Re: [edk2-devel] EmulatorPkg: fixes for NetBSD compilation

[tlaronde@polynum.com]

Hello Pedro,

Le Mon, Nov 21, 2022 at 10:32:51PM +0000, Pedro Falcato a écrit :
> On Mon, Nov 21, 2022 at 9:21 PM <tlaronde@polynum.com> wrote:
> 
> > diff --git a/EmulatorPkg/Unix/Host/BlockIo.c
> > b/EmulatorPkg/Unix/Host/BlockIo.c
> > index cf2d6b4cda..c0c694be55 100644
> > --- a/EmulatorPkg/Unix/Host/BlockIo.c
> > +++ b/EmulatorPkg/Unix/Host/BlockIo.c
> > @@ -133,6 +133,20 @@ EmuBlockIoOpenDevice (
> >
> >        ioctl (Private->fd, DKIOCGETMAXBLOCKCOUNTWRITE,
> > &Private->Media->OptimalTransferLengthGranularity);
> >      }
> > + #elif _NETBSD_SOURCE
> > +    {
> > +      u_int  BlockSize;
> >
> Hi,
> 
> Again, thanks for the patches. Please send them in the way I kind of
> described in my other reply.
> 
> s/u_int/UINT/
> 
> +      off_t  DiskSize;
> >
> I think this off_t is fine, per the other off_t usages, I don't know if the
> maintainers agree.
> 
> > +
> > +      if (ioctl (Private->fd,  DIOCGSECTORSIZE, &BlockSize) == 0) {
> > +        Private->Media->BlockSize = BlockSize;
> > +      }
> > +
> > +      if (ioctl (Private->fd, DIOCGMEDIASIZE, &DiskSize) == 0) {
> > +        Private->NumberOfBlocks   = DivU64x32 (DiskSize,
> > (UINT32)BlockSize);
> > +        Private->Media->LastBlock = Private->NumberOfBlocks - 1;
> > +      }
> > +    }
> >   #else
> >      {
> >        size_t  BlockSize;
> >
> 
> 
> > diff --git a/EmulatorPkg/Unix/Host/Host.c b/EmulatorPkg/Unix/Host/Host.c
> > index 38c01c84af..c505300129 100644
> > --- a/EmulatorPkg/Unix/Host/Host.c
> > +++ b/EmulatorPkg/Unix/Host/Host.c
> > @@ -12,6 +12,12 @@ SPDX-License-Identifier: BSD-2-Clause-Patent
> >  #define MAP_ANONYMOUS  MAP_ANON
> >  #endif
> >
> > +#ifdef _NETBSD_SOURCE
> > +#define MAP_ANON_FD_ (-1)
> > +#else
> > +#define MAP_ANON_FD_ (0)
> > +#endif
> >
> Would there be a harm if we just passed -1 everywhere? It's a bit odd
> NetBSD explicitly requires this, but AFAIK implementations
> either EINVAL on fd != -1 or take whatever since it's anon. The main
> implementations (Linux, FreeBSD, NetBSD, OpenBSD, macOS) seem to
> agree that passing -1 should be safe everywhere (in fact, it's pretty funny
> that all BSD-derived implementations agree with the fd = -1 thing, but I
> digress).

Since MAP_ANON is not specified by POSIX1, I tend to think that what is
not in POSIX should be explicitely set in the code.

So if other OSes (with which edk2 is used) agree on (-1), my personal
taste would be to keep at least the define macro:

#define MAP_ANON_FD_ (-1)

as an indication that the definition is not standard (so that someone
with a system choking on this, could easily spot the culprit). (The
trailing underscore is because leading ones shall be reserved for
standard or system implementation, and the trailing is a way to indicate
that the macro is somehow local. Is there a convention for this in 
EDK2?)

Best,
-- 
        Thierry Laronde <tlaronde +AT+ polynum +dot+ com>
                     http://www.kergis.com/
                    http://kertex.kergis.com/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C