inteltool and sys/io.h

[Trammell Hudson]

When cross compiling inteltool with musl-libc the <sys/io.h> header
is not included due to this test in inteltool.h:

#if defined(__GLIBC__)
#include <sys/io.h>
#endif

Unfortunately I'm not sure what the right test is here, since
the musl libc team is opposed to having a __MUSL__ define:

http://openwall.com/lists/musl/2013/03/29/13

[ron minnich]

why on earth is that test there? what does glibc have to do with whether
you have sys/io.h?

I suggest removing the guard.

On Wed, Feb 14, 2018 at 10:12 AM Trammell Hudson  wrote:
> When cross compiling inteltool with musl-libc the <sys/io.h> header
> is not included due to this test in inteltool.h:
>
> #if defined(__GLIBC__)
> #include <sys/io.h>
> #endif
>
> Unfortunately I'm not sure what the right test is here, since
> the musl libc team is opposed to having a __MUSL__ define:
>
> http://openwall.com/lists/musl/2013/03/29/13

[Jonathan A. Kollasch]

On Wed, Feb 14, 2018 at 11:11:46AM -0700, Trammell Hudson wrote:
> When cross compiling inteltool with musl-libc the <sys/io.h> header
> is not included due to this test in inteltool.h:
>
> #if defined(__GLIBC__)
> #include <sys/io.h>
> #endif

We should provide our own I/O space inlines.  ISTR we do for nvramtool,
superiotool, and flashrom.

For instance, NetBSD libc will not expose these inlines to userland.

This is to say nothing of the NetBSD x86 I/O space write inlines having
reversed argument order compared to their glibc counterparts.

Jonathan Kollasch

[Klemens Nanni]

On Wed, Feb 14, 2018 at 11:11:46AM -0700, Trammell Hudson wrote:
> When cross compiling inteltool with musl-libc the <sys/io.h> header
> is not included due to this test in inteltool.h:
>
> #if defined(__GLIBC__)
> #include <sys/io.h>
> #endif

Same story on other platforms: sys/io.h[1] is implemented through
i386_iopl(2)[2]/amd64_iopl(2)[3] and sysarch(2)[4] on OpenBSD.

Many tools under util/ are developed with Linux/glibc in mind (only).
> Unfortunately I'm not sure what the right test is here, since
> the musl libc team is opposed to having a __MUSL__ define:
>
> http://openwall.com/lists/musl/2013/03/29/13

1: http://man7.org/linux/man-pages/man2/iopl.2.html
2: http://man.openbsd.org/OpenBSD-6.0/./man2/sysarch.2
3: http://man.openbsd.org/OpenBSD-6.0/./man2/i386/i386_iopl.2
4: http://man.openbsd.org/OpenBSD-6.0/./man2/amd64/amd64_iopl.2