Re: [PATCH 01/35] DO NOT APPLY: edk2: turn standard handle types into pointers to non-VOID


Andrew Fish
 



On Sep 18, 2019, at 1:41 AM, Laszlo Ersek <lersek@...> wrote:

On 09/17/19 22:22, Andrew Fish wrote:


On Sep 17, 2019, at 1:06 PM, Ni, Ray <ray.ni@...> wrote:

Laszlo,
Thank you very much for this work.
They are quite helpful to detect potential issues.

But without this specific patch being checked in, future break will still happen.
I don't want it to be checked in ASAP because I know that there are quite a lot of close source code that may get build break due to this change.
Besides that, what prevent you make the decision to check in the changes?


Ray,

I was thinking the same thing. Could we make this an optional feature via a #define? We could always default to the Spec Behavior, and new projects could opt into the stricter version. 

#ifndef STRICTER_UEFI_TYPES
typedef VOID    *EFI_PEI_FV_HANDLE;
#else
struct EFI_PEI_FV_OBJECT;
typedef struct EFI_PEI_FV_OBJECT *EFI_PEI_FV_HANDLE;
#endif

Technically, this would work well.

However, if we wanted to allow new projects to #define
STRICTER_UEFI_TYPES as their normal mode of operation (and not just for
a sanity check in CI), then we'd have to update the UEFI spec too.

Otherwise, code that is technically spec-conformant (albeit semantically
nonsensical), like I mentioned up-thread, would no longer compile:


Laszlo,

I think helping people NOT write nonsensical code is good. It is very good idea and I'd like to add it to the spec but as you point out it would break a lot of existing code so I'm not sure it is possible. I guess we could try to add a strict mode to the spec but given the types are defined in tables that may be problematic. 

We have coding standards that are more strict than what the C spec allows. So I would see the STRICT_UEFI_TYPES as more of a enforce the coding standard kind of thing? 

Thanks,

Andrew Fish

 EFI_HANDLE Foobar;
 UINT64     Val;

 Foobar = &Val;

Thanks
Laszlo


Join devel@edk2.groups.io to automatically receive all group messages.