Alex,
On Sun, Sep 23, 2018 at 12:11:08PM +0200, Alexander Graf wrote:
On 22.09.18 12:34, Heinrich Schuchardt wrote:
On 10/10/2017 02:22 PM, Rob Clark wrote:
From: Leif Lindholm leif.lindholm@linaro.org
Enough implementation of the following protocols to run Shell.efi and SCT.efi:
EfiHiiConfigRoutingProtocolGuid EfiHiiDatabaseProtocol EfiHiiStringProtocol
The i386 Shell.efi also tries to open these before failing:
EfiHiiFontProtocol EfiHiiImageProtocol
That sounds like a bug. The UEFI spec only mandates the following in section 2.6.2:
- If a platform includes a configuration infrastructure, then the
EFI_HII_DATABASE_PROTOCOL, EFI_HII_STRING_PROTOCOL, EFI_HII_CONFIG_ROUTING_PROTOCOL, and EFI_HII_CONFIG_ACCESS_PROTOCOL are required.
Do you think efi implementation on u-boot also wants those protocols? (Apparently, config access protocol can be optional for shell unless we want driver configuration?)
Among others, UEFI spec defines lots of "packages" for HII database: string, guid, font, image, device path, keyboard layout and etc.
Do you expect all of them be supported from the beginning? If not, what criteria do you think is reasonable so that you feel comfortable in accepting an initial commit of HII database support?
Thanks, -Takahiro Akashi
If you support bitmapped fonts, you must support EFI_HII_FONT_PROTOCOL.
Well, in my book the whole fact that the UEFI shell requires HII sounds like a bad design decision, but it should at least not require the font protocol :).
Also, I have built my own x86_64 shell.efi from latest edk2 git manually and that works fine (even in sandbox). Maybe that is only the minimal shell though? Their build system is still a bit of a mystery to me :).
Alex