* Refactor utsname.domainname struct member Autoconf check
Autoconf's AC_CHECK_MEMBERS macro (available since Autoconf 2.50) can be
used instead of the compile check. This was originally implemented for
IRIX compatibility, when Autoconf 2.13 didn't have the struct members
checking macro yet.
Macro by default here defines the HAVE_STRUCT_UTSNAME_DOMAINNAME symbol.
* Remove also redundant DARWIN symbol check
Checking in the configuration step also correctly detects missing struct
member on Darwin systems (macos...).
The imap extension has been moved to PECL. Adding a note about configure
options changes also in the UPGRADING.INTERNALS for easier build scripts
adjustments.
This allows consumers of just the CSPRNG to include a much smaller header. It
also allows to verify at a glance whether a source file might use non-secure
randomness.
This commit includes the new header wherever the CSPRNG is used, possibly
replacing the inclusion of php_random.h if nothing else is used, but also
includes it in the main php_random.h header for compatibility.
Somewhat related to 45f8cfaf10,
2b30f18708, and
b14dd85dca.
When ext/random was initially added in PHP 8.2, these headers started
“forwarding” to the new ext/random/php_random.h to reduce the impact on
existing extensions.
As master already contains some breaking changes of the internal API of
ext/random and as the last PHP version without ext/random will be EOL once
master is released, it appears appropriate to drop these headers now.
Instead of returning the generated `uint64_t` and providing the size (i.e. the
number of bytes of the generated value) out-of-band via the
`last_generated_size` member of the `php_random_status` struct, the `generate`
function is now expected to return a new `php_random_result` struct containing
both the `size` and the `result`.
This has two benefits, one for the developer:
It's no longer possible to forget setting `last_generated_size` to the correct
value, because it now happens at the time of returning from the function.
and the other benefit is for performance:
The `php_random_result` struct will be returned as a register pair, thus the
`size` will be directly available without reloading it from main memory.
Checking a simplified version of `php_random_range64()` on Compiler Explorer
(“Godbolt”) with clang 17 shows a single change in the resulting assembly
showcasing the improvement (https://godbolt.org/z/G4WjdYxqx):
- add rbp, qword ptr [r14]
+ add rbp, rdx
Empirical testing confirms a measurable performance increase for the
`Randomizer::getBytes()` method:
<?php
$e = new Random\Engine\Xoshiro256StarStar(0);
$r = new Random\Randomizer($e);
var_dump(strlen($r->getBytes(100000000)));
goes from 250ms (before the change) to 220ms (after the change). While
generating 100 MB of random data certainly is not the most common use case, it
confirms the theoretical improvement in practice.
This shrinks the struct from 80 bytes to 72 bytes.
This was unused internally, I did not find users externally via GitHub
search.
The intention for this was that it could be used for attaching extra
data as a 3rd party to a node. However, there are better mechanisms for
that like using actual objects.
* Update signature of pcre API
This changes the variables that are bools to actually be bools instead
of ints, which allows some additional optimization by the compiler (e.g.
removing some ternaries and move extensions).
It also gets rid of the use_flags argument because that's just the same
as flags == 0. This reduces the call frame.
* Use zend_string_release_ex where possible
* Remove duplicate symbols from strchr
* Avoid useless value conversions
* Use a raw HashTable* instead of a zval
* Move condition
* Make for loop cheaper by reusing a recently used value as start iteration index
* Remove useless condition
This can't be true if the second condition is true because it would
require the string to occupy the entire address space.
* Upgrading + remark
Commit d86314939c added an additional parameter to Z_PARAM_FUNC_EX.
To maintain compatibility with third-party extensions, we keep
Z_PARAM_FUNC_EX as it used to be, and add Z_PARAM_FUNC_EX2 instead.
This macro is no longer used within php-src since
60ace13f9c, it invokes undefined behavior
depending on the input and the corresponding MT_RAND_PHP mode was deprecated in
PHP 8.3.
Thus remove this macro. Any remaining non-php-src user should just inline it
into their code, but should ideally migrate to a non-biased scaler. In any case
the undefined behavior of the original implementation should be accounted for.
As noted in glibc, the cookie_io_functions_t should be used instead of
internal _IO_cookie_io_functions_t.
The _IO_cookie_io_functions_t was once used as a convenience for not
having the cookie_io_functions_t available (glibc <=2.1.1) as noted in
67bb9d1ae2.
Check in the build system was also always setting the
COOKIE_IO_FUNCTIONS_T to cookie_io_functions_t due to a typo. There is
unused variable have_IO_cookie_io_functions_t vs.
have_cookie_io_functions_t.
- COOKIE_IO_FUNCTIONS_T removed
Closes GH-12236
When we try to load an extension multiple times, we still overwrite the
type, module number, and handle. If the module number is used to
indicate module boundaries (e.g. in reflection and in dom, see e.g.
dom_objects_set_class_ex), then all sorts of errors can happen.
In the case of ext/dom, OP's error happens because the following
happens:
- The property handler is set up incorrectly in
dom_objects_set_class_ex() because the wrong module number is
specified. The class highest in the hierarchy is DOMNode, so the
property handler is incorrectly set to that of DOMNode instead of
DOMDocument.
- The documentElement property doesn't exist on DOMNode, it only exists
on DOMDocument, so it tries to read using zend_std_read_property().
As there is no user property called documentElement, that read
operation returns an undef value.
However, the type is still checked, resulting in the strange exception.
Solve this by changing the API such that the data is only overwritten if
it's owned data.
Closes GH-12246.
When building from sources, someone distributing PHP may want to add a
vendor specific string to the PHP_VERSION so users can differentiate
multiple vendor builds from the same PHP version. For instance, a vendor
backporting a bug fix to a no-longer-supported PHP version could extend
their PHP_EXTRA_VERSION to allow their users to identify that they carry
such fix by checking their PHP_VERSION.
Closes GH-11706
* Implement iteration cache, item cache and length cache for node list iteration
The current implementation follows the spec requirement that the list
must be "live". This means that changes in the document must be
reflected in the existing node lists without requiring the user to
refetch the node list.
The consequence is that getting any item, or the length of the list,
always starts searching from the root element of the node list. This
results in O(n) time to get any item or the length. If there's a for
loop over the node list, this means the iterations will take O(n²) time
in total. This causes real-world performance issues with potential for
downtime (see GH-11308 and its references for details).
We fix this by introducing a caching strategy. We cache the last
iterated object in the iterator, the last requested item in the node
list, and the last length computation. To invalidate the cache, we
simply count the number of modifications made to the containing
document. If the modification number does not match what the number was
during caching, we know the document has been modified and the cache is
invalid. If this ever overflows, we saturate the modification number and
don't do any caching anymore. Note that we don't check for overflow on
64-bit systems because it would take hundreds of years to overflow.
Fixes GH-11308.
I was surprised to see that getting the stricterror property showed in
in the Callgrind profile of some tests. Turns out we sometimes allocate
them. Fix this by returning the default in case no changes were made yet.
Closes GH-11345.
Struct members require some alignment based on their type. This means
that if a struct member is not aligned, there will be a hole created by
the compiler in the struct, which is wasted space. This patch reorders
some of the most commonly used structs, but in such a way that the
fields which were in the same cache line still belong together.
The only exception to this is exception_ignore_args, which was
temporally not close to nearby members, and as such I placed
it further up to close a hole.
On 64-bit Linux this gives us the following shrinks:
* zend_op_array: 248 -> 240
* zend_ssa_var: 56 -> 48
* zend_ssa_var_info: 48 -> 40
* php_core_globals: 672 -> 608
* zend_executor_globals: 1824 -> 1792
On 32-bit, the sizes will either remain the same or will result in
smaller shrinks.
According to @nikic:
> The current guideline for use of bool and zend_result in php-src is
> that bool is an appropriate return value for "is" or "has" style
> functions, which return a yes/no answer. zend_result is an
> appropriate return value for functions that perform some operation
> that may succeed or fail.
Closes GH-10622.
We can't increase the refcount of internal classes during request time.
To work around this problem we simply don't refcount aliases anymore and
add a check in the destruction to skip aliases entirely.
There were also some checks which checked for an alias implicitly by
comparing the refcount, these have been replaced by checking the type of
the zval instead.
