| Age | Commit message (Collapse) | Author |
|---|
|
* lukas/erts/port_monitor_mem_leak/OTP-13818:
erts: Fix port monitor memory leak
|
|
|
|
|
|
* maint-19:
Updated OTP version
Prepare release
Avoid segfault when printing slogan after crashdumping
Fix race causing lost wakeup on receive-after timeout
|
|
* lukas/erts/tracing/fix_sparc_align_issue/OTP-13803:
erts: Align ErtsThrPrgrLaterOp when free'ing tracer
|
|
On Sparc all structs with a double word element in are assumed
to be allocated on a two word boundary. So we have to make sure
that the ErtsThrPrgrLaterOp * points to 8 / 16 byte aligned memory.
As it only costs one word I've done it on all architectures in case
some unknown other arch also needs this.
|
|
* rickard/erl-crash-dump-bug/OTP-13799:
Avoid segfault when printing slogan after crashdumping
|
|
* rickard/proc-tmo-bug/OTP-13798:
Fix race causing lost wakeup on receive-after timeout
|
|
|
|
|
|
|
|
* margnus1/erts/fix-hipe-literal-gc/PR-1122/OTP-13777:
check_process_code: Sweep HiPE stack for literals
|
|
We need to use an initial hash seed for each map pair, i.e. the
hash value. The hash value is reset to the seed for each pair instead
of setting the seed to zero.
If we don't, no additional entropy is given to the system and the
hash collision resolution in maps:from_list/1 would fail.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This type of statistics is now available through the microstate
accounting API.
|
|
* sverker/update_counter-deadlock/ERL-188/OTP-13731:
erts: Add test ets_SUITE:update_counter_table_growth
erts: Fix deadlock in ets:update_counter/4
erts: Optimize db_finalize_dbterm_hash
|
|
|
|
This is needed in order for the heap to never contain
any pointers to invalid memory. This is good for performance
and debugging reasons.
|
|
in 'set' with 'write_concurrency'
when inserting default object causes table to grow
and the bucket to split is protected by same lock as the key.
|
|
Always free term after WUNLOCK_HASH
|
|
Symptom: VM abort "Overrun stack and heap"
Problem: The temporary bignum created in buffer tmp_big[] will be
part of the GC initiated by TestHeapPreserve, but its size is not
included which can cause the GC to overflow if very unlucky.
Solution: Do not include tmp_big in the GC.
|
|
We have to do an extra check if the return_to trace is enabled here
as it may have been cleared by the bif after the flags where
created by the call_trace.
|
|
Because check_process_code neglected checking the HiPE stack for
references to the literal area, such references would survive the purge
and subsequent deletion of a module and its literal area. These dangling
references would then cause incorrect behaviour or even hard crashes of
the VM.
By simply adding a scan of the HiPE stack to check_process_code and
erts_garbage_collect_literals, this problem is fixed.
In order to support full stack walks without deleting the graylimit
trap, a new stack walking interface function,
nstack_walk_init_sdesc_ignore_trap() was introduced.
|
|
|
|
to not block multi threading.
Do the same as erlang:nodes/1, just grab dist table read lock.
|
|
* lukas/erts/nif_allow_port_command_in_non_sched_thread/OTP-13442:
erts: Allow enif_port_command in non-sched thread
|
|
Yes this is an ugly workaround. One approach for a better
solution could be to introduce an internal secret atom
tagged as an atom with a unique index, but impossible
to find by string hash lookup/insert.
|
|
|
|
|
|
It has to be possible to send trace messages to a port
from non-schedulers threads (specifically from the
sys_msg_dispatcher).
|
|
|
|
|
|
* kvakvs/erts/monitor_port/OTP-11384:
erts: Add port monitors
|
|
* erlang:monitor/2 with port argument is added, erlang:demonitor, using port task API and avoiding locking;
* port_info and process_info support for monitored ports (with named port monitors support);
* Exit signals contain type 'process' or 'port';
* Propagation of port exit signals;
* Self-cleaning when origin process dies with monitor on;
* 8 test cases + testcase for port driver crashing;
* Documentation for all of the above (monitor, demonitor, port_info and process_info) updated
|
|
* egil/fix-simple-crashdump-with-maps/OTP-13657:
erts: Don't crash on maps on crash dumps
|
|
* egil/lttng-change-domain:
runtime_tools: Change LTTng dyntrace domain
erts: Change LTTng otp domain
|
|
- Large Maps could cause a stack overrun during crash dump generation.
- This is a simple workaround until a solution has been implemented.
- The error has no impact on a running system.
|
|
* rickard/misc-ds:
Replace enif_is_on_dirty_scheduler() with enif_thread_type()
No GC on dirty IO schedulers
|
|
This reverts commit 7c133fb1094ad1cabbb5cfc157483a43c816c6a9.
|
|
This reverts commit f4bdac18cb9dd45185e911308a5ebd95ff10d7fd.
|
|
|
|
|
|
From 'com_ericsson_otp' to 'org_erlang_otp'.
This domain name is more suitable.
|
|
|
