Re: [PATCH 00/19] OvmfPkg, ArmVirtPkg: leave deprecated interfaces behind

[Laszlo Ersek <lersek@redhat.com>]

On 10/24/16 10:04, Ard Biesheuvel wrote:
> On 21 October 2016 at 22:27, Laszlo Ersek <lersek@redhat.com> wrote:
>> This series intends to solve the following BZs:
>>
>>   <https://bugzilla.tianocore.org/show_bug.cgi?id=166> -- OvmfPkg: Add
>>   the build option "/D DISABLE_NEW_DEPRECATED_INTERFACES" in package DSC
>>   files
>>
>>   <https://bugzilla.tianocore.org/show_bug.cgi?id=165> -- ArmVirtPkg:
>>   Add the build option "/D DISABLE_NEW_DEPRECATED_INTERFACES" in package
>>   DSC files
>>
>> Public branch:
>> <https://github.com/lersek/edk2/commits/no_deprec_ovmf_bz166_armvirt_bz165>.
>>
>> The DISABLE_NEW_DEPRECATED_INTERFACES feature test macro disables the
>> following MdePkg library class APIs:
>>
>>   BaseLib:
>>   - StrCpy, StrnCpy, StrCat, StrnCat, UnicodeStrToAsciiStr,
>>   - AsciiStrCpy, AsciiStrnCpy, AsciiStrCat, AsciiStrnCat,
>>     AsciiStrToUnicodeStr.
>>
>>   PcdLib:
>>   - PcdSet8, PcdSet16, PcdSet32, PcdSet64, PcdSetPtr, PcdSetBool,
>>   - PcdSetEx8, PcdSetEx16, PcdSetEx32, PcdSetEx64, PcdSetExPtr,
>>     PcdSetExBool.
>>
>>   UefiLib:
>>   - GetVariable, GetEfiGlobalVariable.
>>
>> The series gradually weans the OvmfPkg and ArmVirtPkg modules off these.
>> For the 32-bit ARM builds of ArmVirtPkg platforms, I had to dip my toes
>> into ArmPkg a little bit, due to dependencies.
>>
>> I couldn't build-test some changes (for example, the only compiler
>> toolchains I have access to at the moment are GCC48 for Ia32/X64, and
>> GCC5 for ARM/AARCH64). Some changes I could build, but not functionally
>> test (Xen en bloc, 32-bit ARM, RAM-emulated variables in OVMF, -bios
>> flag). For all of these, I liberally sprinkled the patches with Cc's and
>> Notes sections, asking for help. I did make an honest effort to build
>> the ArmVirt and OVMF platforms in as many configurations (-D ...) as I
>> could think of, perusing the various !if directives in the DSC files.
>>
> 
> I tried ArmVirtQemu.dsc in DEBUG mode with RVCTLINUX, and it built fine*
> 
> Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> # RVCT

Thank you!

Then,

> * In general, RVCT tends to fall over quite regularly due to its
> finicky diagnostics, so I did have to apply this patch
> 
> """
> diff --git a/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.c
> b/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.c
> index ca61ac5e1983..1098d9501cc7 100644
> --- a/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.c
> +++ b/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.c
> @@ -891,7 +891,7 @@ NorFlashRead (
>    SEND_NOR_COMMAND (Instance->DeviceBaseAddress, 0, P30_CMD_READ_ARRAY);
> 
>    // Readout the data
> -  AlignedCopyMem (Buffer, (VOID *)StartAddress + Offset, BufferSizeInBytes);
> +  AlignedCopyMem (Buffer, (VOID *)(StartAddress + Offset), BufferSizeInBytes);
> 
>    return EFI_SUCCESS;
>  }

RVCT is correct here; standard C does not permit arithmetic on
pointers-to-void. From C99 for example,

    6.5.6 Additive operators
    ...
    Constraints

  2 For addition, either both operands shall have arithmetic type, or
    one operand shall be a pointer to an object type and the other
    shall have integer type. [...]

"Pointer to void" is not "pointer to object type".

Although in

    6.2.5 Types

we have

  27 A pointer to void shall have the same representation and alignment
     requirements as a pointer to a character type. [...]

but also:

  19 The void type comprises an empty set of values; it is an
     incomplete type that cannot be completed.

So, not an object type.

Can you submit this patch per se? It's a valid and justified change.

> diff --git a/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.inf
> b/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.inf
> index 812dafd065b2..0ef7b8d81bbc 100644
> --- a/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.inf
> +++ b/ArmPlatformPkg/Drivers/NorFlashDxe/NorFlashDxe.inf
> @@ -71,3 +71,7 @@ [Depex]
>    # NorFlashDxe must be loaded before VariableRuntimeDxe in case
> empty flash needs populating with default values
>    #
>    BEFORE gVariableRuntimeDxeFileGuid
> +
> +[BuildOptions]
> +  RVCT:*_*_*_CC_FLAGS = --diag_suppress=6314
> +
> """

No input on this though :)

> but these changes are entirely unrelated to the series, and I will
> follow up with some patches to fix this.
> 

Yes, please.

Thanks!
Laszlo