Previusly, to redce stack usage GC traversed object properties and array
elements in both directions. At first, it seeked backward from the last
element, to find the last refcounted zval. Then, it traversed elements
forward. As result, we had 2 non-spatial memory accesses for each
object/array and usuall 2 expensive CPU cache misses.
Now all the traversal algorithms are refactored to scan elements only in
forward direction. The number of checks and GC stack usage should be the
same. The order of element processing may be differenr, but this
should not be a problem.
For parity with non Windows systems, we should `AC_DEFINE` these macros
so that they are defined in config.w32.h instead of directly passed to
the compiler. This allows extensions which have ext/gd as dependency
to check for these macros. E.g. ext/ps does this[1], and fails to
build on Windows, because `HAVE_GD_BUNDLED` isn't defined.
[1] <https://github.com/steinm/ps/blob/RELEASE_1_4_4/ps.c#L34>
Closes GH-7680.
On x86_64 glibc memrchr() uses SSE/AVX CPU extensions and works much
faster then naive loop. On x86 32-bit we still use inlined version.
memrchr() is a GNU extension. Its prototype becomes available when
<string.h> is included with defined _GNU_SOURCE macro. Previously, we
defined it in "php_config.h", but some sources may include <string.h>
befire it. To avod mess we also pass -D_GNU_SOURCE to C compiler.
While testing the cPanel usage of PHP-FPM, we stumbled on this bug.
Without the fix, the zend_string is corrupted and getting odd filenames
When using FPM we kept getting "No input file specified".
I work for cPanel and we use PHP extensively.
Skip WeakMap lookup check used only for debug assertion in non-debug builds
Use the `zend_hash_lookup` helper to optimize adding a WeakMap entry
if the entry doesn't already exist.
Closes GH-7670
GMP directly implements internal serialize/unserialize handlers
rather than going through the Serializable interface, so it ended
up being missed when adding the new __serialize()/__unserialize()
methods to other classes.
The serialization format is similar to before, but uses hex instead
of decimal encoding and omits the members if not used (which should
be almost always).
In this case we produce additional warnings when preloading
evaluates constants. Possibly we shouldn't be evaluating expressions
with side-effects during preloading, but for now just skip this
test.
Dmitry Stogov