[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180622170625.30688-3-vkuznets@redhat.com>
Date: Fri, 22 Jun 2018 19:06:23 +0200
From: Vitaly Kuznetsov <vkuznets@...hat.com>
To: x86@...nel.org
Cc: devel@...uxdriverproject.org, linux-kernel@...r.kernel.org,
"K. Y. Srinivasan" <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>,
Stephen Hemminger <sthemmin@...rosoft.com>,
Thomas Gleixner <tglx@...utronix.de>,
Ingo Molnar <mingo@...hat.com>,
"H. Peter Anvin" <hpa@...or.com>,
Tianyu Lan <Tianyu.Lan@...rosoft.com>,
"Michael Kelley (EOSG)" <Michael.H.Kelley@...rosoft.com>
Subject: [PATCH 2/4] x86/hyper-v: use 'fast' hypercall for HVCALL_SEND_IPI
Current Hyper-V TLFS (v5.0b) claims that HvCallSendSyntheticClusterIpi
hypercall can't be 'fast' (passing parameters through registers) but
apparently this is not true, Windows always uses 'fast' version. We can
do the same in Linux too.
Signed-off-by: Vitaly Kuznetsov <vkuznets@...hat.com>
---
arch/x86/hyperv/hv_apic.c | 22 ++++++----------------
1 file changed, 6 insertions(+), 16 deletions(-)
diff --git a/arch/x86/hyperv/hv_apic.c b/arch/x86/hyperv/hv_apic.c
index f68855499391..90055f89223b 100644
--- a/arch/x86/hyperv/hv_apic.c
+++ b/arch/x86/hyperv/hv_apic.c
@@ -128,10 +128,8 @@ static bool __send_ipi_mask_ex(const struct cpumask *mask, int vector)
static bool __send_ipi_mask(const struct cpumask *mask, int vector)
{
int cur_cpu, vcpu;
- struct ipi_arg_non_ex **arg;
- struct ipi_arg_non_ex *ipi_arg;
+ struct ipi_arg_non_ex ipi_arg;
int ret = 1;
- unsigned long flags;
if (cpumask_empty(mask))
return true;
@@ -145,16 +143,8 @@ static bool __send_ipi_mask(const struct cpumask *mask, int vector)
if ((ms_hyperv.hints & HV_X64_EX_PROCESSOR_MASKS_RECOMMENDED))
return __send_ipi_mask_ex(mask, vector);
- local_irq_save(flags);
- arg = (struct ipi_arg_non_ex **)this_cpu_ptr(hyperv_pcpu_input_arg);
-
- ipi_arg = *arg;
- if (unlikely(!ipi_arg))
- goto ipi_mask_done;
-
- ipi_arg->vector = vector;
- ipi_arg->reserved = 0;
- ipi_arg->cpu_mask = 0;
+ ipi_arg.vector = vector;
+ ipi_arg.cpu_mask = 0;
for_each_cpu(cur_cpu, mask) {
vcpu = hv_cpu_number_to_vp_number(cur_cpu);
@@ -165,13 +155,13 @@ static bool __send_ipi_mask(const struct cpumask *mask, int vector)
if (vcpu >= 64)
goto ipi_mask_done;
- __set_bit(vcpu, (unsigned long *)&ipi_arg->cpu_mask);
+ __set_bit(vcpu, (unsigned long *)&ipi_arg.cpu_mask);
}
- ret = hv_do_hypercall(HVCALL_SEND_IPI, ipi_arg, NULL);
+ ret = hv_do_fast_hypercall16(HVCALL_SEND_IPI, ipi_arg.vector,
+ ipi_arg.cpu_mask);
ipi_mask_done:
- local_irq_restore(flags);
return ((ret == 0) ? true : false);
}
--
2.14.4
Powered by blists - more mailing lists