copy_file_range() is a Linux-specific system call which allows
efficient copying between two file descriptors, eliminating the need
to transfer data from the kernel to userspace and back. For
networking file systems like NFS and Ceph, it even eliminates copying
data to the client, and local filesystems like Btrfs and XFS can
create shared extents.
On linux -1 means system administrator choosen default or kernel
default, this varies between kernel versions or distributions
it used to be 128 now it is 4096.
Nitpicking, this makes it practically impossible to accidentally use a number with 2 bits set in the future,
it's also my personal preference, its trivial to see "PHP_FILE_NO_DEFAULT_CONTEXT use bit 4" and that "the next unused bit is bit 5"
Closes GH-8233
This fix corrects a behavior of `var_export()` that was mostly "hidden" until PHP 8.1 introduced:
* properties with object initializers
* constants containing object references
* default values of class properties containing `enum`s
Since `var_export(..., true)` is mostly used in conjunction with code generation,
and we cannot make assumptions about the generated code being placed in the root
namespace, we must always provide the FQCN of a class in exported code.
For example:
```php
<?php
namespace MyNamespace { class Foo {} }
namespace { echo "<?php\n\nnamespace Example;\n\n" . var_export(new \MyNamespace\Foo(), true) . ';'; }
```
produces:
```php
<?php
namespace Example;
MyNamespace\Foo::__set_state(array(
));
```
This code snippet is invalid, because `Example\MyNamespace\Foo::__set_state()` (which
does not exist) is called.
With this patch applied, the code looks like following (valid):
```php
<?php
namespace Example;
\MyNamespace\Foo::__set_state(array(
));
```
Ref: https://github.com/php/php-src/issues/8232
Ref: https://github.com/Ocramius/ProxyManager/issues/754
Ref: https://externals.io/message/117466
This change introduces subsequent kill of the process when idle process quit
(SIGQUIT) does not succeed. It can happen in some situations and means that FPM
is not able to scale down in dynamic pm. Using SIGKILL fixes the issue.
Recorded errors may be attached to the wrong cached script when a fatal error
occurs during recording. This happens because the fatal error will cause a
bailout, which may prevent the recorded errors from being freed. If an other
script is compiled after bailout, or if a class is linked after bailout, the
recorded errors will be attached to it.
This change fixes this by freeing recorded errors before executing shutdown
functions.
Fixes GH-8063
`CURLOPT_XFERINFOFUNCTION` is available as of cURL 7.32.0, and
supersedes `CURLOPT_PROGRESSFUNCTION` which is still supported by
latest cURL, though.
Closes GH-7823.
I don't think this is strictly necessary, but I think it makes
sense to check this before interpreting opline->op2.num as an
argument number.
This also adds one more has_known_send_mode() check that I had
missed before.
Abstract functions are now handled in the INIT stage, they do not
affect DO opcodes anymore.
Also add a check for another precondition, namely that the function
is not a trampoline.
We may sometimes create FETCH_*_R opcodes with VAR type (e.g. if a
FUNC_ARG opcode is converted to BP_VAR_R fetch kind). Make sure we
don't infer overly conservative types in that case.
