This regressed in 9a250cc9d6, which allowed static properties to get
overridden by a trait during inheritance. In particular, because of the
change to the loop in zend_update_parent_ce(), it's not guaranteed that
all indirects are after one another.
This means that during persisting the zvals of the static members table,
some static properties may be skipped. In case of the test code, this
means that the array in the trait will keep referring to the old, new
freed, stale value. To solve this, we check the type for IS_INDIRECT,
which is the same as what zend_persist_calc() is already doing anyway.
Since 2543e61aed we can check for IS_INDIRECT to see if it should be
persisted or not.
Closes GH-13794.
This fixes an edge case causing the GC to be triggered repeatedly.
Destructors might add potential garbage to the buffer, so it may happen that num_root it higher than gc_threshold after collection, thus triggering a GC run almost immediately. This can happen by touching enough objects in a destructor, e.g. by iterating over an array. If this happens again in the new run, and the threshold is not updated, the GC may be triggered again.
The edge case requires specific conditions to be triggered and it must happen rarely in practice:
* At least GC_THRESHOLD_TRIGGER (100) objects must be collected during each run for the threshold to not be updated
* At least GC_G(gc_threshold) (initially 10k) objects must be touched (decref'ed to n>0) by any destructor during each run to fill the buffer
The fix is to increase the threshold if GC_G(num_roots) >= GC_G(gc_threshold) after GC. The threshold eventually reaches a point at which the second condition is not met anymore.
The included tests trigger more than 200 GC runs before the fix, and 2 after the fix (dtors always trigger a second run).
A related issue is that zend_gc_check_root_tmpvars() may add potential garbage before the threshold is adjusted, which may trigger GC and exhaust the stack. This is fixed by setting GC_G(active)=1 around zend_gc_check_root_tmpvars().
`if (fpm_shm_size - size > 0)` will be rewritten by the compiler as this: `if (fpm_shm_size != size)`, which is undesirable. The reason this happens is that both variables are size_t, so subtracting them cannot be negative. The only way it can be not > 0, is if they're equal because the result will then be 0. This means that the else branch won't work properly. E.g. if `fpm_shm_size == 50` and `size == 51`, then `fpm_shm_size` will wraparound instead of becoming zero.
To showcase that the compiler actually does this, take a look at this
isolated case: https://godbolt.org/z/azobdWcrY. Here we can see the
usage of the compare instruction + cmove, so the "then" branch
is only done if the variables are equal.
Bumps the minimum required OpenSSL version from 1.0.2 to 1.1.1.
OpenSSL 1.1.1 is an LTS release, but has reached[^1] EOL from upstream. However, Linux distro/OS vendors
continue to ship OpenSSL 1.1.1, so 1.1.1 was picked as the minimum. The current minimum 1.0.2 reached
EOL in 2018.
Bumping the minimum required OpenSSL version makes it possible for ext-openssl to remove a bunch of
conditional code, and assume that TLS 1.3 (shipped with OpenSSL 1.1.1) will be supported everywhere.
- Debian buster: 1.1.1[^2]
- Ubuntu 20.04: 1.1.1[^3]
- CentOS/RHEL 7: 1.0.2
- RHEL 8/Rocky 8/EL 8: 1.1.1
- Fedora 38: 3.0.9 (`openssl11` provides OpenSSL 1.1 as well)
RHEL/CentOS 7 reaches EOL mid 2024, so for PHP 8.4 scheduled towards the end of this year, we can safely
bump the minimum OpenSSL version.
[^1]: https://www.openssl.org/blog/blog/2023/03/28/1.1.1-EOL/index.html
[^2]: https://packages.debian.org/buster/libssl-dev
[^3]: https://packages.ubuntu.com/focal/libssl-dev
And make locatable by via `php-config`. Prior to this, `libphp.*`
would always install to `$prefix/lib`. After this, they will install
to `$libdir`.
In practice, this will make it so that programs embedding libphp can
use `php-config` to determine appropriate compile flags without
guessing.
In `configure.ac`, it seems `$libdir` is mutated in some instances.
Ideally the mutated version would be stored in `$phplibdir` or
something. Instead of tracking down all uses of that variable, I
introduced another variable `$orig_libdir` that holds the original
value passed to the configure script.
This is a no-op for users unless they are compiling with `--libdir`
set to something other than `$prefix/lib`, the default.
Closes GH-12389
Make data_file.c's generator output its array initialization
"by list of strings", instead of "by list of chars": that makes the compiler
happier.
Use strtr() with a precomputed map, instead of a loop over ord(),
to generate in 1/100th the time.
Avoids ambiguous 0x tokens (as specified in C standards), by using octal.
Closes GH-10422.
Shared objects of extensions during the *nix build are copied to the
`modules` directory. It is a practice established since the early days
of the PHP build system. Other build systems may have similar concept of
"library destination directory". On Windows, they are put into the root
build directory. Such directory simplifies collection of the shared
extensions during testing, or when running the cli executable at the end
of the build process.
This change ensures that the directory is consistently created in a
single location, for both the primary PHP build process and when
utilizing `phpize` within community extensions.
The AC_CONFIG_COMMANDS_PRE is executed at the end of the configuration
phase, before creating the config.status script, where also build
directories and global Makefile are created.
The pwd is executed using the recommended $(...) instead of the obsolete
backticks. Autoconf automatically locates the proper shell and
re-executes the configure script if such case is found that $(...) is
not supported (the initial /bin/sh on Solaris 10, for example).
This removes the unused and obsolete Zend/zend_istdiostream.h header and
symbols on Windows:
- HAVE_STDIOSTR_H
- HAVE_CLASS_ISTDIOSTREAM
- istdiostream
