Add FreeBSD armv6 and armv7 targets

Bump to 0.2.48

Bumps libc version to 0.2.48. This release contains the breaking changes performed in #1222 .

fix stack_t on FreeBSD

FreeBSD 10.x is EOL, in FreeBSD 11 and later, ss_sp is actually a void* [1]
dragonflybsd still uses c_char [2]

[1] https://svnweb.freebsd.org/base/releng/11.2/sys/sys/signal.h?revision=334459&view=markup#l438
[2] https://github.com/DragonFlyBSD/DragonFlyBSD/blob/master/sys/sys/signal.h#L339

should be committed with https://github.com/rust-lang/rust/pull/57810

dragonflybsd still uses c_char [2]

[1] https://svnweb.freebsd.org/base/releng/11.2/sys/sys/signal.h?revision=334459&view=markup#l438
[2] https://github.com/DragonFlyBSD/DragonFlyBSD/blob/master/sys/sys/signal.h#L339

Re-enable android build jobs

This performs a `yes | ./sdk/tools/bin/sdkmanager --licenses` - the sole purpose of the `--licenses` flag is to allow the user to accept all licenses a priori, so that all other commands don't show any licenses.

For some reason, after all licenses have been accepted, the install step says that some licenses are still not accepted... so we need to accept those too... Why? I have no idea, and accepting licenses while installing still appears to make things hang a little, but it only does so for one non-accepted license, instead of for 5-6 like before, so now things do work.

Closes #1170 .

Fix CMSG_DATA(3) and friends on BSD

PR #1098 added the CMSG_DATA(3) family of functions into libc.  Because
they're defined as macros in C, they had to be rewritten as Rust
functions for libc.  Also, they can't be tested in CI for the same
reason.  But that PR erroneously used the same definitions in BSD as in
Linux.

This commit corrects the definitions for OSX, FreeBSD, DragonflyBSD,
OpenBSD, and NetBSD.  I renamed a few variables and collapsed a few
macros in order to combine the definitions where possible.

Fixes #1210

PR #1098 added the CMSG_DATA(3) family of functions into libc.  Because
they're defined as macros in C, they had to be rewritten as Rust
functions for libc.  Also, they can't be tested in CI for the same
reason.  But that PR erroneously used the same definitions in BSD as in
Linux.

This commit corrects the definitions for OSX, FreeBSD, DragonflyBSD,
OpenBSD, and NetBSD.  I renamed a few variables and collapsed a few
macros in order to combine the definitions where possible.

Fixes #1210

Fix linking aio_read(3) and friends on NetBSD

The aio functions require librt on NetBSD.

Remove fexecve from netbsdlike as it's not implemented

Sometimes it causes confusion in downstream users of libc, for example it caused nix-rust/nix to fail to compile on OpenBSD 6.4 (see https://github.com/nix-rust/nix/pull/1000).

OpenBSD doesn't implement fexecve. The only reference of it that I can
find in the OpenBSD source is in the man pages of signal(3) and
sigaction(2) (where it's mentioned that it is not implemented).

OpenBSD official source code link:
https://cvsweb.openbsd.org/src/lib/libc/sys/sigaction.2?rev=1.75&content-type=text/x-cvsweb-markup

OpenBSD Github mirror:
https://github.com/openbsd/src/blob/master/lib/libc/sys/sigaction.2#L619

On NetBSD's unistd.h I see that it is under an ifdef. Calling it returns
78 / ENOSYS / Function not implemented.

NetBSD office source code link:
http://cvsweb.netbsd.org/bsdweb.cgi/src/include/unistd.h?rev=1.151&content-type=text/x-cvsweb-markup&only_with_tag=MAIN

NetBSD Github mirror:
https://github.com/NetBSD/src/blob/trunk/include/unistd.h#L319

Tests on OpenBSD 6.4 after the change:
```bash
user@openbsd64 ~/RUST/libc $ cargo build
   Compiling libc v0.2.46 (/home/user/RUST/libc)
    Finished dev [unoptimized + debuginfo] target(s) in 3.88s

user@openbsd64 ~/RUST/libc $ cargo build --release
   Compiling libc v0.2.46 (/home/user/RUST/libc)
    Finished release [optimized] target(s) in 2.21s

user@openbsd64 ~/RUST/libc/libc-test $ cargo test
   Compiling proc-macro2 v0.4.24
   Compiling unicode-xid v0.1.0
   Compiling semver-parser v0.7.0
   Compiling serde v1.0.84
   Compiling libc v0.2.45
   Compiling num-traits v0.2.6
   Compiling ryu v0.2.7
   Compiling cfg-if v0.1.6
   Compiling itoa v0.4.3
   Compiling term v0.4.6
   Compiling bitflags v0.9.1
   Compiling cc v1.0.28
   Compiling libc v0.2.46 (/home/user/RUST/libc)
   Compiling log v0.4.6
   Compiling semver v0.9.0
   Compiling log v0.3.9
   Compiling rustc_version v0.2.3
   Compiling rand v0.4.3
   Compiling extprim v1.6.0
   Compiling quote v0.6.10
   Compiling syn v0.15.23
   Compiling serde_derive v1.0.84
   Compiling syntex_pos v0.59.1
   Compiling serde_json v1.0.34
   Compiling syntex_errors v0.59.1
   Compiling syntex_syntax v0.59.1
   Compiling ctest v0.2.8
   Compiling libc-test v0.1.0 (/home/user/RUST/libc/libc-test)
    Finished dev [unoptimized + debuginfo] target(s) in 2m 16s
     Running /home/user/RUST/libc/target/debug/deps/linux_fcntl-08861a7cd96d1b94
RUNNING ALL TESTS
PASSED 0 tests
     Running /home/user/RUST/libc/target/debug/deps/main-57e266d38aa58cce
RUNNING ALL TESTS
PASSED 7187 tests
```
I've tried running the tests on a NetBSD 8.0 box unfortunately there `libc-test` fails to compile even before this change.

NetBSD 8 adds new address family changing AF_MAX and new ifaddrs flags.

libc-test on NetBSD 8.0  fails without these changes.
ifa_addrflags was added to struct ifaddrs to support AF_CAN.
AF_MAX got bumped up one by AF_CAN.
RT_IFLIST format changed to support ifa_addrflags.

The aio functions require librt on NetBSD.

Bump version to 0.2.47

This will release changes to the Redox module

mincore has been removed from upcoming OpenBSD 6.5

remove test from CI on OpenBSD for "mincore" function.

it has been removed from upcoming 6.5:
- https://marc.info/?l=openbsd-cvs&m=154723400730318
- https://github.com/openbsd/src/commit/54e4f6b9a1dc183e1dc7c4a722c2646aa309246a

I am only removing the test from CI, not touching exported function from Rust libc to avoid a breaking change.

Redox module update

This adds poll, mmap, and stat support

Comment FIOSEEK* until they're in a NetBSD release

Accidentally included two constants that aren't in the current release of NetBSD.

While this doesn't impact libc-consumers, failing the testsuite on latest NetBSD isn't ideal.

I've commented them out; they can be reintroduced when the next NetBSD ships with them included.

glibc x86: add the register offset constants for mcontext_t.gregs

Constants calculated from the [glibc source](https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/x86/sys/ucontext.h;h=3d883c2bf1220ef50cedc6b1491d699c0727c508;hb=HEAD). These are needed when using `mcontext_t`. Thanks!

add FIO* for FreeBSD and NetBSD

Calculated [here](https://repl.it/repls/PaltryTroubledMonitors) from FreeBSD [source](https://github.com/freebsd/freebsd/blob/9867e6d8fe1cac34121e5f2df9357363f5659617/sys/sys/filio.h).

Calculated [here](https://repl.it/repls/ProfitableImpracticalComputeranimation) from NetBSD [source](https://github.com/NetBSD/src/blob/64b8a48e1288eb3902ed73113d157af50b2ec596/sys/sys/filio.h).

Some of these are likely identical across all BSDs – if this can be confirmed they could be elevated to `unix/bsd/mod.rs` alongside `FIOCLEX` and `FIONBIO`.

Remove sys/ioctl_compat.h from DragonFlyBSD tests.

Per the mailing list[1], sys/ioctl_compat.h has been removed.  This should fix #1198.

1: http://lists.dragonflybsd.org/pipermail/commits/2018-April/672079.html

OpenBSD doesn't implement fexecve. The only reference of it that I can
find in the OpenBSD source is in the man pages of signal(3) and
sigaction(2) (where it's mentioned that it is not implemented).

OpenBSD official source code link:
https://cvsweb.openbsd.org/src/lib/libc/sys/sigaction.2?rev=1.75&content-type=text/x-cvsweb-markup

OpenBSD Github mirror:
https://github.com/openbsd/src/blob/master/lib/libc/sys/sigaction.2#L619

On NetBSD's unistd.h I see that it is under an ifdef. Calling it returns
78 / ENOSYS / Function not implemented.

NetBSD office source code link:
http://cvsweb.netbsd.org/bsdweb.cgi/src/include/unistd.h?rev=1.151&content-type=text/x-cvsweb-markup&only_with_tag=MAIN

NetBSD Github mirror:
https://github.com/NetBSD/src/blob/trunk/include/unistd.h#L319

Per the mailing list[1], sys/ioctl_compat.h has been removed.

1: http://lists.dragonflybsd.org/pipermail/commits/2018-April/672079.html

Fix uname on FreeBSD

On FreeBSD, uname is an inline function.  The uname that is present in
libc.so is for FreeBSD 1.0 compatibility.  It expects a buffer of a
different size.

Fixes #1190
Reported-by: Alex Zepeda