| 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The previously used message queue partitioned into two halves has been
replaced by a more general signal queue partitioned into three parts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.
|
|
* sverker/dict-put-immed-opt:
erts: Optimize erlang:put/2 for hash collision lists
erts: Optimize erlang:put/2 for immed values
erts: Refactor erlang:put/2
|
|
|
|
* raimo/stop-encouraging-v4-mapped/ERL-503/OTP-13716:
Stop translating V4MAPPED addresses
Stop returning V4MAPPED addresses
Implement function for IPv4-mapped IPv6 addresses
|
|
* origin/hasse/kernel/rpc_doc_fix/OTP-10551:
kernel: Improve docs of rpc:multicall/5
|
|
as workarounds to avoid blocking auto-connect,
which is now asynchronous in OTP-21.
Also changed old catch to more efficient try-catch.
|
|
to replace DFLAGS_STRICT_ORDER_DELIVERY
and remove that compile time dependency.
|
|
and remove use in dist_util:check_mandatory
|
|
to a simple 'band' operation into ChosenFlags
and no need to remember old 'this_flags' and 'other_flags'.
|
|
for kernel to ask erts about distribution flags
and keep this info in one place.
|
|
No need to clear them as they cannot be set.
|
|
|
|
|
|
* john/erts/allow-opening-device-files/OTP-11462:
Allow opening device files and FIFOs with file:open/2
|
|
* maint:
Skip testing unavailable interfaces in inet_SUITE:getifaddrs
Make re_SUITE:sub_binaries less unstable
|
|
|
|
This test consistently failed on Windows when an interface was
enabled but unplugged, as the interface was 'up' but not 'running'.
|
|
|
|
* lukas/kernel/dist_SUITE_corefile_ignore:
kernel: Ignore cores in erl_distribution_SUITE
|
|
To the best of our knowledge this was introduced since file operations
on device files/FIFO:s could hang the emulator forever back when the
emulator was single-threaded and lacked IO threads; a read operation
could block all progress preventing the write operation it waited for
from occurring.
Granted, this could still happen through starving all dirty IO
schedulers, but the same issue can arise with NFS files which we've
always allowed.
Removing this restriction also lets us remove a stat(2) call that was
added to specifically allow `/dev/null`.
|
|
|
|
Do destructive write of immed value if key exists.
This is an optimization at the expense of get/0
which must now copy all (mutable) key-value tuples.
|
|
|
|
or backslash on Windows.
Purpose: Prevent tricks to get hostile code running.
|
|
* maint:
Updated OTP version
Update release notes
Update version numbers
kernel: Add os:cmd/2 with max_size option
# Conflicts:
# OTP_VERSION
# lib/kernel/doc/src/os.xml
# lib/kernel/src/os.erl
|
|
* maint-20:
Updated OTP version
Update release notes
Update version numbers
erts: Add system_flags(erts_alloc,"+M?sbct *")
erts: Add age order first fit allocator strategies
erts: Refactor erl_ao_firstfit_alloc
erts: Add migration options "acnl" and "acfml"
kernel: Add os:cmd/2 with max_size option
erts: Add more stats for mbcs_pool
erts: Fix alloc_SUITE:migration
stdlib: Make ets_SUITE memory check try again
erts: Improve carrier pool search
erts: Improve alloc_SUITE:migration
erts: Refactor carrier dealloc migration
|
|
|
|
|
|
|
|
* hasse/no_get_stacktrace/OTP-14861:
erts: Update abstract format doc with stacktrace variable
wx: Do not call erlang:get_stacktrace()
tools: Do not call erlang:get_stacktrace()
stdlib: Do not call erlang:get_stacktrace()
sasl: Do not call erlang:get_stacktrace()
runtime_tools: Do not call erlang:get_stacktrace()
reltool: Do not call erlang:get_stacktrace()
parsetools: Do not call erlang:get_stacktrace()
observer: Do not call erlang:get_stacktrace()
mnesia: Do not call erlang:get_stacktrace() (cont)
mnesia: Do not call erlang:get_stacktrace()
kernel: Do not call erlang:get_stacktrace()
inets: Do not call erlang:get_stacktrace()
eunit: Do not call erlang:get_stacktrace()
et: Do not call erlang:get_stacktrace()
dialyzer: Do not call erlang:get_stacktrace()
debugger: Do not call erlang:get_stacktrace()
debugger: Do not try to restore stacktrace
common_test: Do not call erlang:get_stacktrace()
|
|
|
|
The existing wording may be interpreted as saying that embedded mode
eager loads all modules. This revision makes clear embedded mode only
disables module auto loading.
Since I was on it, I have reordered a couple of places to describe
interactive first, and then embedded. It feels natural to cover first
the default and positive mode (auto loads), and then its negation.
|
|
Currently HiPE amd64 assumes the runtime system code is loaded into
the low 2G of the address space. However, this is not the case when
PIE is enabled, it is loaded into a random location. So trampolines
are required to call BIFs, and also we have first to load the address
of sse2_fnegate_mask to a regisiter before xorpd in fchs.
|
|
|
|
|
|
|
|
Some bash versions segfault when a unicode argument is given
so we ignore cores created by nodes spawned by this testcase.
|
|
* maint:
kernel: Correct contracts and a bug in group_history
stdlib: Correct contracts
dialyzer: Optimize handling of a lot of warnings
Conflicts:
lib/kernel/src/erl_boot_server.erl
|
|
|
|
particularly slow erlc when compiler is hipe compiled.
hipe_unified_loader:load did not patch external call sites
and instead caused a double hipe mode switch per call.
hipe_unified_loader:load is only used
for early modules first loaded as beam
and by code:atomic_load and friends.
|
|
|
|
* maint:
stdlib: Correct a filelib test case
stdlib: Let filelib:find_source() search subdirs
|
|
OTP:13713: Add documentation and typespecs for inet:i/0
|
|
|
|
The Design Principles states that an application can have Erlang
source files one level below the "src" directory, and now
filelib:find_source() by default searches one level below "src".
The same applies to "esrc". That directory is only mentioned in
filename(3).
|
|
Previously we accepted trailing NULs, which was backwards compatible
as such usage never resulted in misbehavior in the first place. The
downside is that it prevented erts_native_filename_need from
returning an accurate number of *actual characters*, needlessly
complicating encoding-agnostic code like erts_osenv.
|
|
putenv(3) and friends aren't thread-safe regardless of how you slice
it; a global lock around all environment operations (like before)
keeps things safe as far as our own operations go, but we have
absolutely no control over what libc or a library dragged in by a
driver/NIF does -- they're free to call getenv(3) or putenv(3)
without honoring our lock.
This commit solves this by setting up an "emulated" environment which
can't be touched without going through our interfaces. Third-party
libraries can still shoot themselves in the foot but benign uses of
os:putenv/2 will no longer risk crashing the emulator.
|
|
|
|
|
|
false should be never, false not found in the source code
|
