Nikita Popov [Fri, 26 Jun 2020 09:07:55 +0000 (11:07 +0200)]
Better leak fix for cgi -s / -w
We also need to go through request shutdown. The naming is a bit
confusing, but it's fine to go through fastcgi_request_done even
if not using fastcgi. Whether we loop or not is checked separately.
Nikita Popov [Sat, 15 Feb 2020 16:54:02 +0000 (17:54 +0100)]
Don't include trailing newline in comment token
Don't include a trailing newline in T_COMMENT tokens, instead leave
it for a following T_WHITESPACE token. The newline does not belong
to the comment logically, and this makes for an ugly special case,
as other tokens do not include trailing newlines.
Whitespace-sensitive tooling will want to either forward or backward
emulate this change.
Nikita Popov [Wed, 4 Mar 2020 10:52:55 +0000 (11:52 +0100)]
Make sorting stable
Make user-exposed sorts stable, by storing the position of elements
in the original array, and using those positions as a fallback
comparison criterion. The base sort is still hybrid q/insert.
The use of true/false comparison functions is deprecated (but still
supported) and should be replaced by -1/0/1 comparison functions,
driven by the <=> operator.
Nikita Popov [Thu, 25 Jun 2020 08:30:40 +0000 (10:30 +0200)]
Don't use iterator_funcs_ptr if it is null
This avoids ubsan warnings. Alternatively we could always initialize
iterator_funcs_ptr for aggregates, instead of doing so only for
non-internal ones.
Nikita Popov [Wed, 26 Feb 2020 15:42:49 +0000 (16:42 +0100)]
Introduce InternalIterator
Userland classes that implement Traversable must do so either
through Iterator or IteratorAggregate. The same requirement does
not exist for internal classes: They can implement the internal
get_iterator mechanism, without exposing either the Iterator or
IteratorAggregate APIs. This makes them usable in get_iterator(),
but incompatible with any Iterator based APIs.
A lot of internal classes do this, because exposing the userland
APIs is simply a lot of work. This patch alleviates this issue by
providing a generic InternalIterator class, which acts as an
adapater between get_iterator and Iterator, and can be easily
used by many internal classes. At the same time, we extend the
requirement that Traversable implies Iterator or IteratorAggregate
to internal classes as well.
Nikita Popov [Wed, 4 Mar 2020 09:53:47 +0000 (10:53 +0100)]
Make SimpleXMLElement a RecursiveIterator
Context: https://externals.io/message/108789
This essentially moves the functionality of SimpleXMLIterator into
SimpleXMLElement, and makes SimpleXMLIterator a no-op extension.
Ideally SimpleXMLElement would be an IteratorAggregate, whose
getIterator() method returns SimpleXMLIterator. However, because
SimpleXMLIterator extends SimpleXMLElement (and code depends on
this in non-trivial ways), this is not possible.
The only way to not keep SimpleXMLElement as a magic Traversable
(that implements neither Iterator nor IteratorAggregate) is to
move the SimpleXMLIterator functionality into it.
Alex Dowad [Tue, 23 Jun 2020 14:32:41 +0000 (16:32 +0200)]
Clean house in cryptographic hashing code
- Remove dead code from php_crypt_r.c
This code has been commented out since the file was added in 2008. It's safe to say
that no-one is ever going to use it.
- Fix typo in comment in php_crypt_r.c
- Remove redundant Windows-only implementation of php_md5_crypt_r
There is a portable implementation in the same file, which is selected if not
building for Windows. But why should Windows have its own special implementation
of this function at all? There doesn't seem to be any good reason.
Better to use the portable implementation on all platforms.
- Don't define useless __CONST macro in php_crypt_r.h
This preprocessor macro is not used anywhere.
- Add comment on functions for encoding data as Base64
- Remove dead code from crypt_blowfish.h
- Remove unneeded junk comments from crypt_freesec.c
From Microsoft's `SQLColAttribute()` documentation[1]:
| Please note that some drivers may only write the lower 32-bit or
| 16-bit of a buffer and leave the higher-order bit unchanged.
| Therefore, applications should initialize the value to 0 before
| calling this function.
Nikita Popov [Thu, 14 May 2020 09:51:36 +0000 (11:51 +0200)]
Add flag to forbid dynamic property creation on internal classes
While performing resource -> object migrations, we're adding
defensive classes that are final, non-serializable and non-clonable
(unless they are, of course). This path adds a ZEND_ACC_NO_DYNAMIC_PROPERTIES
flag, that also forbids the creation of dynamic properties on these objects.
This is a subset of #3931 and targeted at internal usage only
(though may be extended to userland at some point in the future).
It's already possible to achieve this (what the removed
WeakRef/WeakMap code does), but there's some caveats: First, this
simple approach is only possible if the class has no declared
properties, otherwise it's necessary to special-case those
properties. Second, it's easy to make it overly strict, e.g. by
forbidding isset($obj->prop) as well. And finally, it requires a
lot of boilerplate code for each class.
Fix #69804: ::getStaticPropertyValue() throws on protected props
`ReflectionClass` allows reading of the values of private and protected
constants, and also to get private and protected static methods.
Therefore getting the values of private and protected static properties
is also permissible, especially since `::getStaticProperties()` already
allows to do so.
We also allow ::setStaticPropertyValue() to modify private and
protected properties, because otherwise this method is useless, as
modifying public properties can be done directly.
Alex Dowad [Wed, 27 May 2020 04:01:22 +0000 (06:01 +0200)]
Simplify `_crypt_extended_init_r`, and fix redundant initialization on Win32/Solaris
Looking at the history of this function, the original implementation had a bug where
it would return from the middle of the function without unlocking the mutex first.
The author attempted to fix this by incrementing the `initialized` flag atomically,
which is not necessary, since the section which modifies the flag is protected by a
mutex.
Coincidentally, at the same time that all this unnecessary 'atomic' machinery was
introduced, the code was also changed so that it didn't return without unlocking the
mutex. So it looks like the bug was fixed by accident.
It's not necessary to declare the flag as `volatile` either, since it is protected
by a mutex.
Further, the 'fixed' implementation was also wrong in another respect: on Windows
and Solaris, the `initialized` flag was not even declared as `static`!! So the
initialization of the static tables for S-boxes, P-boxes, etc. was repeated on
each call to `php_crypt`, completely defeating the purpose of this function.
`atol()` returns a `long` which is not the same as `zend_long` on
LLP64; we use `ZEND_ATOL()` instead.
There is no need for a new test case, since filesize_large.phpt already
tests for that behavior; unfortunately, the FTP test suite relies on
`pcntl_fork()` and therefore cannot be run on Windows.
projects / php / log