public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: Laszlo Ersek <lersek@redhat.com>
To: "haojian.zhuang@linaro.org" <haojian.zhuang@linaro.org>
Cc: "edk2-devel@lists.01.org" <edk2-devel@lists.01.org>,
	"leif.lindholm@linaro.org" <leif.lindholm@linaro.org>,
	"ard.biesheuvel@linaro.org" <ard.biesheuvel@linaro.org>,
	"linaro-uefi@lists.linaro.org" <linaro-uefi@lists.linaro.org>
Subject: Re: [PATCH 1/2] Platform/Hisilicon/HiKey960: add boot options
Date: Thu, 22 Feb 2018 11:56:22 +0100	[thread overview]
Message-ID: <61c94cc9-1b5c-95a7-fdd9-812609c68561@redhat.com> (raw)
In-Reply-To: <ca8e02f6-60f5-bc3b-6a72-5d53baf80619@outlook.com>

On 02/22/18 09:57, Haojian Zhuang wrote:
> On 02/20/2018 11:54 PM, Laszlo Ersek wrote:
>> Hi,
>>
>> On 02/15/18 16:14, Haojian Zhuang wrote:
>>> Add four boot options in emu variable region. They are
>>> "Boot on SD", "Grub", "Android Boot" and "Android Fastboot".
>>>
>>> Contributed-under: TianoCore Contribution Agreement 1.1
>>> Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>
>>> ---
>>>   Platform/Hisilicon/HiKey960/HiKey960.dsc |   3 +
>>>   Platform/Hisilicon/HiKey960/HiKey960.fdf | 241 ++++++++++++++++++++++++++++++-
>>>   2 files changed, 242 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/Platform/Hisilicon/HiKey960/HiKey960.dsc b/Platform/Hisilicon/HiKey960/HiKey960.dsc
>>> index 98289c0..a6864c1 100644
>>> --- a/Platform/Hisilicon/HiKey960/HiKey960.dsc
>>> +++ b/Platform/Hisilicon/HiKey960/HiKey960.dsc
>>> @@ -78,6 +78,9 @@
>>>     gEfiMdeModulePkgTokenSpaceGuid.PcdConOutGopSupport|FALSE
>>>   
>>>   [PcdsFixedAtBuild.common]
>>> +  gEfiMdeModulePkgTokenSpaceGuid.PcdEmuVariableNvStoreReserved|0x1AD88048
>>> +  gEfiMdeModulePkgTokenSpaceGuid.PcdVariableStoreSize|0x7FB8
>>> +
>>>     gEfiMdePkgTokenSpaceGuid.PcdDefaultTerminalType|4
>>>   
>>>     gEfiMdeModulePkgTokenSpaceGuid.PcdFirmwareVersionString|L"Alpha"
>>> diff --git a/Platform/Hisilicon/HiKey960/HiKey960.fdf b/Platform/Hisilicon/HiKey960/HiKey960.fdf
>>> index 655032a..af10430 100644
>>> --- a/Platform/Hisilicon/HiKey960/HiKey960.fdf
>>> +++ b/Platform/Hisilicon/HiKey960/HiKey960.fdf
>>> @@ -26,12 +26,12 @@
>>>   
>>>   [FD.BL33_AP_UEFI]
>>>   BaseAddress   = 0x1AC98000|gArmTokenSpaceGuid.PcdFdBaseAddress  # The base address of the Firmware in NOR Flash.
>>> -Size          = 0x000F0000|gArmTokenSpaceGuid.PcdFdSize         # The size in bytes of the FLASH Device
>>> +Size          = 0x00100000|gArmTokenSpaceGuid.PcdFdSize         # The size in bytes of the FLASH Device
>>>   ErasePolarity = 1
>>>   
>>>   # This one is tricky, it must be: BlockSize * NumBlocks = Size
>>>   BlockSize     = 0x00001000
>>> -NumBlocks     = 0xF0
>>> +NumBlocks     = 0x100
>>>   
>>>   ################################################################################
>>>   #
>>> @@ -53,6 +53,243 @@ NumBlocks     = 0xF0
>>>   gArmTokenSpaceGuid.PcdFvBaseAddress|gArmTokenSpaceGuid.PcdFvSize
>>>   FV = FVMAIN_COMPACT
>>>   
>>> +0x000F0000|0x00008000
>>> +gEfiMdeModulePkgTokenSpaceGuid.PcdFlashNvStorageVariableBase|gEfiMdeModulePkgTokenSpaceGuid.PcdFlashNvStorageVariableSize
>>> +DATA = {
>>> +  ## This is the EFI_FIRMWARE_VOLUME_HEADER
>>> +  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
>>> +  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
>>> +  # FileSystemGuid: gEfiSystemNvDataFvGuid         =
>>> +  0x8D, 0x2B, 0xF1, 0xFF, 0x96, 0x76, 0x8B, 0x4C,
>>> +  0xA9, 0x85, 0x27, 0x47, 0x07, 0x5B, 0x4F, 0x50,
>>> +  # FvLength: 0x8000
>>> +  0x00, 0x80, 0x0, 0x00, 0x00, 0x00, 0x00, 0x00,
>>> +  #Signature "_FVH"       #Attributes
>>> +  0x5f, 0x46, 0x56, 0x48, 0xff, 0xfe, 0x04, 0x00,
>>> +  #HeaderLength #CheckSum #ExtHeaderOffset #Reserved #Revision
>>> +  0x48, 0x00, 0x36, 0x09, 0x00, 0x00, 0x00, 0x02,
>>> +  #Blockmap[0]: 8 Blocks * 0x1000 Bytes / Block
>>> +  0x08, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00,
>>> +  #Blockmap[1]: End
>>> +  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
>>> +  ## end of EFI_FIRMWARE_VOLUME_HEADER.
>>> +
>>> +  ### Offset (0x48) ###
>>> +  ## This is the VARIABLE_STORE_HEADER gEfiVariableGuid
>>> +  0x16, 0x36, 0xcf, 0xdd, 0x75, 0x32, 0x64, 0x41,
>>> +  0x98, 0xb6, 0xfe, 0x85, 0x70, 0x7f, 0xfe, 0x7d,
>>> +  #Size: 0x8000 (gEfiMdeModulePkgTokenSpaceGuid.PcdFlashNvStorageVariableSize) - 0x48 (size of EFI_FIRMWARE_VOLUME_HEADER) = 0x7FB8
>>> +  0xB8, 0x7F, 0x00, 0x00,
>>> +  #FORMATTED: 0x5A #HEALTHY: 0xFE #Reserved: UINT16 #Reserved1: UINT32
>>> +  0x5A, 0xFE, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
>>> +
>>> +  ### Offset (0x64) ###
>>> +  ## This is the VARIABLE_HEADER
>>> +  #StartId: VARIABLE_DATA #State: VAR_ADDED #Reserved
>>> +  0xAA, 0x55, 0x3F, 0x00,
>>> +  #Attributes: EFI_VARIABLE_NON_VOLATILE | EFI_VARIABLE_BOOTSERVICE_ACCESS
>>> +  0x03, 0x00, 0x00, 0x00,
>>> +  #NameSize:
>>> +  0x14, 0x00, 0x00, 0x00,
>>> +  #DataSize:
>>> +  0x18, 0x00, 0x00, 0x00,
>>> +  #VariableGuid: gEfiGlobalVariableGuid
>>> +  0x61, 0xDF, 0xE4, 0x8B, 0xCA, 0x93, 0xD2, 0x11,
>>> +  0xAA, 0x0D, 0x00, 0xE0, 0x98, 0x03, 0x2B, 0x8C,
>>> +  ## End of VARIABLE_HEADER. Offset (0x84)
>>> +  #VariableName: BootOrder
>>> +  0x42, 0x00, 0x6F, 0x00, 0x6F, 0x00, 0x74, 0x00,
>>> +  0x4F, 0x00, 0x72, 0x00, 0x64, 0x00, 0x65, 0x00,
>>> +  0x72, 0x00, 0x00, 0x00,
>>> +  #Data
>>> +  0x00, 0x00, 0x01, 0x00, 0x02, 0x00, 0x03, 0x00,
>>> +  0x04, 0x00, 0x05, 0x00, 0x06, 0x00, 0x07, 0x00,
>>> +  0x08, 0x00, 0x09, 0x00, 0x0a, 0x00, 0x0b, 0x00,
>>> +  ### End of BootOrder.
>>> +
>>> +  ### Offset (0xB0) ###
>>> +  ## This is the VARIABLE_HEADER
>>> +  #StartId: VARIABLE_DATA #State: VAR_ADDED #Reserved
>>> +  0xAA, 0x55, 0x3F, 0x00,
>>> +  #Attributes: EFI_VARIABLE_NON_VOLATILE | EFI_VARIABLE_BOOTSERVICE_ACCESS
>>> +  0x03, 0x00, 0x00, 0x00,
>>> +  #NameSize:
>>> +  0x12, 0x00, 0x00, 0x00,
>>> +  #DataSize:
>>> +  0x42, 0x00, 0x00, 0x00,
>>> +  #VariableGuid: gEfiGlobalVariableGuid
>>> +  0x61, 0xDF, 0xE4, 0x8B, 0xCA, 0x93, 0xD2, 0x11,
>>> +  0xAA, 0x0D, 0x00, 0xE0, 0x98, 0x03, 0x2B, 0x8C,
>>> +  #VariableName: Boot0000
>>> +  0x42, 0x00, 0x6F, 0x00, 0x6F, 0x00, 0x74, 0x00,
>>> +  0x30, 0x00, 0x30, 0x00, 0x30, 0x00, 0x30, 0x00,
>>> +  0x00, 0x00,
>>> +  ###  Offset (0xE2), Size(0x32) ###
>>> +  #Data:
>>> +  #Attributes: LOAD_OPTION_ACTIVE
>>> +  0x01, 0x00, 0x00, 0x00,
>>> +  #FilePathListLength
>>> +  0x26, 0x00,
>>> +  #Description: "Boot on SD"
>>> +  0x42, 0x00, 0x6F, 0x00, 0x6F, 0x00, 0x74, 0x00,      #Boot
>>> +  0x20, 0x00, 0x6F, 0x00, 0x6E, 0x00, 0x20, 0x00,      # on
>>> +  0x53, 0x00, 0x44, 0x00, 0x00, 0x00,                  #SD
>>> +  #FilePathList (0x26)
>>> +  0x01, 0x04, 0x1D, 0x00, 0x5B, 0x90, 0x51, 0x0D,
>>> +  0x7E, 0xB7, 0x2A, 0x45, 0xA2, 0xC0, 0xEC, 0xA0,
>>> +  0xCC, 0x8D, 0x51, 0x4A, 0x00, 0xF0, 0x37, 0xFF,
>>> +  0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x1A, 0x05,
>>> +  0x00, 0x00, 0x7F, 0xFF, 0x04, 0x00,
>>> +  ### Offset (0x124), Size (0x42) ###
>>> +  ### End of Boot0000
>>> +
>>>
>>
>> While in both ArmVirtPkg and OvmfPkg we use a similar DATA definition
>> for *formatting* the pristine variable store (just setting the
>> absolutely necessary headers via a hexdump), I think that adding
>> "business content" like this is a bad idea. For one, if the defaults
>> have to be updated ever again, the patches for the DATA definitions will
>> look terrible. It's hard to review and maintain hexdump like these; we
>> should keep such DATA definitions to the absolute minimum.
>>
>> Such "default" boot options should be set by the platform's
>> PlatformBootManagerLib instance. PlatformBootManagerLib is responsible
>> for generating boot options. In doing that, PlatformBootManagerLib can
>> call helper functions from UefiBootManagerLib, so everything need not be
>> done manually.
>>
>> For example, EfiBootManagerGetLoadOptions() can be used to retrieve the
>> current boot options. Then, you can iterate over the four boot options
>> that you want to ensure exist -- for each:
>>
>> - create a EFI_BOOT_MANAGER_LOAD_OPTION object with
>> EfiBootManagerInitializeLoadOption(),
>> - check if the option already exists, with EfiBootManagerFindLoadOption(),
>> - if the option doesn't exist yet, add it with
>> EfiBootManagerAddLoadOptionVariable(),
>> - free the EFI_BOOT_MANAGER_LOAD_OPTION object with
>> EfiBootManagerFreeLoadOption()
>>
>> Finally, free the originally retrieved set of boot options with
>> EfiBootManagerFreeLoadOptions().
>>
>> You can find example code in
>> "ArmVirtPkg/Library/PlatformBootManagerLib/PlatformBm.c".
>>
>> Thanks,
>> Laszlo
>>
> 
> Hi Laszlo,
> 
> Yes, it's a bit hard to review and maintain. Actually, I implemented a 
> PlatformBootManagerLib in this link 
> (https://github.com/96boards-hikey/OpenPlatformPkg/blob/testing/hikey960_v1.3.4/Platforms/Hisilicon/Library/PlatformBootManagerLib/PlatformBm.c).
> 
> I format it with hard code since I want to re-use 
> PlatformBootManagerLib in ArmPkg. Because these boot options only exist 
> on HiKey/HiKey960 platforms, and most of them are same of the one in 
> ArmPkg. If I append them into ArmPkg, it's not common enough. At last, I 
> hope to implement a non-volatile region on flash devices in the future. 
> It means that implementing a HiKey specific PlatformBootManagerLib is 
> only a temporary solution.

Sorry, I don't understand how this follows. Again: why is it only a
temporary solution to implement a PlatformBootManagerLib instance for HiKey?

The library class says "Platform" in the name. Platforms are supposed to
provide it, one way or another.

If you want to minimize code duplication between ArmPkg and HiKey, it
should be possible to factor out another library class from ArmPkg's
PlatformBootManagerLib instance. It could be a function that returns the
list of platform-specific boot options -- as an array of
EFI_BOOT_MANAGER_LOAD_OPTION elements -- that should always exist. Then
ArmPkg's PlatformBootManagerLib would perform the iteration that I
described above.

Platforms different from HiKey would return an empty array, while HiKey
could return the four options it needs.

Another benefit is that these four boot options would be restored on
every boot, even if the user deleted them. This seems independent of
whether the HiKey platform has functional non-volatile storage or not.

> It seems that I have two options to go ahead.
> 1. Move those hard code boot options into a FV file, and put the FV file
> in edk2-non-osi repository.
> 
> 2. Implement a HiKey related PlatformBootManagerLib.

I think option #2 is far superior. You don't have to duplicate all code
from ArmPkg's PlatformBootManagerLib for that, you can extract another
utility library class / callback function just for providing the options
you always want.

Or perhaps you *don't* want them always, just until NVRAM support
arrives to HiKey? And, in that case, it will be alright for the user to
delete these options permanently?

... I still think extracting a new lib class for these default boot
options would be better. If at some point you'd like to drop the default
boot options, it's easy to switch the new callback to returning zero
elements. The lib class can remain useful to other development platforms.

Thanks
Laszlo


  reply	other threads:[~2018-02-22 10:50 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-02-15 15:14 [PATCH 0/2] add boot options in HiKey Haojian Zhuang
2018-02-15 15:14 ` [PATCH 1/2] Platform/Hisilicon/HiKey960: add boot options Haojian Zhuang
2018-02-20 15:54   ` Laszlo Ersek
2018-02-22  8:57     ` Haojian Zhuang
2018-02-22 10:56       ` Laszlo Ersek [this message]
2018-02-22 11:32         ` Leif Lindholm
2018-02-22 11:49         ` Haojian Zhuang
2018-02-15 15:14 ` [PATCH 2/2] Platform/Hisilicon/HiKey: " Haojian Zhuang

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=61c94cc9-1b5c-95a7-fdd9-812609c68561@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