Pull pci updates from Bjorn Helgaas:
"Enumeration:
- Consolidate duplicated 'next function' scanning and extend to allow
'isolated functions' on s390, similar to existing hypervisors
(Niklas Schnelle)
Resource management:
- Implement pci_iobar_pfn() for sparc, which allows us to remove the
sparc-specific pci_mmap_page_range() and pci_mmap_resource_range().
This removes the ability to map the entire PCI I/O space using
/proc/bus/pci, but we believe that's already been broken since
v2.6.28 (Arnd Bergmann)
- Move common PCI definitions to asm-generic/pci.h and rework others
to be be more specific and more encapsulated in arches that need
them (Stafford Horne)
Power management:
- Convert drivers to new *_PM_OPS macros to avoid need for '#ifdef
CONFIG_PM_SLEEP' or '__maybe_unused' (Bjorn Helgaas)
Virtualization:
- Add ACS quirk for Broadcom BCM5750x multifunction NICs that isolate
the functions but don't advertise an ACS capability (Pavan Chebbi)
Error handling:
- Clear PCI Status register during enumeration in case firmware left
errors logged (Kai-Heng Feng)
- When we have native control of AER, enable error reporting for all
devices that support AER. Previously only a few drivers enabled
this (Stefan Roese)
- Keep AER error reporting enabled for switches. Previously we
enabled this during enumeration but immediately disabled it (Stefan
Roese)
- Iterate over error counters instead of error strings to avoid
printing junk in AER sysfs counters (Mohamed Khalfella)
ASPM:
- Remove pcie_aspm_pm_state_change() so ASPM config changes, e.g.,
via sysfs, are not lost across power state changes (Kai-Heng Feng)
Endpoint framework:
- Don't stop an EPC when unbinding an EPF from it (Shunsuke Mie)
Endpoint embedded DMA controller driver:
- Simplify and clean up support for the DesignWare embedded DMA
(eDMA) controller (Frank Li, Serge Semin)
Broadcom STB PCIe controller driver:
- Avoid config space accesses when link is down because we can't
recover from the CPU aborts these cause (Jim Quinlan)
- Look for power regulators described under Root Ports in DT and
enable them before scanning the secondary bus (Jim Quinlan)
- Disable/enable regulators in suspend/resume (Jim Quinlan)
Freescale i.MX6 PCIe controller driver:
- Simplify and clean up clock and PHY management (Richard Zhu)
- Disable/enable regulators in suspend/resume (Richard Zhu)
- Set PCIE_DBI_RO_WR_EN before writing DBI registers (Richard Zhu)
- Allow speeds faster than Gen2 (Richard Zhu)
- Make link being down a non-fatal error so controller probe doesn't
fail if there are no Endpoints connected (Richard Zhu)
Loongson PCIe controller driver:
- Add ACPI and MCFG support for Loongson LS7A (Huacai Chen)
- Avoid config reads to non-existent LS2K/LS7A devices because a
hardware defect causes machine hangs (Huacai Chen)
- Work around LS7A integrated devices that report incorrect Interrupt
Pin values (Jianmin Lv)
Marvell Aardvark PCIe controller driver:
- Add support for AER and Slot capability on emulated bridge (Pali
Rohár)
MediaTek PCIe controller driver:
- Add Airoha EN7532 to DT binding (John Crispin)
- Allow building of driver for ARCH_AIROHA (Felix Fietkau)
MediaTek PCIe Gen3 controller driver:
- Print decoded LTSSM state when the link doesn't come up (Jianjun
Wang)
NVIDIA Tegra194 PCIe controller driver:
- Convert DT binding to json-schema (Vidya Sagar)
- Add DT bindings and driver support for Tegra234 Root Port and
Endpoint mode (Vidya Sagar)
- Fix some Root Port interrupt handling issues (Vidya Sagar)
- Set default Max Payload Size to 256 bytes (Vidya Sagar)
- Fix Data Link Feature capability programming (Vidya Sagar)
- Extend Endpoint mode support to devices beyond Controller-5 (Vidya
Sagar)
Qualcomm PCIe controller driver:
- Rework clock, reset, PHY power-on ordering to avoid hangs and
improve consistency (Robert Marko, Christian Marangi)
- Move pipe_clk handling to PHY drivers (Dmitry Baryshkov)
- Add IPQ60xx support (Selvam Sathappan Periakaruppan)
- Allow ASPM L1 and substates for 2.7.0 (Krishna chaitanya chundru)
- Add support for more than 32 MSI interrupts (Dmitry Baryshkov)
Renesas R-Car PCIe controller driver:
- Convert DT binding to json-schema (Herve Codina)
- Add Renesas RZ/N1D (R9A06G032) to rcar-gen2 DT binding and driver
(Herve Codina)
Samsung Exynos PCIe controller driver:
- Fix phy-exynos-pcie driver so it follows the 'phy_init() before
phy_power_on()' PHY programming model (Marek Szyprowski)
Synopsys DesignWare PCIe controller driver:
- Simplify and clean up the DWC core extensively (Serge Semin)
- Fix an issue with programming the ATU for regions that cross a 4GB
boundary (Serge Semin)
- Enable the CDM check if 'snps,enable-cdm-check' exists; previously
we skipped it if 'num-lanes' was absent (Serge Semin)
- Allocate a 32-bit DMA-able page to be MSI target instead of using a
driver data structure that may not be addressable with 32-bit
address (Will McVicker)
- Add DWC core support for more than 32 MSI interrupts (Dmitry
Baryshkov)
Xilinx Versal CPM PCIe controller driver:
- Add DT binding and driver support for Versal CPM5 Gen5 Root Port
(Bharat Kumar Gogada)"
* tag 'pci-v5.20-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci: (150 commits)
PCI: imx6: Support more than Gen2 speed link mode
PCI: imx6: Set PCIE_DBI_RO_WR_EN before writing DBI registers
PCI: imx6: Reformat suspend callback to keep symmetric with resume
PCI: imx6: Move the imx6_pcie_ltssm_disable() earlier
PCI: imx6: Disable clocks in reverse order of enable
PCI: imx6: Do not hide PHY driver callbacks and refine the error handling
PCI: imx6: Reduce resume time by only starting link if it was up before suspend
PCI: imx6: Mark the link down as non-fatal error
PCI: imx6: Move regulator enable out of imx6_pcie_deassert_core_reset()
PCI: imx6: Turn off regulator when system is in suspend mode
PCI: imx6: Call host init function directly in resume
PCI: imx6: Disable i.MX6QDL clock when disabling ref clocks
PCI: imx6: Propagate .host_init() errors to caller
PCI: imx6: Collect clock enables in imx6_pcie_clk_enable()
PCI: imx6: Factor out ref clock disable to match enable
PCI: imx6: Move imx6_pcie_clk_disable() earlier
PCI: imx6: Move imx6_pcie_enable_ref_clk() earlier
PCI: imx6: Move PHY management functions together
PCI: imx6: Move imx6_pcie_grp_offset(), imx6_pcie_configure_type() earlier
PCI: imx6: Convert to NOIRQ_SYSTEM_SLEEP_PM_OPS()
...
Pull kvm updates from Paolo Bonzini:
"Quite a large pull request due to a selftest API overhaul and some
patches that had come in too late for 5.19.
ARM:
- Unwinder implementations for both nVHE modes (classic and
protected), complete with an overflow stack
- Rework of the sysreg access from userspace, with a complete rewrite
of the vgic-v3 view to allign with the rest of the infrastructure
- Disagregation of the vcpu flags in separate sets to better track
their use model.
- A fix for the GICv2-on-v3 selftest
- A small set of cosmetic fixes
RISC-V:
- Track ISA extensions used by Guest using bitmap
- Added system instruction emulation framework
- Added CSR emulation framework
- Added gfp_custom flag in struct kvm_mmu_memory_cache
- Added G-stage ioremap() and iounmap() functions
- Added support for Svpbmt inside Guest
s390:
- add an interface to provide a hypervisor dump for secure guests
- improve selftests to use TAP interface
- enable interpretive execution of zPCI instructions (for PCI
passthrough)
- First part of deferred teardown
- CPU Topology
- PV attestation
- Minor fixes
x86:
- Permit guests to ignore single-bit ECC errors
- Intel IPI virtualization
- Allow getting/setting pending triple fault with
KVM_GET/SET_VCPU_EVENTS
- PEBS virtualization
- Simplify PMU emulation by just using PERF_TYPE_RAW events
- More accurate event reinjection on SVM (avoid retrying
instructions)
- Allow getting/setting the state of the speaker port data bit
- Refuse starting the kvm-intel module if VM-Entry/VM-Exit controls
are inconsistent
- "Notify" VM exit (detect microarchitectural hangs) for Intel
- Use try_cmpxchg64 instead of cmpxchg64
- Ignore benign host accesses to PMU MSRs when PMU is disabled
- Allow disabling KVM's "MONITOR/MWAIT are NOPs!" behavior
- Allow NX huge page mitigation to be disabled on a per-vm basis
- Port eager page splitting to shadow MMU as well
- Enable CMCI capability by default and handle injected UCNA errors
- Expose pid of vcpu threads in debugfs
- x2AVIC support for AMD
- cleanup PIO emulation
- Fixes for LLDT/LTR emulation
- Don't require refcounted "struct page" to create huge SPTEs
- Miscellaneous cleanups:
- MCE MSR emulation
- Use separate namespaces for guest PTEs and shadow PTEs bitmasks
- PIO emulation
- Reorganize rmap API, mostly around rmap destruction
- Do not workaround very old KVM bugs for L0 that runs with nesting enabled
- new selftests API for CPUID
Generic:
- Fix races in gfn->pfn cache refresh; do not pin pages tracked by
the cache
- new selftests API using struct kvm_vcpu instead of a (vm, id)
tuple"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (606 commits)
selftests: kvm: set rax before vmcall
selftests: KVM: Add exponent check for boolean stats
selftests: KVM: Provide descriptive assertions in kvm_binary_stats_test
selftests: KVM: Check stat name before other fields
KVM: x86/mmu: remove unused variable
RISC-V: KVM: Add support for Svpbmt inside Guest/VM
RISC-V: KVM: Use PAGE_KERNEL_IO in kvm_riscv_gstage_ioremap()
RISC-V: KVM: Add G-stage ioremap() and iounmap() functions
KVM: Add gfp_custom flag in struct kvm_mmu_memory_cache
RISC-V: KVM: Add extensible CSR emulation framework
RISC-V: KVM: Add extensible system instruction emulation framework
RISC-V: KVM: Factor-out instruction emulation into separate sources
RISC-V: KVM: move preempt_disable() call in kvm_arch_vcpu_ioctl_run
RISC-V: KVM: Make kvm_riscv_guest_timer_init a void function
RISC-V: KVM: Fix variable spelling mistake
RISC-V: KVM: Improve ISA extension by using a bitmap
KVM, x86/mmu: Fix the comment around kvm_tdp_mmu_zap_leafs()
KVM: SVM: Dump Virtual Machine Save Area (VMSA) to klog
KVM: x86/mmu: Treat NX as a valid SPTE bit for NPT
KVM: x86: Do not block APIC write for non ICR registers
...
tl;dr: The Enhanced IBRS mitigation for Spectre v2 does not work as
documented for RET instructions after VM exits. Mitigate it with a new
one-entry RSB stuffing mechanism and a new LFENCE.
== Background ==
Indirect Branch Restricted Speculation (IBRS) was designed to help
mitigate Branch Target Injection and Speculative Store Bypass, i.e.
Spectre, attacks. IBRS prevents software run in less privileged modes
from affecting branch prediction in more privileged modes. IBRS requires
the MSR to be written on every privilege level change.
To overcome some of the performance issues of IBRS, Enhanced IBRS was
introduced. eIBRS is an "always on" IBRS, in other words, just turn
it on once instead of writing the MSR on every privilege level change.
When eIBRS is enabled, more privileged modes should be protected from
less privileged modes, including protecting VMMs from guests.
== Problem ==
Here's a simplification of how guests are run on Linux' KVM:
void run_kvm_guest(void)
{
// Prepare to run guest
VMRESUME();
// Clean up after guest runs
}
The execution flow for that would look something like this to the
processor:
1. Host-side: call run_kvm_guest()
2. Host-side: VMRESUME
3. Guest runs, does "CALL guest_function"
4. VM exit, host runs again
5. Host might make some "cleanup" function calls
6. Host-side: RET from run_kvm_guest()
Now, when back on the host, there are a couple of possible scenarios of
post-guest activity the host needs to do before executing host code:
* on pre-eIBRS hardware (legacy IBRS, or nothing at all), the RSB is not
touched and Linux has to do a 32-entry stuffing.
* on eIBRS hardware, VM exit with IBRS enabled, or restoring the host
IBRS=1 shortly after VM exit, has a documented side effect of flushing
the RSB except in this PBRSB situation where the software needs to stuff
the last RSB entry "by hand".
IOW, with eIBRS supported, host RET instructions should no longer be
influenced by guest behavior after the host retires a single CALL
instruction.
However, if the RET instructions are "unbalanced" with CALLs after a VM
exit as is the RET in #6, it might speculatively use the address for the
instruction after the CALL in #3 as an RSB prediction. This is a problem
since the (untrusted) guest controls this address.
Balanced CALL/RET instruction pairs such as in step #5 are not affected.
== Solution ==
The PBRSB issue affects a wide variety of Intel processors which
support eIBRS. But not all of them need mitigation. Today,
X86_FEATURE_RSB_VMEXIT triggers an RSB filling sequence that mitigates
PBRSB. Systems setting RSB_VMEXIT need no further mitigation - i.e.,
eIBRS systems which enable legacy IBRS explicitly.
However, such systems (X86_FEATURE_IBRS_ENHANCED) do not set RSB_VMEXIT
and most of them need a new mitigation.
Therefore, introduce a new feature flag X86_FEATURE_RSB_VMEXIT_LITE
which triggers a lighter-weight PBRSB mitigation versus RSB_VMEXIT.
The lighter-weight mitigation performs a CALL instruction which is
immediately followed by a speculative execution barrier (INT3). This
steers speculative execution to the barrier -- just like a retpoline
-- which ensures that speculation can never reach an unbalanced RET.
Then, ensure this CALL is retired before continuing execution with an
LFENCE.
In other words, the window of exposure is opened at VM exit where RET
behavior is troublesome. While the window is open, force RSB predictions
sampling for RET targets to a dead end at the INT3. Close the window
with the LFENCE.
There is a subset of eIBRS systems which are not vulnerable to PBRSB.
Add these systems to the cpu_vuln_whitelist[] as NO_EIBRS_PBRSB.
Future systems that aren't vulnerable will set ARCH_CAP_PBRSB_NO.
[ bp: Massage, incorporate review comments from Andy Cooper. ]
Signed-off-by: Daniel Sneddon <daniel.sneddon@linux.intel.com>
Co-developed-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Pull random number generator updates from Jason Donenfeld:
"Though there's been a decent amount of RNG-related development during
this last cycle, not all of it is coming through this tree, as this
cycle saw a shift toward tackling early boot time seeding issues,
which took place in other trees as well.
Here's a summary of the various patches:
- The CONFIG_ARCH_RANDOM .config option and the "nordrand" boot
option have been removed, as they overlapped with the more widely
supported and more sensible options, CONFIG_RANDOM_TRUST_CPU and
"random.trust_cpu". This change allowed simplifying a bit of arch
code.
- x86's RDRAND boot time test has been made a bit more robust, with
RDRAND disabled if it's clearly producing bogus results. This would
be a tip.git commit, technically, but I took it through random.git
to avoid a large merge conflict.
- The RNG has long since mixed in a timestamp very early in boot, on
the premise that a computer that does the same things, but does so
starting at different points in wall time, could be made to still
produce a different RNG state. Unfortunately, the clock isn't set
early in boot on all systems, so now we mix in that timestamp when
the time is actually set.
- User Mode Linux now uses the host OS's getrandom() syscall to
generate a bootloader RNG seed and later on treats getrandom() as
the platform's RDRAND-like faculty.
- The arch_get_random_{seed_,}_long() family of functions is now
arch_get_random_{seed_,}_longs(), which enables certain platforms,
such as s390, to exploit considerable performance advantages from
requesting multiple CPU random numbers at once, while at the same
time compiling down to the same code as before on platforms like
x86.
- A small cleanup changing a cmpxchg() into a try_cmpxchg(), from
Uros.
- A comment spelling fix"
More info about other random number changes that come in through various
architecture trees in the full commentary in the pull request:
https://lore.kernel.org/all/20220731232428.2219258-1-Jason@zx2c4.com/
* tag 'random-6.0-rc1-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/crng/random:
random: correct spelling of "overwrites"
random: handle archrandom with multiple longs
um: seed rng using host OS rng
random: use try_cmpxchg in _credit_init_bits
timekeeping: contribute wall clock to rng on time change
x86/rdrand: Remove "nordrand" flag in favor of "random.trust_cpu"
random: remove CONFIG_ARCH_RANDOM
Pull integrity updates from Mimi Zohar:
"Aside from the one EVM cleanup patch, all the other changes are kexec
related.
On different architectures different keyrings are used to verify the
kexec'ed kernel image signature. Here are a number of preparatory
cleanup patches and the patches themselves for making the keyrings -
builtin_trusted_keyring, .machine, .secondary_trusted_keyring, and
.platform - consistent across the different architectures"
* tag 'integrity-v6.0' of git://git.kernel.org/pub/scm/linux/kernel/git/zohar/linux-integrity:
kexec, KEYS, s390: Make use of built-in and secondary keyring for signature verification
arm64: kexec_file: use more system keyrings to verify kernel image signature
kexec, KEYS: make the code in bzImage64_verify_sig generic
kexec: clean up arch_kexec_kernel_verify_sig
kexec: drop weak attribute from functions
kexec_file: drop weak attribute from functions
evm: Use IS_ENABLED to initialize .enabled
Pull locking updates from Ingo Molnar:
"This was a fairly quiet cycle for the locking subsystem:
- lockdep: Fix a handful of the more complex lockdep_init_map_*()
primitives that can lose the lock_type & cause false reports. No
such mishap was observed in the wild.
- jump_label improvements: simplify the cross-arch support of initial
NOP patching by making it arch-specific code (used on MIPS only),
and remove the s390 initial NOP patching that was superfluous"
* tag 'locking-core-2022-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/lockdep: Fix lockdep_init_map_*() confusion
jump_label: make initial NOP patching the special case
jump_label: mips: move module NOP patching into arch code
jump_label: s390: avoid pointless initial NOP patching
Pull x86 kdump updates from Borislav Petkov:
- Add the ability to pass early an RNG seed to the kernel from the boot
loader
- Add the ability to pass the IMA measurement of kernel and bootloader
to the kexec-ed kernel
* tag 'x86_kdump_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/setup: Use rng seeds from setup_data
x86/kexec: Carry forward IMA measurement log on kexec
Pull x86 build updates from Borislav Petkov:
- Fix stack protector builds when cross compiling with Clang
- Other Kbuild improvements and fixes
* tag 'x86_build_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/purgatory: Omit use of bin2c
x86/purgatory: Hard-code obj-y in Makefile
x86/build: Remove unused OBJECT_FILES_NON_STANDARD_test_nx.o
x86/Kconfig: Fix CONFIG_CC_HAS_SANE_STACKPROTECTOR when cross compiling with clang
Pull misc x86 updates from Borislav Petkov:
- Add a bunch of PCI IDs for new AMD CPUs and use them in k10temp
- Free the pmem platform device on the registration error path
* tag 'x86_misc_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
hwmon: (k10temp): Add support for new family 17h and 19h models
x86/amd_nb: Add AMD PCI IDs for SMN communication
x86/pmem: Fix platform-device leak in error path
Pull x86 cpu updates from Borislav Petkov:
- Remove the vendor check when selecting MWAIT as the default idle
state
- Respect idle=nomwait when supplied on the kernel cmdline
- Two small cleanups
* tag 'x86_cpu_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/cpu: Use MSR_IA32_MISC_ENABLE constants
x86: Fix comment for X86_FEATURE_ZEN
x86: Remove vendor checks from prefer_mwait_c1_over_halt
x86: Handle idle=nomwait cmdline properly for x86_idle
Pull x86 fpu update from Borislav Petkov:
- Add machinery to initialize AMX register state in order for
AMX-capable CPUs to be able to enter deeper low-power state
* tag 'x86_fpu_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
intel_idle: Add a new flag to initialize the AMX state
x86/fpu: Add a helper to prepare AMX state for low-power CPU idle
Pull x86 vmware cleanup from Borislav Petkov:
- A single statement simplification by using the BIT() macro
* tag 'x86_vmware_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/vmware: Use BIT() macro for shifting
Pull RAS update from Borislav Petkov:
"A single RAS change:
- Probe whether hardware error injection (direct MSR writes) is
possible when injecting errors on AMD platforms. In some cases, the
platform could prohibit those"
* tag 'ras_core_for_v6.0_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/mce: Check whether writes to MCA_STATUS are getting ignored
KVM/s390, KVM/x86 and common infrastructure changes for 5.20
x86:
* Permit guests to ignore single-bit ECC errors
* Fix races in gfn->pfn cache refresh; do not pin pages tracked by the cache
* Intel IPI virtualization
* Allow getting/setting pending triple fault with KVM_GET/SET_VCPU_EVENTS
* PEBS virtualization
* Simplify PMU emulation by just using PERF_TYPE_RAW events
* More accurate event reinjection on SVM (avoid retrying instructions)
* Allow getting/setting the state of the speaker port data bit
* Refuse starting the kvm-intel module if VM-Entry/VM-Exit controls are inconsistent
* "Notify" VM exit (detect microarchitectural hangs) for Intel
* Cleanups for MCE MSR emulation
s390:
* add an interface to provide a hypervisor dump for secure guests
* improve selftests to use TAP interface
* enable interpretive execution of zPCI instructions (for PCI passthrough)
* First part of deferred teardown
* CPU Topology
* PV attestation
* Minor fixes
Generic:
* new selftests API using struct kvm_vcpu instead of a (vm, id) tuple
x86:
* Use try_cmpxchg64 instead of cmpxchg64
* Bugfixes
* Ignore benign host accesses to PMU MSRs when PMU is disabled
* Allow disabling KVM's "MONITOR/MWAIT are NOPs!" behavior
* x86/MMU: Allow NX huge pages to be disabled on a per-vm basis
* Port eager page splitting to shadow MMU as well
* Enable CMCI capability by default and handle injected UCNA errors
* Expose pid of vcpu threads in debugfs
* x2AVIC support for AMD
* cleanup PIO emulation
* Fixes for LLDT/LTR emulation
* Don't require refcounted "struct page" to create huge SPTEs
x86 cleanups:
* Use separate namespaces for guest PTEs and shadow PTEs bitmasks
* PIO emulation
* Reorganize rmap API, mostly around rmap destruction
* Do not workaround very old KVM bugs for L0 that runs with nesting enabled
* new selftests API for CPUID
The setup_profiling_timer() is mostly un-implemented by many
architectures. In many places it isn't guarded by CONFIG_PROFILE which is
needed for it to be used. Make it a weak symbol in kernel/profile.c and
remove the 'return -EINVAL' implementations from the kenrel.
There are a couple of architectures which do return 0 from the
setup_profiling_timer() function but they don't seem to do anything else
with it. To keep the /proc compatibility for now, leave these for a
future update or removal.
On ARM, this fixes the following sparse warning:
arch/arm/kernel/smp.c:793:5: warning: symbol 'setup_profiling_timer' was not declared. Should it be static?
Link: https://lkml.kernel.org/r/20220721195509.418205-1-ben-linux@fluff.org
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The archrandom interface was originally designed for x86, which supplies
RDRAND/RDSEED for receiving random words into registers, resulting in
one function to generate an int and another to generate a long. However,
other architectures don't follow this.
On arm64, the SMCCC TRNG interface can return between one and three
longs. On s390, the CPACF TRNG interface can return arbitrary amounts,
with four longs having the same cost as one. On UML, the os_getrandom()
interface can return arbitrary amounts.
So change the api signature to take a "max_longs" parameter designating
the maximum number of longs requested, and then return the number of
longs generated.
Since callers need to check this return value and loop anyway, each arch
implementation does not bother implementing its own loop to try again to
fill the maximum number of longs. Additionally, all existing callers
pass in a constant max_longs parameter. Taken together, these two things
mean that the codegen doesn't really change much for one-word-at-a-time
platforms, while performance is greatly improved on platforms such as
s390.
Acked-by: Heiko Carstens <hca@linux.ibm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Acked-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Instead of the magic numbers 1<<11 and 1<<12 use the constants
from msr-index.h. This makes it obvious where those bits
of MSR_IA32_MISC_ENABLE are consumed (and in fact that Linux
consumes them at all) to simple minds that grep for
MSR_IA32_MISC_ENABLE_.*_UNAVAIL.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20220719174714.2410374-1-pbonzini@redhat.com
When a CPU enters an idle state, a non-initialized AMX register state may
be the cause of preventing a deeper low-power state. Other extended
register states whether initialized or not do not impact the CPU idle
state.
The new helper can ensure the AMX state is initialized before the CPU is
idle, and it will be used by the intel idle driver.
Check the AMX_TILE feature bit before using XGETBV1 as a chain of
dependencies was established via cpuid_deps[]: AMX->XFD->XGETBV1.
Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20220608164748.11864-2-chang.seok.bae@intel.com
On AMD IBRS does not prevent Retbleed; as such use IBPB before a
firmware call to flush the branch history state.
And because in order to do an EFI call, the kernel maps a whole lot of
the kernel page table into the EFI page table, do an IBPB just in case
in order to prevent the scenario of poisoning the BTB and causing an EFI
call using the unprotected RET there.
[ bp: Massage. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20220715194550.793957-1-cascardo@canonical.com
The decision of whether or not to trust RDRAND is controlled by the
"random.trust_cpu" boot time parameter or the CONFIG_RANDOM_TRUST_CPU
compile time default. The "nordrand" flag was added during the early
days of RDRAND, when there were worries that merely using its values
could compromise the RNG. However, these days, RDRAND values are not
used directly but always go through the RNG's hash function, making
"nordrand" no longer useful.
Rather, the correct switch is "random.trust_cpu", which not only handles
the relevant trust issue directly, but also is general to multiple CPU
types, not just x86.
However, x86 RDRAND does have a history of being occasionally
problematic. Prior, when the kernel would notice something strange, it'd
warn in dmesg and suggest enabling "nordrand". We can improve on that by
making the test a little bit better and then taking the step of
automatically disabling RDRAND if we detect it's problematic.
Also disable RDSEED if the RDRAND test fails.
Cc: x86@kernel.org
Cc: Theodore Ts'o <tytso@mit.edu>
Suggested-by: H. Peter Anvin <hpa@zytor.com>
Suggested-by: Borislav Petkov <bp@suse.de>
Acked-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
When RDRAND was introduced, there was much discussion on whether it
should be trusted and how the kernel should handle that. Initially, two
mechanisms cropped up, CONFIG_ARCH_RANDOM, a compile time switch, and
"nordrand", a boot-time switch.
Later the thinking evolved. With a properly designed RNG, using RDRAND
values alone won't harm anything, even if the outputs are malicious.
Rather, the issue is whether those values are being *trusted* to be good
or not. And so a new set of options were introduced as the real
ones that people use -- CONFIG_RANDOM_TRUST_CPU and "random.trust_cpu".
With these options, RDRAND is used, but it's not always credited. So in
the worst case, it does nothing, and in the best case, maybe it helps.
Along the way, CONFIG_ARCH_RANDOM's meaning got sort of pulled into the
center and became something certain platforms force-select.
The old options don't really help with much, and it's a bit odd to have
special handling for these instructions when the kernel can deal fine
with the existence or untrusted existence or broken existence or
non-existence of that CPU capability.
Simplify the situation by removing CONFIG_ARCH_RANDOM and using the
ordinary asm-generic fallback pattern instead, keeping the two options
that are actually used. For now it leaves "nordrand" for now, as the
removal of that will take a different route.
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Borislav Petkov <bp@suse.de>
Acked-by: Heiko Carstens <hca@linux.ibm.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Patch series "cpumask: Fix invalid uniprocessor assumptions", v4.
On uniprocessor builds, it is currently assumed that any cpumask will
contain the single CPU: cpu0. This assumption is used to provide
optimised implementations.
The current assumption also appears to be wrong, by ignoring the fact that
users can provide empty cpumasks. This can result in bugs as explained in
[1] - for_each_cpu() will run one iteration of the loop even when passed
an empty cpumask.
This series introduces some basic tests, and updates the optimisations for
uniprocessor builds.
The x86 patch was written after the kernel test robot [2] ran into a
failed build. I have tried to list the files potentially affected by the
changes to cpumask.h, in an attempt to find any other cases that fail on
!SMP. I've gone through some of the files manually, and ran a few cross
builds, but nothing else popped up. I (build) checked about half of the
potientally affected files, but I do not have the resources to do them
all. I hope we can fix other issues if/when they pop up later.
[1] https://lore.kernel.org/all/20220530082552.46113-1-sander@svanheule.net/
[2] https://lore.kernel.org/all/202206060858.wA0FOzRy-lkp@intel.com/
This patch (of 5):
The maps to keep track of shared caches between CPUs on SMP systems are
declared in asm/smp.h, among them specifically cpu_llc_shared_map. These
maps are externally defined in cpu/smpboot.c. The latter is only compiled
on CONFIG_SMP=y, which means the declared extern symbols from asm/smp.h do
not have a corresponding definition on uniprocessor builds.
The inline cpu_llc_shared_mask() function from asm/smp.h refers to the map
declaration mentioned above. This function is referenced in cacheinfo.c
inside for_each_cpu() loop macros, to provide cpumask for the loop. On
uniprocessor builds, the symbol for the cpu_llc_shared_map does not exist.
However, the current implementation of for_each_cpu() also (wrongly)
ignores the provided mask.
By sheer luck, the compiler thus optimises out this unused reference to
cpu_llc_shared_map, and the linker therefore does not require the
cpu_llc_shared_mask to actually exist on uniprocessor builds. Only on SMP
bulids does smpboot.o exist to provide the required symbols.
To no longer rely on compiler optimisations for successful uniprocessor
builds, move the definitions of cpu_llc_shared_map and cpu_l2c_shared_map
from smpboot.c to cacheinfo.c.
Link: https://lkml.kernel.org/r/cover.1656777646.git.sander@svanheule.net
Link: https://lkml.kernel.org/r/e8167ddb570f56744a3dc12c2149a660a324d969.1656777646.git.sander@svanheule.net
Signed-off-by: Sander Vanheule <sander@svanheule.net>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Marco Elver <elver@google.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: Yury Norov <yury.norov@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Pull x86 fixes from Borislav Petkov:
- Improve the check whether the kernel supports WP mappings so that it
can accomodate a XenPV guest due to how the latter is setting up the
PAT machinery
- Now that the retbleed nightmare is public, here's the first round of
fallout fixes:
* Fix a build failure on 32-bit due to missing include
* Remove an untraining point in espfix64 return path
* other small cleanups
* tag 'x86_urgent_for_v5.19_rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/bugs: Remove apostrophe typo
um: Add missing apply_returns()
x86/entry: Remove UNTRAIN_RET from native_irq_return_ldt
x86/bugs: Mark retbleed_strings static
x86/pat: Fix x86_has_pat_wp()
x86/asm/32: Fix ANNOTATE_UNRET_SAFE use on 32-bit
Pull ACPI fix from Rafael Wysocki:
"Fix more fallout from recent changes of the ACPI CPPC handling on AMD
platforms (Mario Limonciello)"
* tag 'acpi-5.19-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
ACPI: CPPC: Fix enabling CPPC on AMD systems with shared memory
Remove a superfluous ' in the mitigation string.
Fixes: e8ec1b6e08 ("x86/bugs: Enable STIBP for JMP2RET")
Signed-off-by: Kim Phillips <kim.phillips@amd.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
commit 278311e417 ("kexec, KEYS: Make use of platform keyring for
signature verify") adds platform keyring support on x86 kexec but not
arm64.
The code in bzImage64_verify_sig uses the keys on the
.builtin_trusted_keys, .machine, if configured and enabled,
.secondary_trusted_keys, also if configured, and .platform keyrings
to verify the signed kernel image as PE file.
Cc: kexec@lists.infradead.org
Cc: keyrings@vger.kernel.org
Cc: linux-security-module@vger.kernel.org
Reviewed-by: Michal Suchanek <msuchanek@suse.de>
Signed-off-by: Coiby Xu <coxu@redhat.com>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
When commit 72f2ecb7ec ("ACPI: bus: Set CPPC _OSC bits for all
and when CPPC_LIB is supported") was introduced, we found collateral
damage that a number of AMD systems that supported CPPC but
didn't advertise support in _OSC stopped having a functional
amd-pstate driver. The _OSC was only enforced on Intel systems at that
time.
This was fixed for the MSR based designs by commit 8b356e536e
("ACPI: CPPC: Don't require _OSC if X86_FEATURE_CPPC is supported")
but some shared memory based designs also support CPPC but haven't
advertised support in the _OSC. Add support for those designs as well by
hardcoding the list of systems.
Fixes: 72f2ecb7ec ("ACPI: bus: Set CPPC _OSC bits for all and when CPPC_LIB is supported")
Fixes: 8b356e536e ("ACPI: CPPC: Don't require _OSC if X86_FEATURE_CPPC is supported")
Link: https://lore.kernel.org/all/3559249.JlDtxWtqDm@natalenko.name/
Cc: 5.18+ <stable@vger.kernel.org> # 5.18+
Reported-and-tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The build on x86_32 currently fails after commit
9bb2ec608a (objtool: Update Retpoline validation)
with:
arch/x86/kernel/../../x86/xen/xen-head.S:35: Error: no such instruction: `annotate_unret_safe'
ANNOTATE_UNRET_SAFE is defined in nospec-branch.h. And head_32.S is
missing this include. Fix this.
Fixes: 9bb2ec608a ("objtool: Update Retpoline validation")
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/63e23f80-033f-f64e-7522-2816debbc367@kernel.org
Pull lockdep fix for x86 retbleed from Borislav Petkov:
- Fix lockdep complaint for __static_call_fixup()
* tag 'x86_bugs_retbleed' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/static_call: Serialize __static_call_fixup() properly
__static_call_fixup() invokes __static_call_transform() without holding
text_mutex, which causes lockdep to complain in text_poke_bp().
Adding the proper locking cures that, but as this is either used during
early boot or during module finalizing, it's not required to use
text_poke_bp(). Add an argument to __static_call_transform() which tells
it to use text_poke_early() for it.
Fixes: ee88d363d1 ("x86,static_call: Use alternative RET encoding")
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Borislav Petkov <bp@suse.de>
Pull x86 retbleed fixes from Borislav Petkov:
"Just when you thought that all the speculation bugs were addressed and
solved and the nightmare is complete, here's the next one: speculating
after RET instructions and leaking privileged information using the
now pretty much classical covert channels.
It is called RETBleed and the mitigation effort and controlling
functionality has been modelled similar to what already existing
mitigations provide"
* tag 'x86_bugs_retbleed' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (54 commits)
x86/speculation: Disable RRSBA behavior
x86/kexec: Disable RET on kexec
x86/bugs: Do not enable IBPB-on-entry when IBPB is not supported
x86/entry: Move PUSH_AND_CLEAR_REGS() back into error_entry
x86/bugs: Add Cannon lake to RETBleed affected CPU list
x86/retbleed: Add fine grained Kconfig knobs
x86/cpu/amd: Enumerate BTC_NO
x86/common: Stamp out the stepping madness
KVM: VMX: Prevent RSB underflow before vmenter
x86/speculation: Fill RSB on vmexit for IBRS
KVM: VMX: Fix IBRS handling after vmexit
KVM: VMX: Prevent guest RSB poisoning attacks with eIBRS
KVM: VMX: Convert launched argument to flags
KVM: VMX: Flatten __vmx_vcpu_run()
objtool: Re-add UNWIND_HINT_{SAVE_RESTORE}
x86/speculation: Remove x86_spec_ctrl_mask
x86/speculation: Use cached host SPEC_CTRL value for guest entry/exit
x86/speculation: Fix SPEC_CTRL write on SMT state change
x86/speculation: Fix firmware entry SPEC_CTRL handling
x86/speculation: Fix RSB filling with CONFIG_RETPOLINE=n
...
Currently, the only way x86 can get an early boot RNG seed is via EFI,
which is generally always used now for physical machines, but is very
rarely used in VMs, especially VMs that are optimized for starting
"instantaneously", such as Firecracker's MicroVM. For tiny fast booting
VMs, EFI is not something you generally need or want.
Rather, the image loader or firmware should be able to pass a single
random seed, exactly as device tree platforms do with the "rng-seed"
property. Additionally, this is something that bootloaders can append,
with their own seed file management, which is something every other
major OS ecosystem has that Linux does not (yet).
Add SETUP_RNG_SEED, similar to the other eight setup_data entries that
are parsed at boot. It also takes care to zero out the seed immediately
after using, in order to retain forward secrecy. This all takes about 7
trivial lines of code.
Then, on kexec_file_load(), a new fresh seed is generated and passed to
the next kernel, just as is done on device tree architectures when
using kexec. And, importantly, I've tested that QEMU is able to properly
pass SETUP_RNG_SEED as well, making this work for every step of the way.
This code too is pretty straight forward.
Together these measures ensure that VMs and nested kexec()'d kernels
always receive a proper boot time RNG seed at the earliest possible
stage from their parents:
- Host [already has strongly initialized RNG]
- QEMU [passes fresh seed in SETUP_RNG_SEED field]
- Linux [uses parent's seed and gathers entropy of its own]
- kexec [passes this in SETUP_RNG_SEED field]
- Linux [uses parent's seed and gathers entropy of its own]
- kexec [passes this in SETUP_RNG_SEED field]
- Linux [uses parent's seed and gathers entropy of its own]
- kexec [passes this in SETUP_RNG_SEED field]
- ...
I've verified in several scenarios that this works quite well from a
host kernel to QEMU and down inwards, mixing and matching loaders, with
every layer providing a seed to the next.
[ bp: Massage commit message. ]
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: H. Peter Anvin (Intel) <hpa@zytor.com>
Link: https://lore.kernel.org/r/20220630113300.1892799-1-Jason@zx2c4.com
Pull x86 fixes from Borislav Petkov:
- Prepare for and clear .brk early in order to address XenPV guests
failures where the hypervisor verifies page tables and uninitialized
data in that range leads to bogus failures in those checks
- Add any potential setup_data entries supplied at boot to the identity
pagetable mappings to prevent kexec kernel boot failures. Usually,
this is not a problem for the normal kernel as those mappings are
part of the initially mapped 2M pages but if kexec gets to allocate
the second kernel somewhere else, those setup_data entries need to be
mapped there too.
- Fix objtool not to discard text references from the __tracepoints
section so that ENDBR validation still works
- Correct the setup_data types limit as it is user-visible, before 5.19
releases
* tag 'x86_urgent_for_v5.19_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/boot: Fix the setup data types max limit
x86/ibt, objtool: Don't discard text references from tracepoint section
x86/compressed/64: Add identity mappings for setup_data entries
x86: Fix .brk attribute in linker script
x86: Clear .brk area at early boot
x86/xen: Use clear_bss() for Xen PV guests
Some Intel processors may use alternate predictors for RETs on
RSB-underflow. This condition may be vulnerable to Branch History
Injection (BHI) and intramode-BTI.
Kernel earlier added spectre_v2 mitigation modes (eIBRS+Retpolines,
eIBRS+LFENCE, Retpolines) which protect indirect CALLs and JMPs against
such attacks. However, on RSB-underflow, RET target prediction may
fallback to alternate predictors. As a result, RET's predicted target
may get influenced by branch history.
A new MSR_IA32_SPEC_CTRL bit (RRSBA_DIS_S) controls this fallback
behavior when in kernel mode. When set, RETs will not take predictions
from alternate predictors, hence mitigating RETs as well. Support for
this is enumerated by CPUID.7.2.EDX[RRSBA_CTRL] (bit2).
For spectre v2 mitigation, when a user selects a mitigation that
protects indirect CALLs and JMPs against BHI and intramode-BTI, set
RRSBA_DIS_S also to protect RETs for RSB-underflow case.
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
