With the removal of lazy FPU, the full state is always restored on context
switch so stop tracking whether nonlazy xstate is used.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
---
 xen/arch/x86/domctl.c             |  1 -
 xen/arch/x86/hvm/hvm.c            |  1 -
 xen/arch/x86/i387.c               | 12 +-----------
 xen/arch/x86/include/asm/domain.h |  3 ---
 xen/arch/x86/include/asm/xstate.h |  6 ++----
 xen/arch/x86/xstate.c             |  3 ---
 6 files changed, 3 insertions(+), 23 deletions(-)

diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index d9b08182ac1d..a9fbb2d405b7 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -1013,7 +1013,6 @@ long arch_do_domctl(
 
                 v->arch.xcr0 = _xcr0;
                 v->arch.xcr0_accum = _xcr0_accum;
-                v->arch.nonlazy_xstate_used = _xcr0_accum & XSTATE_NONLAZY;
                 compress_xsave_states(v, _xsave_area,
                                       evc->size - PV_XSAVE_HDR_SIZE);
 
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index d4ba82845146..0e7b66d726ae 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -1383,7 +1383,6 @@ static int cf_check hvm_load_cpu_xsave_states(
 
     v->arch.xcr0 = ctxt->xcr0;
     v->arch.xcr0_accum = ctxt->xcr0_accum;
-    v->arch.nonlazy_xstate_used = ctxt->xcr0_accum & XSTATE_NONLAZY;
     compress_xsave_states(v, &ctxt->save_area,
                           size - offsetof(struct hvm_hw_cpu_xsave, save_area));
 
diff --git a/xen/arch/x86/i387.c b/xen/arch/x86/i387.c
index 5e893a2aab94..9acaaf4673df 100644
--- a/xen/arch/x86/i387.c
+++ b/xen/arch/x86/i387.c
@@ -108,25 +108,18 @@ static inline void fpu_fxrstor(struct vcpu *v)
 /*      FPU Save Functions     */
 /*******************************/
 
-static inline uint64_t vcpu_xsave_mask(const struct vcpu *v)
-{
-    return v->arch.nonlazy_xstate_used ? XSTATE_ALL : XSTATE_LAZY;
-}
-
 /* Save x87 extended state */
 static inline void fpu_xsave(struct vcpu *v)
 {
     bool ok;
-    uint64_t mask = vcpu_xsave_mask(v);
 
-    ASSERT(mask);
     /*
      * XCR0 normally represents what guest OS set. In case of Xen itself,
      * we set the accumulated feature mask before doing save/restore.
      */
     ok = set_xcr0(v->arch.xcr0_accum | XSTATE_FP_SSE);
     ASSERT(ok);
-    xsave(v, mask);
+    xsave(v, XSTATE_ALL);
     ok = set_xcr0(v->arch.xcr0 ?: XSTATE_FP_SSE);
     ASSERT(ok);
 }
@@ -202,9 +195,6 @@ void vcpu_restore_fpu(struct vcpu *v)
  */
 static bool _vcpu_save_fpu(struct vcpu *v)
 {
-    if ( !v->arch.nonlazy_xstate_used )
-        return false;
-
     ASSERT(!is_idle_vcpu(v));
 
     /* This can happen, if a paravirtualised guest OS has set its CR0.TS. */
diff --git a/xen/arch/x86/include/asm/domain.h b/xen/arch/x86/include/asm/domain.h
index bd7b02085ef8..385a6666dafa 100644
--- a/xen/arch/x86/include/asm/domain.h
+++ b/xen/arch/x86/include/asm/domain.h
@@ -667,9 +667,6 @@ struct arch_vcpu
      * it explicitly enables it via xcr0.
      */
     uint64_t xcr0_accum;
-    /* This variable determines whether nonlazy extended state has been used,
-     * and thus should be saved/restored. */
-    bool nonlazy_xstate_used;
 
     struct vmce vmce;
 
diff --git a/xen/arch/x86/include/asm/xstate.h b/xen/arch/x86/include/asm/xstate.h
index ca38c43ec1c3..c96d75e38b25 100644
--- a/xen/arch/x86/include/asm/xstate.h
+++ b/xen/arch/x86/include/asm/xstate.h
@@ -33,12 +33,10 @@ extern uint32_t mxcsr_mask;
 #define XSTATE_FP_SSE  (X86_XCR0_X87 | X86_XCR0_SSE)
 #define XCNTXT_MASK    (X86_XCR0_X87 | X86_XCR0_SSE | X86_XCR0_YMM | \
                         X86_XCR0_OPMASK | X86_XCR0_ZMM | X86_XCR0_HI_ZMM | \
-                        XSTATE_NONLAZY)
+                        X86_XCR0_BNDREGS | X86_XCR0_BNDCSR | X86_XCR0_PKRU | \
+                        X86_XCR0_TILE_CFG | X86_XCR0_TILE_DATA)
 
 #define XSTATE_ALL     (~(1ULL << 63))
-#define XSTATE_NONLAZY (X86_XCR0_BNDREGS | X86_XCR0_BNDCSR | X86_XCR0_PKRU | \
-                        X86_XCR0_TILE_CFG | X86_XCR0_TILE_DATA)
-#define XSTATE_LAZY    (XSTATE_ALL & ~XSTATE_NONLAZY)
 #define XSTATE_XSAVES_ONLY         0
 #define XSTATE_COMPACTION_ENABLED  (1ULL << 63)
 
diff --git a/xen/arch/x86/xstate.c b/xen/arch/x86/xstate.c
index 11d390cac985..658f372b8c51 100644
--- a/xen/arch/x86/xstate.c
+++ b/xen/arch/x86/xstate.c
@@ -991,9 +991,6 @@ int handle_xsetbv(u32 index, u64 new_bv)
     curr->arch.xcr0 = new_bv;
     curr->arch.xcr0_accum |= new_bv;
 
-    if ( new_bv & XSTATE_NONLAZY )
-        curr->arch.nonlazy_xstate_used = 1;
-
     mask &= ~XSTATE_FP_SSE;
     if ( mask )
     {
-- 
2.53.0

On 19/03/2026 1:29 pm, Ross Lagerwall wrote:
> With the removal of lazy FPU, the full state is always restored on context
> switch so stop tracking whether nonlazy xstate is used.
>
> Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>

Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>