Age | Commit message (Collapse) | Author |
|
|
|
* bjorn/erts/beam_makeops:
beam_makeops: Turn on warnings and eliminate existing warnings
beam_makeops: Eliminate a deprecation warning
|
|
Conflicts:
erts/etc/common/heart.c
|
|
* egil/ensure-erl_crash.dump/OTP-10422:
test: Relax timeouts for heart_SUITE
erts: Fix crash dump write to port hack
erts: Fix lock check assertion
doc: Document ERL_CRASH_DUMP_SECONDS behaviour
test: Add test for heart restart on crash
test: Add test for heart restart on crash
erts: Change ERL_CRASH_DUMP_SECONDS behaviour
test: Refactor away ?line macros in heart_SUITE
erts: Search for heart in ports that are alive
heart: Refactor heart debugging
erts, heart: Ensure erl_crash.dump is written
|
|
* sverk/valgrind-quoting:
Make cerl -valgrind work with quoted spaces in command line
|
|
Invalid write of size 8
(Address 0x84a63c0 is not stack'd, malloc'd or (recently) free'd)
Suspected: AES_cbc_encrypt:??? (-> 0xD7489CF) [/lib/libcrypto.so.0.9.8]
|
|
More future proof with R16
|
|
|
|
|
|
* heart reboot behaviour
* erl_crash.dump file write behaviour
|
|
* jf/fix_sctp_peeloff_active_true:
Set new peeled off SCTP socket to nonblocking socket
SCTP test case with socket active options once and true
OTP-10491
|
|
to work even when compiler does aggressive function inlining
|
|
Not setting ERL_CRASH_DUMP_SECONDS will now terminate beam
immediately on a crash without writing a crash dump file.
Setting ERL_CRASH_DUMP_SECONDS to 0 will also terminate beam
immediately on a crash without writing a crash dump file, i.e. same as not
setting ERL_CRASH_DUMP_SECONDS environment variable.
Setting ERL_CRASH_DUMP_SECONDS to a negative value will let the beam wait
indefinitely on the crash dump file being written.
Setting ERL_CRASH_DUMP_SECONDS to a positive value will let the beam wait
that many seconds on the crash dump file being written.
A positive value will set both an alarm in beam AND a heart timeout for restart
if heart is running.
This is due to the change of 'heart' behavior when 'heart' is
listening for a crash.
|
|
|
|
|
|
When a crash dump is about to be written and we have
heartbeat enabled on a system. We need time to write it
before heart explicitly kills the beam.
|
|
Perl 5.16.1 (and perhaps other versions) issues the following
warning:
defined(@array) is deprecated at utils/beam_makeops line 1714.
(Maybe you should just omit the defined()?)
for the following line:
$prev_last = pop(@{$gen_transform{$key}})
if defined @{$gen_transform{$key}}; # LINE 1714
The documentation for "defined" says that its use on hashes and
arrays is deprecated and that it may stop working in a future
release.
Simply removing "defined" (as suggested by the warning message)
will not work, as there will be an error when trying to use an
undefined value as an array reference:
Can't use an undefined value as an ARRAY reference at
utils/beam_makeops line 1714.
What we must do is to check whether $gen_transform{$key} is
defined before trying to use it as an array reference.
Noticed-by: Tuncer Ayaz
|
|
|
|
* lukas/erts/dont_break_reductions_skip/OTP-10373:
Skip dont_break_reductions on hipe libs tests
|
|
* lukas/erts/bad_terms_hipe_skip/OTP-10375:
Skip fun corruption when lists is native
|
|
|
|
This is needed as corruption of the index_uniq byte
can cause very strange behaviour when the fun is called
by native code.
|
|
|
|
* egil/fix-boot_combo-test:
test: Fix smoke_test_SUITE
|
|
On architectures where cpu topology was unavailable,
warnings erupted due to unused static functions.
Those warnings are now silenced.
|
|
|
|
* Silence compiler warning
* This is the same style as the rest of the debug printing in hipe.
Keeping to the same style, though changing the printing format
would probably be better. Not a critical change.
|
|
Facts:
crypto nif-lib registers callback functions that openssl uses
for memory management and thread synchronization. The callback
functions can only be set once, openssl does not allow changing the
callback functions.
Problem:
If openssl is dynamicly linked to crypto, you might get s scenario
where the crypto lib is unloaded while leaving openssl loaded
with its old pointers to the unloaded crypto code intact.
If crypto is then reloaded (by init:restart() for example), the crypto
nif-lib might get relocated at a different address. crypto calls
openssl which in turn calls the old invalid callback functions...kaboom.
Solution:
Break apart the callback functions into a separate dynamic lib that
crypto loads with dlopen. When crypto is unloaded the callback lib is
left in place to be reused if/when crypto is loaded again.
|
|
* maint:
Fix bug when making nsis target
|
|
* lukas/erts/whitespace_nsis_fix/OTP-10481:
Fix bug when making nsis target
|
|
* Skip boot_combo tests where cannot start slave nodes
(The binary will not exist)
|
|
|
|
* maint:
Add comment about eproviderfailedinit error
Increase sbct for win64 as block alignment is 16
|
|
|
|
|
|
* egil/remove-unused-files:
Remove unused files from old native dns resolver
|
|
Peeloff feature of SCTP association creates a new socket which is not
set to nonblocking.
Function for receving data is shared with udp which has a default
loop when reading packets which is set to 5.
Calling the receive function more then once, is fine as long as
there are more data to receive or socket are nonblocking.
Set new peeled off socket to be nonblocking to prevent a erlang vm hangup.
|
|
* sverk/code-load-refactor-later-op:
erts: Fix faulty lock check assert
erts: Refactor naming regarding code_write_permission
erts: Refactor tracing to use erts_schedule_thr_prgr_later_op
erts: Allow thr_prgr_later_op to reschedule
erts: Refactor code loading to use erts_schedule_thr_prgr_later_op
erts: Remove some compiler warnings
|
|
Sometime in cross environments the documentation will
create the lib folders, but there will be no ebin so
all undefined functions should then be ignored.
|
|
|
|
|
|
In the erl_crash.dump file, native-compiled modules did not have
any information about attributes and compilation.
The problem is that the code:make_stub_module/3 BIF (which is
internally used when native code is loaded) did not copy the size
field the attribute and compilation info chunks. Those size fields
are only used when writing crash dumps.
|
|
* egil/r16/strengthen-buffer-copies:
Replace sprintf with erts_snprintf in beam
Replace sprintf with erts_snprintf in epmd
Replace sprintf with erts_snprintf in inet_gethost
|
|
|
|
* egil/fix-configure-openbsd/OTP-10395:
Fix linking of shared libraries on OpenBSD
|
|
|
|
|
|
|
|
* sverk/gcc-atomic-config-bug:
erts: Fix gcc atomic bug in ethread.h
OTP-10418
|
|
* sverk/gcc-atomic-config-bug:
erts: Fix gcc atomic bug in ethread.h
|