Re: Assert in ShellPkg with latest tianocore edk2 source on the Reference Platform

From: Laszlo Ersek <lersek@redhat.com>
To: "Kinney, Michael D" <michael.d.kinney@intel.com>,
	Andrew Fish <afish@apple.com>,
	Daniil Egranov <daniil.egranov@arm.com>
Cc: "Carsey, Jaben" <jaben.carsey@intel.com>,
	edk2-devel-01 <edk2-devel@ml01.01.org>,
	Leif Lindholm <Leif.Lindholm@arm.com>
Subject: Re: Assert in ShellPkg with latest tianocore edk2 source on the Reference Platform
Date: Wed, 5 Oct 2016 23:48:53 +0200	[thread overview]
Message-ID: <bd987451-3c44-17e3-07b7-f38b7c2256d9@redhat.com> (raw)
In-Reply-To: <E92EE9817A31E24EB0585FDF735412F56481FD60@ORSMSX113.amr.corp.intel.com>

On 10/05/16 23:34, Kinney, Michael D wrote:
> Laszlo,
> 
> There are many types of UEFI Drivers and they are enumerated in the
> UEFI Driver Writer's Guide.  Some follow the UEFI Driver Model and
> some do not.
> 
> A driver that only consumes/produces UEFI services/protocols without a depex
> should be a UEFI Driver.  If there is a depex or use of any PI services/protocols
> then it must be one of the DXE driver types.
> 
> We actually prefer to categorize drivers as UEFI Driver's if possible, because 
> that maximizes the drivers's compatibility with UEFI conformant platforms.

Thanks!

It's been a while since I last looked at that part of the guide (chapter
6, UEFI Driver Categories). I guess EnglishDxe then qualifies as a
Service Driver.

... Interestingly, when people asked about automatically running
applications (not drivers) for various kinds of init work, without
staying resident, I occasionally thought about recommending a driver
module that would return an error code intentionally, even on success.
And now I see there's actually a category of drivers for this, called
"Initializing Drivers". So now I wonder why we needed SysPrep#### after
all... Probably because their order had to be specified in some way.

The Root Bridge Driver category is also intersting (under "UEFI Driver
Categories"); the core PciHostBridgeDxe in edk2 is a DXE_DRIVER. (It
does have an elaborate depex though, and not just on architectural
protocols.)

Thanks!
Laszlo

>> -----Original Message-----
>> From: edk2-devel [mailto:edk2-devel-bounces@lists.01.org] On Behalf Of Laszlo Ersek
>> Sent: Wednesday, October 5, 2016 2:18 PM
>> To: Andrew Fish <afish@apple.com>; Daniil Egranov <daniil.egranov@arm.com>
>> Cc: Carsey, Jaben <jaben.carsey@intel.com>; edk2-devel-01 <edk2-devel@ml01.01.org>;
>> Leif Lindholm <Leif.Lindholm@arm.com>
>> Subject: Re: [edk2] Assert in ShellPkg with latest tianocore edk2 source on the
>> Reference Platform
>>
>> On 10/05/16 23:05, Andrew Fish wrote:
>>
>>> I'm guessing that your DXE Driver is dispatching prior UEFI Driver that publishes the
>> protocol.
>>
>> BTW... Why is EnglishDxe a UEFI_DRIVER rather than a DXE_DRIVER? It
>> doesn't conform to the UEFI driver model. (I'm not saying it *must* not
>> be a UEFI_DRIVER, but what reason does it have for being one?) It also
>> calls no runtime services, and only the
>> InstallMultipleProtocolInterfaces() boot service; so I don't think it
>> relies on all of the architectural protocols either.
>>
>> Thanks
>> Laszlo
>>
>> _______________________________________________
>> edk2-devel mailing list
>> edk2-devel@lists.01.org
>> https://lists.01.org/mailman/listinfo/edk2-devel