We force users to use the create() methods of StaticResultWaitHandle and StaticExceptionWaitHandle, which properly set m_resultOrException. However, deserialization will use the normal constructor which was failing to initialize the field. The destructor, then, would operate on random data. This broke hphpd when a stack trace had one of these objects on it somewhere. The 'where' command would succeed, but the next command (which deletes the stack trace), would segfault in the wait handle's destructor.
I think it's fair to not serialize this member. It's not exposed thru PHP except via joining with the wait handle, and I think an argument can be made that it's just plain wrong to join with a deserialized copy of a wait handle. So I've just initialized it to a reasonable default.
I was learning from @jdelong and he said that you should use
double quotes for local includes and angle brackets for library
includes. I asked why our code was the way it was, and he said he wanted
to clean it up. I beat him to it :)
Conflicts:
hphp/runtime/base/server/admin_request_handler.cpp
hphp/runtime/vm/named_entity.h
In order to build a dependency graph of continuation execution and data-fetching in PHP-land, we need a few instrumentation points in the asio_ext HHVM extension. There are 4 additions required:
1. Callback when a continuation finishes successfully.
2. Callback when a continuation blocks on a wait_handle.
3. Get array of WaitHandles a GenArrayWaitHandle is waiting on.
4. Get WaitHandle that the SetResultToRefWaitHandle is waiting on.
I don't think this should really affect performance, as in the normal case, nothing has changed, but you never know... I'm also not sure who should be reviewing this, so I've just added @jan for now. If you could pile other people on, that would be cool.
sandcastle appears to be broken.
Move the responsibility of entering/exiting contexts from PHP to the
implementation of $wait_handle->join().
This eliminates possibility of weird situations, like contexts without
any running wait handle. This guarantees that asio_get_current() returns
null only if called completely out of asio framework and simplifies some
logic, such as getCurrentWaitHandleDepth().
Take advantage of previous diff that won't try to construct abstract classes.
Abstract methods now don't need to be implemented, so remove their
dummy implementation.
We only used it to get the values of certain class constants,
and to define the ObjectStaticCallbacks for every class.
We can put the class constants directly into the class_map
(we should have done that before for perf reasons), and then
the only remaining use of ObjectStaticCallbacks is to proxy
the Class* for each builtin class. So just use the Class*
directly.
Once this is in, Im just a small step away from eliminating
make -C hphp/system - so Im leaving a lot of dead code here.
Its going to be easier to delete it en masse, rather than
try to pick and chose now.
enterContext() throws an exception when cross-context cycle is found.
The problem is that it modifies state before the exception is thrown,
assuming that the call will succeed.
When an exception is thrown, a dependency is left in invalid state, with
parent being in more specific context. This breaks exitContext()
algorithm and results in either internal invariant violations as seen
in #2091939, or memory corruptions and crashes as seen in #2125762.
Let's fix it by modifying state after returning back from recursive call
instead of before doing such call. This was previously unsafe in case we
tried to import dependency loop. Once D720506 is committed, dependency
loops will not exist anymore.
Currently, wait handles store pointer to the context they are in. This
pointer is not protected with reference counting, as it is expected that
whenever a context is exited, references to it are cleaned thru
exitContext() mechanism.
If a bug is present that violates this assumption, it is impossible to
guard against invalid pointer access and a hard to debug memory
corruption occurs.
Since the structure of contexts is a simple stack, let's reference them
by index instead of by pointer.
As a bonus, one pointer worth of memory is saved for every non-trivial wait handle.
The actual bugs will be fixed by the next 2 diffs that do:
1. implement online cycle detection
2. do enterContext() atomically and properly handle failure
Avoid dynamic_cast<> on fast paths, use o_instanceof(StaticString)
instead. Saves about ~0.2-0.3% of CPU time.
Once HPHPc is gone, we will convert these calls to HHVM-specific API.
Thanks @bmaurer for discovery and @mwilliams for suggestion how to fix
it.
This change is mostly for FB internal organizational reasons.
Building is not effected beyond the fact that the target now
lands in hphp/hhvm/hhvm rather than src/hhvm/hhvm.
Mike Magruder