Anatol Belski [Thu, 12 Dec 2013 09:17:01 +0000 (10:17 +0100)]
Fixed bug #65486 mysqli_poll() is broken on Win x64
While this issue is visible in mysqli_poll() functions, the cause
lays deeper in the stream to socket casting API. On Win x64 the
SOCKET datatype is a 64 or 32 bit unsigned, while on Linux/Unix-like
it's 32 bit signed integer. The game of casting 32 bit var to/from
64 bit pointer back and forth is the best way to break it.
Further more, while socket and file descriptors are always integers
on Linux, those are different things using different APIs on Windows.
Even though using integer instead of SOCKET might work on Windows, this
issue might need to be revamped more carefully later. By this time
this patch is tested well with phpt and apps and shows no regressions,
neither in mysqli_poll() nor in any other parts.
Ard Biesheuvel [Tue, 10 Dec 2013 11:07:46 +0000 (12:07 +0100)]
Zend: fix overflow handling bug in non-x86 fast_add_function()
The 'result' argument of fast_add_function() may alias with either
of its operands (or both). Take care not to write to 'result' before
reading op1 and op2.
Functions registered using zend_register_functions instead of zend_module_entry.functions are not seen on reflection.
Ex: additional_functions from api_module_entry.
Ex: in CLI, dl, cli_set_process_title and cli_get_process_title
Note:
- also affects functions overrided in extension
(should be be reported in extension, where overrided, not in original extension)
- also allow extension to call zend_register_functions for various list
(instead of having a single bug list)
Dmitry Stogov [Tue, 10 Dec 2013 12:08:46 +0000 (16:08 +0400)]
Merge branch 'PHP-5.5' into PHP-5.6
* PHP-5.5:
Revered func_get_args() changes, because they were not completly transparent and broke some tricky PHP code that worked because of luck before php-5.5.6 (see Bug #66107). The changes are going to be kept in PHP-5.6.
Dmitry Stogov [Tue, 10 Dec 2013 12:02:59 +0000 (16:02 +0400)]
Revered func_get_args() changes, because they were not completly transparent and broke some tricky PHP code that worked because of luck before php-5.5.6 (see Bug #66107). The changes are going to be kept in PHP-5.6.
Ard Biesheuvel [Tue, 10 Dec 2013 11:07:46 +0000 (12:07 +0100)]
Zend: fix overflow handling bug in non-x86 fast_add_function()
The 'result' argument of fast_add_function() may alias with either
of its operands (or both). Take care not to write to 'result' before
reading op1 and op2.
Dmitry Stogov [Tue, 10 Dec 2013 10:19:17 +0000 (14:19 +0400)]
Fixed bug #66252 (Problems in AST evaluation invalidating valid parent:: reference. Constant expessions have to be evaluated in context of defining class).
Ard Biesheuvel [Mon, 9 Dec 2013 15:54:42 +0000 (16:54 +0100)]
cli: don't cast away const in select() timeout argument
The timeout argument to select() is modified to reflect the time
remaining when the function returns on a non-timeout condition.
Passing a pointer to const data and casting away the const-ness is
asking for trouble, but for some reason, this trouble manifests
itself only on non-x86 architectures [whose implementation of select()
in glibc is different from the one supplied for x86]
Fix this by passing a stack copy of the timeout argument to select()
Adam Harvey [Sun, 8 Dec 2013 03:39:40 +0000 (19:39 -0800)]
Switch to using freetype-config for freetype detection.
This fixes GD compilation against libfreetype 2.5.1 and later after they made
the rather interesting decision to change their include directory layout in a
point release.
The original suggestion in the bug was to use pkg-config, but my inclination is
to use freetype-config instead: we should be able to get the same configuration
information without actually needing pkg-config installed, since pkg-config is
by no means guaranteed to exist on many Unices and distros, whereas
freetype-config should always be present if a libfreetype build environment is
installed. Let's try it out and see what happens.
Fixes bug #64405 (Use freetype-config for determining freetype2 dir(s)).
Michael Wallner [Fri, 6 Dec 2013 09:29:24 +0000 (10:29 +0100)]
Fixed bug #61645 (fopen and O_NONBLOCK)
if a mode like "rn" was passed to fopen(), then
php_stream_parse_fopen_modes() would assign O_WRONLY to
flags, because O_NONBLOCK tainted flags for the r/w/+ check
Adam Harvey [Tue, 3 Dec 2013 22:46:15 +0000 (14:46 -0800)]
Remove 128.0.0.0/16 and 191.255.0.0/16 from the reserved list.
These were returned to the general allocation pool by RFC 3330, and hence
shouldn't cause an IP address validation failure due to being reserved. At
least 128.0.0.0/16 is in use on the public Internet today.
Fixes bug #66229 (128.0.0.0/16 isn't reserved any longer).
Michael Wallner [Mon, 2 Dec 2013 15:58:34 +0000 (16:58 +0100)]
Fix bug #65196
Passing DOMDocumentFragment to DOMDocument::saveHTML()
produces invalid markup, because a DocumentFragment is just a container
for child nodes and not a real node itself.
Nikita Popov [Sun, 1 Dec 2013 12:37:56 +0000 (13:37 +0100)]
Fix bug #65764
I'm not exactly sure whether this is the right way to fix it. The
question is whether Generator::throw() on a newborn generator (i.e.
a generator that is not yet at yield expression) should first advance to
the first yield and throw the exception there or whether it should
instead throw the exception in the caller's context.
The old behavior was to throw it at the start of the function (i.e.
the very first opcode), which causes issues like the one in #65764.
Effectively it's impossible to properly handle the exceptions in this
case.
For now I choose the variant where the generator advances to the
first yield before throwing, as that's consistent with how all other
methods on the Generator object currently behave. This does not
necessarily match the behavior in other languages, e.g. Python would throw
the exception in the caller's context. But then our send() method already
has this kind of deviation, so it stays internally consistent at least.
Nikita Popov [Sat, 30 Nov 2013 12:35:33 +0000 (13:35 +0100)]
Cleanup generator closing code a bit
All code dealing with unfinished execution cleanup is now in a separate
function (previously most of it was run even when execution was properly
finished.
Furthermore some code dealing with unclean shutdowns has been removed,
which is no longer necessary, because we no longer try to clean up in
this case.
nikita2206 [Tue, 29 Oct 2013 15:40:17 +0000 (19:40 +0400)]
Fixed bug #65768: DateTimeImmutable::diff does not work
This commit also prevents user classes from directly implementing
DateTimeInterface, because ext/date relies on classes implementing
it to support certain internal structures.
Dmitry Stogov [Thu, 28 Nov 2013 21:00:14 +0000 (01:00 +0400)]
Merge branch 'PHP-5.6' of git.php.net:php-src into PHP-5.6
* 'PHP-5.6' of git.php.net:php-src:
Fixed bug #65199 (Wrong Day of Week) and fixed bug #63391 (Incorrect/inconsistent day of week prior to the year 1600)
Marking those two as XFAIL for now.
Fixed a few DST changeover issues.
Move add/sub to timelib.
Fix backwards transition diffs.
Split up tests into their different categories.
Group those tests a bit better, as some of the bd ones wanted a "ST" feature that doesn't exist yet.
Fixed forward transitions in diffs.
Always reset those parameters.
Fixed "Forward Transition" construction of DateTime objects.
projects / php / log