No idea if Mac or FreeBSD has renameat2(), but glibc refusing to wrap a 10
year old Linux syscal...

We removed catv, less was listed twice, and dd was promoted.

Oliver Webb noticed that selecting allnoconfig then enabling just sh
(instead of "make sh" singl...

In commit 2c96060, the tests for getfattr and setfattr were updated to
only query specific names...

The test system may have extended attributes set by default (for
instance, security.selinux on A...

(If we exit instead of return from scripts/test.sh the subshell doesn't
marshall $FAILCOUNT into...

Remove trailing spaces and leading tabs that crept in.

This version runs every time (much faster), and only includes enabled
structures so the union's ...

Android includes a new note in of type NT_ANDROID_TYPE_PAD_SEGMENT [1],
make readelf -n able to ...

We need to flush after each batch in batch mode.

(Found by inspection, looking for sleeps simil...

It's not scientific notation, it's "exabyte suffix" and then the check
for trailing 'd' meaning ...

using $SED instead of sed anyway (and thus gsed instead of ancient sed
on MacOS). Also, yank chec...

Since this is my second attempt over a fairly short period of time,
I'm not 100% confident this ...

The ULONG_MAX turns into -1 here, which isn't what was intended.
I'll avoid the bikeshed of the ...

now on top of Elliott's.

Ensure that the usage lines include everything mentioned below in
the detailed help, and vice ve...

Add -M option to switch off mlock(), and only touch start of each page
to dirty them (leaving th...

This seems to be a popular reinvention. There are three different
variants in the Android tree, ...

and yank an obsolete comment Elliott spotted.

change xputs*() to fwrite() instead of write() but add a flush to maintain
previous semantics. Th...

I have no idea why Linux 6.7 decided the dtb path now needs "arm" it _twice_,
but the move broke...

(Same as the ifconfig change, but in a separate patch since this is in
pending and more likely t...

I haven't seen these in the wild, but happened to be dealing with the
corresponding header file ...

ASAN finds leaks. The gcc array overlap detector isn't even close to sane,
and we have to switch...

The code is correct (and doesn't seem to bother clang), but gcc can't
tell that we don't touch o...

This doesn't handle refilling toybuf in the case where there's so
much EXIF data before the head...

on 32/64 bit, and switch some 64 bit constants to ULL for 32 bit.

If you build kernel modules, it archives them up separately and then appends
them to the initram...

automatically enable module plumbing if any $MODULES selected.

The fallback ftruncate() call was wrong, because posix_fallocate()
will never shrink a file, but...

Because we statically link the binaries against musl-libc, we need a musl-libc
toolchain for the...

Test with "skeleton -c 8g" for example.

The array of section flags was missing the G flag, so any flags
after it were displayed incorrec...

We're not specifically looking for armv8; we want any aarch64.

I've also changed the 32-bit arm...

select package versions. This means it now builds arm hosted toolchains on
arm, builds x86-64 (no...

(The explicit #include is because musl and bionic don't.)

and move findglobals and showasm into it.

The shell frees it after running a NOFORK and the minor memory leak in
a "nice nohup setsid task...

(It's also kind of silly: it makes uname -m produce different output.
That's it. It doesn't stop...