Message ID | 1552659640-24201-1-git-send-email-andrew.cooper3@citrix.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | xen/x86: Fix cflush()'s parameter tracking | expand |
>>> On 15.03.19 at 15:20, <andrew.cooper3@citrix.com> wrote: > Foring a register operand hides (from the compiler) the fact that clflush > behaves as a read from the memory operand (wrt memory order, faults, etc.). > It also reduces the compilers flexibility with register schedulling. > > Reimplement clfush() (and wbinvd() for consistency) as a static inline rather > than a macro, and have it take a const void pointer. > > In practice, the only generated code which gets modified by this is in > mwait_idle_with_hints(), where a memory operand can now be used. > > While here, I noticed that &mwait_wakeup(cpu) was being calculated twice. > This is caused by the memory clobber in mb(), so take the opportunity to help > the optimiser by calculating it once, ahead of time. bloat-o-meter reports a > delta of -26 as a result of this change. > > Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> Reviewed-by: Jan Beulich <jbeulich@suse.com>
diff --git a/xen/arch/x86/acpi/cpu_idle.c b/xen/arch/x86/acpi/cpu_idle.c index 14b0278..654de24 100644 --- a/xen/arch/x86/acpi/cpu_idle.c +++ b/xen/arch/x86/acpi/cpu_idle.c @@ -401,15 +401,16 @@ void mwait_idle_with_hints(unsigned int eax, unsigned int ecx) { unsigned int cpu = smp_processor_id(); s_time_t expires = per_cpu(timer_deadline, cpu); + const void *monitor_addr = &mwait_wakeup(cpu); if ( boot_cpu_has(X86_FEATURE_CLFLUSH_MONITOR) ) { mb(); - clflush((void *)&mwait_wakeup(cpu)); + clflush(monitor_addr); mb(); } - __monitor((void *)&mwait_wakeup(cpu), 0, 0); + __monitor(monitor_addr, 0, 0); smp_mb(); /* diff --git a/xen/include/asm-x86/system.h b/xen/include/asm-x86/system.h index c665499..3246797 100644 --- a/xen/include/asm-x86/system.h +++ b/xen/include/asm-x86/system.h @@ -11,11 +11,15 @@ __sel; \ }) -#define wbinvd() \ - asm volatile ( "wbinvd" : : : "memory" ) +static inline void wbinvd(void) +{ + asm volatile ( "wbinvd" ::: "memory" ); +} -#define clflush(a) \ - asm volatile ( "clflush (%0)" : : "r"(a) ) +static inline void clflush(const void *p) +{ + asm volatile ( "clflush %0" :: "m" (*(const char *)p) ); +} #define xchg(ptr,v) \ ((__typeof__(*(ptr)))__xchg((unsigned long)(v),(ptr),sizeof(*(ptr))))
Foring a register operand hides (from the compiler) the fact that clflush behaves as a read from the memory operand (wrt memory order, faults, etc.). It also reduces the compilers flexibility with register schedulling. Reimplement clfush() (and wbinvd() for consistency) as a static inline rather than a macro, and have it take a const void pointer. In practice, the only generated code which gets modified by this is in mwait_idle_with_hints(), where a memory operand can now be used. While here, I noticed that &mwait_wakeup(cpu) was being calculated twice. This is caused by the memory clobber in mb(), so take the opportunity to help the optimiser by calculating it once, ahead of time. bloat-o-meter reports a delta of -26 as a result of this change. Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> --- CC: Jan Beulich <JBeulich@suse.com> CC: Wei Liu <wei.liu2@citrix.com> CC: Roger Pau Monné <roger.pau@citrix.com> --- xen/arch/x86/acpi/cpu_idle.c | 5 +++-- xen/include/asm-x86/system.h | 12 ++++++++---- 2 files changed, 11 insertions(+), 6 deletions(-)