From: Laszlo Ersek <lersek@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: edk2-devel@lists.01.org,
Jordan Justen <jordan.l.justen@intel.com>,
Ard Biesheuvel <ard.biesheuvel@linaro.org>
Subject: Re: [PATCH 1/2] OvmfPkg: create a separate PlatformDebugLibIoPort instance for SEC
Date: Thu, 16 Nov 2017 19:43:13 +0100 [thread overview]
Message-ID: <01286b14-d1b0-04e0-086e-ed2d93d11384@redhat.com> (raw)
In-Reply-To: <20171116104716.15144-2-pbonzini@redhat.com>
On 11/16/17 11:47, Paolo Bonzini wrote:
> The next patch will want to add a global variable to
> PlatformDebugLibIoPort, but this is not suitable for the SEC
> phase, because SEC runs from read-only flash. The solution is
> to have two library instances, one for SEC and another
> for all other firmware phases. This patch adds the "plumbing"
> for the SEC library instance, separating the INF files and
> moving the constructor to a separate C source file.
>
> Contributed-under: TianoCore Contribution Agreement 1.1
> Cc: Laszlo Ersek <lersek@redhat.com>
> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> Cc: Jordan Justen (Intel address) <jordan.l.justen@intel.com>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
> OvmfPkg/Library/PlatformDebugLibIoPort/DebugLib.c | 15 -------
> .../PlatformDebugLibIoPort/DebugLibDetect.c | 32 +++++++++++++
> .../PlatformDebugLibIoPort/DebugLibDetectRom.c | 32 +++++++++++++
> .../PlatformDebugLibIoPort.inf | 1 +
> .../PlatformRomDebugLibIoPort.inf | 52 ++++++++++++++++++++++
> OvmfPkg/OvmfPkgIa32.dsc | 2 +-
> OvmfPkg/OvmfPkgIa32X64.dsc | 2 +-
> OvmfPkg/OvmfPkgX64.dsc | 2 +-
> 8 files changed, 120 insertions(+), 18 deletions(-)
> create mode 100644 OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetect.c
> create mode 100644 OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetectRom.c
> create mode 100644 OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
Very nice!
I have a number of superficial comments. I've considered squashing these changes into your patch myself, but I didn't know what you'd prefer, so first I'll just list them.
> diff --git a/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLib.c b/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLib.c
> index 5435767c1c..a5572a8eeb 100644
> --- a/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLib.c
> +++ b/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLib.c
> @@ -29,21 +29,6 @@
> //
> #define MAX_DEBUG_MESSAGE_LENGTH 0x100
>
> -/**
> - This constructor function does not have to do anything.
> -
> - @retval EFI_SUCCESS The constructor always returns RETURN_SUCCESS.
> -
> -**/
> -RETURN_STATUS
> -EFIAPI
> -PlatformDebugLibIoPortConstructor (
> - VOID
> - )
> -{
> - return EFI_SUCCESS;
> -}
> -
> /**
> Prints a debug message to the debug output device if the specified error level is enabled.
>
(1) This is a pre-existent wart in the code. The return type is RETURN_STATUS, but we return EFI_SUCCESS instead of RETURN_SUCCESS.
Numerically they are the same, but the message is different. RETURN_* is for library instances that have MODULE_TYPE=BASE, meaning that they can be built into modules that are more "primitive" than DXE_DRIVERs (such as SEC, PEIM, PEI_CORE, ...). This distinction primarily influences the constructor function's parameter list -- for MODULE_TYPE=BASE library instances, it is just VOID --, but it also dictates whether we should use EFI_xxx or RETURN_xxx as status codes.
Can you please prepend a patch to the series that replaces EFI_SUCCESS with RETURN_SUCCESS in this function?
(2) Similarly, as a "precursor cleanup", <Uefi.h> should not be included anywhere in this library instance. Namely, "MdePkg/Include/Uefi.h" has the following leading comment (rewrapped here for readability):
> Root include file for Mde Package UEFI, UEFI_APPLICATION type modules.
>
> This is the include file for any module of type UEFI and
> UEFI_APPLICATION. Uefi modules only use types defined via this include
> file and can be ported easily to any environment.
UEFI_DRIVER and UEFI_APPLICATION modules are the *least* primitive module types, while this library instance is on the other end of the spectrum.
Now, "MdePkg/Include/Uefi.h" is just a convenience header for including:
#include <Uefi/UefiBaseType.h>
#include <Uefi/UefiSpec.h>
>From these, "MdePkg/Include/Uefi/UefiBaseType.h" is totally fine to include wherever, while "MdePkg/Include/Uefi/UefiSpec.h" is the one that should be kept out of BASE lib classes.
Thus, in the same prepended patch, can you please also try to replace <Uefi.h> with <Uefi/UefiBaseType.h>? The lib instance should continue building just the same. (If it breaks, that means we have a layering violation in the code somewhere.)
Consequently:
> diff --git a/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetect.c b/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetect.c
> new file mode 100644
> index 0000000000..fee908861b
> --- /dev/null
> +++ b/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetect.c
> @@ -0,0 +1,32 @@
> +/** @file
> + Constructor code for QEMU debug port library.
> + Non-SEC instance.
> +
> + Copyright (c) 2017, Red Hat, Inc.<BR>
> + This program and the accompanying materials
> + are licensed and made available under the terms and conditions of the BSD License
> + which accompanies this distribution. The full text of the license may be found at
> + http://opensource.org/licenses/bsd-license.php.
> +
> + THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
> + WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
> +
> +**/
> +
> +#include <Base.h>
> +#include <Uefi.h>
> +
> +/**
> + This constructor function does not have anything to do.
> +
> + @retval EFI_SUCCESS The constructor always returns RETURN_SUCCESS.
> +
> +**/
> +RETURN_STATUS
> +EFIAPI
> +PlatformDebugLibIoPortConstructor (
> + VOID
> + )
> +{
> + return EFI_SUCCESS;
> +}
> diff --git a/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetectRom.c b/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetectRom.c
> new file mode 100644
> index 0000000000..407fe613ff
> --- /dev/null
> +++ b/OvmfPkg/Library/PlatformDebugLibIoPort/DebugLibDetectRom.c
> @@ -0,0 +1,32 @@
> +/** @file
> + Constructor code for QEMU debug port library.
> + SEC instance.
> +
> + Copyright (c) 2017, Red Hat, Inc.<BR>
> + This program and the accompanying materials
> + are licensed and made available under the terms and conditions of the BSD License
> + which accompanies this distribution. The full text of the license may be found at
> + http://opensource.org/licenses/bsd-license.php.
> +
> + THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
> + WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
> +
> +**/
> +
> +#include <Base.h>
> +#include <Uefi.h>
> +
> +/**
> + This constructor function does not have to do anything.
> +
> + @retval EFI_SUCCESS The constructor always returns RETURN_SUCCESS.
> +
> +**/
> +RETURN_STATUS
> +EFIAPI
> +PlatformRomDebugLibIoPortConstructor (
> + VOID
> + )
> +{
> + return EFI_SUCCESS;
> +}
(3) In the above two files, <Uefi.h> should not be included, plus the constructors should return RETURN_SUCCESS.
(4) It's a bit higher up in the patch (I'm mentioning it here only for keeping the previous points focused): if you diff the two new files against each other, you get
- This constructor function does not have anything to do.
+ This constructor function does not have to do anything.
Please consider unifying this.
Then, we get to the INF files:
> diff --git a/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf b/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> index 0e74fe94cb..65d8683f1f 100644
> --- a/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> +++ b/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> @@ -30,6 +30,7 @@
>
> [Sources]
> DebugLib.c
> + DebugLibDetect.c
>
> [Packages]
> MdePkg/MdePkg.dec
> diff --git a/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf b/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
> new file mode 100644
> index 0000000000..93763d47dd
> --- /dev/null
> +++ b/OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
> @@ -0,0 +1,52 @@
> +## @file
> +# Instance of Debug Library for the QEMU debug console port.
> +# It uses Print Library to produce formatted output strings.
> +#
> +# Copyright (c) 2006 - 2015, Intel Corporation. All rights reserved.<BR>
> +# Copyright (c) 2017, Red Hat, Inc.<BR>
> +#
> +# This program and the accompanying materials
> +# are licensed and made available under the terms and conditions of the BSD License
> +# which accompanies this distribution. The full text of the license may be found at
> +# http://opensource.org/licenses/bsd-license.php.
> +# THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
> +# WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
> +#
> +#
> +##
> +
> +[Defines]
> + INF_VERSION = 0x00010005
> + BASE_NAME = PlatformRomDebugLibIoPort
> + FILE_GUID = CEB0D9D3-328F-4C24-8C02-28FA1986AE1B
> + MODULE_TYPE = BASE
> + VERSION_STRING = 1.0
> + LIBRARY_CLASS = DebugLib
> + CONSTRUCTOR = PlatformRomDebugLibIoPortConstructor
> +
> +#
> +# VALID_ARCHITECTURES = IA32 X64 IPF EBC
> +#
> +
> +[Sources]
> + DebugLib.c
> + DebugLibDetectRom.c
> +
> +[Packages]
> + MdePkg/MdePkg.dec
> + OvmfPkg/OvmfPkg.dec
> +
> +[LibraryClasses]
> + BaseMemoryLib
> + IoLib
> + PcdLib
> + PrintLib
> + BaseLib
> + DebugPrintErrorLevelLib
> +
> +[Pcd]
> + gUefiOvmfPkgTokenSpaceGuid.PcdDebugIoPort ## CONSUMES
> + gEfiMdePkgTokenSpaceGuid.PcdDebugClearMemoryValue ## CONSUMES
> + gEfiMdePkgTokenSpaceGuid.PcdDebugPropertyMask ## CONSUMES
> + gEfiMdePkgTokenSpaceGuid.PcdFixedDebugPrintErrorLevel ## CONSUMES
> +
(5) This is totally my fault, but last night I couldn't think of it:
The LIBRARY_CLASS assignment in the [Defines] section supports a syntax that restricts the library instance to specific client module types. Such as:
LIBRARY_CLASS = DebugLib|SEC
LIBRARY_CLASS = DebugLib|PEI_CORE PEIM DXE_CORE DXE_DRIVER DXE_RUNTIME_DRIVER SMM_CORE DXE_SMM_DRIVER UEFI_DRIVER UEFI_APPLICATION
(The module types need not be sorted "logically" within LIBRARY_CLASS, but I did that above for a pleasing read.)
Such restrictions are helpful because then an invalid / unintended library class resolution in a DSC file is caught at build time; it won't get to cause weird behavior at runtime.
Please consider adding these lists in v3.
> diff --git a/OvmfPkg/OvmfPkgIa32.dsc b/OvmfPkg/OvmfPkgIa32.dsc
> index c2f534fdbf..7ccb61147f 100644
> --- a/OvmfPkg/OvmfPkgIa32.dsc
> +++ b/OvmfPkg/OvmfPkgIa32.dsc
> @@ -207,7 +207,7 @@
> !ifdef $(DEBUG_ON_SERIAL_PORT)
> DebugLib|MdePkg/Library/BaseDebugLibSerialPort/BaseDebugLibSerialPort.inf
> !else
> - DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> + DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
> !endif
> ReportStatusCodeLib|MdeModulePkg/Library/PeiReportStatusCodeLib/PeiReportStatusCodeLib.inf
> ExtractGuidedSectionLib|MdePkg/Library/BaseExtractGuidedSectionLib/BaseExtractGuidedSectionLib.inf
> diff --git a/OvmfPkg/OvmfPkgIa32X64.dsc b/OvmfPkg/OvmfPkgIa32X64.dsc
> index 9f300a2e6f..237ec71b5e 100644
> --- a/OvmfPkg/OvmfPkgIa32X64.dsc
> +++ b/OvmfPkg/OvmfPkgIa32X64.dsc
> @@ -212,7 +212,7 @@
> !ifdef $(DEBUG_ON_SERIAL_PORT)
> DebugLib|MdePkg/Library/BaseDebugLibSerialPort/BaseDebugLibSerialPort.inf
> !else
> - DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> + DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
> !endif
> ReportStatusCodeLib|MdeModulePkg/Library/PeiReportStatusCodeLib/PeiReportStatusCodeLib.inf
> ExtractGuidedSectionLib|MdePkg/Library/BaseExtractGuidedSectionLib/BaseExtractGuidedSectionLib.inf
> diff --git a/OvmfPkg/OvmfPkgX64.dsc b/OvmfPkg/OvmfPkgX64.dsc
> index 1ffcf37f8b..a5047fa38e 100644
> --- a/OvmfPkg/OvmfPkgX64.dsc
> +++ b/OvmfPkg/OvmfPkgX64.dsc
> @@ -212,7 +212,7 @@
> !ifdef $(DEBUG_ON_SERIAL_PORT)
> DebugLib|MdePkg/Library/BaseDebugLibSerialPort/BaseDebugLibSerialPort.inf
> !else
> - DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> + DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
> !endif
> ReportStatusCodeLib|MdeModulePkg/Library/PeiReportStatusCodeLib/PeiReportStatusCodeLib.inf
> ExtractGuidedSectionLib|MdePkg/Library/BaseExtractGuidedSectionLib/BaseExtractGuidedSectionLib.inf
>
This is fine, but I'll use the opportunity to ask you to tweak the git config of your edk2 clone a bit :) The below hints are from <https://github.com/tianocore/tianocore.github.io/wiki/Laszlo's-unkempt-git-guide-for-edk2-contributors-and-maintainers>.
(6) Looking at the above hunks, it is not readily apparent to the reviewer what section of the DSC file is being patched. It can be improved:
(6a) run:
git config diff.ini.xfuncname '^\[[A-Za-z0-9_., ]+]'
(6b) add the following to ".git/info/attributes":
*.dec diff=ini
*.dsc diff=ini
*.dsc.inc diff=ini
*.fdf diff=ini
*.fdf.inc diff=ini
*.inf diff=ini
As a result, the hunks will look like:
> diff --git a/OvmfPkg/OvmfPkgIa32.dsc b/OvmfPkg/OvmfPkgIa32.dsc
> index c2f534fdbf3b..7ccb61147f27 100644
> --- a/OvmfPkg/OvmfPkgIa32.dsc
> +++ b/OvmfPkg/OvmfPkgIa32.dsc
> @@ -207,7 +207,7 @@ [LibraryClasses.common.SEC]
> !ifdef $(DEBUG_ON_SERIAL_PORT)
> DebugLib|MdePkg/Library/BaseDebugLibSerialPort/BaseDebugLibSerialPort.inf
> !else
> - DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformDebugLibIoPort.inf
> + DebugLib|OvmfPkg/Library/PlatformDebugLibIoPort/PlatformRomDebugLibIoPort.inf
> !endif
> ReportStatusCodeLib|MdeModulePkg/Library/PeiReportStatusCodeLib/PeiReportStatusCodeLib.inf
> ExtractGuidedSectionLib|MdePkg/Library/BaseExtractGuidedSectionLib/BaseExtractGuidedSectionLib.inf
Note "[LibraryClasses.common.SEC]" after the "@@" marker.
(Yes, we could carry a .gitattributes file in the tree, but that would require us to agree upon it.)
(7) Consider creating a file called "edk2.diff.order", with the following contents:
*.dec
*.dsc.inc
*.dsc
*.fdf.inc
*.fdf
*.inf
*.h
*.vfr
*.c
and setting the "diff.orderFile" config knob to the pathname of that file.
Then the reviewer will get to see hunks in the following order:
- package declaration file changes,
- platform DSC and flash description changes,
- module INF changes
- C header changes,
- VFR (visual form representation) changes,
- finally, C language changes.
(Yes, we could carry an order file in the tree as well, but my attempt to do it for QEMU had failed, so I didn't try posting it for edk2.)
I hope you aren't overly annoyed by the above wall of text. Do you have time for a v3?
Thank you!
Laszlo
next prev parent reply other threads:[~2017-11-16 18:39 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-11-16 10:47 [PATCH v2 0/2] OvmfPkg: save on I/O port accesses when the debug port is not in use Paolo Bonzini
2017-11-16 10:47 ` [PATCH 1/2] OvmfPkg: create a separate PlatformDebugLibIoPort instance for SEC Paolo Bonzini
2017-11-16 18:43 ` Laszlo Ersek [this message]
2017-11-16 10:47 ` [PATCH 2/2] OvmfPkg: save on I/O port accesses when the debug port is not in use Paolo Bonzini
2017-11-16 20:03 ` Laszlo Ersek
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=01286b14-d1b0-04e0-086e-ed2d93d11384@redhat.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