1
<?xml version="1.0" encoding="latin1" ?>
2
<!DOCTYPE chapter SYSTEM "chapter.dtd">
9
<holder>Ericsson AB, All Rights Reserved</holder>
12
The contents of this file are subject to the Erlang Public License,
13
Version 1.1, (the "License"); you may not use this file except in
14
compliance with the License. You should have received a copy of the
15
Erlang Public License along with this software. If not, it can be
16
retrieved online at http://www.erlang.org/.
18
Software distributed under the License is distributed on an "AS IS"
19
basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
20
the License for the specific language governing rights and limitations
23
The Initial Developer of the Original Code is Ericsson AB.
26
<title>ERTS Release Notes</title>
27
<prepared>otp_appnotes</prepared>
32
<p>This document describes the changes made to the ERTS application.</p>
34
<section><title>Erts 5.6.5</title>
36
<section><title>Fixed Bugs and Malfunctions</title>
40
A bug in inet_drv concerning gen_tcp:connect has been
41
corrected. A connect towards a non-open port through open
42
firewalls could sometimes erroneously be successful. Any
43
subsequent operation would fail, though.</p>
49
Floating point arithmetics in drivers could cause a
50
runtime system crash and/or unexpected results on runtime
51
systems with floating point exceptions enabled. Floating
52
point exceptions are disabled unless explicitly enabled
53
or if hipe is enabled.</p>
59
A bug when many sockets got signalled simultaneously
60
causing the emulator to panic with the message
61
"Inconsistent, why isnt io reported?" is now corrected.</p>
67
Starting erl with option "-detached" now disconnects
68
correctly from terminal session on Unix.</p>
74
Mended gdb etp-commands for ETS access.</p>
80
<c>erlang:decode_packet/3</c> allows white space between
81
HTTP header tag and colon according to RFC2616.</p>
87
An emulator compiled for SCTP now starts even if the
88
dynamic libraries are not present. The SCTP driver is
89
then of course not loaded.</p>
94
<p>To build on Mac OS X, 10.3.0 or later is now required
95
because of fixes for two problems:</p>
96
<p>There would be a resource leak when <c>erl_ddl</c>
97
attempted to unload a driver. This problem has been
98
corrected by using <c>dlopen()</c> (which works on all
99
modern Unix-like platforms) instead of the Mac OS X
100
specific API calls.</p>
101
<p>Signal handling in the run-time system for HiPE has
102
been updated to work on later versions of Mac OS X than
103
10.2.x. Therefore, <c>--enable-hipe</c> now works on Mac
104
OS X with Intel CPUs.</p>
105
<p>Thanks to Geoff Cant for the patches.</p>
110
<p>Corrected some information about the protocol between
111
EPMD and Erlang nodes. (Thanks to Michael Regen.)</p>
118
<c>erlang:system_monitor(Pid,{long_gc,Time})</c>, and the
119
GC time exceeded 1 second, it sometimes erroneously
120
showed up as about 4300 seconds. (This bug was corrected
121
in R9C, but re-introduced in R12B.) (Thanks to Chris
124
Own Id: OTP-7622 Aux Id: OTP-4903, seq8379 </p>
130
<section><title>Improvements and New Features</title>
134
The driver entry of a dynamically loaded driver is now
135
copied when loaded which enables some internal
136
optimizations. Note that drivers that modify the driver
137
entry during execution will not work anymore. Such a
138
miss-use of the driver interface is however not
145
The split function is now added to the re library.
146
Exceptions and errors from both run, replace and split
147
are made more consistent.</p>
149
Own Id: OTP-7514 Aux Id: OTP-7494 </p>
153
Fixed harmless compiler warnings when building the
154
emulator and minor build changes in order to avoid
155
unnecessary rebuilds.</p>
161
There is now experimental support for loading of code
162
from archive files. See the documentation of <c>code</c>,
163
<c>init</c>, <c>erl_prim_loader </c> and <c>escript</c>
166
The error handling of <c>escripts</c> has been improved.</p>
168
An <c>escript</c> may now set explicit arguments to the
169
emulator, such as <c>-smp enabled</c>.</p>
171
An <c>escript</c> may now contain a precompiled beam
174
An <c>escript</c> may now contain an archive file
175
containing one or more applications (experimental).</p>
177
The internal module <c>code_aux</c> has been removed.</p>
179
Own Id: OTP-7548 Aux Id: otp-6622 </p>
183
The reallocation functionality part of the ERTS internal
184
memory allocators, now consider current block in
185
combination with surrounding free blocks as an
186
alternative location for a reallocation.</p>
191
<p>There could remain false references from a process to
192
a module that has been called earlier, so that the
193
process would be killed if the module was reloaded.
194
(Thanks to Richard Carlsson.)</p>
195
<p>Also, the fix for this bug also made it possible to
196
make stack backtraces (as returned from
197
<c>erlang:get_stacktrace/0</c> and other functions) more
198
correct in that the immediate caller is always included
199
in the stack backtrace (it could sometimes be
206
Improved locking in IO-handling for better smp
213
Improved BIF rescheduling functionanality.</p>
219
Loading a module compiled with Erlang/OTP R9C and calling
220
<c>module_info/0</c> in the module would crash the
221
emulator. The emulator now refuses to load any module
222
compiled with R9C or earlier. (Note: only trivial modules
223
compiled with R10B or earlier could be loaded anyway.)
224
(Thanks to Martin Kjellin.)</p>
233
<section><title>Erts 5.6.4.2</title>
235
<section><title>Fixed Bugs and Malfunctions</title>
239
A process calling one of the following BIFs could under
240
very rare conditions deadlock in the runtime system with
241
SMP support: <c>check_process_code/2</c>,
242
<c>garbage_collect/1</c>, <c>process_info/[1,2]</c>,
243
<c>system_flag/2</c>, and
244
<c>erlang:suspend_process/[1,2]</c>.</p>
250
A couple of statistics variables were not managed in a
251
thread safe manner in the runtime system with SMP
258
An extremely rare race condition when terminating a
259
process could potentially cause a runtime system crash.</p>
265
Under certain conditions and when using run_erl/to_erl,
266
the terminal Erlang driver (ttsl_drv) could crash the
267
emulator by doing a division by zero due to incorrect
268
handling of terminals reporting a zero width. For
269
terminals reporting zero width, the driver now fallbacks
270
to a default width of 80 and a default height of 24
271
(vt100), as a fallback behaviour. This fixes the crashes
272
and also makes output on "dumb" terminals much more
275
Own Id: OTP-7592 Aux Id: seq11073 </p>
282
<section><title>Erts 5.6.4.1</title>
284
<section><title>Improvements and New Features</title>
288
A new <c>erts_alloc</c> parameter
289
<c>+M<S>rmbcmt</c> (relative multiblock carrier
290
move threshold) has been added. It determines when to
291
force a moving realloc in a multiblock carrier when a
292
block is shrunk. For more information see the
293
<c>erts_alloc(3)</c> documentation.</p>
298
<p>The new option <c>+d</c> can be given to <c>erl</c> to
299
suppress the crash dump generation if an internal error
300
is detected. As a result, a more useful core dump is
303
Own Id: OTP-7578 Aux Id: seq11052 </p>
310
<section><title>Erts 5.6.4</title>
312
<section><title>Fixed Bugs and Malfunctions</title>
316
Double backslashes in format string passed to the erts
317
internal printf implementation produced erroneous
318
results. No such format strings were passed to the erts
319
internal printf implementation, i.e., the bug was
320
therefore completely harmless. (Thanks to Perry Smith.)</p>
326
Large files are now handled on Windows, where the
327
filesystem supports it.</p>
333
Bug fixed for <c>{packet,http}</c> when space follows
340
The trace option <c>running</c> could cause an emulator
341
crash if the current function couldn't be determined.</p>
347
Using 16#ffffFFFF as a timeout value in receive...after
348
would often cause a timeout almost at once due to an
349
32-bit integer overflow. (Thanks to Serge Aleynikov and
356
For the process that an escript runs in, the
357
<c>trap_exit</c> process flag is now <c>false</c> instead
358
of <c>true</c> (as in previous releases). Scripts that
359
depend on the previous (counter-intuitive) behaviour
360
might not work. (Thanks to Bengt Kleberg.)</p>
362
*** POTENTIAL INCOMPATIBILITY ***</p>
368
A bug in the <c>string:to_integer/1</c> builtin made the
369
emulator unstable. This is now corrected. (Thanks to Lev
378
<section><title>Improvements and New Features</title>
381
<p>Performance for ETS intensive applications should now
382
be better in the SMP emulator. Also, ETS table
383
identifiers (as returned by <c>ets:new/2</c>) are now
384
spread out in wider range than before (using 28 bits in a
385
32-bit emulator) to make sure that the table identifier
386
for a deleted table will not be quickly re-used.</p>
387
<p>NOTE: Table identifiers can now be negative integers.
388
Programs that (incorrectly) assume that table identifiers
389
can only be positive integers might stop to work. (The
390
type of a table identifier is intentionally not
391
documented, and may change in a future release.)</p>
397
New BIF <c>erlang:decode_packet/3</c> that extracts a
398
protocol packet from a binary. Similar to the socket
399
option <c>{packet, Type}</c>. Also documented the socket
400
packet type <c>http</c> and made it official.
401
<em>NOTE</em>: The tuple format for <c>http</c> packets
402
sent from an active socket has been changed in an
403
incompatible way.</p>
405
*** POTENTIAL INCOMPATIBILITY ***</p>
411
The source code for the documentation for some
412
applications (erts, kernel, stdlib, and several others)
413
are now included in the source tar ball. There is
414
currently no Makefile support for building HTML files
415
from the source (such support will be included in a
422
A lot of frequently accessed memory counters (erts
423
internal) have been removed. This since they hurt
424
performance on the runtime system with SMP support. As a
425
result <c>erlang:memory/[0,1]</c> will only deliver a
426
result if all <c>erts_alloc(3)</c> allocators are enabled
427
(default). The result delivered when all
428
<c>erts_alloc(3)</c> allocators are enabled are both more
429
accurate and less accurate than before. More memory than
430
before are included in the result, but the different
431
parts that are summed are not gathered atomically. A call
432
to <c>erlang:memory/[0,1]</c> is much cheaper for the
433
system than before. This since the information isn't
434
gathered atomically anymore which was very expensive.</p>
436
*** POTENTIAL INCOMPATIBILITY ***</p>
442
Pre-allocators used for, for example, timers, and
443
messages have been rewritten to be scheduler specific.
444
That is, different schedulers will now allocate from
445
different pools which reduces lock contention.</p>
451
On Mac OS X, file:sync/1 now guarantees that all
452
filesystem buffers are written to the disk by using the
453
fcntl() with F_FULLFSYNC option. Previously, file:sync/1
454
called fsync(), which only guaranteed that the data had
455
been transferred to the disk drive. (Thanks to Jan
462
Termination of a process that takes a long time can now
463
be preempted, i.e., the terminating process will be
464
rescheduled for later continuation of termination so that
465
other processes can execute. Termination of a process
466
that owns many and/or large ets tables typically takes a
473
A new trace option <c>exiting</c> has been added. The
474
<c>exiting</c> trace is similar to the <c>running</c>
475
trace, but for exiting processes. For more information
476
see the erlang(3) documentation.</p>
478
The <c>erlang:trace/3</c> bif now doesn't block other
479
scheduler threads if only one tracee is specified in the
480
call to <c>erlang:trace/3</c>.</p>
486
The re module is exdended with repetitive matches (global
487
option) and replacement function.</p>
489
Own Id: OTP-7494 Aux Id: OTP-7181 </p>
493
In the section about binary construction, the reference
494
manual now mentions what happens when an integer value
495
does not fit into an integer segment of size N (namely,
496
that the N least significant bits will be put into into
497
the binary and that the most significant bits will be
498
silently discarded). (Thanks to Edwin Fine.)</p>
504
Setting the <c>{active,once}</c> for a socket (using
505
inets:setopts/2) is now specially optimized (because the
506
<c>{active,once}</c> option is typically used much more
507
frequently than other options).</p>
515
<section><title>Known Bugs and Problems</title>
519
Floating point arithmetics in drivers can cause a runtime
520
system crash and/or unexpected results on runtime systems
521
with floating point exceptions enabled. Floating point
522
exceptions are disabled unless explicitly enabled or if
532
<section><title>Erts 5.6.3.3</title>
534
<section><title>Fixed Bugs and Malfunctions</title>
538
Binary construction with an integer field of size 0 at
539
the end of the constructed binary (and the size given in
540
a variable), would cause a write of one byte outside the
541
memory reserved for the binary, which in turn could cause
542
an emulator crash.</p>
548
A race condition in the dynamic driver implementation
549
could cause an emulator crash. (Thanks to Paul Fisher)</p>
555
Calls to <c>erlang:system_info(allocated_areas)</c> could
556
cause the runtime system with SMP support to crash.</p>
562
The <c>env</c> option to <c>open_port()</c> could cause
563
the runtime system with SMP support to crash.</p>
571
<section><title>Improvements and New Features</title>
575
Operations that needed to block other threads in the
576
runtime system with SMP support unnecessarily waited for
577
async threads to block. Most important the
578
<c>erlang:memory/[0,1]</c> bif, code loading, and the
579
<c>erlang:trace/3</c> bif.</p>
588
<section><title>Erts 5.6.3.2</title>
590
<section><title>Fixed Bugs and Malfunctions</title>
594
Calls to <c>erlang:memory/[0,1]</c> could cause the
595
runtime system with SMP support to crash.</p>
604
<section><title>Erts 5.6.3.1</title>
606
<section><title>Fixed Bugs and Malfunctions</title>
610
Doing local call trace on bit syntax matching code that
611
has been optimized with delayed sub-binary creation could
612
crash the emulator.</p>
614
Own Id: OTP-7399 Aux Id: seq10978 </p>
621
<section><title>Erts 5.6.3</title>
623
<section><title>Fixed Bugs and Malfunctions</title>
627
Only one to_erl process at a time is allowed to connect
628
to the same run_erl pipe. Prevents buggy behaviour when
629
IO from several to_erl's get interleaved.</p>
635
IPv6 name resolving has now been fixed to use
636
getaddrinfo() patch (thorughly reworked) courtesy of Love
637
H�rnquist-�strand submitted by Fredrik Thulin. It also
638
can use gethostname2() patch (also reworked) courtesy of
639
Mikael Magnusson for debian submitted by Sergei Golovan.</p>
645
Improved error handling in run_erl</p>
650
<p>A permanent fix for the deadlock issue temporarily
651
fixed by OTP-7260.</p> <taglist><tag>OTP-7260</tag><item>
652
The runtime system with SMP support could under rare
653
circumstances deadlock when a distribution channel was
654
taken down while multiple simultaneous operations were
655
performed on it. </item></taglist>
657
Own Id: OTP-7267 Aux Id: OTP-7260 </p>
661
./configure has been improved to find 64-bit OpenSSL
668
A terminating process could under very rare circumstances
669
trigger a bug which could crash the runtime system with
676
SCTP_ADDR_CONFIRMED events are now handled by gen_sctp.</p>
682
binary_to_term/1 would crash the emulator if the binary
683
data contained an external fun with non-atom module
684
and/or function. Corrected to generate a badarg
691
On Mac OS 10.5 (Leopard), sending to socket which the
692
other end closes could cause the emulator to consume 100%
693
CPU time. (Thanks to Matthias Radestock.)</p>
699
The vanilla driver used on Windows could crash the
700
emulator and sometimes produced corrupt files. The
701
vanilla driver is the driver that is used when one only
702
pass a filename as first argument to <c>open_port/2</c>.
703
<em>NOTE</em>: This use of <c>open_port/2</c> is
704
<em>obsolete</em>, and the documentation of this use has
705
previously been removed. The functionality is only
706
present for backward compatibility reasons and
707
<em>will</em> eventually be removed.</p>
713
Faulty matching in binaries larger than 512Mb on 64bit
714
machines fixed.(On 32bit, the size limit for binaries is
715
still 512Mb). Thanks to Edwin Fine and Per Gustafsson for
716
finding fault and fix.</p>
722
crypto:start() on Windows caused emulator to hang on
723
error popup window if openssl DLL was not found. Windows
724
error popups now suppressed.</p>
730
Configuration option <c>witout-termcap</c> can be used to
731
disable the use of termcap libraries for terminal cursor
732
control in the shell.</p>
738
to_erl reports its terminal window size to run_erl in
739
order to get output formatted accordingly</p>
745
On Solaris, the <c>compressed</c> option for file
746
operations did not work if the file descriptor happened
747
to be greater than 255 (a problem with fopen() and
748
friends in Solaris itself).</p>
750
Own Id: OTP-7343 Aux Id: seq10949 </p>
754
A race condition in the runtime system with SMP support
755
causing an erroneous removal of a newly created ets table
758
The race occurred when a process removed a table during
759
termination simultaneously as another process removed the
760
same table via <c>ets:delete/1</c> and a third process
761
created a table that accidently got the same internal
762
table index as the table being removed.</p>
768
<c>zlib:inflate</c> failed when the size of the inflated
769
data was an exact multiple of the internal buffer size
770
(4000 bytes by default).</p>
776
If the total number of allowed atoms is exceeded, there
777
will now be a controlled termination of the emulator with
778
a crash dump file. The emulator used to simply crash.
779
(Thanks Howard Yeh and Thomas Lindgren.)</p>
785
The break handler in werl on Windows could cause the
786
emulator to hang or crash, that is now corrected.</p>
788
Own Id: OTP-7394 Aux Id: seq10969 </p>
792
The configure script now tests for an serious
793
optimization bug in gcc-4.3.0. If the bug is present, the
794
configure script will abort (if this happens, the only
795
way to build Erlang/OTP is to change to another version
796
of gcc). (Thanks to Mikael Pettersson.)</p>
804
<section><title>Improvements and New Features</title>
808
On Unix, the emulator now notices when the width of the
809
terminal has changed. (Thanks to Matthew Dempsky and
810
Patrick Mahoney.)</p>
816
There is a new function <c>init:stop/1</c> which can be
817
used to shutdown the system cleanly AND generate a
818
non-zero exit status or crash dump. (Thanks to Magnus
825
process_info(Pid, garbage_collection) now returns more
832
The <c>hide</c> option for <c>open_port/2</c> is now
833
documented. (Thanks to Richard Carlsson.)</p>
841
<section><title>Known Bugs and Problems</title>
845
Floating point arithmetics in drivers can cause a runtime
846
system crash on runtime systems with floating point
847
exceptions enabled. Floating point exceptions are
848
disabled unless explicitly enabled or if hipe is enabled.</p>
858
<section><title>Erts 5.6.2</title>
860
<section><title>Fixed Bugs and Malfunctions</title>
864
The maximum length of an atom of 255 characters is now
865
strictly enforced. <c>binary_to_term/1</c> will now fail
866
with a badarg if an encoded term contains an atom longer
867
than 255 characters. Atoms created by drivers will now be
868
truncated to 255 characters if necessary. (Thanks to
869
Matthew Dempsky.)</p>
875
A bug in "bignum handling" on some 64bit architectures
876
could cause rem and div operations on large numbers to
877
hang indefinitely. Rem operations involving the smallest
878
negative number representable in 28 bits or 60 bits could
879
also cause access violation and emulator crash. Both
880
errors are corrected.</p>
886
When doing the initial garbage collection after waking a
887
hibernated process, a fullsweep garbage collection was
888
unnecessarily triggered.</p>
893
<p>The beta testing module <c>gen_sctp</c> now supports
894
active mode as stated in the documentation. Active mode
895
is still rather untested, and there are some issues about
896
what should be the right semantics for
897
<c>gen_sctp:connect/5</c>. In particular: should it be
898
blocking or non-blocking or choosable. There is a high
899
probability it will change semantics in a (near) future
900
patch.</p> <p>Try it, give comments and send in bug
907
Invalid arguments to <c>ets:update_counter/3</c> were not
908
handled correctly. A tuple position (<c>Pos</c>) less
909
than 1 caused the element directly following the key to
910
be updated (as if no position at all had been specified).
911
All invalid values for <c>Pos</c> will now fail with
918
The runtime system with SMP support could under rare
919
circumstances deadlock when a distribution channel was
920
taken down while multiple simultaneous operations were
929
<section><title>Improvements and New Features</title>
933
More checksum/hash algorithms from the zlib library are
934
now available as built in functions (like md5 hashes has
935
been for a long time).</p>
941
Minor improvements in the garbage collector.</p>
943
Own Id: OTP-7139 Aux Id: OTP-7132 </p>
947
The switch "-detached" to the windows werl program now
948
can create an erlang virtual machine without any main
949
window and without a temporary console showing.</p>
954
<p><c>erlang:system_info/1</c> now accepts the
955
<c>logical_processors</c>, and <c>debug_compiled</c>
956
arguments. For more info see the, <c>erlang(3)</c>
957
documentation.</p> <p>The scale factor returned by
958
<c>test_server:timetrap_scale_factor/0</c> is now also
959
effected if the emulator uses a larger amount of
960
scheduler threads than the amount of logical processors
967
A new BIF ets:update_element/3. To update individual
968
elements within an ets-tuple, without having to read,
969
update and write back the entire tuple.</p>
975
A process executing the <c>processes/0</c> BIF can now be
976
preempted by other processes during its execution. This
977
in order to disturb the rest of the system as little as
978
possible. The returned result is, of course, still a
979
consistent snapshot of existing processes at a time
980
during the call to <c>processes/0</c>.</p>
982
The documentation of the <c>processes/0</c> BIF and the
983
<c>is_process_alive/1</c> BIF have been updated in order
984
to clarify the difference between an existing process and
985
a process that is alive.</p>
995
<section><title>Erts 5.6.1.1</title>
997
<section><title>Fixed Bugs and Malfunctions</title>
1001
Not enough parameters were passed when sending an error
1002
report in erl_check_io.c (Thanks to Matthew Dempsky).</p>
1004
Own Id: OTP-7176</p>
1008
In rare circumstances, complex binary matching code could
1009
cause the emulator to crash or not match when it should.
1010
(Thanks to Rory Byrne.)</p>
1012
Own Id: OTP-7198</p>
1018
<section><title>Improvements and New Features</title>
1021
<p>The <c>{allocator_sizes, Alloc}</c> and
1022
<c>alloc_util_allocators</c> arguments are now accepted
1023
by <c>erlang:system_info/1</c>. For more information see
1024
the <c>erlang(3)</c> documentation.</p>
1026
Own Id: OTP-7167</p>
1030
The finishing reallocation of the heap block when
1031
hibernating a process is now always moving the heap block
1032
since it drastically reduces memory fragmentation when
1033
hibernating large amounts of processes.</p>
1035
Own Id: OTP-7187</p>
1042
<section><title>Erts 5.6.1</title>
1044
<section><title>Fixed Bugs and Malfunctions</title>
1048
The SMP emulator on sparc64 erroneously used the sparc32
1049
atomic and the sparc32 spinlock implementations which
1050
caused it to crash.</p>
1052
Own Id: OTP-7006</p>
1056
Call tracing the new guard BIFs <c>byte_size</c>,
1057
<c>bit_size</c>, or <c>tuple_size</c> and the loading a
1058
module that uses one of those functions, could cause the
1059
emulator to terminate.</p>
1061
Own Id: OTP-7008</p>
1065
configuring --enable-darwin-universal or
1066
--enable-darwin-64bit on MacOSX could result in a non
1067
optimized emulator. Top level configure script now
1070
Own Id: OTP-7014</p>
1074
configuring --with-gd did not produce correct include
1075
flags for percept.</p>
1077
Own Id: OTP-7015</p>
1081
Environment variables weren't handled in thread safe
1082
manner in the runtime system with SMP support on Windows.</p>
1084
<c>erl_drv_putenv()</c>, and <c>erl_drv_getenv()</c> has
1085
been introduced for use in drivers. Do <em>not</em> use
1086
putenv(), or getenv() directly in drivers. For more
1087
information see the <c>erl_driver</c> documentation.</p>
1089
Own Id: OTP-7035</p>
1093
HIPE: Corrected the choice of interface to the send/3 and
1094
setnode/3 BIFs for native-compiled code. Using the
1095
incorrect interface could, in unusual circumstances, lead
1096
to random runtime errors.</p>
1098
Own Id: OTP-7067</p>
1102
Garbage collections could become extremely slow when
1103
there were many keys in the process dictionary. (Thanks
1104
to Fredrik Svahn.)</p>
1106
Own Id: OTP-7068</p>
1110
The duplicate documentation directory in the windows
1111
installation is removed.</p>
1113
Own Id: OTP-7070</p>
1116
<p>Documentation bugfixes and clarifications.</p> (Thanks
1117
to Joern (opendev@gmail.com), Matthias Lang, and Richard
1120
Own Id: OTP-7079</p>
1124
The runtime system with SMP support <em>not</em> using
1125
the native atomic integer implementation part of OTP
1126
could deadlock when run on a system with more than one
1127
logical processor. That is, only the runtime system with
1128
SMP support on <em>other</em> hardware platforms than
1129
x86, x86_64, sparc32, and powerpc32 were effected by this
1132
Own Id: OTP-7080</p>
1136
The break handling code (run when Ctrl-C is hit) could
1137
could potentially deadlock the runtime system with SMP
1140
Own Id: OTP-7104</p>
1144
The sctp driver has been updated to work against newer
1145
lksctp packages e.g 1.0.7 that uses the API spelling
1146
change adaption -> adaptation. Older lksctp (1.0.6) still
1147
work. The erlang API in gen_sctp.erl and inet_sctp.hrl
1148
now spells 'adaptation' regardless of the underlying C
1151
*** POTENTIAL INCOMPATIBILITY ***</p>
1153
Own Id: OTP-7120</p>
1156
<p>A bug in <c>erlang:phash2/1</c> on 64-bit platforms
1157
has been fixed. (Thanks to Scott Lystig Fritchie.)</p>
1159
Own Id: OTP-7127</p>
1163
The emulator could under rare circumstances crash while
1164
garbage collecting.</p>
1166
Own Id: OTP-7132</p>
1172
<section><title>Improvements and New Features</title>
1175
<p>The documentation has been updated so as to reflect
1176
the last updates of the Erlang shell as well as the minor
1177
modifications of the control sequence <c>p</c> of the
1178
<c>io_lib</c> module.</p> <p>Superfluous empty lines have
1179
been removed from code examples and from Erlang shell
1182
Own Id: OTP-6944 Aux Id: OTP-6554, OTP-6911 </p>
1186
Bit syntax construction with a small integer in a
1187
non-byte aligned field wider than the CPU's word size
1188
could cause garbage bits in the beginning of the field.</p>
1190
Own Id: OTP-7085</p>
1194
All Windows versions older than Windows 2000 are now
1195
<em>not supported</em> by the Erlang runtime system. This
1196
since there was a need for usage of features introduced
1197
in Windows 2000.</p>
1199
Own Id: OTP-7086</p>
1202
<p>Memory management improvements especially for the
1203
runtime system with SMP support:</p> <list> <item> The
1204
runtime system with SMP support can now use multiple,
1205
thread specific instances of most memory allocators. This
1206
improves performance since it reduces lock contention in
1207
the memory allocators. It may however increase memory
1208
usage for some applications. The runtime system with SMP
1209
support will by default enable this feature on most
1210
allocators. The amount of instances used can be
1211
configured. </item> <item> <c>driver_alloc()</c>,
1212
<c>driver_realloc()</c>, and <c>driver_free()</c> now use
1213
their own erts specific memory allocator instead of the
1214
default <c>malloc()</c> implementation on the system.
1215
</item> <item> The default configuration of some
1216
allocators have been changed to fit applications that use
1217
much memory better. </item> <item> Some new
1218
<c>erts_alloc</c> configuration parameters have been
1219
added. </item> <item> <c>erts_alloc_config</c> has been
1220
modified to be able to create configurations suitable for
1221
multiple instances of allocators. </item> <item> The
1222
returned value from <c>erlang:system_info({allocator,
1223
Alloc})</c> has been changed. This since an allocator may
1224
now run in multiple instances. </item> </list> <p>If you
1225
for some reason want the memory allocators to be
1226
configured as before, you can pass the <c>+Mea r11b</c>
1227
command-line argument to <c>erl</c>.</p> <p>For more
1228
information see the <c>erts_alloc(3)</c>, the
1229
<c>erts_alloc_config(3)</c>, and the <c>erlang(3)</c>
1232
*** POTENTIAL INCOMPATIBILITY ***</p>
1234
Own Id: OTP-7100</p>
1238
On Unix, denormalized floating point numbers could not be
1239
created using <c>list_to_float/1</c> or
1240
<c>binary_to_term/1</c>. (Thanks to Matthew Dempsky.)</p>
1242
Own Id: OTP-7122</p>
1246
Native atomic integers and spin-locks are now also
1247
available for the runtime system with SMP support on
1250
Own Id: OTP-7130</p>
1254
FP exceptions support for sparc64 userspace on Linux has
1255
been added. Note that FP exception support is now turned
1256
off by default, so to actually enable it you need to do
1257
'<c>./configure --enable-fp-exceptions</c>'.</p>
1259
Own Id: OTP-7131</p>
1266
<section><title>Erts 5.6</title>
1268
<section><title>Fixed Bugs and Malfunctions</title>
1272
A bug for raw files when reading 0 bytes returning 'eof'
1273
instead of empty data has been corrected.</p>
1275
Own Id: OTP-6291 Aux Id: OTP-6967 </p>
1279
All exported functions in gzio.c have now been renamed to
1280
avoid conflict with drivers that are indirectly linked
1281
with an external zlib library.</p>
1283
Own Id: OTP-6816 Aux Id: OTP-6591 </p>
1287
On the 64-bit Erlang emulator, bit syntax construction
1288
with integers containing more than 60 bits ("big
1289
numbers") into fields with more than 60 bits could
1290
produce incorrect results.</p>
1292
Own Id: OTP-6833</p>
1296
When the runtime system failed to allocate memory for
1297
binaries, it could dead lock while writing the
1298
<c>erl_crash.dump</c>.</p>
1300
Own Id: OTP-6848</p>
1304
The runtime system with SMP support could deadlock if a
1305
process called the <c>erlang:suspend_process(Pid)</c> BIF
1306
or the <c>erlang:garbage_collect(Pid)</c> BIF while the
1307
process identified by <c>Pid</c> was currently running
1308
and the process calling the BIFs was terminated during
1309
the call to the BIFs.</p>
1311
Processes suspending each other via the
1312
<c>erlang:suspend_process/1</c> BIF or garbage collecting
1313
each other via the <c>erlang:garbage_collect/1</c> BIF
1314
could deadlock each other when the runtime system with
1315
SMP support was used.</p>
1317
Own Id: OTP-6920</p>
1321
<c>dbg</c> could leave traced processes in a suspended
1322
state if the tracer process was killed with exit reason
1325
<c>erlang:suspend_process/2</c> has been introduced which
1326
accepts an option list as second argument. For more
1327
information see the <c>erlang(3)</c> documentation.</p>
1329
Processes suspended via
1330
<c>erlang:suspend_process/[1,2]</c> will now be
1331
automatically resumed if the process that called
1332
<c>erlang:suspend_process/[1,2]</c> terminates.</p>
1334
Processes could previously be suspended by one process
1335
and resumed by another unless someone was tracing the
1336
suspendee. This is <em>not</em> possible anymore. The
1337
process resuming a process <em>has</em> to be the one
1338
that suspended it.</p>
1340
*** POTENTIAL INCOMPATIBILITY ***</p>
1342
Own Id: OTP-6946</p>
1345
<p>file:write_file/3, file:write/2 and file:read/2 could
1346
crash (contrary to documentation) for odd enough file
1347
system problems, e.g write to full file system. This bug
1348
has now been corrected.</p> <p>In this process the file
1349
module has been rewritten to produce better error codes.
1350
Posix error codes now originate from the OS file system
1351
calls or are generated only for very similar causes (for
1352
example 'enomem' is generated if a memory allocation
1353
fails, and 'einval' is generated if the file handle in
1354
Erlang is a file handle but currently invalid).</p>
1355
<p>More Erlang-ish error codes are now generated. For
1356
example <c>{error,badarg}</c> is now returned from
1357
<c>file:close/1</c> if the argument is not of a file
1358
handle type. See file(3).</p> <p>The possibility to write
1359
a single byte using <c>file:write/2</c> instead of a list
1360
or binary of one byte, contradictory to the
1361
documentation, has been removed.</p>
1363
*** POTENTIAL INCOMPATIBILITY ***</p>
1365
Own Id: OTP-6967 Aux Id: OTP-6597 OTP-6291 </p>
1369
Monitor messages produced by the system monitor
1370
functionality, and garbage collect trace messages could
1371
contain erroneous heap and/or stack sizes when the actual
1372
heaps and/or stacks were huge.</p>
1374
As of erts version 5.6 the <c>large_heap</c> option to
1375
<c>erlang:system_monitor/[1,2]</c> has been modified. The
1376
monitor message is sent if the sum of the sizes of all
1377
memory blocks allocated for all heap generations is equal
1378
to or larger than the specified size. Previously the
1379
monitor message was sent if the memory block allocated
1380
for the youngest generation was equal to or larger than
1381
the specified size.</p>
1383
*** POTENTIAL INCOMPATIBILITY ***</p>
1385
Own Id: OTP-6974 Aux Id: seq10796 </p>
1389
<c>inet:getopts/2</c> returned random values on Windows
1392
Own Id: OTP-7003</p>
1398
<section><title>Improvements and New Features</title>
1402
The emulator internal process lock implementation has
1403
been rewritten and optimized. A slight risk of starvation
1404
existed in the previous implementation. This risk has
1405
also been eliminated in the new implementation.</p>
1407
Own Id: OTP-6500</p>
1411
Bitstrings (bit-level) binaries and binary comprehensions
1412
are now part of the language. See the Reference Manual.</p>
1414
Own Id: OTP-6558</p>
1418
The windows version of erlang now has SMP support. The
1419
SMP emulator is run by default on machines which shows
1420
more than one virtual or physical processor.</p>
1422
Own Id: OTP-6560 Aux Id: OTP-6925 </p>
1426
The details of the compressed term format has been
1427
documented in erl_ext_dist.txt. (Thanks to Daniel
1430
Own Id: OTP-6755</p>
1434
The runtime system with SMP support is now started by
1435
default if more than one logical processor are detected.
1436
For more information, see the <c>erl(3)</c>
1439
Own Id: OTP-6756</p>
1443
The external format for Erlang terms and the distribution
1444
protocol are now documented in ERTS User's Guide.</p>
1446
Own Id: OTP-6779</p>
1450
New BIF's erlang:system_profile/1 and
1451
erlang:system_profile/2. These BIF's controls concurrency
1452
profiling options for processes, ports and schedulers.</p>
1454
Own Id: OTP-6783 Aux Id: OTP-6285 </p>
1458
The <c>ErlDrvTermData</c> term types used by
1459
<c>driver_output_term()</c> and <c>driver_send_term()</c>
1460
have been extended with the term types
1461
<c>ERL_DRV_BUF2BINARY</c>, <c>ERL_DRV_EXT2TERM</c>, and
1462
<c>ERL_DRV_UINT</c>. <c>ERL_DRV_BUF2BINARY</c> is used
1463
for passing and creating a binary,
1464
<c>ERL_DRV_EXT2TERM</c> is used for passing terms encoded
1465
with the external term format, and <c>ERL_DRV_UINT</c> is
1466
used for passing unsigned integers.</p>
1468
Also the data types <c>ErlDrvUInt</c> and
1469
<c>ErlDrvSInt</c> have been added which makes it more
1470
obvious how arguments to term types are interpreted with
1471
regards to width and signedness.</p>
1473
The incorrect data types <c>ErlDriverTerm</c>,
1474
<c>ErlDriverBinary</c>, and <c>ErlDriverPort</c> in the
1475
<c>erl_driver(3)</c> documentation have been replaced
1476
with the correct data types <c>ErlDrvTermData</c>,
1477
<c>ErlDrvBinary</c>, and <c>ErlDrvPort</c>.</p>
1479
For more information see the <c>erl_driver(3)</c>
1482
Own Id: OTP-6823</p>
1486
Miscellaneous improvements of the erts internal thread
1489
It now support optimized atomic operations and spin-locks
1492
Fall-backs based on mutexes and/or spin-locks for missing
1493
optimized atomic operations, spin-locks, or rwlocks has
1494
been implemented. This makes it possible to compile the
1495
runtime system with SMP support on a lot more platforms.</p>
1497
Default stack size on OpenBSD has been increased to 256
1500
Own Id: OTP-6831 Aux Id: OTP-6560 </p>
1503
<p>Many bit syntax operations, both construction and
1504
matching, are faster. For further information, see the
1505
Efficiency Guide.</p>
1507
Own Id: OTP-6838</p>
1510
<p>Literal lists, tuples, and binaries are no longer
1511
constructed at run-time as they used to be, but are
1512
stored in a per-module constant pool. Literals that are
1513
used more than once are stored only once.</p>
1514
<p>This is not a change to the language, only in the
1515
details of its implementation. Therefore, the
1516
implications of this change is described in the
1517
Efficiency Guide.</p>
1518
<p>Example 1: In the expression <c>element(BitNum-1,
1519
{1,2,4,8,16,32,64,128})</c>, the tuple used to be
1520
constructed every time the expression was executed, which
1521
could be detrimental to performance in two ways if the
1522
expression was executed in a loop: the time to build the
1523
tuple itself and the time spent in garbage collections
1524
because the heap filled up with garbage faster.</p>
1525
<p>Example 2: Literal strings, such as <c>"abc"</c>, used
1526
to be stored in the compiled code compactly as a byte
1527
string and expanded to a list at run-time. Now all
1528
strings will be stored expanded to lists (such as
1529
<c>[$a,$b,$c]</c>) in the constant pool. That means that
1530
the string will be faster to use at run-time, but that it
1531
will require more space even when not used. If space is
1532
an issue, you might want to use binary literals (that is,
1533
<c><<"abc">></c>) instead of string literals for
1534
infrequently used long strings (such as error
1537
Own Id: OTP-6850</p>
1540
<p>The Erlang driver API has been extended with a
1541
portable POSIX thread like API for multi-threading. The
1542
Erlang driver thread API provides:</p> <list>
1543
<item>Threads</item> <item>Mutexes</item> <item>Condition
1544
variables</item> <item>Read/Write locks</item>
1545
<item>Thread specific data</item> </list> <p>For more
1546
information see the <c>erl_driver(3)</c>
1549
Own Id: OTP-6857</p>
1553
Recursive calls now usually consume less stack than in
1554
R11B. See the Efficiency Guide.</p>
1556
Own Id: OTP-6862 Aux Id: seq10746 </p>
1560
The deprecated BIFs <c>erlang:old_binary_to_term/1</c>
1561
and <c>erlang:info/1</c> have been removed.</p>
1563
*** POTENTIAL INCOMPATIBILITY ***</p>
1565
Own Id: OTP-6876</p>
1569
Calls to driver call-backs triggered by external events
1570
are now scheduled and interleaved with execution of
1571
Erlang processes also on the runtime system without SMP
1574
Own Id: OTP-6878</p>
1578
Faster arithmetic of integers of more than 27 bits signed
1579
(or 60 bits signed on an 64-bit CPU), and also faster
1580
integer multiplication. (Thanks to Tony Rogvall.)</p>
1582
Own Id: OTP-6891</p>
1585
<p>Significant improvements of the <c>process_info</c>
1586
BIFs:</p> <list> <item> <c>process_info/2</c> can now be
1587
called with a list of items as second argument in order
1588
to atomically retrieve information about multiple items.
1589
</item> <item> <c>process_info/[1,2]</c> has been
1590
optimized in the runtime system with SMP support. The
1591
whole scheduler could previously be blocked for a
1592
significant period of time in <c>process_info/[1,2]</c>
1593
waiting for a lock on the process being inspected. The
1594
Erlang process calling <c>process_info/[1,2]</c> can
1595
still be blocked for a significant period of time waiting
1596
for the lock, but the scheduler will now be able to run
1597
other processes while the process calling
1598
<c>process_info/[1,2]</c> waits for the lock. </item>
1599
<item> <c>process_info/2</c> now accept a few more items
1600
than before. </item> <item> The documentation of
1601
<c>process_info/[1,2]</c> has been improved. </item>
1602
</list> <p>For more information see the <c>erlang(3)</c>
1605
Own Id: OTP-6899</p>
1609
<c>open_port({}, [])</c> could crash the emulator.
1610
(Thanks to Matthew Dempsky.)</p>
1612
Own Id: OTP-6901</p>
1615
<p>Two new guard BIFs have been introduced as a
1616
recommended replacement for <c>size/1</c>. (The
1617
<c>size/1</c> BIF will be removed no earlier than in
1618
R14B.) The BIFs are <c>tuple_size/1</c> to calculate the
1619
size of a tuple and <c>byte_size/1</c> to calculate the
1620
number of bytes needed for the contents of the binary or
1621
bitstring (rounded up to the nearest number of bytes if
1623
<p>There is also a new <c>bit_size/1</c> BIF that returns
1624
the exact number of bits that a binary or bitstring
1627
Own Id: OTP-6902</p>
1630
<p>The <c>ets:fixtable/2</c> function, which has been
1631
deprecated for several releases, has been removed.</p>
1632
<p>The <c>ets:info/1</c> function has been reimplemented
1633
as a BIF, which guarantees that information returned is
1635
<p>The <c>ets:info/2</c> function now fails with reason
1636
<c>badarg</c> if the second argument is invalid.
1637
(Dialyzer can be used to find buggy code where the second
1638
argument is misspelled.)</p>
1640
*** POTENTIAL INCOMPATIBILITY ***</p>
1642
Own Id: OTP-6906</p>
1646
As the linux kernel may generate a minor fault when
1647
tracing with CPU timestamps, and there exists no patch to
1648
the Linux kernel that fixes the problem, cpu timestamps
1649
are disabled on Linux for now.</p>
1651
*** POTENTIAL INCOMPATIBILITY ***</p>
1653
Own Id: OTP-6922</p>
1656
<p>The functions io:columns/0, io:columns/1, io:rows/0
1657
and io:rows/1 are added to allow the user to get
1658
information about the terminal geometry. The shell takes
1659
some advantage of this when formatting output. For
1660
regular files and other io-devices where height and width
1661
are not applicable, the functions return
1662
{error,enotsup}.</p>
1663
<p>Potential incompatibility: If one has written a custom
1664
io-handler, the handler has to either return an error or
1665
take care of io-requests regarding terminal height and
1666
width. Usually that is no problem as io-handlers, as a
1667
rule of thumb, should give an error reply when receiving
1668
unknown io-requests, instead of crashing.</p>
1670
*** POTENTIAL INCOMPATIBILITY ***</p>
1672
Own Id: OTP-6933</p>
1676
<c>driver_caller()</c> can now also be used from the
1677
<c>start</c> callback of a driver.</p>
1679
Own Id: OTP-6951</p>
1683
The emulator can now be compiled for 64bit intel, as well
1684
as a 32bit universal binary on darwin/MacOSX 10.4 and
1687
Own Id: OTP-6961</p>
1691
If <c>open_port</c> fails because all available ports are
1692
already in use, it will now throw a <c>system_limit</c>
1693
exception instead of an <c>enfile</c> exception.
1694
(<c>enfile</c> might still be thrown if the operating
1695
system would return ENFILE.)</p>
1697
Own Id: OTP-6968</p>
1701
The <c>spawn_monitor/1</c> and <c>spawn_monitor/3</c> BIFs
1702
are now auto-imported (i.e. they no longer need an
1703
<c>erlang:</c> prefix).</p>
1705
Own Id: OTP-6975</p>
1709
On Windows, the werl window now handles resize, so that
1710
the whole window can be utilized. Text selection is also
1711
updated to be line oriented instead of rectangle oriented
1712
as in earlier versions.</p>
1714
Own Id: OTP-6994 Aux Id: OTP-6933 </p>
1718
Kqueue support (kernel-poll) has been enabled on FreeBSD.
1719
The problem with kqueue not detecting writes on a pipe on
1720
FreeBSD was actually not a kqueue issue, but a writev on
1721
pipes issue. Neither poll(), nor select() detected the
1722
write when the bug hit. NetBSD and DragonFlyBSD probably
1723
have or have had the same bug. This bug should have been
1724
fixed in FreeBSD 6.3 and FreeBSD 7.0 thanks to
1725
Jean-Sebastien Pedron.</p>
1727
Own Id: OTP-7001</p>
1735
<section><title>Erts 5.5.5.5</title>
1737
<section><title>Fixed Bugs and Malfunctions</title>
1741
Hanging writes on temporarilly unavailable NFS
1742
filesystems could cause the execution of (not file
1743
related) erlang code to get blocked even though I/O
1744
threads were used. This is now corrected.</p>
1746
Own Id: OTP-6907 Aux Id: seq10771 </p>
1753
<section><title>Erts 5.5.5.4</title>
1755
<section><title>Fixed Bugs and Malfunctions</title>
1759
Data passed to a driver via <c>erlang:port_call</c> could
1760
be corrupted when the runtime system with SMP support was
1761
used. (Thanks to YAMASHINA Hio.)</p>
1763
Own Id: OTP-6879</p>
1767
In the SMP emulator, if serveral processes called
1768
ets:update_counter/3 (even for different tables) when the
1769
counter values exceeded 27 bits, the counter values could
1770
be corrupted or the emulator could crash.</p>
1772
Own Id: OTP-6880 Aux Id: seq10760 </p>
1780
<section><title>Erts 5.5.5.3</title>
1782
<section><title>Fixed Bugs and Malfunctions</title>
1785
<p>Creating a named table using <c>ets:new/2</c> or
1786
renaming a named table using <c>ets:rename/2</c> could in
1787
rare circumstances succed, meaning that there would be
1788
two or more tables with the same name. Now such call will
1789
fail with a <c>badarg</c> exception as it is supposed to
1791
<p><c>ets:delete/1</c> used on a named table now removes
1792
the name immediately so that a new table with the same
1793
name can be created.</p>
1794
<p>Turning on call trace on the internal BIF that
1795
implements <c>ets:delete/1</c> would crash the
1797
<p>SMP emulator only: Using <c>ets:rename/2</c> on a
1798
table that <c>ets:safe_fixtable/2</c> has been used on
1799
could casue an emulator crash or undefined behaviour
1800
because of a missing lock.</p>
1802
Own Id: OTP-6872 Aux Id: seq10756, seq10757 </p>
1809
<section><title>Erts 5.5.5.2</title>
1811
<section><title>Known Bugs and Problems</title>
1815
ets:select/3 on ordered_set and with a chunksize a
1816
multiple of 1000 gave all elements instead of just 1000.
1817
Also ets:slot/2 on ordered set could give unexpected
1818
results on SMP emulator. Both problems are corrected.</p>
1820
Own Id: OTP-6842</p>
1827
<section><title>Erts 5.5.5.1</title>
1829
<section><title>Fixed Bugs and Malfunctions</title>
1833
All exported functions in gzio.c have now been renamed to
1834
avoid conflict with drivers that are indirectly linked
1835
with an external zlib library.</p>
1837
Own Id: OTP-6816 Aux Id: OTP-6591 </p>
1841
Calling binary_to_term/1 with certain invalid binaries
1842
would crash the emulator.</p>
1844
Own Id: OTP-6817</p>
1850
<section><title>Improvements and New Features</title>
1854
Restored speed of bit-syntax matching of 32 bits
1857
Own Id: OTP-6789 Aux Id: seq10688 </p>
1865
<title>Erts 5.5.5</title>
1868
<title>Fixed Bugs and Malfunctions</title>
1869
<list type="bulleted">
1871
<p>The functions in gzio.c have been renamed to avoid
1872
conflict with drivers that indirectly linked with an
1873
external zlib library.</p>
1874
<p>Own Id: OTP-6591</p>
1877
<p>The emulator without SMP support dumped core if an
1878
async-thread finished a job after the emulator had begun
1879
writing an <c><![CDATA[erl_crash.dump]]></c>. </p>
1880
<p>Own Id: OTP-6685</p>
1883
<p>In bit syntax matching, integer fields with greater
1884
size than 16Mb would fail to match. (Thanks to Bertil
1885
Karlsson and Francesco Pierfederici.)</p>
1886
<p>Matching out a 32-bit integer not aligned on a byte
1887
boundary from a binary could cause an heap overflow (and
1888
subsequent termination of the emulator).</p>
1889
<p>A module that contained bit syntax construction with a
1890
literal field size greater than 16Mb would fail to
1892
<p>Several other similar bugs having to do with huge
1893
field sizes were eliminated.</p>
1894
<p>Attempting to construct a binary longer than 536870911
1895
bytes will now fail with a <c><![CDATA[system_limit]]></c> exception
1896
(rather than fail in mysterious ways or construct an
1897
binary with incorrect contents). Similarily, attempting
1898
to match a binary longer than 536870911 bytes will now
1899
fail (instead of producing an incorrect result). This
1900
limitation has been documented in the Efficiency Guide.
1901
(The limit is in the 32-bit emulator; use the 64-bit
1902
emulator if you need to handle larger binaries than
1904
<p>Own Id: OTP-6686</p>
1907
<p>Bugs in rem and div of very large numbers are corrected.</p>
1908
<p>Own Id: OTP-6692</p>
1911
<p><c><![CDATA[erlang:system_info({allocator, Alloc})]]></c> didn't
1912
allocate enough heap when a bignum was part of the result
1913
which could cause an emulator crash.</p>
1914
<p>Own Id: OTP-6693</p>
1917
<p>It was previously not possible to pass
1918
<c><![CDATA[erts_alloc]]></c> the same configuration via the
1919
command-line, as used by default. </p>
1920
<p>A <c><![CDATA[+M*]]></c> command-line argument that configure a
1921
size of some sort can now be passed a value that equals
1922
the size of the address space. The value used, in this
1923
case, will be <c><![CDATA["the size of the address space" - 1]]></c>.</p>
1924
<p>Own Id: OTP-6699</p>
1927
<p><c><![CDATA[SysIOVec* driver_peekq(ErlDrvPort port, int *vlen)]]></c> did not update <c><![CDATA[*vlen]]></c> if <c><![CDATA[port]]></c> was
1928
invalid. <c><![CDATA[*vlen]]></c> is now set to <c><![CDATA[-1]]></c> if the
1929
<c><![CDATA[port]]></c> is invalid.</p>
1930
<p>The <c><![CDATA[efile]]></c> driver
1931
expects <c><![CDATA[*vlen]]></c> to be updated also when the
1932
<c><![CDATA[port]]></c> is invalid. This situation occurs seldom, but
1933
when the runtime system has async-threads enabled and
1934
ports are killed it can. When it occured the runtime
1936
<p>Own Id: OTP-6729</p>
1942
<title>Improvements and New Features</title>
1943
<list type="bulleted">
1945
<p>For scripts written using <c><![CDATA[escript]]></c>, there is a new
1946
function <c><![CDATA[escript:script_name/0]]></c>, which can be used
1947
to retrieve the pathame of the script. The documentation
1948
has been clarified regarding pre-defined macros such as
1949
?MODULE and the module name.</p>
1950
<p>Own Id: OTP-6593</p>
1953
<p>The section Guards in the chapter The Abstract Format
1954
of the ERTS User's Guide has been updated.</p>
1955
<p>Own Id: OTP-6600</p>
1958
<p>Corrected protocol layer flue for socket options
1959
SO_LINGER, SO_SNDBUF and SO_RCVBUF, for SCTP.</p>
1960
<p>Own Id: OTP-6625 Aux Id: OTP-6336 </p>
1963
<p>The behaviour of the inet option {active,once} on peer
1964
close is improved and documented.</p>
1965
<p>Own Id: OTP-6681</p>
1968
<p>The inet option send_timeout for connection oriented
1969
sockets is added to allow for timeouts in communicating
1970
send requests to the underlying TCP stack.</p>
1971
<p>Own Id: OTP-6684 Aux Id: seq10637 OTP-6681 </p>
1974
<p>The command line flag <c><![CDATA[-args_file FileName]]></c>, and
1975
the environment variables <c><![CDATA[ERL_AFLAGS]]></c>, and
1976
<c><![CDATA[ERL_ZFLAGS]]></c> for the <c><![CDATA[erl]]></c> command have been
1977
added. For more information see the <c><![CDATA[erl(1)]]></c>
1979
<p>Own Id: OTP-6697</p>
1982
<p>The <c><![CDATA[is_constant/1]]></c> type test has been deprecated.
1983
<c><![CDATA[is_constant/1]]></c> is improperly named and almost
1984
entirely undocumented.</p>
1985
<p>Own Id: OTP-6731</p>
1992
<title>Erts 5.5.4.3</title>
1995
<title>Fixed Bugs and Malfunctions</title>
1996
<list type="bulleted">
1998
<p><c><![CDATA[process_flag(trap_exit, Bad)]]></c> where <c><![CDATA[Bad]]></c>
1999
was a term not equal to <c><![CDATA[true]]></c> or <c><![CDATA[false]]></c>,
2000
didn't fail with <c><![CDATA[badarg]]></c> as it should; instead, the
2001
failure was silently ignored. This bug was introduced in
2002
<c><![CDATA[erts-5.5.2]]></c>.</p>
2003
<p>Own Id: OTP-6627 Aux Id: OTP-6160 </p>
2006
<p>The minimum and default stack size for async-threads has
2007
been increased to 16 kilowords. This since the previous
2008
minimum and default stack size of 8 kilowords proved to
2009
be too small (introduced in <c><![CDATA[erts-5.5.4.2]]></c>).</p>
2010
<p>Own Id: OTP-6628 Aux Id: OTP-6580, Seq10633 </p>
2016
<title>Improvements and New Features</title>
2017
<list type="bulleted">
2019
<p>process_flag/2 accepts the new flag <c><![CDATA[sensitive]]></c>.</p>
2020
<p>Own Id: OTP-6592 Aux Id: seq10555 </p>
2027
<title>Erts 5.5.4.2</title>
2030
<title>Fixed Bugs and Malfunctions</title>
2031
<list type="bulleted">
2033
<p>When a port steals control over a file descriptor from
2034
another port, the stealing port tests if the other port
2035
is alive. This in order to be able to give an accurate
2036
error message. In the runtime system with SMP support,
2037
this test was done without appropriate locks held. This
2038
could in worst case lead to an erroneous error message;
2039
therefore, this bug is to be considered harmless.</p>
2040
<p>Own Id: OTP-6602</p>
2046
<title>Improvements and New Features</title>
2047
<list type="bulleted">
2049
<p>The default stack size for threads in the async-thread
2050
pool has been shrunk to 8 kilowords, i.e., 32 KB on
2051
32-bit architectures. This small default size has been
2052
chosen since the amount of async-threads might be quite
2053
large. The default stack size is enough for drivers
2054
delivered with Erlang/OTP, but might not be sufficiently
2055
large for other dynamically linked in drivers that use
2056
the <c><![CDATA[driver_async()]]></c> functionality. A suggested
2057
stack size for threads in the async-thread pool can be
2058
configured via the <c><![CDATA[+a]]></c> command line argument of
2059
<c><![CDATA[erl(1)]]></c>.</p>
2060
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
2061
<p>Own Id: OTP-6580</p>
2068
<title>Erts 5.5.4.1</title>
2071
<title>Fixed Bugs and Malfunctions</title>
2072
<list type="bulleted">
2074
<p>Setting the time on the system while using heart on a
2075
linux machine where the emulator was built with
2076
clock_gettime support (default from Linux 2.6/erts-5.5.4
2077
and upwards), could make the heart command fire. This was
2078
due to bug in the heart executable which is now
2080
<p>Own Id: OTP-6598 Aux Id: seq10614 </p>
2087
<title>Erts 5.5.4</title>
2090
<title>Fixed Bugs and Malfunctions</title>
2091
<list type="bulleted">
2093
<p>Corrected misspelling of '<c><![CDATA[-pz]]></c>' in the help text
2094
for <c><![CDATA[erlc]]></c>. (Thanks to Ulf Wiger.)</p>
2095
<p>Own Id: OTP-6433</p>
2098
<p>The MD5 calculation of a BEAM file done by
2099
<c><![CDATA[code:module_md5/1]]></c>, <c><![CDATA[beam_lib:md5/1]]></c>, and by
2100
the compiler for the default value of the <c><![CDATA[vsn]]></c>
2101
attribute have all been changed so that its result will
2102
be the same on all platforms; modules containing funs
2103
could get different MD5s on different platforms.</p>
2104
<p>Own Id: OTP-6459</p>
2107
<p>The emulator could dump core while writing an
2108
<c><![CDATA[erl_crash.dump]]></c> file if there were funs with a
2109
large terms in its environment. Since there is no way to
2110
inspect a fun's environment in the crashdump_viewer
2111
application anyway, a variables in the environment are
2112
now set to [] before dumping the fun. (Thanks to
2113
Jean-Sebastien Pedron.)</p>
2114
<p>Own Id: OTP-6504</p>
2117
<p><c><![CDATA[{Port, {exit_status, S}}]]></c> messages from ports
2118
opened with the <c><![CDATA[exit_status]]></c> option could under
2119
rare circumstances be delayed. This bug was present on
2120
Erlang runtime systems without SMP support on all unix
2121
operating systems other than SunOS.</p>
2122
<p>Own Id: OTP-6528</p>
2125
<p>A bug in linuxthreads could cause the emulator to dump
2126
core when dlerror() was called before the first call to
2127
dlopen(). As a workaround the emulator always makes a
2128
call to dlopen() on initialization when linuxthreads is
2129
used as thread library.</p>
2130
<p>Own Id: OTP-6530</p>
2133
<p>file:sync/1 did not do anything on Windows. Now it calls
2134
the system function for flushing buffers
2135
(FlushFileBuffers()). (Thanks to Matthew Sackman.)</p>
2136
<p>Own Id: OTP-6531</p>
2139
<p><c><![CDATA[open_port/2]]></c> could on the runtime system with SMP
2140
support fail with the wrong exit reason when a port
2141
couldn't be created. When this happened the exit reason
2142
was typically <c><![CDATA[eintr]]></c>, or <c><![CDATA[ebusy]]></c> instead of
2143
<c><![CDATA[eagain]]></c>.</p>
2144
<p>Own Id: OTP-6536</p>
2147
<p>The file driver (efile_drv) did not flush data written
2148
using the option 'delayed_write' after the set timeout
2149
time, rather at the next file operation. This bug has now
2151
<p>Own Id: OTP-6538</p>
2157
<title>Improvements and New Features</title>
2158
<list type="bulleted">
2160
<p>An interface towards the SCTP Socket API Extensions
2161
has been implemented.It is an Open Source patch courtesy
2162
of Serge Aleynikov and Leonid Timochouk. The Erlang code
2163
parts has been adapted by the OTP team, changing the
2164
Erlang API somewhat.</p>
2165
<p>The Erlang interface consists of the module
2166
<c><![CDATA[gen_sctp]]></c> and an include file
2167
<c><![CDATA[-include_lib("kernel/include/inet_sctp.hrl").]]></c> for
2168
option record definitions. The <c><![CDATA[gen_sctp]]></c> module is
2170
<p>The delivered Open Source patch, before the OTP team
2171
rewrites, was written according to
2172
http://tools.ietf.org/html/draft-ietf-tsvwg-sctpsocket-13
2173
and was claimed to work fine, tested on Linux Fedora Core
2174
5.0 (kernel 2.6.15-2054 or later) and on Solaris 10 and
2175
11. The OTP team rewrites used the same standard document
2176
but might have accidentally broken some functionality. If
2177
so it will soon be patched to working state. The tricky
2178
parts in C and the general design has essentially not
2179
changed. During the rewrites the code was hand tested on
2180
SuSE Linux Enterprise Server 10, and briefly on Solaris
2181
10. Feedbach on code and docs is very much
2183
<p>The SCTP interface is in beta state. It has only been
2184
hand tested and has no automatic test suites in OPT
2185
meaning everything is most certainly not tested. Socket
2186
active mode is broken. IPv6 is not tested. The
2187
documentation has been reworked due to the API changes,
2188
but has not been proofread after this.</p>
2189
<p>Thank you from the OTP team to Serge Aleynikov and
2190
Leonid Timochouk for a valuable contribution. We hope we
2191
have not messed it up too much.</p>
2192
<p>Own Id: OTP-6336</p>
2195
<p>A <c><![CDATA[{minor_version,Version}]]></c> option is now recognized
2196
by <c><![CDATA[term_to_binary/2]]></c>. {minor_version,1} will cause
2197
floats to be encoded in an exact and more space-efficient
2198
way compared to the previous encoding.</p>
2199
<p>Own Id: OTP-6434</p>
2202
<p>There is a new <c><![CDATA[escript]]></c> program that can be used
2203
for writing scripts in Erlang. Erlang scripts don't need
2204
to be compiled and any arguments can be passed to them
2205
without risk that they are interpreted by the Erlang
2207
<p>Own Id: OTP-6505</p>
2210
<p>Setting and getting socket options in a "raw" fashion is
2211
now allowed. Using this feature will inevitably produce
2212
non portable code, but will allow setting ang getting
2213
arbitrary uncommon options on TCP stacks that do have
2215
<p>Own Id: OTP-6519</p>
2218
<p>Miscellaneous signal handling improvements on the Erlang
2219
runtime system with SMP support.</p>
2220
<p>The fallback implementation of spin locks and atomic
2221
operations are now implemented using pthread spin locks
2222
when pthread spin locks are found on the system.</p>
2223
<p>The Erlang runtime system with SMP support can now run on
2224
Linux systems that has Linuxthreads instead of NPTL
2225
(Native POSIX Thread Library). Note that the SMP support
2226
hasn't been as thoroughly tested with Linuxthreads as
2227
with NPTL. A runtime system with SMP support will
2228
therefore not be built by default on Linux when NPTL
2229
isn't found. In order to force a build of the runtime
2230
system with SMP support, pass <c><![CDATA[--enable-smp-support]]></c>
2231
to <c><![CDATA[configure]]></c> when building OTP.</p>
2232
<p>Own Id: OTP-6525</p>
2239
<title>Erts 5.5.3.1</title>
2242
<title>Fixed Bugs and Malfunctions</title>
2243
<list type="bulleted">
2245
<p><c><![CDATA[erlang:system_flag(multi_scheduling, block)]]></c> could
2246
cause the emulator with SMP support to deadlock.</p>
2247
<p>Own Id: OTP-6431 Aux Id: OTP-6403 </p>
2250
<p>The runtime system with SMP support failed to call the
2251
driver timeout callback of ports in state closing. This
2252
could cause ports to fail flushing their I/O queues.</p>
2253
<p>Own Id: OTP-6432</p>
2256
<p>The <c><![CDATA[{Port, closed}]]></c> message from a closed port
2257
could arrive at the port owner before <c><![CDATA[Port]]></c> had
2258
been removed from the result of <c><![CDATA[erlang:ports/0]]></c> in
2259
the runtime system with SMP support.</p>
2260
<p>Own Id: OTP-6437</p>
2263
<p>The async id of async jobs created via
2264
<c><![CDATA[driver_async]]></c> wasn't created in a thread safe
2265
manner in the runtime system with SMP support. This could
2266
in worst case cause <c><![CDATA[driver_async_cancel()]]></c> to
2267
cancel another async job than intended.</p>
2268
<p>Own Id: OTP-6438</p>
2271
<p>Under rare circumstances a terminating connection between
2272
two nodes could cause an instantaneous reconnect between
2273
the two nodes to fail on the runtime system with SMP
2275
<p>Own Id: OTP-6447</p>
2278
<p>In the documentation of the driver entry field
2279
<c><![CDATA[extended_marker]]></c> of erts version 5.5.3
2280
(<c><![CDATA[driver_entry(3)]]></c>), the following is stated: "The
2281
following fields are ignored if this field is equal to
2282
<c><![CDATA[0]]></c>". This is a documentation bug and has been
2283
changed to: "If this field is equal to <c><![CDATA[0]]></c>, all the
2284
fields following this field also <em>have</em> to be
2285
<c><![CDATA[0]]></c>, or <c><![CDATA[NULL]]></c> in case it is a pointer field".</p>
2286
<p>The runtime check for detection of old incompatible
2287
drivers made when loading drivers has been improved. The
2288
emulator can, however, not make sure that a driver that
2289
doesn't use the extended driver interface isn't
2290
incompatible. Therefore, when loading a driver that
2291
doesn't use the extended driver interface, there is a
2292
risk that it will be loaded also when the driver is
2293
incompatible. When the driver use the extended driver
2294
interface, the emulator can verify that it isn't of an
2295
incompatible driver version. You are therefore advised to
2296
use the extended driver interface.</p>
2297
<p>For more information see the <c><![CDATA[erl_driver(3)]]></c> and
2298
<c><![CDATA[driver_entry(3)]]></c> documentation.</p>
2299
<p>Own Id: OTP-6452 Aux Id: OTP-6330 </p>
2302
<p>When terminating ports on the runtime system with SMP
2303
support, removal of links to processes was done without
2304
locking the link lock on processes. This could cause an
2306
<p>Own Id: OTP-6475</p>
2309
<p>The emulator with SMP support could crash when a port
2310
flushed its I/O queue by calling <c><![CDATA[driver_async()]]></c>
2311
from the timeout driver callback.</p>
2312
<p>Own Id: OTP-6479</p>
2315
<p>Large exit reasons could under rare circumstances cause
2316
the runtime system with SMP support to crash.</p>
2317
<p>Own Id: OTP-6521</p>
2323
<title>Improvements and New Features</title>
2324
<list type="bulleted">
2326
<p>Faster system calls for keeping the time accurate are
2327
used on newer Linux kernels, which can result in a
2328
significant speed-up of the emulator on those systems.</p>
2329
<p>Own Id: OTP-6430</p>
2332
<p>Added number of async threads and number of scheduler
2333
threads to the system information that can be retrieved
2334
via <c><![CDATA[driver_system_info()]]></c>. For more information see
2335
the <c><![CDATA[erl_driver(3)]]></c> documentation.</p>
2336
<p>Own Id: OTP-6440</p>
2339
<p>When <c><![CDATA[SIGUSR1]]></c> is received by the runtime system
2340
with SMP support, the <c><![CDATA[erl_crash.dump]]></c> is now
2341
written by a special thread, instead of as before
2342
directly from the signal handler.</p>
2343
<p>Own Id: OTP-6465</p>
2346
<p>term_to_binary/2 with compression is now faster.</p>
2348
<p>term_to_binary/2 now accepts the option
2349
'<c><![CDATA[{compressed,Level}]]></c>' for specifying the
2350
compression level. <c><![CDATA[Level]]></c> must be in the range 0
2351
(no compression) through 9 (highest compression level).
2353
<p>Future compatibility bugfix: binary_to_term/1 did not
2354
handle the <c><![CDATA[Uniq]]></c> and <c><![CDATA[Index]]></c> fields
2356
<p>Own Id: OTP-6494</p>
2359
<p>Removed unnecessary reallocation when initializing
2360
kernel-poll set.</p>
2361
<p>Own Id: OTP-6516</p>
2368
<title>Erts 5.5.3</title>
2371
<title>Fixed Bugs and Malfunctions</title>
2372
<list type="bulleted">
2374
<p>Node and fun table entries could under rare circumstances
2375
be deallocated multiple times on the emulator with SMP
2377
<p>Own Id: OTP-6369</p>
2380
<p>epoll_wait() can repeat entries for the same file
2381
descriptor in the result array. This could cause the
2382
ready_input, ready_output, or event callbacks of a driver
2383
to unintentionally be called multiple times. We have only
2384
noted repeated entries when an error condition has
2385
occurred on the file descriptor. In this case, the
2386
repeated entries should normally not be a problem for the
2387
driver since it should detect the error and de-select the
2388
file descriptor. Therefore this bug should be considered
2389
harmless. The emulator now coalesce repeated entries into
2391
<p>You are only affected by this bug if you are using
2392
erts-5.5.2.* and the kernel-poll feature on linux.</p>
2393
<p>Own Id: OTP-6376 Aux Id: OTP-6222 </p>
2396
<p>If a process that was waiting in gen_tcp:accept/1 was
2397
killed, calling gen_tcp:accept/1 again on the same listen
2398
socket would fail with '<c><![CDATA[{error,einval}]]></c>'.</p>
2399
<p>Own Id: OTP-6381 Aux Id: seq10535 </p>
2402
<p>The emulator failed to start on Solaris 10 when
2403
kernel-poll support was enabled and the maximum number of
2404
filedescriptors configured was less than or equal to 256.</p>
2405
<p>Own Id: OTP-6384 Aux Id: OTP-6222 </p>
2408
<p>The R10B compiler could generate unsafe
2409
<c><![CDATA[bs_save/bs_restore]]></c> instructions that could cause
2410
memory corruption. (The R11B compiler does not have that
2411
problem.) The erlang emulator will now refuse to load
2412
R10B-compiled modules that contain such unsafe
2413
<c><![CDATA[bs_save/bs_restore]]></c> instructions. In addition, the
2414
beam_validator module in the compiler will also reject
2415
such instructions (in case it is used to validate R10B
2416
code). (Thanks to Matthew Reilly.)</p>
2417
<p>Own Id: OTP-6386</p>
2420
<p>Process and port timers could fail to work properly on
2421
the runtime system with SMP support. Many thanks to
2422
Dmitriy Kargapolov and Serge Aleynikov who tracked down
2424
<p>Own Id: OTP-6387</p>
2427
<p>Bit syntax code compiled by an R10B compiler that matched
2428
out a floating point number would not properly check that
2429
the floating point number was valid; if the float was,
2430
for instance, NaN the emulator could crash.</p>
2431
<p>Own Id: OTP-6395</p>
2434
<p>statistics(runtime) on Windows used to return the elapsed
2435
system time for the process, instead of the user time.
2436
Corrected. (Thanks to Robert Virding.)</p>
2437
<p>Own Id: OTP-6407</p>
2440
<p>A loadable driver (loaded by erl_ddll) which used
2441
driver_async() would cause the emulator to crash. (Thanks
2442
to Dmitriy Kargapolov.)</p>
2443
<p>Own Id: OTP-6410</p>
2446
<p>Under rare circumstances the emulator on unix platforms
2447
could fail to terminate the Erlang port corresponding to
2448
a port program opened with the <c><![CDATA[exit_status]]></c> option.</p>
2449
<p>Own Id: OTP-6411</p>
2452
<p>A link removed via <c><![CDATA[unlink/1]]></c> could under rare
2453
circumstances transfer exit signals for a short period of
2454
time after the call to <c><![CDATA[unlink/1]]></c> had returned when
2455
the runtime system with SMP support was used.</p>
2456
<p>Own Id: OTP-6425 Aux Id: OTP-6160 </p>
2462
<title>Improvements and New Features</title>
2463
<list type="bulleted">
2465
<p>In the runtime system with SMP support, ports are now
2466
being scheduled on the scheduler threads interleaved with
2467
processes instead of being run in a separate I/O thread.</p>
2468
<p>Own Id: OTP-6262</p>
2471
<p>More interfaces are added in erl_ddll, to support
2472
different usage scenarios.</p>
2473
<p>Own Id: OTP-6307 Aux Id: OTP-6234 </p>
2476
<p>In the runtime system with SMP support, the global
2477
I/O lock has been replaced with a more fine grained port
2478
locking scheme. Port locking is either done on driver
2479
level, i.e., all ports executing the same driver share a
2480
lock, or on port level, i.e., each port has its own lock.
2481
Currently the inet driver, the efile driver, and the
2482
spawn driver use port level locking and all other
2483
statically linked in drivers use driver level locking. By
2484
default dynamically linked in drivers will use locking on
2485
driver level. For more information on how to enable port
2486
level locking see the <c><![CDATA[erl_driver(3)]]></c> and the
2487
<c><![CDATA[driver_entry(3)]]></c> man pages. </p>
2489
version 5.5.3 the driver interface has been extended. The
2490
extended interface introduce version management, the
2491
possibility to pass capability flags to the runtime
2492
system at driver initialization, and some new driver API
2493
functions. For example, the <c><![CDATA[driver_system_info()]]></c>
2494
function which can be used to determine if the driver is
2495
run in a runtime system with SMP support or not. The
2496
extended interface doesn't have to be used, <em>but</em>
2497
dynamically linked in driver <em>have</em> to be
2498
recompiled. For information see the <c><![CDATA[erl_driver(3)]]></c>
2499
and the <c><![CDATA[driver_entry(3)]]></c> man pages. </p>
2500
<p><em>NOTE:</em> Dynamically linked in drivers
2501
<em>have</em> to be recompiled. </p>
2502
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
2503
<p>Own Id: OTP-6330 Aux Id: OTP-6262 </p>
2506
<p>A test and debug feature which modifies the timing of the
2507
runtime system has been added. For more information, see
2508
documentation of the <c><![CDATA[+T<Level>]]></c> command line
2509
argument of <c><![CDATA[erl(1)]]></c>.</p>
2510
<p>Own Id: OTP-6382</p>
2513
<p>The version of zlib (http://zlib.net) linked into
2514
run-time system has been updated to version 1.2.3.</p>
2515
<p>Own Id: OTP-6394</p>
2518
<p>The <c><![CDATA[erlc]]></c> program now passes on the <c><![CDATA[-smp]]></c>
2519
and <c><![CDATA[-hybrid]]></c> options to the Erlang emulator it
2520
starts. This is mainly useful for compiling native code,
2521
because native code must be compiled with same type of
2522
run-time system as in which it will be run.</p>
2523
<p>If the <c><![CDATA[-s]]></c> option is given, <c><![CDATA[erlc]]></c> now
2524
prints a warning that it is deprecated and that it will
2525
be removed in R12B.</p>
2526
<p>Own Id: OTP-6398</p>
2529
<p>The <c><![CDATA[schedulers]]></c> option of
2530
<c><![CDATA[erlang:system_flag/2]]></c> has been removed, i.e., the
2531
number of scheduler threads cannot be changed after
2532
emulator boot time any more.</p>
2533
<p>A <c><![CDATA[multi_scheduling]]></c> option has been added to
2534
<c><![CDATA[erlang:system_flag/2]]></c>. This option can be used for
2535
blocking and unblocking multi-scheduling. For more
2536
information see the <c><![CDATA[erlang(3)]]></c> documentation.</p>
2537
<p>Own Id: OTP-6403</p>
2540
<p>A port program that had been started with the
2541
<c><![CDATA[exit_status]]></c> option and closed one of the pipes
2542
used for communication with the emulator caused the
2543
emulator to continuously poll for termination of the port
2544
program. This only became a problem when the emulator
2545
didn't have other things to do and the port program
2546
closed a pipe much earlier than it terminated. When the
2547
emulator had other things to do, such as running Erlang
2548
processes, the emulator polled for termination in between
2549
scheduling of processes.</p>
2550
<p>Now the emulator doesn't poll for termination of the port
2551
program at all; instead, it waits for the child signal
2552
from the terminated port program to arrive and then
2553
schedules the Erlang port for termination.</p>
2554
<p>The termination of any port programs have also been
2555
optimized. Previously the termination of any port program
2556
did always cause a scan of a table of the size equal to
2557
maximum number of file descriptors. If the maximum number
2558
of file descriptors was large, this scan could be quite
2559
expensive. Now the search have been reduced to the set of
2560
ports started with the <c><![CDATA[exit_status]]></c> option.</p>
2561
<p>Note, all of the above only applies to Erlang emulators
2562
on Unix platforms.</p>
2563
<p>Own Id: OTP-6412 Aux Id: seq10194 </p>
2566
<p>* BEAM: added support for floating-point exceptions on
2567
FreeBSD (32-bit x86)</p>
2568
<p>* SMP: made locking procedures work even when native
2569
lock operations aren't implemented</p>
2570
<p>* SMP: improved timing accuracy in the timer thread
2572
<p>Own Id: OTP-6424</p>
2579
<title>Erts 5.5.2.2</title>
2582
<title>Fixed Bugs and Malfunctions</title>
2583
<list type="bulleted">
2585
<p>A bug in the kernel poll implementation could cause the
2586
emulator to omit polling for events on file descriptors.
2587
The bug was only present when using the kernel poll
2588
implementation based on epoll or kqueue. This bug was
2589
introduced in erts-5.5.2.</p>
2590
<p>Own Id: OTP-6344 Aux Id: OTP-6222 </p>
2597
<title>Erts 5.5.2.1</title>
2600
<title>Fixed Bugs and Malfunctions</title>
2601
<list type="bulleted">
2603
<p>The run_erl program now aquires its psedo-ttys using
2604
openpty(3), so that it will work on newer Linux
2605
distributions that don't have the traditional pseudo-tty
2606
devices in the file system. On platforms that don't have
2607
openpty(3), run_erl will still search for pseudo-tty
2608
devices in the file system.</p>
2609
<p>The run_erl program will now wait using waitpid(3) to
2610
prevent the program it spawned to become defunct. run_erl
2611
will also terminate after a delay of 5 seconds (to allow
2612
any pending output to be written to the log file) if the
2613
spawned program terminates even if some child of it still
2614
holds stdin and/or stdout open.</p>
2615
<p>Own Id: OTP-6225 Aux Id: seq10500 </p>
2618
<p>A bug in ordered_set ETS datatyp caused ets:select (and
2619
match) to return faulty results when the table contained
2620
process id's from another node.</p>
2621
<p>Own Id: OTP-6338</p>
2628
<title>Erts 5.5.2</title>
2631
<title>Fixed Bugs and Malfunctions</title>
2632
<list type="bulleted">
2634
<p>erlc: A typo in the help text for '-pa path' was
2636
<p>Own Id: OTP-6218</p>
2639
<p>Failure in port command/control driver callbacks could
2640
crash the non-SMP emulator. This bug was introduced in
2641
the 5.5 version of erts.</p>
2642
<p>Own Id: OTP-6224</p>
2645
<p>Erroneous "<c><![CDATA[Input driver gone away without deselecting!]]></c>" error reports could sometimes occur
2646
when a driver instance terminated in the ready_output()
2647
callback of the driver. This bug was only present in
2648
emulators that used poll(). Note, that this bug was
2649
harmless, the only effect it had was the erroneous error
2651
<p>Own Id: OTP-6229 Aux Id: OTP-3993, Seq5266, Seq7247,
2655
<p>The emulator could cause a type assertion failure while
2656
writing an erl_crash.dump, causing the erl_crash.dump to
2657
be terminated and a core dump generated.</p>
2658
<p>Own Id: OTP-6235 Aux Id: seq10444 </p>
2661
<p>The registered name of a process is now the last
2662
observable resource removed before links and monitors are
2663
triggered when a process terminates.</p>
2664
<p>Previously ets tables were removed after the registered
2665
name. This could cause problems on the runtime system
2666
with SMP support for code that expected that ets tables
2667
owned by a specific process had been removed if the name
2668
of the process had been removed.</p>
2669
<p>Own Id: OTP-6237</p>
2672
<p>Failure to fork() a new (os) process could cause the
2673
emulator to deadlock. This bug affect all emulators with
2674
SMP support, and emulators with async thread support on
2676
<p>Own Id: OTP-6241 Aux Id: OTP-3906 </p>
2679
<p>Fprof traces could become truncated for the SMP emulator.
2680
This bug has now been corrected.</p>
2681
<p>Own Id: OTP-6246</p>
2684
<p>The undocumented functions inet:getiflist/0,1
2685
inet:ifget/2,3 and inet:getif/1 were completely broken on
2686
Windows. That has been fixed.</p>
2687
<p>Own Id: OTP-6255</p>
2690
<p>Behavior in case of disappeared nodes when using the
2691
dist_auto_connect once got changed in R11B-1. The
2692
timeouts regarding normal distributed operationes is now
2693
reverted to the old (pre R11B-1).</p>
2694
<p>Own Id: OTP-6258 Aux Id: OTP-6200, seq10449 </p>
2697
<p>The bsl and bsr operators could cause the emulator to
2698
crash if given invalid arguments. (Thanks to datacompboy
2699
and Per Gustafsson.)</p>
2700
<p>Own Id: OTP-6259</p>
2703
<p>driver_cancel_timer() could under certain circumstances
2704
fail to cancel the timer on the runtime system with SMP
2706
<p>Own Id: OTP-6261</p>
2709
<p>A call to erlang:system_info(info) could deadlock the
2710
runtime system with SMP support.</p>
2711
<p>Own Id: OTP-6268</p>
2717
<title>Improvements and New Features</title>
2718
<list type="bulleted">
2720
<p>Exit signals are now truly asynchronous in the runtime
2721
system with SMP support. This simplifies locking in bifs
2722
a lot, and makes process termination cheaper.</p>
2723
<p>Own Id: OTP-6160</p>
2726
<p>When tracing on the runtime system with SMP support it
2727
can be difficult to know when a trace has been delivered
2728
to the tracer. A new built in function
2729
<c><![CDATA[erlang:trace_delivered/1]]></c> has been introduced in
2730
order to make it easier to know when the trace has been
2731
delivered. See the erlang(3) man page for more
2733
<p>Own Id: OTP-6205 Aux Id: OTP-6269 </p>
2736
<p>Kernel poll support can now be combined with SMP support.
2737
Currently the following kernel poll versions exist:
2738
<c><![CDATA[/dev/poll]]></c>, <c><![CDATA[epoll]]></c>, and <c><![CDATA[kqueue]]></c>. Linux
2739
<c><![CDATA[kpoll]]></c> has been replaced with <c><![CDATA[epoll]]></c>. Some
2740
time in the future there will also be a kernel poll
2741
version using Solaris event ports.</p>
2742
<p>The "check io" implementation for unix has been
2743
completely rewritten. The current kernel poll
2744
implementation reduce the amount of system calls needed
2745
compared to the old kernel poll implementation.</p>
2746
<p>When <c><![CDATA[epoll]]></c> or <c><![CDATA[kqueue]]></c> is used either
2747
<c><![CDATA[poll]]></c> or <c><![CDATA[select]]></c> is used as fallback.
2748
Previously only <c><![CDATA[poll]]></c> could be used as fallback.
2749
Since <c><![CDATA[select]]></c> now can be used as fallback, kernel
2750
poll support is now also available on newer MacOSX. Note
2751
however, when <c><![CDATA[select]]></c> is used as fallback, the
2752
maximum number of file descriptors is limited to
2753
<c><![CDATA[FD_SETSIZE]]></c>. </p>
2754
<p>Kernel poll support is now enabled by default if
2755
<c><![CDATA[/dev/poll]]></c>, <c><![CDATA[epoll]]></c>, or <c><![CDATA[kqueue]]></c> is found
2756
when building OTP, i.e. you do not have to pass the
2757
<c><![CDATA[--enable-kernel-poll]]></c> argument to <c><![CDATA[configure]]></c>.
2758
As before, kernel poll is disabled by default in the
2759
runtime system. In order to enable it, pass the
2760
<c><![CDATA[+Ktrue]]></c> command line argument to <c><![CDATA[erl]]></c>.</p>
2761
<p>Note: <c><![CDATA[configure]]></c> will refuse to enable kernel poll
2762
support on FreeBSD since <c><![CDATA[kqueue]]></c> have problems with
2763
(at least) pipes on all version of FreeBSD that we have
2765
<p>Own Id: OTP-6222 Aux Id: seq10380 </p>
2768
<p>The <c><![CDATA[erl_ddll]]></c> module and the code in the emulator
2769
have been completely rewritten; several bugs were fixed.</p>
2770
<p>Own Id: OTP-6234</p>
2773
<p>The SMP emulator now avoids locking for the following
2774
operations (thus making them as fast as in the UP
2775
emulator): <c><![CDATA[atom_to_list/1]]></c>, atom comparision, atom
2776
hashing, <c><![CDATA[erlang:apply/3]]></c>.</p>
2777
<p>Own Id: OTP-6252</p>
2780
<p>There are new BIFs <c><![CDATA[erlang:spawn_monitor/1,3]]></c>,
2781
and the new option <c><![CDATA[monitor]]></c> for
2782
<c><![CDATA[spawn_opt/2,3,4,5]]></c>.</p>
2783
<p>The <c><![CDATA[observer_backend]]></c> module has been updated to
2784
handle the new BIFs.</p>
2785
<p>Own Id: OTP-6281</p>
2792
<title>Erts 5.5.1.1</title>
2795
<title>Improvements and New Features</title>
2796
<list type="bulleted">
2798
<p>There is now an option read_packets for UDP sockets that
2799
sets the maximum number of UDP packets that will be read
2800
for each invocation of the socket driver.</p>
2801
<p>Own Id: OTP-6249 Aux Id: seq10452 </p>
2808
<title>Erts 5.5.1</title>
2811
<title>Fixed Bugs and Malfunctions</title>
2812
<list type="bulleted">
2814
<p>Complex pattern matching of strings would fail in the 64
2815
bits emulator because of a bug in the loader. (Thanks to
2816
Igor Goryachev.)</p>
2817
<p>Own Id: OTP-6142</p>
2820
<p><c><![CDATA[-134217728 div 134217728]]></c> and <c><![CDATA[-134217728 rem 134217728]]></c> would be calculated incorrectly.
2821
<c><![CDATA[abs(-2147483648)]]></c> could in unlucky circumstances
2822
cause a heap overflow, as could size(Binary) when size of
2823
the binary was larger than 128Mb.</p>
2824
<p>Own Id: OTP-6154</p>
2827
<p>erlang:display/1 displayed erroneous values for negative
2829
<p>Big integers (both positive and negative) were previously
2830
displayed in hexadecimal form while small integers were
2831
displayed in decimal form. All integers are now displayed
2832
in decimal form.</p>
2833
<p>NOTE: erlang:display/1 should only be used for debugging.</p>
2834
<p>Own Id: OTP-6156</p>
2837
<p>A call to erlang:trace/3 with erroneous flags caused the
2838
SMP emulator to deadlock instead of exiting the calling
2839
process with badarg.</p>
2840
<p>Own Id: OTP-6175</p>
2843
<p>A bug causing the emulator to hang when exiting a process
2844
that is exception traced has been fixed.</p>
2845
<p>Own Id: OTP-6180</p>
2848
<p>ets:rename/1 could deadlock, or crash the SMP emulator
2849
when the table wasn't a named table.</p>
2850
<p>ets:next/2, and ets:prev/2 could return erroneous results
2851
on the SMP emulator.</p>
2852
<p>Own Id: OTP-6198 Aux Id: seq10392, seq10415 </p>
2855
<p>A memory allocation bug could cause the SMP emulator to
2856
crash when a process had executed a <c><![CDATA[receive after]]></c>
2857
with a larger timeout than 10 minutes.</p>
2858
<p>Own Id: OTP-6199</p>
2861
<p>The runtime system with SMP support did not slowly adjust
2862
it's view of time when the system time suddenly changed.</p>
2863
<p>Timeouts could sometimes timeout too early on the runtime
2864
system with SMP support.</p>
2865
<p>Own Id: OTP-6202</p>
2871
<title>Improvements and New Features</title>
2872
<list type="bulleted">
2874
<p>The smp runtime system now automatically detects the
2875
number of logical processors on MacOSX (darwin) and
2877
<p>The smp runtime system is now built by default on MacOSX
2878
(darwin) on x86.</p>
2879
<p>Own Id: OTP-6119</p>
2882
<p>The <c><![CDATA[-smp]]></c> command line argument now take the
2883
following options: <c><![CDATA[enable]]></c>, <c><![CDATA[auto]]></c>, or
2884
<c><![CDATA[disable]]></c>.</p>
2885
<p>Especially the <c><![CDATA[-smpauto]]></c> argument is useful since
2886
it starts the Erlang runtime system with SMP support if
2887
it is available and more than one logical processor are
2888
detected; otherwise, it starts the Erlang runtime system
2889
without SMP support. For more information see the
2890
<c><![CDATA[erl(1)]]></c> man page.</p>
2891
<p>Own Id: OTP-6126</p>
2894
<p>Increased the reduction cost for sending messages in the
2895
SMP emulator so it behaves more like the non-SMP
2897
<p>Own Id: OTP-6196</p>
2900
<p>A port running a dynamically linked-in driver that exits
2901
due to the driver being unloaded now exits with exit
2902
reason <c><![CDATA[driver_unloaded]]></c>. Previously the port exited
2903
with exit reason <c><![CDATA[-1]]></c>.</p>
2904
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
2905
<p>Own Id: OTP-6204</p>
2908
<p>Changed name of the <c><![CDATA[erlang:system_info/1]]></c> argument
2909
<c><![CDATA[scheduler]]></c> to <c><![CDATA[scheduler_id]]></c>. This since the
2910
<c><![CDATA[scheduler]]></c> argument so easily could be mixed up
2911
with the <c><![CDATA[schedulers]]></c> argument (both returning
2913
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
2914
<p>Own Id: OTP-6208</p>
2917
<p>The changes below were made by Mikael Pettersson, HiPE.</p>
2918
<p>HiPE runtime system:</p>
2919
<p>Reduce overheads in the HiPE runtime system's BIF glue
2921
<p>Fix bug when exceptions are thrown from BEAM to HiPE.</p>
2922
<p>Support SPARC on Linux.</p>
2923
<p>Support x86 on FreeBSD.</p>
2924
<p>Floating-point exceptions:</p>
2925
<p>Reduce overheads in checking results of floating-point
2927
<p>Minor bug fix in SSE2 floating-point exception
2929
<p>Support SSE2 floating-point exceptions on 32-bit x86
2931
<p>Make FP exceptions work in the SMP runtime system on
2933
<p>Support floating-point exceptions on SPARCs running
2935
<p>Runtime system:</p>
2936
<p>Minor scheduler optimisation in the non-SMP runtime
2938
<p>Substantial reduction of I/O thread overheads in the
2939
SMP runtime system if the separate timer thread is used.
2940
(In R11B-1, the separate timer thread is not used.)</p>
2941
<p>Own Id: OTP-6211</p>
2948
<title>ERTS 5.5</title>
2951
<title>Fixed Bugs and Malfunctions</title>
2952
<list type="bulleted">
2954
<p>Previously <c><![CDATA[unlink/1]]></c> and <c><![CDATA[erlang:demonitor/2]]></c>
2955
behaved completely asynchronous. This had one undesirable
2956
effect, though. You could never know when you were
2957
guaranteed <em>not</em> to be affected by a link that you
2958
had unlinked or a monitor that you had demonitored.</p>
2959
<p>The new behavior of <c><![CDATA[unlink/1]]></c> and
2960
<c><![CDATA[erlang:demonitor/2]]></c> can be viewed as two operations
2961
performed atomically. Asynchronously send an unlink
2962
signal or a demonitor signal, and ignore any future
2963
results of the link or monitor.</p>
2964
<p><em>NOTE</em>: This change can cause some obscure code
2965
to fail which previously did not. For example, the
2966
following code might hang:</p>
2967
<code type="none"><![CDATA[
2968
Mon = erlang:monitor(process, Pid),
2970
\011 exit(Pid, bang),
2971
erlang:demonitor(Mon),
2973
\011 {'DOWN', Mon , process, Pid, _} -> ok
2974
\011 %% We were previously guaranteed to get a down message
2975
\011 %% (since we exited the process ourself), so we could
2976
\011 %% in this case leave out:
2977
\011 %% after 0 -> ok
2980
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
2981
<p>Own Id: OTP-5772</p>
2984
<p>Two bugs fixed: If the environment variable <c><![CDATA[ERL_FLAGS]]></c>
2985
was set, its contents would be appended to the end of the
2986
command line even if the command line had an <c><![CDATA[-extra]]></c>
2987
options. Changed to place the options from <c><![CDATA[ERL_FLAGS]]></c>
2988
just before <c><![CDATA[-extra]]></c>. Also, the <c><![CDATA[-smp]]></c> and
2989
<c><![CDATA[-hybrid]]></c> flags no longer have any effect if placed
2990
after <c><![CDATA[-extra]]></c>.</p>
2991
<p>Own Id: OTP-6054</p>
2997
<title>Improvements and New Features</title>
2998
<list type="bulleted">
3000
<p>The documentation for writing drivers in the ERTS User's
3001
Guide has been expanded and updated.</p>
3002
<p>Own Id: OTP-5192</p>
3005
<p>The <c><![CDATA[andalso]]></c> and <c><![CDATA[orelse]]></c> operators are
3006
now allowed to be used in guards. That also applies to
3007
match specifications.</p>
3008
<p>Own Id: OTP-5894 Aux Id: OTP-5149 </p>
3011
<p>There is a new trace match spec function
3012
<c><![CDATA[{exception_trace}]]></c> and a corresponding trace
3013
message <c><![CDATA[exception_from]]></c> that can be used to trace
3014
on any exit from a function, both normal function return
3015
and exception return. See the documentation for
3017
<p>The <c><![CDATA[return_to]]></c> trace message is now also
3018
generated when the execution returns to a function due to
3019
catching an exception.</p>
3020
<p>Own Id: OTP-5956</p>
3023
<p>Erlang runtime system with SMP (symmetric multi processing)
3025
<p>The runtime system with SMP support is in this release
3026
focused on stability and there are a number of steps
3027
with optimizations to follow before it will take
3028
full advantage of multi processor systems.
3029
The released system is however thruly multi threaded
3030
and you will notice increased performance
3031
for many applications already.
3032
We recommend that you evaluate your application on
3033
the SMP version of the runtime system and wait for some
3034
more optimizations before you use it in a real product.
3035
You will then discover if there are any problems in
3036
your application that needs to be fixed in order for
3037
it to work properly in a multi threaded environment.
3038
More optimized versions of the runtime system
3039
with SMP support will be included in the R11B
3040
maintenance releases.</p>
3041
<p>The SMP enabled runtime system will be started if
3042
the <c><![CDATA[-smp]]></c> command line argument is passed to
3043
the <c><![CDATA[erl]]></c> command. In order to make use of more than
3044
one processor core, multiple scheduler threads are used. By
3045
default, the number of scheduler threads will equal
3046
the number of processor cores. The number of scheduler
3047
threads can be set with the <c><![CDATA[+S]]></c> command line argument.
3048
For more information see the <c><![CDATA[erl(1)]]></c> man page.</p>
3049
<p>A runtime system with SMP support is by default built on
3050
the following platforms if posix threads, and a gcc
3051
compiler of at least version 2.95 is found:</p>
3052
<list type="bulleted">
3054
<p>Linux with at least kernel version 2.6 and the Native
3055
POSIX Thread Library on x86, x86_64, and 32-bits
3059
<p>Solaris of at least version 8 on 32-bits SPARC-V9.</p>
3062
<p>MacOSX of at least version 10.4 (Darwin 8.0) on
3063
32-bits PowerPC.</p>
3066
<p>The runtime system with SMP support is known <em>not</em>
3068
<list type="bulleted">
3073
<p>Linux with kernel versions less than 2.4, or without
3074
the Native POSIX Thread Library.</p>
3077
<p>Other hardware platforms than x86, x86_64, 32-bits
3078
SPARC-V9 and 32-bits PowerPC.</p>
3081
<p>Windows will be supported in a future release.</p>
3082
<p>The runtime system with SMP support might build on other
3083
operating systems in combination with supported hardware.
3084
In order to force a build of a runtime system with SMP
3085
support, pass the <c><![CDATA[--enable-smp-support]]></c> command line
3086
argument to configure. Note, however, that it is not enough
3087
that it builds. The underlying thread library and operating
3088
system has to provide SMP support as well. If the thread
3089
library does not distribute scheduler threads over multiple
3090
processor cores then the runtime system will only seemingly
3091
provide SMP support. If the runtime system is not built by
3092
default on a specific platform, we have <em>not</em> tested
3093
it on that platform.</p>
3094
<p><em>NOTE</em>: The design of SMP support for drivers is
3095
ongoing. There will probably be incompatible driver
3096
changes (only affecting drivers run on the runtime system
3097
with SMP support) released as patches for R11B.</p>
3098
<p><em>Potential incompatibility</em>: Previously, specific
3099
driver call-backs were always called from the same thread.
3100
This is <em>not</em> true in the runtime system with SMP
3101
support. Calls to call-backs will be made from different
3102
threads, e.g., two consecutive calls to exactly the same
3103
call-back can be made from two different threads. This
3104
will in <em>most</em> cases not be a problem. All calls
3105
to call-backs are synchronized, i.e., only one call-back
3106
will be called at a time.</p>
3107
<p>In the future the default behavior will <em>probably</em>
3108
be the following: Calls to call-backs will, as now, be
3109
made from different threads. Calls to call-backs in the
3110
same driver <em>instance</em> will be synchronized. It
3111
will probably be possible to configure so that all calls
3112
to call-backs in all driver instances of a specific
3113
driver type will be synchronized. It may be possible to
3114
configure so that all calls to call-backs of a driver
3115
instance or a of a specific driver type will be made from
3116
the same thread.</p>
3117
<p>Parallelism in the Erlang code executed is a necessity
3118
for the Erlang runtime system to be able to take
3119
advantage of multi-core or multi-processor hardware.
3120
There need to be at least as many Erlang processes
3121
runnable as processor cores for the Erlang runtime system
3122
to be able to take advantage of all processor cores.</p>
3123
<p>An Erlang runtime system with SMP support with only one
3124
Erlang process runnable all the time will almost always be
3125
slower than the same Erlang runtime system without SMP
3126
support. This is due to thread synchronization overhead.</p>
3127
<p>Known major bottleneck in the Erlang runtime system:</p>
3128
<list type="bulleted">
3130
<p>Currently the I/O system uses one "big lock", i.e. only
3131
one thread can do I/O at a time (with the exception of
3132
async threads and threads created by users own linked-in
3133
drivers). This is high on the list of things to
3134
optimize. Note, code that does not do I/O can be executed
3135
at the same time as one thread does I/O.</p>
3138
<p>Some pitfalls which might cause Erlang programs that work on
3139
the non-SMP runtime system to fail on the SMP runtime
3141
<list type="bulleted">
3143
<p>A newly spawned process will often begin executing
3144
immediately. Code that expects that the parent process
3145
will be able to execute for a while before the child
3146
process begins executing is likely to fail.</p>
3149
<p>High priority processes could previously provide
3150
mutual exclusion (bad programming style) by preventing
3151
normal and low priority processes from being run. High
3152
priority processes cannot be used this way to provide
3153
mutual exclusion.</p>
3156
<p><c><![CDATA[erlang:yield()]]></c> could be used to provide some
3157
kind of temporary mutual exclusion (also bad programming
3158
style). <c><![CDATA[erlang:yield()]]></c> cannot be used to provide
3159
any kind of mutual exclusion.</p>
3162
<p>Obscure pitfall, only if a process being traced also
3163
sends normal messages to the tracer:<br></br>
3164
The order between trace messages and normal
3165
messages is undefined. I.e. the order between normal
3166
messages sent from a tracee to a tracer and the trace
3167
messages generated from the same tracee to the same
3168
tracer is undefined. The internal order of normal
3169
messages and the internal order of trace messages will,
3170
of course, be preserved as before.</p>
3173
<p>The kernel poll feature is currently not supported by
3174
the runtime system with SMP support. It will probably be
3175
supported in a future release.</p>
3176
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
3177
<p>Own Id: OTP-6006 Aux Id: OTP-6095 </p>
3180
<p>Linked-in driver modifications.</p>
3181
<list type="bulleted">
3183
<p>Linked-in drivers must be recompiled.</p>
3186
<p>The <c><![CDATA[refc]]></c> field in the <c><![CDATA[ErlDrvBinary]]></c> type
3187
has been removed. The reference count can be accessed
3188
via API functions. For more information see
3189
the <c><![CDATA[erl_driver(1)]]></c> man page.</p>
3192
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
3193
<p>Own Id: OTP-6095 Aux Id: OTP-6006 </p>
3200
<title>ERTS 5.4.13</title>
3203
<title>Fixed Bugs and Malfunctions</title>
3204
<list type="bulleted">
3206
<p>Large files (more than 2 GBytes) are now handled on
3208
<p>Own Id: OTP-5849 Aux Id: seq10157 </p>
3211
<p>A failing bit syntax construction could fail with the
3212
PREVIOUS exeption reason that had occured in the process
3213
(instead of with <c><![CDATA[badarg]]></c>).</p>
3214
<p>Own Id: OTP-5911</p>
3217
<p>When building OTP, the Kernel application was built in
3218
both the primary and secondary bootstrap steps, which
3219
would cause problems if OTP including its bootstrap is
3220
checked into a version control system (such as CVS).
3221
(Thanks to Sebastian Strollo.)</p>
3222
<p>Own Id: OTP-5921</p>
3225
<p><c><![CDATA[binary_to_term(<<131,109,255,255,255,255)]]></c> and
3226
similar expressions used to crash the emulator instead of
3227
causing a <c><![CDATA[badarg]]></c> exception. (Thanks to Matthias
3229
<p>Own Id: OTP-5933</p>
3232
<p><c><![CDATA[erlang:hibernate/3]]></c> could sometimes crash the emulator
3233
when no heap was needed.</p>
3234
<p>Own Id: OTP-5940</p>
3237
<p>Execution of match specs could under rare circumstances
3238
cause the emulator to dump core.</p>
3239
<p>Execution of match specs could cause memory leaks in the
3240
hybrid emulator.</p>
3241
<p>Own Id: OTP-5955</p>
3244
<p>A bug in <c><![CDATA[erlang:trace_info/2]]></c> when getting info for a
3245
function in a deleted module resulting in an emulator
3246
crash, has been corrected.</p>
3247
<p>Own Id: OTP-5957</p>
3250
<p>Different (and old) <c><![CDATA[config.guess]]></c> files in the
3251
ERTS and Erl_Interface applications would cause build
3252
problems on the new Intel-based iMacs.
3253
(Thanks to Sebastion Strollo.)</p>
3254
<p>Own Id: OTP-5967</p>
3257
<p>pthread header and library mismatch on Linux systems (at
3258
least some SuSE and Debian) with both NPTL and
3259
Linuxthreads libraries installed.</p>
3260
<p>Own Id: OTP-5981</p>
3266
<title>Improvements and New Features</title>
3267
<list type="bulleted">
3269
<p>The driver_set_timer did not change the previous timeout
3270
if called a second time. Now it works as specified,
3271
changing the timeout.</p>
3272
<p>Own Id: OTP-5942</p>
3275
<p>The undocumented <c><![CDATA[{packet,http}]]></c> option (for the
3276
<c><![CDATA[gen_tcp]]></c> module) did not not work correctly when
3277
there were multiple continuation lines. (Thanks to Per
3279
<p>Own Id: OTP-5945</p>
3282
<p>The setuid_socket_wrap program was corrected to work for
3283
C compilers that treat the <c><![CDATA[char]]></c> type as unsigned.
3284
(Thanks to Magnus Henoch.)</p>
3285
<p>Own Id: OTP-5946</p>
3292
<title>ERTS 5.4.12</title>
3295
<title>Fixed Bugs and Malfunctions</title>
3296
<list type="bulleted">
3298
<p>Fixed a minor build problem on Windows.</p>
3299
<p>Own Id: OTP-5819 Aux Id: OTP-5382 OTP-5540 OTP-5577 </p>
3302
<p>The <c><![CDATA[delay_send]]></c> option for <c><![CDATA[gen_tcp]]></c> was
3303
broken on Windows.</p>
3304
<p>Own Id: OTP-5822</p>
3307
<p>If there were user-defined variables in the boot
3308
script, and their values were not provided using the
3309
<c><![CDATA[-boot_var]]></c> option, the emulator would refuse to
3310
start with a confusing error message. Corrected to show a
3311
clear, understandable message.</p>
3312
<p>The <c><![CDATA[prim_file]]></c> module was modified to not depend
3313
on the <c><![CDATA[lists]]></c> module, to make it possible to start
3314
the emulator using a user-defined loader. (Thanks to
3315
Martin Bjorklund.)</p>
3316
<p>Own Id: OTP-5828 Aux Id: seq10151 </p>
3322
<title>Improvements and New Features</title>
3323
<list type="bulleted">
3325
<p>The HiPE compiler identifies more leaf functions,
3326
giving slightly faster code.</p>
3327
<p>Corrected problems in HiPE's coalescing register
3328
allocating that would cause it to fail when compiling
3329
very large functions (e.g. some of parse modules in the
3330
Megaco application).</p>
3331
<p>Own Id: OTP-5853</p>
3338
<title>ERTS 5.4.11</title>
3341
<title>Fixed Bugs and Malfunctions</title>
3342
<list type="bulleted">
3344
<p>Timers could sometimes timeout too early. This bug has
3346
<p>Automatic cancellation of timers created by
3347
<c><![CDATA[erlang:send_after(Time,pid(),Msg)]]></c> and
3348
<c><![CDATA[erlang:start_timer(Time,pid(),Msg)]]></c> has been
3349
introduced. Timers created with the receiver specified by a
3350
pid, will automatically be cancelled when the receiver
3351
exits. For more information see the <c><![CDATA[erlang(3)]]></c> man
3353
<p>In order to be able to maintain a larger amount of timers
3354
without increasing the maintenance cost, the internal
3355
timer wheel and bif timer table have been enlarged.</p>
3356
<p>Also a number of minor bif timer optimizations have been
3358
<p>Own Id: OTP-5795 Aux Id: OTP-5090, seq8913, seq10139,
3362
<p><c><![CDATA[erlang:monitor(process,Pid)]]></c> hanged if <c><![CDATA[Pid]]></c>
3363
referred to a process on a non-existing node with the same
3364
nodename as the nodename of node on which the call was made.
3365
This bug has now been fixed.</p>
3366
<p>Own Id: OTP-5827</p>
3372
<title>Improvements and New Features</title>
3373
<list type="bulleted">
3375
<p>By setting Unix environment variables, the priority for
3376
the emulator can be lowered when it is writing crash
3377
dumps and the time allowed for finishing writing a crash
3378
dump can be set to a certain number of seconds. See the
3379
documentation for <c><![CDATA[erl]]></c> in the ERTS application.
3380
(Also, a few other previously undocumented environment
3381
variables are now documented.)</p>
3382
<p>Own Id: OTP-5818</p>
3385
<p>Documentation improvements:</p>
3386
<p>- documentation for <c><![CDATA[erlang:link/1]]></c> corrected</p>
3387
<p>- command line flag <c><![CDATA[-code_path_cache]]></c> added</p>
3388
<p>- <c><![CDATA[erl]]></c> command line flags clarifications</p>
3389
<p>- <c><![CDATA[net_kernel(3)]]></c> clarifications</p>
3390
<p>Own Id: OTP-5847</p>
3397
<title>ERTS 5.4.10</title>
3400
<title>Fixed Bugs and Malfunctions</title>
3401
<list type="bulleted">
3403
<p>-D_GNU_SOURCE is now always passed on the compile command
3404
line on linux. This in order to ensure that all included
3405
system headers see _GNU_SOURCE defined.</p>
3406
<p>_GNU_SOURCE is now also defined on linux in configure
3407
when looking for features.</p>
3408
<p>Some minor (harmless) configure bugs were also fixed.</p>
3409
<p>Own Id: OTP-5749</p>
3412
<p>Some compiler warnings and Dialyzer warnings were
3413
eliminated in the Tools application.</p>
3414
<p>When tracing to a port (which <c><![CDATA[fprof]]></c> does),
3415
there could be fake schedule out/schedule in messages
3416
sent for a process that had exited.</p>
3417
<p>Own Id: OTP-5757</p>
3423
<title>Improvements and New Features</title>
3424
<list type="bulleted">
3426
<p>The BIFs <c><![CDATA[iolist_size/1]]></c> and <c><![CDATA[iolist_to_binary/1]]></c>
3428
<p>The BIF <c><![CDATA[list_to_existing_atom/1]]></c> has been added.</p>
3429
<p>Minor bug fix: The exception reason could be changed
3430
to <c><![CDATA[error]]></c> inside nested try/catch constructs if the
3431
<c><![CDATA[erlang:raise/3]]></c> BIF was called with an empty
3432
stacktrace. (Calling <c><![CDATA[erlang:raise/3]]></c> with an empty
3433
stacktrace is NOT recommended.)</p>
3434
<p>Minor bugfix: On Windows, <c><![CDATA[file:open/2]]></c> will now
3435
return the documented error reason <c><![CDATA[{error,eisdir}]]></c>
3436
if the filename refers to a directory (it used to return
3437
<c><![CDATA[{error,eacces}]]></c>).</p>
3438
<p>The message in the documentation for
3439
<c><![CDATA[erlang:system_monitor/2]]></c>, description of
3440
<c><![CDATA[busy_dist_port]]></c>, was corrected.</p>
3441
<p>Own Id: OTP-5709 Aux Id: seq10100 </p>
3444
<p>The previously undocumented and UNSUPPORTED <c><![CDATA[zlib]]></c>
3445
module has been updated in an incompatible way and many
3446
bugs have been corrected. It is now also documented.</p>
3447
<p>*** POTENTIAL INCOMPATIBILITY ***</p>
3448
<p>Own Id: OTP-5715</p>
3451
<p>New socket options <c><![CDATA[priority]]></c> and <c><![CDATA[tos]]></c> for
3452
platforms that support them (currently only Linux).</p>
3453
<p>Own Id: OTP-5756</p>
3456
<p>Only the emulator is now linked with termcap library in
3457
order to decrease library dependencies for other otp
3459
<p>Own Id: OTP-5758</p>
3466
<title>ERTS 5.4.9.2</title>
3469
<title>Fixed Bugs and Malfunctions</title>
3470
<list type="bulleted">
3472
<p>The native resolver has gotten an control API for
3473
extended debugging and soft restart. It is:
3474
<c><![CDATA[inet_gethost_native:control(Control)]]></c><br></br><c><![CDATA[Control = {debug_level,Level} | soft_restart]]></c><br></br><c><![CDATA[Level = integer() in the range 0-4]]></c>.</p>
3475
<p>Own Id: OTP-5751 Aux Id: EABln25013 </p>
3482
<title>ERTS 5.4.9.1</title>
3485
<title>Improvements and New Features</title>
3486
<list type="bulleted">
3488
<p>On VxWorks, epmd did not handle file descriptors with
3489
higher numbers than 63. Also, if epmd should get a file
3490
descriptor with a number >= FD_SETSIZE, it will close a
3491
the file descritport and write a message to the log
3492
(instead of mysteriously fail); the Erlang node that
3493
tried to register will fail with a duplicate_name error
3494
(unfortunately, epmd has no way to indicate to the Erlang
3495
node why the register attempt failed).</p>
3496
<p>Own Id: OTP-5716 Aux Id: seq10070 </p>
3503
<title>ERTS 5.4.9</title>
3506
<title>Fixed Bugs and Malfunctions</title>
3507
<list type="bulleted">
3509
<p>Timezone data is now initialized better. (was a problem
3510
at least on NetBSD 2.0.2) Thanks to Rich Neswold.</p>
3511
<p>Own Id: OTP-5621</p>
3514
<p>The hybrid-heap emulator ('erl -hybrid') is much more
3515
stable. We have corrected all known bugs that caused it
3516
to dump core while running our test suites.</p>
3517
<p>Own Id: OTP-5634</p>
3520
<p>Fixed rare memory leaks in <c><![CDATA[erlang:demonitor/1]]></c> when
3521
distributed monitors were removed.</p>
3522
<p>Own Id: OTP-5692</p>
3525
<p>Processes were sometimes unnecessarily garbage collected
3526
when terminating. These unnecessary garbage collections
3527
have now been eliminated.</p>
3528
<p>Own Id: OTP-5693</p>
3534
<title>Improvements and New Features</title>
3535
<list type="bulleted">
3537
<p>The <c><![CDATA[c:i/0]]></c> function will now run in a paged mode if
3538
there are more than 100 processes in the system. (Thanks
3540
<p><c><![CDATA[erlang:system_info(process_count)]]></c> has
3541
been optimized and does now return exactly the same value
3542
as <c><![CDATA[length(processes())]]></c>. Previously
3543
<c><![CDATA[erlang:system_info(process_count)]]></c> did not include
3544
exiting processes which are included in
3545
<c><![CDATA[length(processes())]]></c>.</p>
3546
<p>The <c><![CDATA[+P]]></c> flag for <c><![CDATA[erl]]></c>, which sets the maximum
3547
number of processes allowed to exist at the same, no longer
3548
accepts values higher than 134217727. (You will still
3549
probably run out of memory before you'll be able to reach
3551
<p>Own Id: OTP-5645 Aux Id: seq9984 </p>
3554
<p>The term-building driver functions
3555
<c><![CDATA[driver_output_term()]]></c> and <c><![CDATA[driver_send_term()]]></c>
3556
have been updated:</p>
3557
<p>The ERL_DRV_FLOAT type has been added.</p>
3558
<p>For the ERL_DRV_BINARY type, the length and offset are
3559
now validated against the length of the driver binary.</p>
3560
<p>The ERL_DRV_PID type is now implemented (it was
3561
documented, but not implemented).</p>
3562
<p>Own Id: OTP-5674</p>
3569
<title>ERTS 5.4.8</title>
3572
<title>Fixed Bugs and Malfunctions</title>
3573
<list type="bulleted">
3575
<p><c><![CDATA[ets:delete/1]]></c> now allows other Erlang process to run
3576
when a large table is being deleted.</p>
3577
<p>Own Id: OTP-5572</p>
3580
<p>A bug regarding tcp sockets which results in hanging
3581
<c><![CDATA[gen_tcp:send/2]]></c> has been corrected. To encounter
3582
this bug you needed one process that read from a socket,
3583
one that wrote more date than the reader read out so the
3584
sender got suspended, and then the reader closed the
3585
socket. (Reported and diagnosed by Alexey Shchepin.)</p>
3586
<p>Corrected a bug in the (undocumented and unsupported)
3587
option <c><![CDATA[{packet,http}]]></c> for <c><![CDATA[gen_tcp.]]></c>
3588
(Thanks to Claes Wikstrom and Luke Gorrie.)</p>
3589
<p>Updated the documentation regarding the second argument to
3590
<c><![CDATA[gen_tcp:recv/2]]></c>, the <c><![CDATA[Length]]></c> to receive.</p>
3591
<p>Own Id: OTP-5582 Aux Id: seq9839 </p>
3594
<p>Erlang/OTP will now build on Mac OS X 10.4 "Tiger" (the
3595
problem was that 10.4 has a partially implemented poll()
3596
function that can't handle devices). Also, on Mac OS X
3597
10.3 "Panther", Erlang/OTP will now use select() instead
3598
of poll() (because poll() on Mac OS X 10.3 is implemented
3599
using select()).</p>
3600
<p>Own Id: OTP-5595</p>
3603
<p>A bug in the file driver when opening a file in
3604
compressed mode, and the returned allocated pointer from
3605
the compressing library was in very high memory (>= 2GB),
3606
causing e.g. <c><![CDATA[file:read/2]]></c> to return
3607
<c><![CDATA[{error,ebadf}]]></c>, has been corrected.</p>
3608
<p>Own Id: OTP-5618</p>
3614
<title>Improvements and New Features</title>
3615
<list type="bulleted">
3617
<p>The new fun <c><![CDATA[M:F/A]]></c> construct creates a fun that
3618
refers to the latest version of <c><![CDATA[M:F/A.]]></c> This syntax is
3619
meant to replace tuple funs <c><![CDATA[{M,F}]]></c> which have many
3621
<p>The new type test <c><![CDATA[is_function(Fun,A)]]></c> (which may be
3622
used in guards) test whether <c><![CDATA[Fun]]></c> is a fun that can be
3623
applied with <c><![CDATA[A]]></c> arguments. (Currently, <c><![CDATA[Fun]]></c>
3624
can also be a tuple fun.)</p>
3625
<p>Own Id: OTP-5584</p>
3628
<p>In the HiPE application, there's a new experimental
3629
register allocator (optimistic coalescing), and the
3630
linear scan register allocator is now also available on
3631
ppc. Plus lots of cleanups.</p>
3632
<p>Minor hybrid heap corrections.</p>
3633
<p>The maximum size of a heap used to be articially
3634
limited so that the size of a heap would fit in 28 bits;
3635
that limitation could cause the emulator to terminate in
3636
a garbage collection even if there still was available
3637
memory. Now the largest heap size for a 32 bit CPU is
3638
1,699,221,830 bytes. (Thanks to Jesper Wilhelmsson.)</p>
3639
<p>Also removed the undocumented <c><![CDATA[+H]]></c> emulator option.</p>
3640
<p>Own Id: OTP-5596</p>
3647
<title>ERTS 5.4.7</title>
3650
<title>Fixed Bugs and Malfunctions</title>
3651
<list type="bulleted">
3653
<p><c><![CDATA[to_erl]]></c> could close unexpectedly on Linux systems.
3654
Also, output from the emulator could be lost. Corrected.</p>
3655
<p>Own Id: OTP-5561</p>
3661
<title>Improvements and New Features</title>
3662
<list type="bulleted">
3664
<p>The <c><![CDATA[cpu_timestamp]]></c> option for <c><![CDATA[erlang:trace/3]]></c>
3665
is now also supported on Linux.</p>
3666
<p>Own Id: OTP-5532 Aux Id: seq9813 </p>
3669
<p>The last known <c><![CDATA[werl]]></c> window size/position is now saved
3670
correctly when <c><![CDATA[werl]]></c> is stopped with the window
3671
minimized. A problem with the placement not being saved if
3672
the emulator is halted or stopped from the JCL menu has also
3674
<p>Own Id: OTP-5544 Aux Id: OTP-5522 </p>
3681
<title>ERTS 5.4.6</title>
3684
<title>Fixed Bugs and Malfunctions</title>
3685
<list type="bulleted">
3687
<p>Some math libraries do not always throw floating-point
3688
exceptions on errors. In order to be able to use these
3689
libraries, floating-point errors are now more thoroughly
3691
<p>Misc floating-point fixes for Linux and MacOSX.</p>
3692
<p>Own Id: OTP-5467</p>
3695
<p>An internal buffer was sometimes not cleared which caused
3696
garbage to appear in error messages sent to the error
3698
<p><c><![CDATA[errno]]></c> was sometimes clobbered which caused erroneous
3699
error reports about <c><![CDATA[poll()]]></c> errors.</p>
3700
<p>Only emulators on unix platforms were affected by these
3702
<p>Own Id: OTP-5492</p>
3705
<p>The ethread library now works on OpenBSD.</p>
3706
<p>Own Id: OTP-5515</p>
3709
<p>Corrected a bug in the (undocumented and unsupported)
3710
option <c><![CDATA[{packet,http}]]></c> for <c><![CDATA[gen_tcp]]></c>.
3711
(Thanks to Claes Wikstrom and Luke Gorrie.)</p>
3712
<p>Own Id: OTP-5519</p>
3718
<title>Improvements and New Features</title>
3719
<list type="bulleted">
3721
<p><c><![CDATA[binary_to_term/1]]></c> could cause the emulator to crash
3722
when given invalid pids or funs.</p>
3723
<p>Own Id: OTP-5484 Aux Id: seq9801 </p>
3726
<p>Some more stability problems were fixed in the
3727
hybrid-heap emulator.</p>
3728
<p>Own Id: OTP-5489</p>
3731
<p>After <c><![CDATA[werl]]></c> was closed with the window minimized, it
3732
was not possible to restart <c><![CDATA[werl]]></c> with an open
3733
window. A temporary solution has so far been implemented
3734
that restores the initial window settings every time
3735
<c><![CDATA[werl]]></c> is started.</p>
3736
<p>Own Id: OTP-5522</p>
3743
<title>ERTS 5.4.5</title>
3746
<title>Fixed Bugs and Malfunctions</title>
3747
<list type="bulleted">
3749
<p>If a process had node links (created by
3750
<c><![CDATA[monitor_node/2]]></c>), excuting
3751
<c><![CDATA[process_info(Pid,memory)]]></c> for that process would
3752
crash the emulator.</p>
3753
<p>Own Id: OTP-5420</p>
3756
<p>Minor corrections to the help text printed by <c><![CDATA[erlc -help]]></c>. The documentation for <c><![CDATA[erlc]]></c> was also
3757
slightly updated.</p>
3758
<p>Own Id: OTP-5428</p>
3761
<p>32-bit words were used for offsets in the garbage
3762
collector. This caused the emulator to crash on 64-bit
3763
machines when heaps were moved more than 4 GB during
3764
garbage collection.</p>
3765
<p>Own Id: OTP-5430</p>
3768
<p><c><![CDATA[is_boolean(42.5)]]></c> failed to load if optimization was
3769
explicitly turned off.</p>
3770
<p>Own Id: OTP-5448</p>
3773
<p>If there was a call to <c><![CDATA[Module:foo/X]]></c> from any loaded
3774
module, the returned by <c><![CDATA[M:module_info(exports)]]></c> would
3775
always include <c><![CDATA[{foo,X}]]></c> (even though
3776
<c><![CDATA[Module:foo/X]]></c> if was not defined).</p>
3777
<p>Own Id: OTP-5450 Aux Id: seq9722 </p>
3783
<title>Improvements and New Features</title>
3784
<list type="bulleted">
3786
<p>The <c><![CDATA[c]]></c> option for the <c><![CDATA[+B]]></c> flag has been
3787
introduced which makes it possible to use Ctrl-C
3788
(Ctrl-Break on Windows) to interrupt the shell process
3789
rather than to invoke the emulator break handler. All new
3790
<c><![CDATA[+B]]></c> options are also supported on Windows
3791
(<c><![CDATA[werl]]></c>) as of now. Furthermore, Ctrl-C on Windows has
3792
now been reserved for copying text (what Ctrl-Ins was used
3793
for previously). Ctrl-Break should be used for break
3794
handling. Lastly, the documentation of the system flags has
3796
<p>Own Id: OTP-5388</p>
3803
<title>ERTS 5.4.4</title>
3806
<title>Fixed Bugs and Malfunctions</title>
3807
<list type="bulleted">
3809
<p>The function <c><![CDATA[erlang:float/1]]></c> can now be used in
3810
match specifications and is recognized by
3811
<c><![CDATA[dbg:fun2ms/1]]></c> and <c><![CDATA[ets:fun2ms/1]]></c>. This
3812
addition is part of the work to "harmonize" match
3813
specification guards with Erlang guards.</p>
3814
<p>Own Id: OTP-5297 Aux Id: OTP-4927 </p>
3817
<p>The <c><![CDATA[register/2]]></c> BIF would return <c><![CDATA[true]]></c> even if
3818
the second argument was not a pid for living process.
3819
Corrected to cause an exception.</p>
3820
<p>Own Id: OTP-5324 Aux Id: seq9698 </p>
3823
<p>In the 'bag' type of ets tables, elements having the same
3824
key were supposed to be order in insertion order. The
3825
would be wrong if a rehash occurred.</p>
3826
<p>Own Id: OTP-5340 Aux Id: seq9704 </p>
3829
<p>Linked in drivers in the Crypto and Asn1 applications
3830
are now compiled with the <c><![CDATA[-D_THREAD_SAFE]]></c> and
3831
<c><![CDATA[-D_REENTRANT]]></c> switches on unix when the emulator has
3832
thread support enabled.</p>
3833
<p>Linked in drivers on MacOSX are not compiled with the
3834
undocumented <c><![CDATA[-lbundle1.o]]></c> switch anymore. Thanks to
3835
Sean Hinde who sent us a patch.</p>
3836
<p>Linked in driver in Crypto, and port programs in SSL, now
3837
compiles on OSF1.</p>
3838
<p>Minor makefile improvements in Runtime_Tools.</p>
3839
<p>Own Id: OTP-5346</p>
3842
<p>Fixed a bug in the hybrid heap in distributed send
3844
<p>Own Id: OTP-5361</p>
3847
<p>A BIF <c><![CDATA[erlang:raise/3]]></c> has been added. See the manual
3848
for details. It is intended for internal system programming
3849
only, advanced error handling.</p>
3850
<p>Own Id: OTP-5376 Aux Id: OTP-5257 </p>
3853
<p>Mikael Pettersson (HiPE) corrected a few bugs in the
3854
emulator that caused problems when compiled with the
3855
experimental gcc-4.0.0.</p>
3856
<p>Own Id: OTP-5386</p>
3862
<title>Improvements and New Features</title>
3863
<list type="bulleted">
3865
<p>Minor update of the internal documentation of
3866
the <c><![CDATA[epmd]]></c> protocol.</p>
3867
<p>The listen port of <c><![CDATA[epmd]]></c> has now been registered at
3869
<url href="http://www.iana.org/assignments/port-numbers">http://www.iana.org/assignments/port-numbers</url>.</p>
3870
<p>Own Id: OTP-5234</p>
3873
<p><c><![CDATA[run_erl.c]]></c> now works on Mac OS X and FreeBSD.</p>
3874
<p>Own Id: OTP-5384</p>
3877
<p>A few bugs were corrected in the HiPE application.</p>
3878
<p>Own Id: OTP-5385</p>
3885
<title>ERTS 5.4.3</title>
3888
<title>Fixed Bugs and Malfunctions</title>
3889
<list type="bulleted">
3891
<p>Font and color preferences for <c><![CDATA[werl.exe]]></c> now can be
3892
saved even after the first time you run <c><![CDATA[werl.exe]]></c>.
3893
The window position and size is also saved. Patch from
3894
James Hague who did all the hard work.</p>
3895
<p>Own Id: OTP-5250</p>
3898
<p>OTP archive libraries, e.g. the <c><![CDATA[erl_interface]]></c>
3899
library, on MacOSX could not be used without first
3900
rerunning <c><![CDATA[ranlib]]></c> on them. This bug has now been
3902
<p>Own Id: OTP-5274</p>
3905
<p>Bugs in <c><![CDATA[erlang:hash/2]]></c> and <c><![CDATA[erlang:phash/2]]></c> on
3906
64-bit platforms have been fixed.</p>
3907
<p>Own Id: OTP-5292</p>
3913
<title>Improvements and New Features</title>
3914
<list type="bulleted">
3916
<p>Corrected a crash in the 64-bits emulator.</p>
3917
<p>Corrected a problem in the hybrid heap emulator.</p>
3918
<p>In the chapter about the abstract format in the ERTS User's
3919
Guide, updated the last section about how the abstract
3920
format is stored in BEAM files.</p>
3921
<p>Own Id: OTP-5262</p>
3924
<p>The maximum number of concurrent threads in the internal
3925
ethread thread package has been limited to 2048.</p>
3926
<p>Own Id: OTP-5280</p>
3933
<title>ERTS 5.4.2.1</title>
3936
<title>Improvements and New Features</title>
3937
<list type="bulleted">
3939
<p>If Erlang/OTP was installed in a short directory name,
3940
such as <c><![CDATA[C:\\Program\\erl5.4.2]]></c>, the emulator would not
3942
<p>Own Id: OTP-5254</p>
3949
<title>ERTS 5.4.2</title>
3952
<title>Fixed Bugs and Malfunctions</title>
3953
<list type="bulleted">
3955
<p>If one used <c><![CDATA[select/3]]></c> and <c><![CDATA[select/1]]></c> on a
3956
non-fixed ETS table and deleted objects simultaneously,
3957
the emulator could crash. Note that the result of such
3958
simultaneous operations on tables that are not in a fixed
3959
state is still undefined, but the emulator crash is,
3960
needless to say, fixed.</p>
3961
<p>Own Id: OTP-5209 Aux Id: seq9198 </p>
3964
<p>Arithmetic with big numbers could crash the emulator.</p>
3965
<p>The HiPE native code compiler and run-time code in the
3966
emulator has been updated. (Note: Native code is still
3968
<p>Eliminated a few bugs that could crash the hybrid emulator
3969
(which is not supported).</p>
3970
<p>Own Id: OTP-5233 Aux Id: seq9587 </p>
3977
<title>ERTS 5.4.1</title>
3980
<title>Fixed Bugs and Malfunctions</title>
3981
<list type="bulleted">
3983
<p>The ethread library was unnecessarily rebuilt multiple
3984
times during the build process, also a debug version of
3985
the library was build during the install phase. These
3986
unnecessary builds have now been removed. Note, the
3987
content of the installed Erlang/OTP system is not
3988
effected at all by this change.</p>
3989
<p>Own Id: OTP-5203</p>
3992
<p>The emulator could fail to clear the memory segment
3993
cache. This could potentially cause memory allocation to
3994
unnecessarily fail when memory usage was close to its
3995
maximum. This bug has now been fixed.</p>
3996
<p>Own Id: OTP-5211</p>
3999
<p>std_alloc (std short for standard) was sometimes called
4000
def_alloc (def short for default). std_alloc is now
4001
everywhere refered to as std_alloc.</p>
4002
<p>Own Id: OTP-5216</p>
4005
<p>A documentation bug has been corrected in
4006
the <c><![CDATA[erts_alloc(3)]]></c> documentation. It was stated that
4007
some of the memory allocators present were by default
4008
disabled. This is true for Erlang/OTP R9C, but is not true
4009
for Erlang/OTP R10B. In R10B all memory allocators present
4010
are enabled by default.</p>
4011
<p>Own Id: OTP-5217</p>
4017
<title>Improvements and New Features</title>
4018
<list type="bulleted">
4020
<p>The emulator now close all open files and sockets
4021
immediately after receiving an USR1 signal. This causes
4022
the emulator to unregister at <c><![CDATA[epmd]]></c> as early as
4024
<p>Own Id: OTP-5221 Aux Id: OTP-4985, seq9514 </p>
4027
<p>Try/catch support in the emulator slightly updated.</p>
4028
<p>Own Id: OTP-5229</p>