| Message ID | 1454397064-29268-3-git-send-email-shuai.ruan@linux.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
>>> On 02.02.16 at 08:11, <shuai.ruan@linux.intel.com> wrote: > --- a/xen/arch/x86/xstate.c > +++ b/xen/arch/x86/xstate.c > @@ -248,24 +248,26 @@ void xsave(struct vcpu *v, uint64_t mask) > uint32_t hmask = mask >> 32; > uint32_t lmask = mask; > int word_size = mask & XSTATE_FP ? (cpu_has_fpu_sel ? 8 : 0) : -1; > +#define XSAVE(pfx) \ > + alternative_io_3(".byte " pfx "0x0f,0xae,0x27\n", \ > + ".byte " pfx "0x0f,0xae,0x37\n", \ > + X86_FEATURE_XSAVEOPT, \ > + ".byte " pfx "0x0f,0xc7,0x27\n", \ > + X86_FEATURE_XSAVEC, \ > + ".byte " pfx "0x0f,0xc7,0x37\n", \ > + X86_FEATURE_XSAVES, \ > + "=m" (*ptr), \ > + "a" (lmask), "d" (hmask), "D" (ptr)) > > if ( word_size <= 0 || !is_pv_32bit_vcpu(v) ) > { > typeof(ptr->fpu_sse.fip.sel) fcs = ptr->fpu_sse.fip.sel; > typeof(ptr->fpu_sse.fdp.sel) fds = ptr->fpu_sse.fdp.sel; > > - if ( cpu_has_xsaves ) > - asm volatile ( ".byte 0x48,0x0f,0xc7,0x2f" > - : "=m" (*ptr) > - : "a" (lmask), "d" (hmask), "D" (ptr) ); > - else if ( cpu_has_xsavec ) > - asm volatile ( ".byte 0x48,0x0f,0xc7,0x27" > - : "=m" (*ptr) > - : "a" (lmask), "d" (hmask), "D" (ptr) ); > - else if ( cpu_has_xsaveopt ) > + if ( cpu_has_xsaveopt || cpu_has_xsaves ) > { > /* > - * xsaveopt may not write the FPU portion even when the respective > + * xsaveopt/xsaves may not write the FPU portion even when the respective Apart from this line now being too long and hence the entire comment needing re-formatting Reviewed-by: Jan Beulich <jbeulich@suse.com> Jan
>>> On 03.02.16 at 12:40, <JBeulich@suse.com> wrote: >>>> On 02.02.16 at 08:11, <shuai.ruan@linux.intel.com> wrote: >> --- a/xen/arch/x86/xstate.c >> +++ b/xen/arch/x86/xstate.c >> @@ -248,24 +248,26 @@ void xsave(struct vcpu *v, uint64_t mask) >> uint32_t hmask = mask >> 32; >> uint32_t lmask = mask; >> int word_size = mask & XSTATE_FP ? (cpu_has_fpu_sel ? 8 : 0) : -1; >> +#define XSAVE(pfx) \ >> + alternative_io_3(".byte " pfx "0x0f,0xae,0x27\n", \ >> + ".byte " pfx "0x0f,0xae,0x37\n", \ >> + X86_FEATURE_XSAVEOPT, \ >> + ".byte " pfx "0x0f,0xc7,0x27\n", \ >> + X86_FEATURE_XSAVEC, \ >> + ".byte " pfx "0x0f,0xc7,0x37\n", \ >> + X86_FEATURE_XSAVES, \ >> + "=m" (*ptr), \ >> + "a" (lmask), "d" (hmask), "D" (ptr)) >> >> if ( word_size <= 0 || !is_pv_32bit_vcpu(v) ) >> { >> typeof(ptr->fpu_sse.fip.sel) fcs = ptr->fpu_sse.fip.sel; >> typeof(ptr->fpu_sse.fdp.sel) fds = ptr->fpu_sse.fdp.sel; >> >> - if ( cpu_has_xsaves ) >> - asm volatile ( ".byte 0x48,0x0f,0xc7,0x2f" >> - : "=m" (*ptr) >> - : "a" (lmask), "d" (hmask), "D" (ptr) ); >> - else if ( cpu_has_xsavec ) >> - asm volatile ( ".byte 0x48,0x0f,0xc7,0x27" >> - : "=m" (*ptr) >> - : "a" (lmask), "d" (hmask), "D" (ptr) ); >> - else if ( cpu_has_xsaveopt ) >> + if ( cpu_has_xsaveopt || cpu_has_xsaves ) >> { >> /* >> - * xsaveopt may not write the FPU portion even when the > respective >> + * xsaveopt/xsaves may not write the FPU portion even when the > respective > > Apart from this line now being too long and hence the entire > comment needing re-formatting > Reviewed-by: Jan Beulich <jbeulich@suse.com> Withdrawn. There's a bug here, and since I'm re-doing patch 1 from scratch I'll send out the result later. Jan
diff --git a/xen/arch/x86/xstate.c b/xen/arch/x86/xstate.c index 4e87ab3..832f4ad 100644 --- a/xen/arch/x86/xstate.c +++ b/xen/arch/x86/xstate.c @@ -248,24 +248,26 @@ void xsave(struct vcpu *v, uint64_t mask) uint32_t hmask = mask >> 32; uint32_t lmask = mask; int word_size = mask & XSTATE_FP ? (cpu_has_fpu_sel ? 8 : 0) : -1; +#define XSAVE(pfx) \ + alternative_io_3(".byte " pfx "0x0f,0xae,0x27\n", \ + ".byte " pfx "0x0f,0xae,0x37\n", \ + X86_FEATURE_XSAVEOPT, \ + ".byte " pfx "0x0f,0xc7,0x27\n", \ + X86_FEATURE_XSAVEC, \ + ".byte " pfx "0x0f,0xc7,0x37\n", \ + X86_FEATURE_XSAVES, \ + "=m" (*ptr), \ + "a" (lmask), "d" (hmask), "D" (ptr)) if ( word_size <= 0 || !is_pv_32bit_vcpu(v) ) { typeof(ptr->fpu_sse.fip.sel) fcs = ptr->fpu_sse.fip.sel; typeof(ptr->fpu_sse.fdp.sel) fds = ptr->fpu_sse.fdp.sel; - if ( cpu_has_xsaves ) - asm volatile ( ".byte 0x48,0x0f,0xc7,0x2f" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); - else if ( cpu_has_xsavec ) - asm volatile ( ".byte 0x48,0x0f,0xc7,0x27" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); - else if ( cpu_has_xsaveopt ) + if ( cpu_has_xsaveopt || cpu_has_xsaves ) { /* - * xsaveopt may not write the FPU portion even when the respective + * xsaveopt/xsaves may not write the FPU portion even when the respective * mask bit is set. For the check further down to work we hence * need to put the save image back into the state that it was in * right after the previous xsaveopt. @@ -277,14 +279,9 @@ void xsave(struct vcpu *v, uint64_t mask) ptr->fpu_sse.fip.sel = 0; ptr->fpu_sse.fdp.sel = 0; } - asm volatile ( ".byte 0x48,0x0f,0xae,0x37" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); } - else - asm volatile ( ".byte 0x48,0x0f,0xae,0x27" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); + + XSAVE("0x48,"); if ( !(mask & ptr->xsave_hdr.xstate_bv & XSTATE_FP) || /* @@ -315,24 +312,10 @@ void xsave(struct vcpu *v, uint64_t mask) } else { - if ( cpu_has_xsaves ) - asm volatile ( ".byte 0x0f,0xc7,0x2f" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); - else if ( cpu_has_xsavec ) - asm volatile ( ".byte 0x0f,0xc7,0x27" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); - else if ( cpu_has_xsaveopt ) - asm volatile ( ".byte 0x0f,0xae,0x37" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); - else - asm volatile ( ".byte 0x0f,0xae,0x27" - : "=m" (*ptr) - : "a" (lmask), "d" (hmask), "D" (ptr) ); + XSAVE(""); word_size = 4; } +#undef XSAVE if ( word_size >= 0 ) ptr->fpu_sse.x[FPU_WORD_SIZE_OFFSET] = word_size; }
This patch use alternavtive asm on the xsave side. As xsaves use modified optimization like xsaveopt, xsaves may not writing the FPU portion of the save image too. So xsaves also need some extra tweaks. Signed-off-by: Shuai Ruan <shuai.ruan@linux.intel.com> --- xen/arch/x86/xstate.c | 49 ++++++++++++++++--------------------------------- 1 file changed, 16 insertions(+), 33 deletions(-)