Fiber switching was disabled during destructor execution due to conflicts
with the garbage collector. This unfortunately introduces a function color
problem: destructors can not call functions that may switch Fibers.
In this change we update the GC so that Fiber switching during GC is safe. In
turn we allow Fiber switching during destrutor execution.
The GC executes destructors in a dedicated Fiber. If a destructor suspends, the
Fiber is owned by userland and a new dedicated Fiber is created to execute the
remaining destructors. Destructor suspension results in a resurection of the
object, which is handled as usual: The object is not considered garbage anymore,
but may be collected in a later run.
When the GC is executed in the main context (not in a Fiber), then destructors
are executed in the main context as well because there is no risk of conflicting
with GC in this case (main context can not suspend).
Fixes GH-11389
Closes GH-13460
You cannot return or yield a reference to a nullsafe chain. This was
checked already in zend_compile_return but not yet in
zend_compile_yield.
Closes GH-14716.
- PHP_CHECK_FUNC -> AC_SEARCH_LIBS
- Redundant symbols HAVE_LIBRT and HAVE_TIMER_CREATE removed
- The rt library for some older systems (Solaris <= 10, older Linux)
appended as needed
- This uses AC_ and AS_* macros and relies more on Autoconf shell
code handling
- Help texts updated and synced
- Check added into a separate M4 macro
- AC_CACHE_CHECK used for running the test program with cache variable
for optional edge-case overrides
- Help texts updated
- If check fails for some reason, the configure step emits error as
also done in the Zend C code
- Cross-compilation values updated with type casts as done in the
current conftest file output
- AC_CACHE_CHECK simplifies message checkings and can speed up the
consequtive configure -C checks
- AC_DEFINE descriptions style synced
- Used php_cv_* cache variables style
- Reduced over-quoted arguments to AC_LANG_PROGRAM
- Sync macro help text
This wraps the Zend signal handling check in its own M4 macro for
readability and uses AS_VAR_IF, AS_VAR_APPEND for CFLAGS variable,
updates help texts and fixes CS.
Values retrieved from zend_getenv should be freed.
Note: The only possible value for `zend_getenv` is `sapi_getenv` which uses
zend alloc to duplicate the string that it reads from the SAPI module.
Closes GH-14708.
This adds ZEND_CHECK_GLOBAL_REGISTER_VARIABLES M4 macro to properly
sort the --enable-gcc-global-regs in configure --help output, fixes
over-quoted arguments, syncs CS and updates help texts.
is_zend_ptr() expected zend_mm_heap.huge_list to be circular, but it's in fact NULL-terminated. It could crash when at least one huge block exists and the ptr did not belong to any block.
The --no-generation-date flag is a common re2c flag used in all re2c
invocations. This adds the 2nd optional argument to PHP_PROG_RE2C M4
macro in BC manner to set the default re2c command-line options and sets
the default RE2C_FLAGS similarly on Windows.
`start_memory_manager()` calls `zend_mm_init()` via `alloc_globals_ctor()`
before setting `REAL_PAGE_SIZE` to the right value. Moving the `REAL_PAGE_SIZE`
setting block before the call to `alloc_globals_ctor()` makes the allocator
behave properly on systems with a page size different than 4k.
Suggested-by: arnaud-lb
The m4_normalize is for Autoconf < 2.70 (on 2.70 and later versions a
blank-or-newline separated items can be expanded without using
backslash-newline).
* zend_compile: Rename `string_placeholder_count` to `placeholder_count` in `zend_compile_func_sprintf()`
This is intended to make the diff of a follow-up commit smaller.
* zend_compile: Add support for `%d` to `sprintf()` optimization
This extends the existing `sprintf()` optimization by support for the `%d`
placeholder, which effectively equivalent to an `(int)` cast followed by a
`(string)` cast.
For a synthetic test using:
<?php
$a = 'foo';
$b = 42;
for ($i = 0; $i < 100_000_000; $i++) {
sprintf("%s-%d", $a, $b);
}
This optimization yields a 1.3× performance improvement:
$ hyperfine 'sapi/cli/php -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php' \
'/tmp/unoptimized -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php'
Benchmark 1: sapi/cli/php -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php
Time (mean ± σ): 3.296 s ± 0.094 s [User: 3.287 s, System: 0.005 s]
Range (min … max): 3.213 s … 3.527 s 10 runs
Benchmark 2: /tmp/unoptimized -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php
Time (mean ± σ): 4.300 s ± 0.025 s [User: 4.290 s, System: 0.007 s]
Range (min … max): 4.266 s … 4.334 s 10 runs
Summary
sapi/cli/php -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php ran
1.30 ± 0.04 times faster than /tmp/unoptimized -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php
* Fix sprintf_rope_optimization_003.phpt test expecation for 32-bit integers
* zend_compile: Indent switch-case labels in zend_compile_func_sprintf()
* Add GMP test to sprintf() rope optimization
* Add `%s` test case to sprintf() GMP test
Inline the lookup whether a function is observed at all.
This strategy is also used for FRAMELESS calls. If the frameless call is observed, we instead allocate a call frame and push the arguments, to call the the function afterwards.
Doing so is still a performance benefit as opposed to executing individual INIT_FCALL+SEND_VAL ops. Thus, even if the frameless call turns out to be observed, the call overhead is slightly lower than before.
If the internal function is not observed at all, the unavoidable overhead is fetching the FLF zend_function pointer and the run-time cache needs to be inspected.
As part of this work, it turned out to be most viable to put the result operand on the ZEND_OP_DATA instead of ZEND_FRAMELESS_ICALL_3, allowing seamless interoperability with the DO_ICALL opcode.
This is a bit unusual in comparison to all other ZEND_OP_DATA usages, but seems to not pose problems overall.
There is also a small issue resolved: trampolines would always use the ZEND_CALL_TRAMPOLINE_SPEC_OBSERVER function due to zend_observer_fcall_op_array_extension being set to -1 too late.
* zend_compile: Add `zend_compile_rope_finalize()`
This just extracts the implementation as-is into a dedicated function to make
it reusable in preparation of a future commit.
* zend_compile: Use clearer parameter names for `zend_compile_rope_finalize()`
* zend_compile: Fix `zend_compile_rope_finalize()` for ropes containing a single constant string
Without this Opcache will trigger a use-after-free in
`zend_optimizer_compact_literals()`.
Co-authored-by: Ilija Tovilo <ilija.tovilo@me.com>
* zend_compile: Optimize `sprintf()` into a rope
This optimization will compile `sprintf()` using only `%s` placeholders into a
rope at compile time, effectively making those calls equivalent to the use of
string interpolation, with the added benefit of supporting arbitrary
expressions instead of just expressions starting with a `$`.
For a synthetic test using:
<?php
$a = 'foo';
$b = 'bar';
for ($i = 0; $i < 100_000_000; $i++) {
sprintf("%s-%s", $a, $b);
}
This optimization yields a 2.1× performance improvement:
$ hyperfine 'sapi/cli/php -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php' \
'/tmp/unoptimized -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php'
Benchmark 1: sapi/cli/php -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php
Time (mean ± σ): 1.869 s ± 0.033 s [User: 1.865 s, System: 0.003 s]
Range (min … max): 1.840 s … 1.945 s 10 runs
Benchmark 2: /tmp/unoptimized -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php
Time (mean ± σ): 4.011 s ± 0.034 s [User: 4.006 s, System: 0.005 s]
Range (min … max): 3.964 s … 4.079 s 10 runs
Summary
sapi/cli/php -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php ran
2.15 ± 0.04 times faster than /tmp/unoptimized -d zend_extension=php-src/modules/opcache.so -d opcache.enable_cli=1 test.php
This optimization comes with a small and probably insignificant behavioral
change: If one of the values cannot be (cleanly) converted to a string, for
example when attempting to insert an object that is not `Stringable`, the
resulting Exception will naturally not show the `sprintf()` call in the
resulting stack trace, because there is no call to `sprintf()`.
Nevertheless it will correctly point out the line of the `sprintf()` call as
the source of the Exception, pointing the user towards the correct location.
* zend_compile: Eagerly handle empty format strings in `sprintf()` optimization
* zend_compile: Add additional explanatory comments to zend_compile_func_sprintf()
* Add zero-argument test to sprintf_rope_optimization_001.phpt
---------
Co-authored-by: Ilija Tovilo <ilija.tovilo@me.com>
We keep track of free slots by organizing them in a linked list, with the
first word of every free slot being a pointer to the next one.
In order to make corruptions more difficult to exploit, we check the consistency
of these pointers before dereference by comparing them with a shadow. The shadow
is a copy of the pointer, stored at the end of the slot.
Before this change, an off-by-1 write is enough to produce a valid freelist
pointer. After this change, a bigger out of bound write is required for that.
The difficulty is increase further by mangling the shadow with a secret, and
byte-swapping it, which increases the minimal required out of bound write
length.
Closes GH-14054
These are either undefined or defined (to value 1):
- __DragonFly__
- __FreeBSD__
- HAS_MCAST_EXT
- HAVE_GETCWD
- HAVE_GETWD
- HAVE_GLIBC_ICONV
- HAVE_JIT
- HAVE_LCHOWN
- HAVE_NL_LANGINFO
- HAVE_RL_CALLBACK_READ_CHAR
- HAVE_RL_ON_NEW_LINE
- HAVE_SQL_EXTENDED_FETCH
- HAVE_UTIME
Follow up of GH-5526 (-Wundef)
GCC doesn't pessimize the error cases correctly:
https://godbolt.org/z/Pa6xsKMWc
This speeds up the happy case and in this case the code size is also
smaller, so it's a double-win.
