| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
|\
| |
| |
| |
| |
| |
| |
| |
| | |
into maint-24
* bmk/kernel/20220427/gen_tcp_socket_shutdown_24.3.3/OTP-18080:
[kernel|gen-tcp|test] Tweaked shutdown (api) test case(s)
[kernel|gen-tcp-socket] Changed the behaviour of shutdown
[erts|esock] Add caller pid to the initial debug printout for close
|
| |
| |
| |
| | |
OTP-18080
|
| |
| |
| |
| |
| |
| |
| |
| | |
The behaviour of shutdown was changed to be more compatible with
the "old" gen_tcp (that is, gen_tcp used with inet_backend = inet,
which is *currently* default).
OTP-18080
|
| |
| |
| |
| | |
OTP-18080
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | | |
'bmk/kernel/20220427/gen_tcp_socket_raw_option_24.3.3/OTP-18078' into maint-24
* bmk/kernel/20220427/gen_tcp_socket_raw_option_24.3.3/OTP-18078:
[kernel] Fixed handling of raw option for gen-tcp-socket
|
| |/
| |
| |
| |
| |
| |
| | |
When using inet-backend socket, getopts of the raw option
failed badly.
OTP-18078
|
|\ \
| | |
| | |
| | |
| | | |
* big-r81/epmd-nonfatal-loopback/OTP-17970:
epmd: don't fail if non-critical interfaces fail to bind
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When specifying bind address(es) via ERL_EPMD_ADDRESS or the -address
option, epmd will bind to the IPv4/IPv6 loopback intefaces in addition
to the user-supplied address(es).
However, it is possible that one of the loopback interfaces does not
exist. For example, if the host system has ipv6 disabled via the
disable_ipv6 sysctl, there will be no loopback interface "::1", and
attmepting to bind to "::1" will fail.
Under these circumstances, log the failure but do not consider the
failure to bind to the loopback interface as fatal.
If any of the user-supplied addresses fail to bind, however, that is
still considered fatal and epmd will terminate.
|
|\ \ \
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
* lukas/erts/fix-off-heap-change-gc/OTP-18075:
erts: Fix gc to fetch sig queue at mqd change
# Conflicts:
# erts/emulator/beam/erl_gc.c
# erts/emulator/test/message_queue_data_SUITE.erl
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When a process is changing from on heap to off heap
message queue data, there may still be messages in the
external queue with data on the heap. So we need to move
all signals in the external queue to the internal queue
before doing the GC.
Closes #5933
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
* lukas/24/erts/fix-dist-fragment-exit-leak/OTP-18077:
erts: Fix faulty alloc assert on 32-bit
erts: Add testcase for exit distr fragments
erts: Fix size calculation of suspend monitor
erts: Add etp-links for inspecting process links
erts: Fix fragmented send to finish before exiting
|
| |\ \ \ \
| | |_|_|/
| |/| | |
| | | | | |
lukas/24/erts/fix-dist-fragment-exit-leak/OTP-18077
|
| | |\ \ \
| | | |_|/
| | |/| |
| | | | | |
lukas/23/erts/fix-dist-fragment-exit-leak/OTP-18077
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The total block size can be > 2 GB, so the ASSERT would trigger
on working code. So we only check the blocks size on 64-bit
systems.
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | |/
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If a process is suspended doing a fragmented send and then
receives an exit signal it was terminated before it could
finish sending the message leading to a memory leak on the
receiving side.
This change fixes that so that the message is allowed to
finish being sent before the process exits.
Closes #5876
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | | |
* lukas/erts/remove-child-setup-assert:
erts: Remove faulty child setup assert
|
| | |_|/
| |/| | |
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | | |
* lukas/22/erts/fix-localtime-tz-change/OTP-18076:
erts: Fix localtime_r summer/winter-time change bug
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
According to POSIX, localtime_r does not have to update the process
internal tz data when called. So if time went from winter to
summertime erts is running localtime_r may return an incorrect time.
So we need to make sure to call tzset before each call to localtime_r.
localtime is guaranteed to update the tz information, so no need to
call it when using localtime.
|
|\ \ \ \
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
* lukas/add-github-actions:
gh: Fix ubuntu base to use base argument
gh: Use lastest lts ubuntu for build and tests
gh: Fix debian base image
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The wx webview package on debian seems to have been broken
for cross installs, so we disable it for now.
|
|\ \ \ \ \
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
* bjorn/compiler/setelement/OTP-18082:
Fix incorrect optimization of setelement/3
|
| | |_|_|/
| |/| | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
`beam_call_types` would assume that a call to `setelement/3`
would always fail for a tuple with known exact size and a position
that could be both inside and outside the tuple. That would make
the call to `setelement/3` tail-recursive and all code following
the call would be ignored.
This bug was introduced in 2ea04617d18b. It has already been fixed on
the `master` branch (in #5820) with an updated test case. To avoid
cumbersome merge conflicts, this commit don't include any test case.
|
|\ \ \ \ \
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
* sverker/24/erts/fix-factory-undo/OTP-18027:
erts: Make erts_factory_undo leave empty heap fragment
erts: Fix bug in erts_factory_proc_prealloc_init
erts: Ensure setting p->live_hf_end at first trap with disabled GC
|
| |\ \ \ \ \
| | |_|_|_|/
| |/| | | |
| | | | | | |
sverker/24/erts/fix-factory-undo
|
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
if it was empty when we started.
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
causing erts_factory_undo() to leave garbage in already existing
heap fragment.
|
| | |/ / / |
|
|\ \ \ \ \
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
* kuba/inets/fix_all_handling_mod_esi/OTP-18069:
inets: fix 'all' handling for erl_script_alias
|
| |/ / / /
| | | | |
| | | | |
| | | | | |
- convert Module string to a known atom before generating webpage
|
|\ \ \ \ \
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
* john/stdlib/fix-shell-docs-spec/OTP-18081:
stdlib: Fix specs for shell_docs:get_type_doc/3
stdlib: Fix specs for shell_docs:get_doc/3 and get_callback_doc/3
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
... I missed this in the previous commit, and merged forward to
`maint` and `master` before noticing it. :-(
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The spec said that `Metadata` must be the empty map, which is not
what we meant.
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
* ingela/ssl/inet-backend/GH-5930/OTP-18071:
ssl: Remove is_port guards for sockets
|
| | |/ / / /
| |/| | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Sockets may have other types then port, remove these checks as it
violates socket abstraction and breaks the code for gen_tcp sockets
with new inet_backend.
Closes #5930
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
maint-24
* ingela/ssl/decode-cert-request-algs/GH-5835/OTP-18028:
Update lib/ssl/src/tls_gen_connection.erl
Update lib/ssl/src/tls_gen_connection.erl
Update lib/ssl/src/tls_gen_connection.erl
Update lib/ssl/src/tls_gen_connection.erl
ssl: Fix version mismatch
|
| | | | | | | |
|
| | | | | | | |
|
| | | | | | | |
|
| | | | | | |
| | | | | | |
| | | | | | | |
Co-authored-by: Kenneth Lundin <kenneth.lundin@gmail.com>
|
| |/ / / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The special case handled by the tls_gen_connection:effective_version
function is only valid in the hello and wait_sh state.
Closes #5835
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
* john/erts/fix-persistent-term-get-put-race/maint-24:
erts: Fix a race in persistent_term:get/1,2
|