From 70c9d989107c6ac964bb437c5a4ea6ffe3214e45 Mon Sep 17 00:00:00 2001
From: Miroslav Rezanina <mrezanin@redhat.com>
Date: Mon, 10 Aug 2020 07:52:28 +0200
Subject: [PATCH] UefiCpuPkg/PiSmmCpuDxeSmm: pause in WaitForSemaphore() before
 re-fetch
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

RH-Author: Laszlo Ersek <lersek@redhat.com>
Message-id: <20200731141037.1941-2-lersek@redhat.com>
Patchwork-id: 98121
O-Subject: [RHEL-8.3.0 edk2 PATCH 1/1] UefiCpuPkg/PiSmmCpuDxeSmm: pause in WaitForSemaphore() before re-fetch
Bugzilla: 1861718
RH-Acked-by: Vitaly Kuznetsov <vkuznets@redhat.com>
RH-Acked-by: Eduardo Habkost <ehabkost@redhat.com>

Most busy waits (spinlocks) in "UefiCpuPkg/PiSmmCpuDxeSmm/MpService.c"
already call CpuPause() in their loop bodies; see SmmWaitForApArrival(),
APHandler(), and SmiRendezvous(). However, the "main wait" within
APHandler():

>     //
>     // Wait for something to happen
>     //
>     WaitForSemaphore (mSmmMpSyncData->CpuData[CpuIndex].Run);

doesn't do so, as WaitForSemaphore() keeps trying to acquire the semaphore
without pausing.

The performance impact is especially notable in QEMU/KVM + OVMF
virtualization with CPU overcommit (that is, when the guest has
significantly more VCPUs than the host has physical CPUs). The guest BSP
is working heavily in:

  BSPHandler()                  [MpService.c]
    PerformRemainingTasks()     [PiSmmCpuDxeSmm.c]
      SetUefiMemMapAttributes() [SmmCpuMemoryManagement.c]

while the many guest APs are spinning in the "Wait for something to
happen" semaphore acquisition, in APHandler(). The guest APs are
generating useless memory traffic and saturating host CPUs, hindering the
guest BSP's progress in SetUefiMemMapAttributes().

Rework the loop in WaitForSemaphore(): call CpuPause() in every iteration
after the first check fails. Due to Pause Loop Exiting (known as Pause
Filter on AMD), the host scheduler can favor the guest BSP over the guest
APs.

Running a 16 GB RAM + 512 VCPU guest on a 448 PCPU host, this patch
reduces OVMF boot time (counted until reaching grub) from 20-30 minutes to
less than 4 minutes.

The patch should benefit physical machines as well -- according to the
Intel SDM, PAUSE "Improves the performance of spin-wait loops". Adding
PAUSE to the generic WaitForSemaphore() function is considered a general
improvement.

Cc: Eric Dong <eric.dong@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Cc: Rahul Kumar <rahul1.kumar@intel.com>
Cc: Ray Ni <ray.ni@intel.com>
Ref: https://bugzilla.redhat.com/show_bug.cgi?id=1861718
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20200729185217.10084-1-lersek@redhat.com>
Reviewed-by: Eric Dong <eric.dong@intel.com>
(cherry picked from commit 9001b750df64b25b14ec45a2efa1361a7b96c00a)
Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
---
 UefiCpuPkg/PiSmmCpuDxeSmm/MpService.c | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/UefiCpuPkg/PiSmmCpuDxeSmm/MpService.c b/UefiCpuPkg/PiSmmCpuDxeSmm/MpService.c
index 57e788c..4bcd217 100644
--- a/UefiCpuPkg/PiSmmCpuDxeSmm/MpService.c
+++ b/UefiCpuPkg/PiSmmCpuDxeSmm/MpService.c
@@ -40,14 +40,18 @@ WaitForSemaphore (
 {
   UINT32                            Value;
 
-  do {
+  for (;;) {
     Value = *Sem;
-  } while (Value == 0 ||
-           InterlockedCompareExchange32 (
-             (UINT32*)Sem,
-             Value,
-             Value - 1
-             ) != Value);
+    if (Value != 0 &&
+        InterlockedCompareExchange32 (
+          (UINT32*)Sem,
+          Value,
+          Value - 1
+          ) == Value) {
+      break;
+    }
+    CpuPause ();
+  }
   return Value - 1;
 }
 
-- 
1.8.3.1