Since 1.9 with commit b20aa9eef3 ("MAJOR: tasks: create per-thread wait
queues") a task bound to a single thread will not use locks when being
queued or dequeued because the wait queue is assumed to be the owner
thread's.
But there exists a rare situation where this is not true: the health
check tasks may be running on one thread waiting for a response, and
may in parallel be requeued by another thread calling health_adjust()
after a detecting a response error in traffic when "observe l7" is set,
and "fastinter" is lower than "inter", requiring to shorten the running
check's timeout. In this case, the task being requeued was present in
another thread's wait queue, thus opening a race during task_unlink_wq(),
and gets requeued into the calling thread's wait queue instead of the
running one's, opening a second race here.
This patch aims at protecting against the risk of calling task_unlink_wq()
from one thread while the task is queued on another thread, hence unlocked,
by introducing a new TASK_SHARED_WQ flag.
This new flag indicates that a task's position in the wait queue may be
adjusted by other threads than then one currently executing it. This means
that such WQ manipulations must be performed under a lock. There are two
types of such tasks:
- the global ones, using the global wait queue (technically speaking,
those whose thread_mask has at least 2 bits set).
- some local ones, which for now will be placed into the global wait
queue as well in order to benefit from its lock.
The flag is automatically set on initialization if the task's thread mask
indicates more than one thread. The caller must also set it if it intends
to let other threads update the task's expiration delay (e.g. delegated
I/Os), or if it intends to change the task's affinity over time as this
could lead to the same situation.
Right now only the situation described above seems to be affected by this
issue, and it is very difficult to trigger, and even then, will often have
no visible effect beyond stopping the checks for example once the race is
met. On my laptop it is feasible with the following config, chained to
httpterm:
global
maxconn 400 # provoke FD errors, calling health_adjust()
defaults
mode http
timeout client 10s
timeout server 10s
timeout connect 10s
listen px
bind :8001
option httpchk /?t=50
server sback 127.0.0.1:8000 backup
server-template s 0-999 127.0.0.1:8000 check port 8001 inter 100 fastinter 10 observe layer7
This patch will automatically address the case for the checks because
check tasks are created with multiple threads bound and will get the
TASK_SHARED_WQ flag set.
If in the future more tasks need to rely on this (multi-threaded muxes
for example) and the use of the global wait queue becomes a bottleneck
again, then it should not be too difficult to place locks on the local
wait queues and queue the task on its bound thread.
This patch needs to be backported to 2.1, 2.0 and 1.9. It depends on
previous patch "MINOR: task: only check TASK_WOKEN_ANY to decide to
requeue a task".
Many thanks to William Dauchy for providing detailed traces allowing to
spot the problem.
Tasklets may be woken up to run on the calling thread or by a specific thread
(the owner). But since we use a non-thread safe mechanism when the calling
thread is also the for the owner, there may sometimes be collisions when two
threads decide to wake the same tasklet up at the same time and one of them
is the owner.
This is more of a matter of usage than code, in that a tasklet usually is
designed to be woken up and executed on the calling thread only (most cases)
or on a specific thread. Thus it is a property of the tasklet itself as this
solely depends how the code is constructed around it.
This patch performs a small change to address this. By default tasklet_new()
creates a "local" tasklet, which will run on the calling thread, like in 2.0.
This is done by setting tl->tid to a negative value. If the caller wants the
tasklet to run exclusively on a specific thread, it just has to set tl->tid,
which is already what shared tasklet callers do anyway.
No backport is needed.
As using an mt_list for the tasklet list is costly, instead use a regular list,
but add an mt_list for tasklet woken up by other threads, to be run on the
current thread. At the beginning of process_runnable_tasks(), we just take
the new list, and merge it into the task_list.
This should give us performances comparable to before we started using a
mt_list, but allow us to use tasklet_wakeup() from other threads.
The aim is to rassemble all scheduler information related to the current
thread. It simply points to task_per_thread[tid] without having to perform
the operation at each time. We save around 1.2 kB of code on performance
sensitive paths and increase the request rate by almost 1%.
Change the tasklet code so that the tasklet list is now a mt_list.
That means that tasklet now do have an associated tid, for the thread it
is expected to run on, and any thread can now call tasklet_wakeup() for
that tasklet.
One can change the associated tid with tasklet_set_tid().
Instead of using the same type for regular linked lists and "autolocked"
linked lists, use a separate type, "struct mt_list", for the autolocked one,
and introduce a set of macros, similar to the LIST_* macros, with the
MT_ prefix.
When we use the same entry for both regular list and autolocked list, as
is done for the "list" field in struct connection, we know have to explicitely
cast it to struct mt_list when using MT_ macros.
The CPU time accounting field called "cpu_time" is used only by tasks
and not tasklets, yet it used to be stored into the TASK_COMMON part,
which doesn't make sense and wastes tasklet memory. In addition, moving
it to tasks also helps better group the various parts in cache lines.
Sometimes we need to delegate some list processing to a function running
on another thread. In this case the list element will simply be queued
into a dedicated self-locked list and the task responsible for this list
will be woken up, calling the associated function which will run over the
list.
This is what work_list does. Such lists will be dedicated to a limited
type of work but will significantly ease such remote handling. A function
is provided to create these per-thread lists, their tasks and to properly
bind each task to a distinct thread, so that the caller only has to store
the resulting pointer to the start of the structure.
These structures should not be abused though as each head will consume
4 pointers per thread, hence 32 bytes per thread or 2 kB for 64 threads.
There's a very small but existing uncertainty window when waking another
thread up where it is possible for task_wakeup() not to wake the other
task up because it's still running while this once is in the process of
finishing and loses its TASK_RUNNING flag. In this case the wakeup will
be missed.
The problem is that we have a single flag to store 3 states, since the
transition from running to sleeping isn't atomic. Thus we need to have
another flag to cover this part. This patch introduces TASK_QUEUED to
mention that the task is already in the run queue, running or not. This
bit will be removed while TASK_RUNNING is kept once dequeued, and will
be used when removing TASK_RUNNING to check if the task has been requeued.
It might be possible to slightly improve this but the occurrence rate
is quite low and we don't really need to complexify the scheduler to
optimize for a rare case.
The impact with the current code is very low since we have few inter-
thread wakeups. Most of them are caused by checks killing sessions.
This must be backported to 1.9.
Right now we measure for each task the cumulated time spent waiting for
the CPU and using it. The timestamp uses a 64-bit integer to report a
nanosecond-level date. This is only enabled when "profiling.tasks" is
enabled, and consumes less than 1% extra CPU on x86_64 when enabled.
The cumulated processing time and wait time are reported in "show sess".
The task's counters are also reset when an HTTP transaction is reset
since the HTTP part pretends to restart on a fresh new stream. This
will make sure we always report correct numbers for each request in
the logs.
By removing the reason code for the wakeup we can gain 8 extra bits to
encode the task's state. The reason code was never used at all and is
wrong by design since subsequent calls will OR this value anyway. Let's
say it goodbye and leave the room for more precious bits. The woken bits
were moved to the higher byte so that the most important bits can stay
grouped together.
Introduce tasklets, lightweight tasks. They have no notion of priority,
they are just run as soon as possible, and will probably be used for I/O
later.
For the moment they're used to replace the temporary thread-local list
that was used in the scheduler. The first part of the struct is common
with tasks so that tasks can be cast to tasklets and queued in this list.
Once a task is in the tasklet list, it has its leaf_p set to 0x1 so that
it cannot accidently be confused as not in the queue.
Pure tasklets are identifiable by their nice value of -32768 (which is
normally not possible).
In preparation for thread-specific runqueues, change the task API so that
the callback takes 3 arguments, the task itself, the context, and the state,
those were retrieved from the task before. This will allow these elements to
change atomically in the scheduler while the application uses the copied
value, and even to have NULL tasks later.
This macro should be used to declare variables or struct members depending on
the USE_THREAD compile option. It avoids the encapsulation of such declarations
between #ifdef/#endif. It is used to declare all lock variables.
Currently the task scheduler suffers from an O(n) lookup when
skipping tasks that are not for the current thread. The reason
is that eb32_lookup_ge() has no information about the current
thread so it always revisits many tasks for other threads before
finding its own tasks.
This is particularly visible with HTTP/2 since the number of
concurrent streams created at once causes long series of tasks
for the same stream in the scheduler. With only 10 connections
and 100 streams each, by running on two threads, the performance
drops from 640kreq/s to 11.2kreq/s! Lookup metrics show that for
only 200000 task lookups, 430 million skips had to be performed,
which means that on average, each lookup leads to 2150 nodes to
be visited.
This commit backports the principle of scope lookups for ebtrees
from the ebtree_v7 development tree. The idea is that each node
contains a mask indicating the union of the scopes for the nodes
below it, which is fed during insertion, and used during lookups.
Then during lookups, branches that do not contain any leaf matching
the requested scope are simply ignored. This perfectly matches a
thread mask, allowing a thread to only extract the tasks it cares
about from the run queue, and to always find them in O(log(n))
instead of O(n). Thus the scheduler uses tid_bit and
task->thread_mask as the ebtree scope here.
Doing this has recovered most of the performance, as can be seen on
the test below with two threads, 10 connections, 100 streams each,
and 1 million requests total :
Before After Gain
test duration : 89.6s 4.73s x19
HTTP requests/s (DEBUG) : 11200 211300 x19
HTTP requests/s (PROD) : 15900 447000 x28
spin_lock time : 85.2s 0.46s /185
time per lookup : 13us 40ns /325
Even when going to 6 threads (on 3 hyperthreaded CPU cores), the
performance stays around 284000 req/s, showing that the contention
is much lower.
A test showed that there's no benefit in using this for the wait queue
though.
It was a leftover from the last cleaning session; this mask applies
to threads and calling it process_mask is a bit confusing. It's the
same in fd, task and applets.
2 global locks have been added to protect, respectively, the run queue and the
wait queue. And a process mask has been added on each task. Like for FDs, this
mask is used to know which threads are allowed to process a task.
For many tasks, all threads are granted. And this must be your first intension
when you create a new task, else you have a good reason to make a task sticky on
some threads. This is then the responsibility to the process callback to lock
what have to be locked in the task context.
Nevertheless, all tasks linked to a session must be sticky on the thread
creating the session. It is important that I/O handlers processing session FDs
and these tasks run on the same thread to avoid conflicts.
These notification management function and structs are generic and
it will be better to move in common parts.
The notification management functions and structs have names
containing some "lua" references because it was written for
the Lua. This patch removes also these references.
In order to authorize call of task_wakeup on running task:
- from within the task handler itself.
- in futur, from another thread.
The lookups on runqueue and waitqueue are re-worked
to prepare multithread stuff.
If task_wakeup is called on a running task, the woken
message flags are savec in the 'pending_state' attribute of
the state. The real wakeup is postponed at the end of the handler
process and the woken messages are copied from pending_state
to the state attribute of the task.
It's important to note that this change will cause a very minor
(though measurable) performance loss but it is necessary to make
forward progress on a multi-threaded scheduler. Most users won't
ever notice.
Just by reordering the struct task, we could shrink it by 8 bytes from
120 to 112 bytes. A careful reordering allowed each part to be located
closer to the hot parts it's used with, resulting in another performance
increase of about 0.5%.
The two new functions below make it possible to register any number
of functions or tasks to a system signal. They will be called in the
registration order when the signal is received.
struct sig_handler *signal_register_fct(int sig, void (*fct)(struct sig_handler *), int arg);
struct sig_handler *signal_register_task(int sig, struct task *task, int reason);
All files referencing the previous ebtree code were changed to point
to the new one in the ebtree directory. A makefile variable (EBTREE_DIR)
is also available to use files from another directory.
The ability to build the libebtree library temporarily remains disabled
because it can have an impact on some existing toolchains and does not
appear worth it in the medium term if we add support for multi-criteria
stickiness for instance.
Timers are unsigned and used as tree positions. Ticks are signed and
used as absolute date within current time frame. While the two are
normally equal (except zero), it's important not to confuse them in
the code as they are not interchangeable.
We add two inline functions to turn each one into the other.
The comments have also been moved to the proper location, as it was
not easy to understand what was a tick and what was a timer unit.
All the tasks callbacks had to requeue the task themselves, and update
a global timeout. This was not convenient at all. Now the API has been
simplified. The tasks callbacks only have to update their expire timer,
and return either a pointer to the task or NULL if the task has been
deleted. The scheduler will take care of requeuing the task at the
proper place in the wait queue.
In many situations, we wake a task on an I/O event, then queue it
exactly where it was. This is a real waste because we delete/insert
tasks into the wait queue for nothing. The only reason for this is
that there was only one tree node in the task struct.
By adding another tree node, we can have one tree for the timers
(wait queue) and one tree for the priority (run queue). That way,
we can have a task both in the run queue and wait queue at the
same time. The wait queue now really holds timers, which is what
it was designed for.
The net gain is at least 1 delete/insert cycle per session, and up
to 2-3 depending on the workload, since we save one cycle each time
the expiration date is not changed during a wake up.
It's very frequent to require some information about the
reason why a task is running. Some flags have been added
so that a task now knows if it got woken up due to I/O
completion, timeout, etc...
This is the first attempt at moving all internal parts from
using struct timeval to integer ticks. Those provides simpler
and faster code due to simplified operations, and this change
also saved about 64 bytes per session.
A new header file has been added : include/common/ticks.h.
It is possible that some functions should finally not be inlined
because they're used quite a lot (eg: tick_first, tick_add_ifset
and tick_is_expired). More measurements are required in order to
decide whether this is interesting or not.
Some function and variable names are still subject to change for
a better overall logics.
The run queue scheduler now considers task->nice to queue a task and
to pick a task out of the queue. This makes it possible to boost the
access to statistics (both via HTTP and UNIX socket). The UNIX socket
receives twice as much a boost as the HTTP socket because it is more
sensible.
We now insert tasks in a certain sequence in the run queue.
The sorting key currently is the arrival order. It will now
be possible to apply a "nice" value to any task so that it
goes forwards or backwards in the run queue.
The calls to wake_expired_tasks() and maintain_proxies()
have been moved to the main run_poll_loop(), because they
had nothing to do in process_runnable_tasks().
The task_wakeup() function is not inlined anymore, as it was
only used at one place.
The qlist member of the task structure has been removed now.
The run_queue list has been replaced for an integer indicating
the number of tasks in the run queue.
The ultree code has been removed in favor of a simpler and
cleaner ebtree implementation. The eternity queue does not
need to exist anymore, and the pool_tree64 has been removed.
The ebtree node is stored in the task itself. The qlist list
header is still used by the run-queue, but will be able to
disappear once the run-queue uses ebtree too.
The timeout functions were difficult to manipulate because they were
rounding results to the millisecond. Thus, it was difficult to compare
and to check what expired and what did not. Also, the comparison
functions were heavy with multiplies and divides by 1000. Now, all
timeouts are stored in timevals, reducing the number of operations
for updates and leading to cleaner and more efficient code.
The rbtree-based wait queue consumes a lot of CPU. Use the ul2tree
instead. Lots of cleanups and code reorganizations made it possible
to reduce the task struct and simplify the code a bit.
This patch from Sin Yu makes use of an rbtree for the wait queue,
which will solve the slowdown problem encountered when timeouts
are heterogenous in the configuration. The next step will be to
turn maintain_proxies() into a per-proxy task so that we won't
have to scan them all after each poll() loop.
The files are now stored under :
- include/haproxy for the generic includes
- include/types.h for the structures needed within prototypes
- include/proto.h for function prototypes and inline functions
- src/*.c for the C files
Most include files are now covered by LGPL. A last move still needs
to be done to put inline functions under GPL and not LGPL.
Version has been set to 1.3.0 in the code but some control still
needs to be done before releasing.
