| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The maple_calibrate_decr() have been removed since
commit 10f7e7c15e6c ("[PATCH] ppc64: consolidate calibrate_decr
implementations"), and now it is useless, so remove it.
Signed-off-by: Gaosheng Cui <[email protected]>
Signed-off-by: Michael Ellerman <[email protected]>
Link: https://msgid.link/[email protected]
|
|
Handling FEAT_ATS1A (which provides the AT S1E{1,2}A instructions)
is pretty easy, as it is just the usual AT without the permission
check.
This basically amounts to plumbing the instructions in the various
dispatch tables, and handling FEAT_ATS1A being disabled in the
ID registers.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Hooray, we're done. Plug the AT traps into the system instruction
table, and let it rip.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
FEAT_PAN3 added a check for executable permissions to FEAT_PAN2.
Add the required SCTLR_ELx.EPAN and descriptor checks to handle
this correctly.
Reviewed-by: Alexandru Elisei <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Ensure that SCTLR_EL1.EPAN is RES0 when FEAT_PAN3 isn't supported.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
In order to plug the brokenness of our current AT implementation,
we need a SW walker that is going to... err.. walk the S1 tables
and tell us what it finds.
Of course, it builds on top of our S2 walker, and share similar
concepts. The beauty of it is that since it uses kvm_read_guest(),
it is able to bring back pages that have been otherwise evicted.
This is then plugged in the two AT S1 emulation functions as
a "slow path" fallback. I'm not sure it is that slow, but hey.
Reviewed-by: Alexandru Elisei <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Make this helper visible to at.c, we are going to need it.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
On the face of it, AT S12E{0,1}{R,W} is pretty simple. It is the
combination of AT S1E{0,1}{R,W}, followed by an extra S2 walk.
However, there is a great deal of complexity coming from combining
the S1 and S2 attributes to report something consistent in PAR_EL1.
This is an absolute mine field, and I have a splitting headache.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Similar to our AT S1E{0,1} emulation, we implement the AT S1E2
handling.
This emulation of course suffers from the same problems, but is
somehow simpler due to the lack of PAN2 and the fact that we are
guaranteed to execute it from the correct context.
Co-developed-by: Jintack Lim <[email protected]>
Signed-off-by: Jintack Lim <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Building on top of our primitive AT S1E{0,1}{R,W} emulation,
add minimal support for the FEAT_PAN2 instructions, momentary
context-switching PSTATE.PAN so that it takes effect in the
context of the guest.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Emulating AT instructions is one the tasks devolved to the host
hypervisor when NV is on.
Here, we take the basic approach of emulating AT S1E{0,1}{R,W}
using the AT instructions themselves. While this mostly work,
it doesn't *always* work:
- S1 page tables can be swapped out
- shadow S2 can be incomplete and not contain mappings for
the S1 page tables
We are not trying to handle these case here, and defer it to
a later patch. Suitable comments indicate where we are in dire
need of better handling.
Co-developed-by: Jintack Lim <[email protected]>
Signed-off-by: Jintack Lim <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
If our guest has been configured without PAN2, make sure that
AT S1E1{R,W}P will generate an UNDEF.
Reviewed-by: Anshuman Khandual <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
The upper_attr attribute has been badly named, as it most of the
time carries the full "last walked descriptor".
Rename it to "desc" and make ti contain the full 64bit descriptor.
This will be used by the S1 PTW.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Despite KVM not using the contiguous bit for anything related to
TLBs, the spec does require that the alignment defined by the
contiguous bit for the page size and the level is enforced.
Add the required checks to offset the point where PA and VA merge.
Fixes: 61e30b9eef7f ("KVM: arm64: nv: Implement nested Stage-2 page table walk logic")
Reported-by: Alexandru Elisei <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Although we have helpers that encode the level of a given fault
type, the Address Size fault type is missing it.
While we're at it, fix the bracketting for ESR_ELx_FSC_ACCESS_L()
and ESR_ELx_FSC_PERM_L().
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Although we already have the primitives to set PSTATE.PAN with an
immediate, we don't have a way to read the current state nor set
it ot an arbitrary value (i.e. we can generally save/restore it).
Thankfully, all that is missing for this is the definition for
the PAN pseudo system register, here named SYS_PSTATE_PAN.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
As KVM is about to grow a full emulation for the AT instructions,
add the layout of the PAR_EL1 register in its non-D128 configuration.
Note that the constants are a bit ugly, as the register has two
layouts, based on the state of the F bit.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
Although Linux doesn't make use of hierarchical permissions (TFFT!),
KVM needs to know where the various bits related to this feature
live in the TCR_ELx registers as well as in the page tables.
Add the missing bits.
Signed-off-by: Marc Zyngier <[email protected]>
|
|
To allow using newer instructions that current assemblers don't know about,
replace the `at` instruction with the underlying SYS instruction.
Signed-off-by: Joey Gouly <[email protected]>
Cc: Oliver Upton <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
Reviewed-by: Marc Zyngier <[email protected]>
Reviewed-by: Anshuman Khandual <[email protected]>
Acked-by: Will Deacon <[email protected]>
Signed-off-by: Marc Zyngier <[email protected]>
|
|
The implementation of _get_SP() was removed in commit f4db196717c6
("[POWERPC] Remove _get_SP"), remove the now obsolete declaration.
Signed-off-by: Gaosheng Cui <[email protected]>
Reviewed-by: Christophe Leroy <[email protected]>
[mpe: Update change log to refer to correct commit per Christophe]
Signed-off-by: Michael Ellerman <[email protected]>
Link: https://msgid.link/[email protected]
|
|
Add Amlogic C3 C308L AW419 board. The corresponding binding
has been applied, therefore, this series does not need to
add a binding corresponding to the AW419 board.
Reviewed-by: Neil Armstrong <[email protected]>
Signed-off-by: Xianwei Zhao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Neil Armstrong <[email protected]>
|
|
Add some device nodes for SoC C3, including periphs clock controller
node, PLL clock controller node, SPICC node, regulator node, NAND
controller node, sdcard node, Ethernet MAC and PHY node.
The sdacrd depends on regulator and pinctrl(select), so some
property fields are placed at the board level. The nand chip
is placed on the board, So some property fields about SPIFC
and NAND controller node are placed at the board level.
THe Ethernet MAC support outchip PHY, so place this property
field(select PHY) at the board level.
Signed-off-by: Xianwei Zhao <[email protected]>
Reviewed-by: Neil Armstrong <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Neil Armstrong <[email protected]>
|
|
Add a fastpath for HLT VM-Exits by immediately re-entering the guest if
it has a pending wake event. When virtual interrupt delivery is enabled,
i.e. when KVM doesn't need to manually inject interrupts, this allows KVM
to stay in the fastpath run loop when a vIRQ arrives between the guest
doing CLI and STI;HLT. Without AMD's Idle HLT-intercept support, the CPU
generates a HLT VM-Exit even though KVM will immediately resume the guest.
Note, on bare metal, it's relatively uncommon for a modern guest kernel to
actually trigger this scenario, as the window between the guest checking
for a wake event and committing to HLT is quite small. But in a nested
environment, the timings change significantly, e.g. rudimentary testing
showed that ~50% of HLT exits where HLT-polling was successful would be
serviced by this fastpath, i.e. ~50% of the time that a nested vCPU gets
a wake event before KVM schedules out the vCPU, the wake event was pending
even before the VM-Exit.
Link: https://lore.kernel.org/all/[email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Shuffle code around in x86.c so that the various helpers related to vCPU
blocking/running logic are (a) located near each other and (b) ordered so
that HLT emulation can use kvm_vcpu_has_events() in a future path.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Exit to userspace if a fastpath handler triggers such an exit, which can
happen when skipping the instruction, e.g. due to userspace
single-stepping the guest via KVM_GUESTDBG_SINGLESTEP or because of an
emulation failure.
Fixes: 404d5d7bff0d ("KVM: X86: Introduce more exit_fastpath_completion enum values")
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Now that the WRMSR fastpath for x2APIC_ICR and TSC_DEADLINE are identical,
ignoring the backend MSR handling, consolidate the common bits of skipping
the instruction and setting the return value.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Re-enter the guest in the fastpath if WRMSR emulation for x2APIC's ICR is
successful, as no additional work is needed, i.e. there is no code unique
for WRMSR exits between the fastpath and the "!= EXIT_FASTPATH_NONE" check
in __vmx_handle_exit().
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
The host save area is a VMCB, track it as such to help readers follow
along, but mostly to cleanup/simplify the retrieval of the SEV-ES host
save area.
Note, the compile-time assertion that
offsetof(struct vmcb, save) == EXPECTED_VMCB_CONTROL_AREA_SIZE
ensures that the SEV-ES save area is indeed at offset 0x400 (whoever added
the expected/architectural VMCB offsets apparently likes decimal).
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Incorporate the _host_ SEV-ES save area into the VMCB as a union with the
legacy save area. The SEV-ES variant used to save/load host state is
larger than the legacy save area, but resides at the same offset. Prefix
the field with "host" to make it as obvious as possible that the SEV-ES
variant in the VMCB is only ever used for host state. Guest state for
SEV-ES VMs is stored in a completely separate page (VMSA), albeit with
the same layout as the host state.
Add a compile-time assert to ensure the VMCB layout is correct, i.e. that
KVM's layout matches the architectural definitions.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Add __sme_pa_to_page() to pair with __sme_page_pa() and use it to replace
open coded equivalents, including for "iopm_base", which previously
avoided having to do __sme_clr() by storing the raw PA in the global
variable.
Opportunistically convert __sme_page_pa() to a helper to provide type
safety.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Rewrite the comment in FNAME(fetch) to explain why KVM needs to check that
the gPTE is still fresh before continuing the shadow page walk, even if
KVM already has a linked shadow page for the gPTE in question.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Drop the pointless and poorly named "out_gpte_changed" label, in
FNAME(fetch), and instead return RET_PF_RETRY directly.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Combine the back-to-back if-statements for synchronizing children when
linking a new indirect shadow page in order to decrease the indentation,
and to make it easier to "see" the logic in its entirety.
No functional change intended.
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Re-introduce the "split" x2APIC ICR storage that KVM used prior to Intel's
IPI virtualization support, but only for AMD. While not stated anywhere
in the APM, despite stating the ICR is a single 64-bit register, AMD CPUs
store the 64-bit ICR as two separate 32-bit values in ICR and ICR2. When
IPI virtualization (IPIv on Intel, all AVIC flavors on AMD) is enabled,
KVM needs to match CPU behavior as some ICR ICR writes will be handled by
the CPU, not by KVM.
Add a kvm_x86_ops knob to control the underlying format used by the CPU to
store the x2APIC ICR, and tune it to AMD vs. Intel regardless of whether
or not x2AVIC is enabled. If KVM is handling all ICR writes, the storage
format for x2APIC mode doesn't matter, and having the behavior follow AMD
versus Intel will provide better test coverage and ease debugging.
Fixes: 4d1d7942e36a ("KVM: SVM: Introduce logic to (de)activate x2AVIC mode")
Cc: [email protected]
Cc: Maxim Levitsky <[email protected]>
Cc: Suravee Suthikulpanit <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Hoist kvm_x2apic_icr_write() above kvm_apic_write_nodecode() so that a
local helper to _read_ the x2APIC ICR can be added and used in the
nodecode path without needing a forward declaration.
No functional change intended.
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
Inject a #GP on a WRMSR(ICR) that attempts to set any reserved bits that
are must-be-zero on both Intel and AMD, i.e. any reserved bits other than
the BUSY bit, which Intel ignores and basically says is undefined.
KVM's xapic_state_test selftest has been fudging the bug since commit
4b88b1a518b3 ("KVM: selftests: Enhance handling WRMSR ICR register in
x2APIC mode"), which essentially removed the testcase instead of fixing
the bug.
WARN if the nodecode path triggers a #GP, as the CPU is supposed to check
reserved bits for ICR when it's partially virtualized.
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Sean Christopherson <[email protected]>
|
|
vcc3v3-sd-s0-regulator used enable-active-low. According the binding
of the fixed regulator, that is the assumed mode of operation if
enable-active-high is not specified. So this is property is not part
of the binding, therefore remove it.
Signed-off-by: Heiko Stuebner <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Heiko Stuebner <[email protected]>
|
|
regulator-init-microvolt is used in the vendor-kernel, but not part
of the specification.
Signed-off-by: Heiko Stuebner <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Heiko Stuebner <[email protected]>
|
|
A remant from moving from the vendor kernel, the regulator is using
a fairchild fcs prefix instead of rockchip,* in the mainline kernel
according to its binding.
Signed-off-by: Heiko Stuebner <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Heiko Stuebner <[email protected]>
|
|
Add three counters to follow and understand hiperdispatch behavior;
* adjustment_count (amount of capacity adjustments triggered)
* greedy_time_ms (time spent while all cpus are on high capacity)
* conservative_time_ms (time spent while only entitled cpus are on high
capacity)
These counters can be found under /sys/kernel/debug/s390/hiperdispatch/
Time counters are in <msec> format and only cover the time spent
when hiperdispatch is active.
Acked-by: Vasily Gorbik <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
Add two attributes for debug purposes. They can be found under;
/sys/devices/system/cpu/hiperdispatch/
* hd_stime_threshold : allows user to adjust steal time threshold
* hd_delay_factor : allows user to adjust delay factor of hiperdispatch
work (after topology updates, delayed work is
always delayed extra by this factor)
hd_stime_threshold can have values between 0-100 as it represents a
percentage value.
hd_delay_factor can have values greater than 1. It is multiplied with
the default delay to achieve a longer interval, pushing back the next
hiperdispatch adjustment after a topology update.
Ex:
if delay interval is 250ms and the delay factor is 4;
delayed interval is now 1000ms(1sec). After each capacity adjustment
or topology change, work has a delayed interval of 1 sec for one
interval.
Acked-by: Vasily Gorbik <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
Expose hiperdispatch controls via sysctl. The user can now toggle
hiperdispatch via assigning 0 or 1 to s390.hiperdispatch attribute.
When hiperdipatch is toggled on, it tries to adjust CPU capacities,
while system is in vertical polarization to gain performance benefits
from different CPU polarizations. Disabling hiperdispatch reverts the
CPU capacities to their default (HIGH_CAPACITY) and stops the dynamic
adjustments.
Introduce a kconfig option HIPERDISPATCH_ON which allows users to
use hiperdispatch by default on vertical polarization. Using the
sysctl attribute s390.hiperdispatch would overwrite this behavior.
Acked-by: Vasily Gorbik <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
Add trace events to debug hiperdispatch behavior and track domain
rebuilding. Two events provide information about the decision making of
hiperdispatch and the adjustments made.
Acked-by: Vasily Gorbik <[email protected]>
Co-developed-by: Tobias Huschle <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
The measurements done by hiperdispatch can have sudden spikes and dips
during run time. To prevent these outliers effecting the decision making
process and causing adjustment overhead, use weighted average of the
steal time.
Acked-by: Vasily Gorbik <[email protected]>
Co-developed-by: Tobias Huschle <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
When LPAR is in vertical polarization, CPUs get different polarization
values, namely vertical high, vertical medium and vertical low. These
values represent the likelyhood of the CPU getting physical runtime.
Vertical high CPUs will always get runtime and others get varying
runtime depending on the load the CEC is under.
Vertical high and vertical medium CPUs are considered the CPUs which the
current LPAR has the entitlement to run on. The vertical lows are on the
other hand are borrowed CPUs which would only be given to the LPAR by
hipervisor when the other LPARs are not utilizing them.
Using the CPU capacities, hint linux scheduler when it should prioritise
vertical high and vertical medium CPUs over vertical low CPUs.
By tracking various system statistics hiperdispatch determines when to
adjust cpu capacities.
After each adjustment, rebuilding of scheduler domains is necessary to
notify the scheduler about capacity changes but since this operation is
costly it should be done as sparsely as possible.
Acked-by: Vasily Gorbik <[email protected]>
Co-developed-by: Tobias Huschle <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
Linux scheduler allows architectures to assign capacity values to
individual CPUs. This hints scheduler the performance difference between
CPUs and allows more efficient task distribution them. Implement
helper methods to set and get CPU capacities for s390. This is
particularly helpful in vertical polarization configurations of LPARs.
On vertical polarization an LPARs CPUs can get different polarization
values depending on the CEC configuration. CPUs with different
polarization values can perform different from each other, using CPU
capacities this can be reflected to linux scheduler.
Acked-by: Vasily Gorbik <[email protected]>
Signed-off-by: Mete Durlu <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
By default, all systems on s390 start in horizontal cpu polarization.
Selecting the new config option SCHED_TOPOLOGY_VERTICAL allows to build
a kernel that switches to vertical polarization during boot.
Acked-by: Heiko Carstens <[email protected]>
Tested-by: Mete Durlu <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
Provide an additional path to set the polarization of the system, such
that a user no longer relies on the sysfs interface only and is able
configure the polarization for every reboot via sysctl control files.
The new sysctl can be set as follows:
- s390.polarization=0 for horizontal polarization
- s390.polarization=1 for vertical polarization
Acked-by: Heiko Carstens <[email protected]>
Tested-by: Mete Durlu <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
Introduce a new debug file which allows to determine how many warning
track grace periods were missed on each CPU.
The new file can be found as /sys/kernel/debug/s390/wti
It is formatted as:
CPU0 CPU1 [...] CPUx
xyz xyz [...] xyz
Acked-by: Heiko Carstens <[email protected]>
Reviewed-by: Mete Durlu <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|
|
A virtual CPU that has received a warning-track interrupt may fail to
acknowledge the interrupt within the warning-track grace period.
While this is usually not a problem, it will become necessary to
investigate if there is a large number of such missed warning-track
interrupts. Therefore, it is necessary to track these events.
The information is tracked through the s390 debug facility and can be
found under /sys/kernel/debug/s390dbf/wti/.
The hex_ascii output is formatted as:
<pid> <symbol>
The values pid and current psw are collected when a warning track
interrupt is received. Symbol is either the kernel symbol matching the
collected psw or redacted to <user> when running in user space.
Each line represents the currently executing process when a warning
track interrupt was received which was then not acknowledged within its
grace period.
Acked-by: Heiko Carstens <[email protected]>
Reviewed-by: Mete Durlu <[email protected]>
Signed-off-by: Tobias Huschle <[email protected]>
Signed-off-by: Vasily Gorbik <[email protected]>
|