public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "Michael D Kinney" <michael.d.kinney@intel.com>
To: "devel@edk2.groups.io" <devel@edk2.groups.io>,
	"derek.lin2@hpe.com" <derek.lin2@hpe.com>,
	"Kinney, Michael D" <michael.d.kinney@intel.com>
Cc: "Justen, Jordan L" <jordan.l.justen@intel.com>,
	"afish@apple.com" <afish@apple.com>, "Ni, Ray" <ray.ni@intel.com>,
	Kitty Chen <kitty.chen@hpe.com>
Subject: Re: [edk2-devel] [PATCH] EmulatorPkg: TimerLib QueryPerformance functions
Date: Thu, 30 Jan 2020 17:30:40 +0000	[thread overview]
Message-ID: <E92EE9817A31E24EB0585FDF735412F5B9E818D9@ORSMSX113.amr.corp.intel.com> (raw)
In-Reply-To: <20200130071214.16472-1-derek.lin2@hpe.com>

Hi Derik,

The Linux service clock_gettime() returns a structure
with seconds and nanoseconds and you are retuning the
number of nanoseconds from QueryPerformanceCounter().
Doesn't that imply that the rate of the performance
counter is 1 ns?  Does this allow you to remove the
calibration code?  Depending on the clock source 
used by clock_gettime(), there may be large jumps
in the number of ns.  You can use clock_getres() to
see if the resolution is acceptable.  It looks
like clock_gettime() optionally supports other time
sources that may be based on features such like TSC
and those may require calibration.

Mike



> -----Original Message-----
> From: devel@edk2.groups.io <devel@edk2.groups.io> On
> Behalf Of Lin, Derek (HPS SW)
> Sent: Wednesday, January 29, 2020 11:12 PM
> To: derek.lin2@hpe.com; devel@edk2.groups.io
> Cc: Justen, Jordan L <jordan.l.justen@intel.com>;
> afish@apple.com; Ni, Ray <ray.ni@intel.com>; Kitty Chen
> <kitty.chen@hpe.com>
> Subject: [edk2-devel] [PATCH] EmulatorPkg: TimerLib
> QueryPerformance functions
> 
> Implement QueryPerformanceCounter() and
> QueryPerformanceFrequency()
> for both Unix and Windows.
> 
> This has been tested in both Unix and Windows in an
> application using
> TimerLib.
> 
> Signed-off-by: Derek Lin <derek.lin2@hpe.com>
> Signed-off-by: Kitty Chen <kitty.chen@hpe.com>
> ---
>  EmulatorPkg/Unix/Host/EmuThunk.c | 40
> +++++++++++++++++++++-----------
>  EmulatorPkg/Win/Host/WinThunk.c  | 10 +++++---
>  2 files changed, 34 insertions(+), 16 deletions(-)
> 
> diff --git a/EmulatorPkg/Unix/Host/EmuThunk.c
> b/EmulatorPkg/Unix/Host/EmuThunk.c
> index 1e9dc99187..92703d3088 100644
> --- a/EmulatorPkg/Unix/Host/EmuThunk.c
> +++ b/EmulatorPkg/Unix/Host/EmuThunk.c
> @@ -11,6 +11,7 @@
> 
>  Copyright (c) 2004 - 2019, Intel Corporation. All
> rights reserved.<BR>
>  Portions copyright (c) 2008 - 2011, Apple Inc. All
> rights reserved.<BR>
> +(C) Copyright 2020 Hewlett Packard Enterprise
> Development LP<BR>
>  SPDX-License-Identifier: BSD-2-Clause-Patent
> 
>  **/
> @@ -33,6 +34,7 @@ struct timeval settimer_timeval;
>  UINTN  settimer_callback = 0;
> 
>  BOOLEAN gEmulatorInterruptEnabled = FALSE;
> +UINT64 mPerformanceFrequency = 0;
> 
> 
>  UINTN
> @@ -236,16 +238,6 @@ SecInterruptEanbled (void)
>    return gEmulatorInterruptEnabled;
>  }
> 
> -
> -UINT64
> -QueryPerformanceFrequency (
> -  VOID
> -  )
> -{
> -  // Hard code to nanoseconds
> -  return 1000000000ULL;
> -}
> -
>  UINT64
>  QueryPerformanceCounter (
>    VOID
> @@ -274,12 +266,34 @@ QueryPerformanceCounter (
> 
>    return (Start * sTimebaseInfo.numer) /
> sTimebaseInfo.denom;
>  #else
> -  // Need to figure out what to do for Linux?
> -  return 0;
> +  int status;
> +  struct timespec time;
> +  status = clock_gettime(CLOCK_REALTIME, &time);
> +  return MultU64x32 (time.tv_sec, 1000000000) +
> time.tv_nsec;
> +
>  #endif
>  }
> 
> -
> +UINT64
> +QueryPerformanceFrequency (
> +  VOID
> +  )
> +{
> +  UINT64 Counter1;
> +  UINT64 Counter2;
> +  if (mPerformanceFrequency) {
> +    return mPerformanceFrequency;
> +  }
> +  //
> +  // Don't know how to query performance frequency in
> Linux,
> +  // so instead, sleep 0.1 second and calculate it.
> +  //
> +  Counter1 = QueryPerformanceCounter();
> +  SecSleep (100* 1000 * 1000);
> +  Counter2 = QueryPerformanceCounter();
> +  mPerformanceFrequency = (Counter2-Counter1) * 10;
> +  return mPerformanceFrequency;
> +}
> 
>  VOID
>  SecSleep (
> diff --git a/EmulatorPkg/Win/Host/WinThunk.c
> b/EmulatorPkg/Win/Host/WinThunk.c
> index a77be2a64b..41b5cffe18 100644
> --- a/EmulatorPkg/Win/Host/WinThunk.c
> +++ b/EmulatorPkg/Win/Host/WinThunk.c
> @@ -1,6 +1,7 @@
>  /**@file
> 
>  Copyright (c) 2006 - 2018, Intel Corporation. All
> rights reserved.<BR>
> +(C) Copyright 2020 Hewlett Packard Enterprise
> Development LP<BR>
>  SPDX-License-Identifier: BSD-2-Clause-Patent
> 
>  Module Name:
> @@ -438,8 +439,9 @@ SecQueryPerformanceFrequency (
>    VOID
>    )
>  {
> -  // Hard code to nanoseconds
> -  return 1000000000ULL;
> +  UINT64 Frequency;
> +  QueryPerformanceFrequency ((LARGE_INTEGER *)
> &Frequency);
> +  return Frequency;
>  }
> 
>  UINT64
> @@ -447,7 +449,9 @@ SecQueryPerformanceCounter (
>    VOID
>    )
>  {
> -  return 0;
> +  UINT64 PerformanceCount;
> +  QueryPerformanceCounter ((LARGE_INTEGER *)
> &PerformanceCount);
> +  return PerformanceCount;
>  }
> 
> 
> --
> 2.17.0.windows.1
> 
> 
> 


      reply	other threads:[~2020-01-30 17:30 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-30  7:12 [PATCH] EmulatorPkg: TimerLib QueryPerformance functions Lin, Derek (HPS SW)
2020-01-30 17:30 ` Michael D Kinney [this message]

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-list 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=E92EE9817A31E24EB0585FDF735412F5B9E818D9@ORSMSX113.amr.corp.intel.com \
    --to=devel@edk2.groups.io \
    /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