From: Vitaly Kuznetsov <vkuznets@redhat.com>
To: x86@kernel.org
Cc: devel@linuxdriverproject.org, linux-kernel@vger.kernel.org,
"K. Y. Srinivasan" <kys@microsoft.com>,
Haiyang Zhang <haiyangz@microsoft.com>,
Stephen Hemminger <sthemmin@microsoft.com>,
Thomas Gleixner <tglx@linutronix.de>,
Ingo Molnar <mingo@redhat.com>, "H. Peter Anvin" <hpa@zytor.com>,
Tianyu Lan <Tianyu.Lan@microsoft.com>,
"Michael Kelley (EOSG)" <Michael.H.Kelley@microsoft.com>
Subject: [PATCH 3/4] x86/hyper-v: use cheaper HVCALL_SEND_IPI hypercall when possible
Date: Fri, 22 Jun 2018 19:06:24 +0200 [thread overview]
Message-ID: <20180622170625.30688-4-vkuznets@redhat.com> (raw)
In-Reply-To: <20180622170625.30688-1-vkuznets@redhat.com>
When there is no need to send an IPI to a CPU with VP number > 64
we can do the job with fast HVCALL_SEND_IPI hypercall.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
arch/x86/hyperv/hv_apic.c | 29 ++++++++++++++++++++---------
1 file changed, 20 insertions(+), 9 deletions(-)
diff --git a/arch/x86/hyperv/hv_apic.c b/arch/x86/hyperv/hv_apic.c
index 90055f89223b..ee962784d25b 100644
--- a/arch/x86/hyperv/hv_apic.c
+++ b/arch/x86/hyperv/hv_apic.c
@@ -99,6 +99,9 @@ static bool __send_ipi_mask_ex(const struct cpumask *mask, int vector)
int nr_bank = 0;
int ret = 1;
+ if (!(ms_hyperv.hints & HV_X64_EX_PROCESSOR_MASKS_RECOMMENDED))
+ return false;
+
local_irq_save(flags);
arg = (struct ipi_arg_ex **)this_cpu_ptr(hyperv_pcpu_input_arg);
@@ -140,8 +143,18 @@ static bool __send_ipi_mask(const struct cpumask *mask, int vector)
if ((vector < HV_IPI_LOW_VECTOR) || (vector > HV_IPI_HIGH_VECTOR))
return false;
- if ((ms_hyperv.hints & HV_X64_EX_PROCESSOR_MASKS_RECOMMENDED))
- return __send_ipi_mask_ex(mask, vector);
+ /*
+ * From the supplied CPU set we need to figure out if we can get away
+ * with cheaper HVCALL_SEND_IPI hypercall. This is possible when the
+ * highest VP number in the set is < 64. As VP numbers are usually in
+ * ascending order and match Linux CPU ids, here is an optimization:
+ * we check the VP number for the highest bit in the supplied set first
+ * so we can quickly find out if using HVCALL_SEND_IPI_EX hypercall is
+ * a must. We will also check all VP numbers when walking the supplied
+ * CPU set to remain correct in all cases.
+ */
+ if (hv_cpu_number_to_vp_number(cpumask_last(mask)) >= 64)
+ goto do_ex_hypercall;
ipi_arg.vector = vector;
ipi_arg.cpu_mask = 0;
@@ -153,16 +166,17 @@ static bool __send_ipi_mask(const struct cpumask *mask, int vector)
* only target upto 64 CPUs.
*/
if (vcpu >= 64)
- goto ipi_mask_done;
+ goto do_ex_hypercall;
__set_bit(vcpu, (unsigned long *)&ipi_arg.cpu_mask);
}
ret = hv_do_fast_hypercall16(HVCALL_SEND_IPI, ipi_arg.vector,
ipi_arg.cpu_mask);
-
-ipi_mask_done:
return ((ret == 0) ? true : false);
+
+do_ex_hypercall:
+ return __send_ipi_mask_ex(mask, vector);
}
static bool __send_ipi_one(int cpu, int vector)
@@ -218,10 +232,7 @@ static void hv_send_ipi_self(int vector)
void __init hv_apic_init(void)
{
if (ms_hyperv.hints & HV_X64_CLUSTER_IPI_RECOMMENDED) {
- if ((ms_hyperv.hints & HV_X64_EX_PROCESSOR_MASKS_RECOMMENDED))
- pr_info("Hyper-V: Using ext hypercalls for IPI\n");
- else
- pr_info("Hyper-V: Using IPI hypercalls\n");
+ pr_info("Hyper-V: Using IPI hypercalls\n");
/*
* Set the IPI entry points.
*/
--
2.14.4
next prev parent reply other threads:[~2018-06-22 17:06 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-22 17:06 [PATCH 0/4] x86/hyper-v: optimize PV IPIs Vitaly Kuznetsov
2018-06-22 17:06 ` [PATCH 1/4] x86/hyper-v: implement hv_do_fast_hypercall16 Vitaly Kuznetsov
2018-07-03 7:03 ` [tip:x86/hyperv] x86/hyper-v: Implement hv_do_fast_hypercall16 tip-bot for Vitaly Kuznetsov
2018-06-22 17:06 ` [PATCH 2/4] x86/hyper-v: use 'fast' hypercall for HVCALL_SEND_IPI Vitaly Kuznetsov
2018-06-25 23:54 ` Michael Kelley (EOSG)
2018-07-03 7:04 ` [tip:x86/hyperv] x86/hyper-v: Use " tip-bot for Vitaly Kuznetsov
2018-06-22 17:06 ` Vitaly Kuznetsov [this message]
2018-06-26 0:03 ` [PATCH 3/4] x86/hyper-v: use cheaper HVCALL_SEND_IPI hypercall when possible Michael Kelley (EOSG)
2018-07-03 7:04 ` [tip:x86/hyperv] x86/hyper-v: Use " tip-bot for Vitaly Kuznetsov
2018-06-22 17:06 ` [PATCH 4/4] x86/hyper-v: trace PV IPI send Vitaly Kuznetsov
2018-07-03 7:05 ` [tip:x86/hyperv] x86/hyper-v: Trace " tip-bot for Vitaly Kuznetsov
2018-06-27 0:49 ` [PATCH 0/4] x86/hyper-v: optimize PV IPIs Wanpeng Li
2018-06-27 9:25 ` Vitaly Kuznetsov
2018-06-27 10:27 ` Wanpeng Li
2018-06-27 9:32 ` Vitaly Kuznetsov
2018-06-28 16:27 ` Vitaly Kuznetsov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180622170625.30688-4-vkuznets@redhat.com \
--to=vkuznets@redhat.com \
--cc=Michael.H.Kelley@microsoft.com \
--cc=Tianyu.Lan@microsoft.com \
--cc=devel@linuxdriverproject.org \
--cc=haiyangz@microsoft.com \
--cc=hpa@zytor.com \
--cc=kys@microsoft.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=sthemmin@microsoft.com \
--cc=tglx@linutronix.de \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox