered by a S1 page table walk by not
always classifying it as a write, as this breaks on R/O memslots
- Document why we cannot exit with KVM_EXIT_MMIO when taking a write
fault from a S1 PTW on a R/O memslot
- Put the Apple M2 on the naughty list for not being able to
correctly implement the vgic SEIS feature, just like the M1 before
it
- Reviewer updates: Alex is stepping down, replaced by Zenghui
x86:
- Fix various rare locking issues in Xen emulation and teach lockdep
to detect them
- Documentation improvements
- Do not return host topology information from KVM_GET_SUPPORTED_CPUID"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: x86/xen: Avoid deadlock by adding kvm->arch.xen.xen_lock leaf node lock
KVM: Ensure lockdep knows about kvm->lock vs. vcpu->mutex ordering rule
KVM: x86/xen: Fix potential deadlock in kvm_xen_update_runstate_guest()
KVM: x86/xen: Fix lockdep warning on "recursive" gpc locking
Documentation: kvm: fix SRCU locking order docs
KVM: x86: Do not return host topology information from KVM_GET_SUPPORTED_CPUID
KVM: nSVM: clarify recalc_intercepts() wrt CR8
MAINTAINERS: Remove myself as a KVM/arm64 reviewer
MAINTAINERS: Add Zenghui Yu as a KVM/arm64 reviewer
KVM: arm64: vgic: Add Apple M2 cpus to the list of broken SEIS implementations
KVM: arm64: Convert FSC_* over to ESR_ELx_FSC_*
KVM: arm64: Document the behaviour of S1PTW faults on RO memslots
KVM: arm64: Fix S1PTW handling on RO memslots
KVM: arm64: PMU: Fix PMCR_EL0 reset value
Pull EFI fixes from Ard Biesheuvel:
- avoid a potential crash on the efi_subsys_init() error path
- use more appropriate error code for runtime services calls issued
after a crash in the firmware occurred
- avoid READ_ONCE() for accessing firmware tables that may appear
misaligned in memory
* tag 'efi-fixes-for-v6.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
efi: tpm: Avoid READ_ONCE() for accessing the event log
efi: rt-wrapper: Add missing include
efi: fix userspace infinite retry read efivars after EFI runtime services page fault
efi: fix NULL-deref in init error path
Pull arm64 fixes from Will Deacon:
"Here's a sizeable batch of Friday the 13th arm64 fixes for -rc4. What
could possibly go wrong?
The obvious reason we have so much here is because of the holiday
season right after the merge window, but we've also brought back an
erratum workaround that was previously dropped at the last minute and
there's an MTE coredumping fix that strays outside of the arch/arm64
directory.
Summary:
- Fix PAGE_TABLE_CHECK failures on hugepage splitting path
- Fix PSCI encoding of MEM_PROTECT_RANGE function in UAPI header
- Fix NULL deref when accessing debugfs node if PSCI is not present
- Fix MTE core dumping when VMA list is being updated concurrently
- Fix SME signal frame handling when SVE is not implemented by the
CPU
- Fix asm constraints for cmpxchg_double() to hazard both words
- Fix build failure with stack tracer and older versions of Clang
- Bring back workaround for Cortex-A715 erratum 2645198"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: Fix build with CC=clang, CONFIG_FTRACE=y and CONFIG_STACK_TRACER=y
arm64/mm: Define dummy pud_user_exec() when using 2-level page-table
arm64: errata: Workaround possible Cortex-A715 [ESR|FAR]_ELx corruption
firmware/psci: Don't register with debugfs if PSCI isn't available
firmware/psci: Fix MEM_PROTECT_RANGE function numbers
arm64/signal: Always allocate SVE signal frames on SME only systems
arm64/signal: Always accept SVE signal frames on SME only systems
arm64/sme: Fix context switch for SME only systems
arm64: cmpxchg_double*: hazard against entire exchange variable
arm64/uprobes: change the uprobe_opcode_t typedef to fix the sparse warning
arm64: mte: Avoid the racy walk of the vma list during core dump
elfcore: Add a cprm parameter to elf_core_extra_{phdrs,data_size}
arm64: mte: Fix double-freeing of the temporary tag storage during coredump
arm64: ptrace: Use ARM64_SME to guard the SME register enumerations
arm64/mm: add pud_user_exec() check in pud_user_accessible_page()
arm64/mm: fix incorrect file_map_count for invalid pmd
Pull s390 fixes from Heiko Carstens:
- Add various missing READ_ONCE() to cmpxchg() loops prevent the
compiler from potentially generating incorrect code. This includes a
rather large change to the s390 specific hardware sampling code and
its current use of cmpxchg_double().
Do the fix now to get it out of the way of Peter Zijlstra's
cmpxchg128() work, and have something that can be backported. The
added new code includes a private 128 bit cmpxchg variant which will
be removed again after Peter's rework is available. Also note that
this 128 bit cmpxchg variant is used to implement 128 bit
READ_ONCE(), while strictly speaking it wouldn't be necessary, and
_READ_ONCE() should also be sufficient; even though it isn't obvious
for all converted locations that this is the case. Therefore use this
implementation for for the sake of clarity and consistency for now.
- Fix ipl report address handling to avoid kdump failures/hangs.
- Fix misuse of #(el)if in kernel decompressor.
- Define RUNTIME_DISCARD_EXIT to fix link error with GNU ld < 2.36,
caused by the recently changed discard behaviour.
- Make sure _edata and _end symbols are always page aligned.
- The current header guard DEBUG_H in one of the s390 specific header
files is too generic and conflicts with the ath9k wireless driver.
Add an _ASM_S390_ prefix to the guard to make it unique.
- Update defconfigs.
* tag 's390-6.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
s390: update defconfigs
KVM: s390: interrupt: use READ_ONCE() before cmpxchg()
s390/percpu: add READ_ONCE() to arch_this_cpu_to_op_simple()
s390/cpum_sf: add READ_ONCE() semantics to compare and swap loops
s390/kexec: fix ipl report address for kdump
s390: fix -Wundef warning for CONFIG_KERNEL_ZSTD
s390: define RUNTIME_DISCARD_EXIT to fix link error with GNU ld < 2.36
s390: expicitly align _edata and _end symbols on page boundary
s390/debug: add _ASM_S390_ prefix to header guard
Pull xen fixes from Juergen Gross:
- two cleanup patches
- a fix of a memory leak in the Xen pvfront driver
- a fix of a locking issue in the Xen hypervisor console driver
* tag 'for-linus-6.2-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
xen/pvcalls: free active map buffer on pvcalls_front_free_map
hvc/xen: lock console list traversal
x86/xen: Remove the unused function p2m_index()
xen: make remove callback of xen driver void returned
Pull perf events hw enablement from Ingo Molnar:
- More hardware-enablement for Intel Meteor Lake & Emerald Rapid
systems: pure model ID enumeration additions that do not affect other
systems.
* tag 'perf-urgent-2023-01-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/x86/intel/uncore: Add Emerald Rapids
perf/x86/msr: Add Emerald Rapids
perf/x86/msr: Add Meteor Lake support
perf/x86/cstate: Add Meteor Lake support
Commit 3f4c8211d9 ("x86/mm: Use mm_alloc() in poking_init()") broke
the kernel for running as Xen PV guest.
It seems as if the new address space is never activated before being
used, resulting in Xen rejecting to accept the new CR3 value (the PGD
isn't pinned).
Fix that by adding the now missing call of paravirt_arch_dup_mmap() to
poking_init(). That call was previously done by dup_mm()->dup_mmap() and
it is a NOP for all cases but for Xen PV, where it is just doing the
pinning of the PGD.
Fixes: 3f4c8211d9 ("x86/mm: Use mm_alloc() in poking_init()")
Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20230109150922.10578-1-jgross@suse.com
In commit 14243b3871 ("KVM: x86/xen: Add KVM_IRQ_ROUTING_XEN_EVTCHN
and event channel delivery") the clever version of me left some helpful
notes for those who would come after him:
/*
* For the irqfd workqueue, using the main kvm->lock mutex is
* fine since this function is invoked from kvm_set_irq() with
* no other lock held, no srcu. In future if it will be called
* directly from a vCPU thread (e.g. on hypercall for an IPI)
* then it may need to switch to using a leaf-node mutex for
* serializing the shared_info mapping.
*/
mutex_lock(&kvm->lock);
In commit 2fd6df2f2b ("KVM: x86/xen: intercept EVTCHNOP_send from guests")
the other version of me ran straight past that comment without reading it,
and introduced a potential deadlock by taking vcpu->mutex and kvm->lock
in the wrong order.
Solve this as originally suggested, by adding a leaf-node lock in the Xen
state rather than using kvm->lock for it.
Fixes: 2fd6df2f2b ("KVM: x86/xen: intercept EVTCHNOP_send from guests")
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Message-Id: <20230111180651.14394-4-dwmw2@infradead.org>
[Rebase, add docs. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The kvm_xen_update_runstate_guest() function can be called when the vCPU
is being scheduled out, from a preempt notifier. It *opportunistically*
updates the runstate area in the guest memory, if the gfn_to_pfn_cache
which caches the appropriate address is still valid.
If there is *contention* when it attempts to obtain gpc->lock, then
locking inside the priority inheritance checks may cause a deadlock.
Lockdep reports:
[13890.148997] Chain exists of:
&gpc->lock --> &p->pi_lock --> &rq->__lock
[13890.149002] Possible unsafe locking scenario:
[13890.149003] CPU0 CPU1
[13890.149004] ---- ----
[13890.149005] lock(&rq->__lock);
[13890.149007] lock(&p->pi_lock);
[13890.149009] lock(&rq->__lock);
[13890.149011] lock(&gpc->lock);
[13890.149013]
*** DEADLOCK ***
In the general case, if there's contention for a read lock on gpc->lock,
that's going to be because something else is either invalidating or
revalidating the cache. Either way, we've raced with seeing it in an
invalid state, in which case we would have aborted the opportunistic
update anyway.
So in the 'atomic' case when called from the preempt notifier, just
switch to using read_trylock() and avoid the PI handling altogether.
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Message-Id: <20230111180651.14394-2-dwmw2@infradead.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In commit 5ec3289b31 ("KVM: x86/xen: Compatibility fixes for shared runstate
area") we declared it safe to obtain two gfn_to_pfn_cache locks at the same
time:
/*
* The guest's runstate_info is split across two pages and we
* need to hold and validate both GPCs simultaneously. We can
* declare a lock ordering GPC1 > GPC2 because nothing else
* takes them more than one at a time.
*/
However, we forgot to tell lockdep. Do so, by setting a subclass on the
first lock before taking the second.
Fixes: 5ec3289b31 ("KVM: x86/xen: Compatibility fixes for shared runstate area")
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Message-Id: <20230111180651.14394-1-dwmw2@infradead.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
