| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Communication between Erlang processes has conceptually always been
performed through asynchronous signaling. The runtime system
implementation has however previously preformed most operation
synchronously. In a system with only one true thread of execution, this
is not problematic (often the opposite). In a system with multiple threads
of execution (as current runtime system implementation with SMP support)
it becomes problematic. This since it often involves locking of structures
when updating them which in turn cause resource contention. Utilizing
true asynchronous communication often avoids these resource contention
issues.
The case that triggered this change was contention on the link lock due
to frequent updates of the monitor trees during communication with a
frequently used server. The signal order delivery guarantees of the
language makes it hard to change the implementation of only some signals
to use true asynchronous signaling. Therefore the implementations
of (almost) all signals have been changed.
Currently the following signals have been implemented as true
asynchronous signals:
- Message signals
- Exit signals
- Monitor signals
- Demonitor signals
- Monitor triggered signals (DOWN, CHANGE, etc)
- Link signals
- Unlink signals
- Group leader signals
All of the above already defined as asynchronous signals in the
language. The implementation of messages signals was quite
asynchronous to begin with, but had quite strict delivery constraints
due to the ordering guarantees of signals between a pair of processes.
The previously used message queue partitioned into two halves has been
replaced by a more general signal queue partitioned into three parts
that service all kinds of signals. More details regarding the signal
queue can be found in comments in the erl_proc_sig_queue.h file.
The monitor and link implementations have also been completely replaced
in order to fit the new asynchronous signaling implementation as good
as possible. More details regarding the new monitor and link
implementations can be found in the erl_monitor_link.h file.
|
|
|
|
Just to simplify and get 4 distinctive states
IDLE, PENDING, CONNECTED and EXITING.
The old possible flag combos were:
0
PENDING
CONNECTED
CONNECTED|EXITING
EXITING
The two EXITING states did not serve any purpose
other then as a slight optimization in monitor_node(_,false,_)
to shortcut EXITING when there can be no monitors.
|
|
for kernel to ask erts about distribution flags
and keep this info in one place.
|
|
|
|
to make sure it's kept alive.
|
|
|
|
When finalizing outgoing distribution messages
we transcode them into using tuple fallbacks if the
receiver does not support bitstrings and export-funs.
This can only happen if the message was first encoded toward
a pending connection when the receiver was unknown.
It's an optimistic approach optmimized for modern beam nodes,
that expect real bitstrings and funs (since <R13).
Only erl_interface/jinterface lack this support.
|
|
to avoid tuple fallbacks for export funs and bitstrings.
ToDo: Re-encode if receiver turn out to be erl_interface/jinterface.
|
|
|
|
|
|
to handle "lookup without refc++" correctly
which was introduced in 4dcb2ae7810a507b701a30072b2f514cab7ebbdb.
When decrementing refc to zero (in try_delete or prepare_try_delete)
we must always wait thread progress to make sure no thread has done
lookup without refc++ and is just about to do refc++ and thereby revive
the DistEntry. That is, we wait for a potential other thread to either
do refc++ or drop its pointer to the DistEntry.
And if that other thread does refc++ (in erts_ref_dist_entry)
it must also do the extra refc++ for the scheduled pending delete.
|
|
Conflicts:
erts/emulator/beam/bif.c
erts/emulator/beam/dist.c
erts/emulator/beam/dist.h
erts/emulator/beam/erl_bif_info.c
erts/emulator/beam/erl_node_tables.c
erts/emulator/beam/erl_node_tables.h
erts/emulator/beam/external.c
|
|
|
|
|
|
|
|
This refactor was done using the unifdef tool like this:
for file in $(find erts/ -name *.[ch]); do unifdef -t -f defile -o $file $file; done
where defile contained:
#define ERTS_SMP 1
#define USE_THREADS 1
#define DDLL_SMP 1
#define ERTS_HAVE_SMP_EMU 1
#define SMP 1
#define ERL_BITS_REENTRANT 1
#define ERTS_USE_ASYNC_READY_Q 1
#define FDBLOCK 1
#undef ERTS_POLL_NEED_ASYNC_INTERRUPT_SUPPORT
#define ERTS_POLL_ASYNC_INTERRUPT_SUPPORT 0
#define ERTS_POLL_USE_WAKEUP_PIPE 1
#define ERTS_POLL_USE_UPDATE_REQUESTS_QUEUE 1
#undef ERTS_HAVE_PLAIN_EMU
#undef ERTS_SIGNAL_STATE
|
|
The implementation is still hidden behind ERTS_ENABLE_LOCK_COUNT, and
all categories are still enabled by default, but the actual counting can be
toggled at will.
OTP-13170
|
|
|
|
|
|
|
|
Conflicts:
erts/emulator/beam/erl_binary.h
erts/emulator/beam/erl_monitors.c
erts/emulator/beam/erl_nif.c
erts/emulator/beam/global.h
erts/emulator/test/nif_SUITE_data/nif_SUITE.c
|
|
|
|
|
|
Magic references are *intentionally* indistinguishable from ordinary
references for the Erlang software. Magic references do not change
the language, and are intended as a pure runtime internal optimization.
An ordinary reference is typically used as a key in some table. A
magic reference has a direct pointer to a reference counted magic
binary. This makes it possible to implement various things without
having to do lookups in a table, but instead access the data directly.
Besides very fast lookups this can also improve scalability by
removing a potentially contended table. A couple of examples of
planned future usage of magic references are ETS table identifiers,
and BIF timer identifiers.
Besides future optimizations using magic references it should also
be possible to replace the exposed magic binary cludge with magic
references. That is, magic binaries that are exposed as empty
binaries to the Erlang software.
|
|
NIF resources was not handled in a thread-safe manner in the runtime
system without SMP support.
As a consequence of this fix, the following driver functions are now
thread-safe also in the runtime system without SMP support:
- driver_free_binary()
- driver_realloc_binary()
- driver_binary_get_refc()
- driver_binary_inc_refc()
- driver_binary_dec_refc()
|
|
Instead of passing around a file descriptor
use a function pointer to facilitate more advanced
backend write logic such as size limitation or compression.
|
|
* henrik/update-copyrightyear:
update copyright-year
|
|
from future nodes.
|
|
|
|
|
|
This is mostly a pure refactoring.
Except for the buggy cases when calling erlang:halt() with a positive
integer in the range -(INT_MIN+2) to -INT_MIN that got confused with
ERTS_ABORT_EXIT, ERTS_DUMP_EXIT and ERTS_INTR_EXIT.
Outcome OLD erl_exit(n, ) NEW erts_exit(n, )
------- ------------------- -------------------------------------------
exit(Status) n = -Status <= 0 n = Status >= 0
crashdump+abort n > 0, ignore n n = ERTS_ERROR_EXIT < 0
The outcome of the old ERTS_ABORT_EXIT, ERTS_INTR_EXIT and
ERTS_DUMP_EXIT are the same as before (even though their values have
changed).
|
|
* lukas/erts/forker: (28 commits)
erts: Never abort in the forked child
erts: Mend ASSERT makro for erl_child_setup
erts: Allow enomem failures in port_SUITE
erts: iter_port sleep longer on freebsd
erts: Allow one dangling fd if there is a gethost port
erts: Only use forker StackAck on freebsd
erts: It is not possible to exit the forker driver
erts: Add forker StartAck for port start flowcontrol
erts: Fix large open_port arg segfault for win32
erts: Fix memory leak at async open port
kernel: Remove cmd server for unix os:cmd
erts: Add testcase for huge port environment
erts: Move os_pid to port hash to child setup
erts: Handle all EINTR and EAGAIN cases in child setup
erts: Make child_setup work with large environments
erts: Fix forker driver ifdefs for win32
erts: Fix uds socket handling for os x
erts: Fix dereferencing of unaligned integer for sparc
erts: Flatten too long io vectors in uds write
erts: Add fd count test for spawn_driver
...
Conflicts:
erts/emulator/beam/erl_node_tables.c
erts/preloaded/src/erts_internal.erl
|
|
|
|
|
|
broken by 949de78331b9c4ecb9.
|
|
|
|
* The youngest generation of the heap can now consist of multiple
blocks. Heap fragments and message fragments are added to the
youngest generation when needed without triggering a GC. After
a GC the youngest generation is contained in one single block.
* The off_heap_message_queue process flag has been added. When
enabled all message data in the queue is kept off heap. When
a message is selected from the queue, the message fragment (or
heap fragment) containing the actual message is attached to the
youngest generation. Messages stored off heap is not part of GC.
|
|
that could lead strange things to happen when renaming a node with
a name that already exist in node and dist tables.
Problem:
erts_set_this_node() is a bit brutal and does not handle the case
when an old remote node name is reused as the new local node name.
Solution:
Treat erts_this_node and erts_this_dist_entry as all the others
with correct reference counting.
|
|
Not needed as it is always set to erts_this_dist_entry (on net_kernel).
|
|
|
|
|
|
* rickard/delayed-delete-node/OTP-12802:
Delayed node table GC
|
|
* rickard/node-table-read-lock/OTP-12777:
Only read lock node table when reading info
|
|
|
|
|
|
* jlouis/cocci-noderef-fixes:
Correct usage of sizeof() for pointer types
OTP-12771
|
|
Given some pointer *x, calling sizeof(x) will give us
the size of the pointer (4/8 bytes) not the size fo the
underlying dereferenced structure. Use coccinelle to
search for these occurrences in the source code, and
correct them one by one. In the case of
erl_node_tables.c, the erts_snprintf() calls used a
much too small buffer.
- run_erl.c: Use the size of the signal type, not
its pointer.
- erl_node_tables.c: Use the size of the _BUFFER in
erts_snprintf() to make sure we can use the full space.
|
|
|
