The nxt_murmur_hash2() generated 4-byte hash that was stored in uintptr_t,
which was 8 bytes long on 64-bit systems. At each iteration, it took the
previous key and hashed it again.
The problem was that it took only the first 4 bytes of the key, and these
4 bytes were always zero on 64-bit big-endian system. That resulted in
equal keys at each iteration.
The bug was discovered on IBM/S390x.
The nxt_conf_map_object() function used nxt_int_t for NXT_CONF_MAP_INT, which
was 8 bytes long on 64-bit systems.
But the nxt_port_main_start_worker_handler() used it to map into the int field
of the nxt_common_app_conf_t structure, which was 4 bytes. As the result, on
a 64-bit big-endian system all the meaningful module type numbers were assigned
into the gap above the "type" field.
The bug was discovered on IBM/S390x.
The issue was caused by misplacement of allocated blocks in rbtree due
to broken comparison function if the distance between two allocations
did not fit into intptr_t. As the result, nxt_mp_free() could have
failed to find the allocation.
In particular, it was mostly observed when Unit was compiled with
musl C library on 32-bits systems.
This closes#118 issue on GitHub.
Timers that don't require maximum precision (most of them, actually) can be
triggered earlier or later within the bias interval.
To reduce wakeups by timers, the expire function now triggers not only all
timers that fall within the elapsed time, but also those whose bias falls
within this interval.
Delete/insert operation complexity for a red-black tree is O(log n),
where n is the total number of tree elements.
If all delete operations are performed before all insert operations,
the average number of tree elements during an operation will be
lower than in the mixed-operations case.
To optimize rbtree operations, all changes are stored in array and later
processed in batches.
The previous implementation of this mechanics had a number of design flaws.
Each change was saved in a new array entry; until the changes were applied,
the timer remained in an intermediate state (NXT_TIMER_CHANGING).
This intermediate state didn't allow to identify if time was going to be
disabled or enabled. However, the nxt_conn_io_read() function relied on
this information; as a result, in some cases the read timeout wasn't set.
Also, the nxt_timer_delete() function did not reliably track whether a timer
was added to the work queue. It checked the NXT_TIMER_ENQUEUED state of
a timer, but this state could be reset to NXT_TIMER_DISABLED by a
nxt_timer_disable() call or another nxt_timer_delete() call.
Now, instead of keeping the whole history of the timer's changes, the new
implementation updates the timer state immediately, and only one operation
is added to the array to add or delete timer in the rbtree according
to its final state.
Previously, passing 0 resulted in reading the whole body and all negative
values raised an exception.
Now the behaviour is in consistentance with io.RawIOBase.read() interface,
and passing 0 returns empty (byte) string, while -1 results in reading the
whole body.
Overriding LANG might not work, since it has less precedence than LC_* settings.
LC_ALL has the highest precedence.
This should resolve issue #121 on GitHub.
It has the following structure and default values:
{
"http": {
"header_read_timeout": 30,
"body_read_timeout": 30,
"send_timeout": 30,
"idle_timeout": 180,
"max_body_size": 8388608
}
}
The implementation of module was based on the assumption that PHP reads request
body and headers in the particular order. For the POST request the body goes
before headers and vice versa for all other requests.
But as it appeared later, this order is unspecified and depends on many factors,
including the particular code of PHP application. Among other factors those
can affect ordering:
- presence of "Content-Type" header;
- "variables_order" php.ini setting;
- "enable_post_data_reading" php.ini setting;
- reading php://input by application;
and this list can be incomplete.
As a temporary workaround, request body now is always put before headers and it
is gracefully skipped whenever PHP wants to get headers.
This closes#144 issue on GitHub.
The previous method changed PHP options only for the first request.
On the request completion the options were rolled back.
This closes#145 issue on GitHub.
Allowing characters up to 0xFF doesn't conflict with RFC 7230.
Particularly, this make it possible to pass unencoded UTF-8 data
through HTTP headers, which can be useful.
In order to reopen access log, the router process creates a memory pool
and allocates a buffer to send RPC message for the main process.
Previously, the memory pool was destroyed when RPC response handler was
called. It appeared, that the buffer completion handler could be not
triggered by that time and still remained in a queue.
Now the memory pool is destroyed only after both events are happen.
This also fixes#101 issue on GitHub. The function previously used to
parse IPv6 address of control socket was broken. Now the working function
is used instead.
Valentin Bartenev