A simple implementation compiles very well with gcc and clang on ppc and
x86, and will handle any-aligned input.

"engla" on IRC reported that ccanlint on linux/powerpc would loop
infinitely calling dep_failed after all tests are run and the score is
output.

Reproduced, and discovered that cannot_run() takes a container_of(),
except our top node is not a struct ccanlint.  The result was harmless on
x86, but set dep_failed to the return address on powerpc, causing that
to be called repeatedly.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

This is good form for datastructures, so caller can track weird bugs.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We use TAL_USE_TALLOC to use libtalloc as the backend: you can test that with
ccanlint --compiler="cc -DTAL_USE_TALLOC".
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We can't really tell if non-ccan dependencies are used, so don't try.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

David Gibson points out that a project may already be using talloc, so
this gives a simple adapter if one module uses tal.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

It may interfere with other at_exit() calls, so let them call it manually.
Also, use memset to zero, which really does make valgrind notice any leaks.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

In this case, CCAN_CFLAGS should be "".
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Since we never post-add a length property, we can save some cycles by
allocating it together with the object itself:

Before:
	$ ./samba-allocs talloc.dump --tal
	Tal time:                6058997-6215211(6102645)ns
	Tal_free time:           4791112-4968964(4824814)ns
	Single tal_free time:    3173647-3331584(3200183)ns
	$ ./samba-allocs talloc.dump --tal-size
	Virtual size = 9809920, RSS = 3952640

After:
	$ ./samba-allocs talloc.dump --tal
	Tal time:                5911994-6043808(5956914)ns
	Tal_free time:           4676026-4834598(4719908)ns
	Single tal_free time:    2888195-3067365(2922298)ns
	$ ./samba-allocs talloc.dump --tal-size
	Virtual size = 9809920, RSS = 3948544
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Since we added the length property, this penalizes us with an extra
alloc.  Unfortunately, we can't tell from the talloc.dump which are
array tallocs and which are single tallocs, but grepping the sources
gives a rate of 1 in 27.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

The first ok() line was actually failing, because sigaction returns 0
on success, but because the program exit(0) out the abort handler,
that wasn't being seen.

So remove tap and just abort and assert our way through the test,
which I think is clearer.

We use macros to mug the talloc entry points, since talloc_set_allocator()
is not in mainline talloc.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Now it's a useful little command line utility, with tests.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We weren't using the compiler and cflags options in tools/compile.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We can use tal_count.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

eg:
	tal/str: Module's tests cover all the code (tests_coverage): PASS (+3/6)
	tal/str: Total score: 55/58
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Slowly removing the talloc dependency.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Tests and example copied from talloc_link, which I also wrote.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

They are far too convenient, despite their inefficiency.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Shorter, sweeter, and less legacy.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Useful for joining two strings.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Helper to reallocate and append to an array.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

They don't need anything internal to tal, they're just helpers.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

It's short, and much clearer.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Unlike talloc, we don't have a count on everything, just on array
allocations.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

You don't need write access to the context to attach a destructor;
it's meta.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

$ for i in `seq 10`; do ./samba-allocs talloc.dump --tal; done | stats.c --trim-outliers
Before:
Read 25998 nodes
Tal time:                2857988-2981158(2904324)ns
Tal_free time:           2417170-2582088(2470931)ns
Single tal_free time:    1551654-1609583(1575984)ns

After:
Read 25998 nodes
Tal time:                2780974-2901323(2839352)ns
Tal_free time:           2241725-2355309(2285165)ns
Single tal_free time:    1549908-1612485(1566609)ns
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>