Patch "arm64/fpsimd: Have KVM explicitly say which FP registers to save" has been added to the 6.1-stable tree
gregkh at linuxfoundation.org
gregkh at linuxfoundation.org
Mon Apr 21 23:43:53 PDT 2025
This is a note to let you know that I've just added the patch titled
arm64/fpsimd: Have KVM explicitly say which FP registers to save
to the 6.1-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
arm64-fpsimd-have-kvm-explicitly-say-which-fp-registers-to-save.patch
and it can be found in the queue-6.1 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable at vger.kernel.org> know about it.
>From stable+bounces-128302-greg=kroah.com at vger.kernel.org Fri Apr 4 15:28:08 2025
From: Mark Brown <broonie at kernel.org>
Date: Fri, 04 Apr 2025 14:23:36 +0100
Subject: arm64/fpsimd: Have KVM explicitly say which FP registers to save
To: Catalin Marinas <catalin.marinas at arm.com>, Will Deacon <will at kernel.org>, Marc Zyngier <maz at kernel.org>, James Morse <james.morse at arm.com>, Suzuki K Poulose <suzuki.poulose at arm.com>, Oliver Upton <oliver.upton at linux.dev>, Oleg Nesterov <oleg at redhat.com>, Greg Kroah-Hartman <gregkh at linuxfoundation.org>
Cc: linux-arm-kernel at lists.infradead.org, linux-kernel at vger.kernel.org, kvmarm at lists.linux.dev, kvmarm at lists.cs.columbia.edu, Mark Brown <broonie at kernel.org>, stable at vger.kernel.org, Mark Rutland <mark.rutland at arm.com>
Message-ID: <20250404-stable-sve-6-1-v1-3-cd5c9eb52d49 at kernel.org>
From: Mark Brown <broonie at kernel.org>
[ Upstream commit deeb8f9a80fdae5a62525656d65c7070c28bd3a4 ]
In order to avoid needlessly saving and restoring the guest registers KVM
relies on the host FPSMID code to save the guest registers when we context
switch away from the guest. This is done by binding the KVM guest state to
the CPU on top of the task state that was originally there, then carefully
managing the TIF_SVE flag for the task to cause the host to save the full
SVE state when needed regardless of the needs of the host task. This works
well enough but isn't terribly direct about what is going on and makes it
much more complicated to try to optimise what we're doing with the SVE
register state.
Let's instead have KVM pass in the register state it wants saving when it
binds to the CPU. We introduce a new FP_STATE_CURRENT for use
during normal task binding to indicate that we should base our
decisions on the current task. This should not be used when
actually saving. Ideally we might want to use a separate enum for
the type to save but this enum and the enum values would then
need to be named which has problems with clarity and ambiguity.
In order to ease any future debugging that might be required this patch
does not actually update any of the decision making about what to save,
it merely starts tracking the new information and warns if the requested
state is not what we would otherwise have decided to save.
Signed-off-by: Mark Brown <broonie at kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas at arm.com>
Reviewed-by: Marc Zyngier <maz at kernel.org>
Link: https://lore.kernel.org/r/20221115094640.112848-4-broonie@kernel.org
Signed-off-by: Will Deacon <will at kernel.org>
[ Mark: trivial backport ]
Signed-off-by: Mark Rutland <mark.rutland at arm.com>
Signed-off-by: Mark Brown <broonie at kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh at linuxfoundation.org>
---
arch/arm64/include/asm/fpsimd.h | 3 ++-
arch/arm64/include/asm/processor.h | 1 +
arch/arm64/kernel/fpsimd.c | 27 ++++++++++++++++++++++++---
arch/arm64/kvm/fpsimd.c | 9 ++++++++-
4 files changed, 35 insertions(+), 5 deletions(-)
--- a/arch/arm64/include/asm/fpsimd.h
+++ b/arch/arm64/include/asm/fpsimd.h
@@ -61,7 +61,8 @@ extern void fpsimd_kvm_prepare(void);
extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state,
void *sve_state, unsigned int sve_vl,
void *za_state, unsigned int sme_vl,
- u64 *svcr, enum fp_type *type);
+ u64 *svcr, enum fp_type *type,
+ enum fp_type to_save);
extern void fpsimd_flush_task_state(struct task_struct *target);
extern void fpsimd_save_and_flush_cpu_state(void);
--- a/arch/arm64/include/asm/processor.h
+++ b/arch/arm64/include/asm/processor.h
@@ -123,6 +123,7 @@ enum vec_type {
};
enum fp_type {
+ FP_STATE_CURRENT, /* Save based on current task state. */
FP_STATE_FPSIMD,
FP_STATE_SVE,
};
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -126,6 +126,7 @@ struct fpsimd_last_state_struct {
unsigned int sve_vl;
unsigned int sme_vl;
enum fp_type *fp_type;
+ enum fp_type to_save;
};
static DEFINE_PER_CPU(struct fpsimd_last_state_struct, fpsimd_last_state);
@@ -356,7 +357,8 @@ void task_set_vl_onexec(struct task_stru
* but userspace is discouraged from relying on this.
*
* task->thread.sve_state does not need to be non-NULL, valid or any
- * particular size: it must not be dereferenced.
+ * particular size: it must not be dereferenced and any data stored
+ * there should be considered stale and not referenced.
*
* * SVE state - FP_STATE_SVE:
*
@@ -369,7 +371,9 @@ void task_set_vl_onexec(struct task_stru
* task->thread.uw.fpsimd_state should be ignored.
*
* task->thread.sve_state must point to a valid buffer at least
- * sve_state_size(task) bytes in size.
+ * sve_state_size(task) bytes in size. The data stored in
+ * task->thread.uw.fpsimd_state.vregs should be considered stale
+ * and not referenced.
*
* * FPSR and FPCR are always stored in task->thread.uw.fpsimd_state
* irrespective of whether TIF_SVE is clear or set, since these are
@@ -459,6 +463,21 @@ static void fpsimd_save(void)
vl = last->sve_vl;
}
+ /*
+ * Validate that an explicitly specified state to save is
+ * consistent with the task state.
+ */
+ switch (last->to_save) {
+ case FP_STATE_CURRENT:
+ break;
+ case FP_STATE_FPSIMD:
+ WARN_ON_ONCE(save_sve_regs);
+ break;
+ case FP_STATE_SVE:
+ WARN_ON_ONCE(!save_sve_regs);
+ break;
+ }
+
if (system_supports_sme()) {
u64 *svcr = last->svcr;
@@ -1709,6 +1728,7 @@ static void fpsimd_bind_task_to_cpu(void
last->sme_vl = task_get_sme_vl(current);
last->svcr = ¤t->thread.svcr;
last->fp_type = ¤t->thread.fp_type;
+ last->to_save = FP_STATE_CURRENT;
current->thread.fpsimd_cpu = smp_processor_id();
/*
@@ -1733,7 +1753,7 @@ static void fpsimd_bind_task_to_cpu(void
void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state,
unsigned int sve_vl, void *za_state,
unsigned int sme_vl, u64 *svcr,
- enum fp_type *type)
+ enum fp_type *type, enum fp_type to_save)
{
struct fpsimd_last_state_struct *last =
this_cpu_ptr(&fpsimd_last_state);
@@ -1748,6 +1768,7 @@ void fpsimd_bind_state_to_cpu(struct use
last->sve_vl = sve_vl;
last->sme_vl = sme_vl;
last->fp_type = type;
+ last->to_save = to_save;
}
/*
--- a/arch/arm64/kvm/fpsimd.c
+++ b/arch/arm64/kvm/fpsimd.c
@@ -130,9 +130,16 @@ void kvm_arch_vcpu_ctxflush_fp(struct kv
*/
void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu)
{
+ enum fp_type fp_type;
+
WARN_ON_ONCE(!irqs_disabled());
if (vcpu->arch.fp_state == FP_STATE_GUEST_OWNED) {
+ if (vcpu_has_sve(vcpu))
+ fp_type = FP_STATE_SVE;
+ else
+ fp_type = FP_STATE_FPSIMD;
+
/*
* Currently we do not support SME guests so SVCR is
* always 0 and we just need a variable to point to.
@@ -141,7 +148,7 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm
vcpu->arch.sve_state,
vcpu->arch.sve_max_vl,
NULL, 0, &vcpu->arch.svcr,
- &vcpu->arch.fp_type);
+ &vcpu->arch.fp_type, fp_type);
clear_thread_flag(TIF_FOREIGN_FPSTATE);
update_thread_flag(TIF_SVE, vcpu_has_sve(vcpu));
Patches currently in stable-queue which might be from broonie at kernel.org are
queue-6.1/kvm-arm64-remove-host-fpsimd-saving-for-non-protected-kvm.patch
queue-6.1/spi-cadence-qspi-fix-probe-on-am62a-lp-sk.patch
queue-6.1/asoc-qdsp6-q6asm-dai-fix-q6asm_dai_compr_set_params-error-path.patch
queue-6.1/asoc-qdsp6-q6apm-dai-fix-capture-pipeline-overruns.patch
queue-6.1/kvm-arm64-mark-some-header-functions-as-inline.patch
queue-6.1/kvm-arm64-eagerly-switch-zcr_el-1-2.patch
queue-6.1/kvm-arm64-unconditionally-save-flush-host-fpsimd-sve-sme-state.patch
queue-6.1/asoc-amd-add-dmi-quirk-for-acp6x-mic-support.patch
queue-6.1/kvm-arm64-refactor-exit-handlers.patch
queue-6.1/asoc-qdsp6-q6apm-dai-set-10-ms-period-and-buffer-alignment.patch
queue-6.1/asoc-codecs-lpass-wsa-macro-fix-vi-feedback-rate.patch
queue-6.1/arm64-fpsimd-track-the-saved-fpsimd-state-type-separately-to-tif_sve.patch
queue-6.1/kvm-arm64-remove-vhe-host-restore-of-cpacr_el1.zen.patch
queue-6.1/kvm-arm64-remove-vhe-host-restore-of-cpacr_el1.smen.patch
queue-6.1/asoc-fsl_audmix-register-card-device-depends-on-dais.patch
queue-6.1/arm64-fpsimd-have-kvm-explicitly-say-which-fp-registers-to-save.patch
queue-6.1/kvm-arm64-discard-any-sve-state-when-entering-kvm-guests.patch
queue-6.1/arm64-fpsimd-stop-using-tif_sve-to-manage-register-saving-in-kvm.patch
queue-6.1/asoc-codecs-lpass-wsa-macro-fix-logic-of-enabling-vi-channels.patch
queue-6.1/kvm-arm64-calculate-cptr_el2-traps-on-activating-traps.patch
More information about the linux-arm-kernel
mailing list