| Age | Commit message (Collapse) | Author |
|---|
|
The compiler does not generate select_val instructions that only
selects one value, but the loader may previously have created such
an instruction when it splitted a select_val instruction that
selected on bignums.
|
|
Combine the put_tuple/2 and all following put/1 instructions
to one i_put_tuple/2 instruction. In general, that will reduce
the number of instruction words by 50 percent.
Measurements seem to indicate that the speed is about the same.
|
|
|
|
|
|
In many (not all) cases, the value for the 'I' type will
fit into 32 bits.
|
|
|
|
|
|
|
|
Introduce a new 'Q' type, similar to 'P' except that it
can be packed.
|
|
In the 32-bit BEAM emulator, it is only possible to pack
3 register operands into one word. Therefore, the move2
instruction (that has 4 operands) needs two words for its
operands.
Take advantage of the larger wordsize in the 64-bit emulator
and pack up to 4 operands into a single word.
|
|
In the transformation engine in the loader, an is_eq/1 instruction
is currently always preceded by an is_type/1 instruction. Therefore,
save a word and slight amount of time by combining those
instructions into an is_type_eq/2 instruction.
|
|
The i_jump_on_val_zero/3 and i_select_tuple_arity/3 instructions
were not disassembled correctly.
|
|
|
|
It would only really work in simple case like:
select_val S=q Fail=f Size=u Rest=* => ...
where all operands for a single instruction where bound to
variables, and not for more complicated cases such as:
i_put_tuple Dst Arity Puts=* | put PutSrc => ...
|
|
There was a version of the BEAM loader and emulator that
had two versions of the fmove/2 instruction, one version
that allocated heap space internally and a newer version that
assumed that a previous test_heap/2 instruction had already
allocated the heap space.
Though the allocating fmove/2 instruction is no longer
supported, some vestiges of it still remains.
|
|
erts_debug:instructions/0 is useful for finding which specific
instructions that are not used at all.
|
|
* pan/unicode-filenames/OTP-8887: (27 commits)
Test and correct filelib and filename
Add documentation to erlang.xml and slight correction to unicode_usage.xml
Add section about Unicode file names to stdlib users guide
Correct bug in file_name_SUITE making it fail on Unix instead of Windows7
Add documentation about raw filenames and Unicode file name translation mode
Make filelib not crash on re codepoints beyond 255 in re when filename is raw
Mend on_load_embedded testcase which did not handle windows links
Correct testcase regarding windows versions supporting soft links.
Teach filelib to use re in unicode mode when filenames are not raw
Treat soft links on Windows correctly in file_name_SUITE
Adapt new soft and hard link routines on Windos to Unicode
Corrected testcases broken by unicode filenames
Update preloaded prim_file
Teach prim_file not to accept atoms and not to throw exceptions
Adapt inet_drv to Visual Studio 2008
Teach spawn_executable about Unicode
Convert filenames read on MacOSX to canonical form
Teach file to accept codepoints beyond 255.
Add testcases
Correct shell utilities to handle unicode and possibly binaries
...
|
|
* rickard/load-balance-wrap/OTP-8950:
Prevent wrapping of values used during load balancing
|
|
Some integer values used during load balancing could
under rare circumstances wrap causing a load unbalance
between schedulers.
|
|
|
|
Also corrected compressed files on Windows
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The compressed format is using a slighty modified variant of the extern format
(term_to_binary). To not worsen key lookup's too much, the top tuple itself
and the key element are not compressed. Table objects with only immediate
non-key elements will therefor not gain anything (but actually consume one
extra word for "alloc_size").
|
|
* rickard/cpu-groups/OTP-8861:
Generalize reader groups
Move cpu topology functionality into erl_cpu_topology.[ch]
Do not use more reader groups for schedulers than schedulers
Conflicts:
erts/emulator/beam/erl_init.c
|
|
Reader groups have been generalized to cpu groups which can be
used for implementing reader groups, but also for implementing
other functionality in the future.
|
|
|
|
When the runtime system had fewer schedulers than logical processors,
the system could get an unnecessarily large amount reader groups.
|
|
* rickard/sys_schedule_debug:
Verify that no outstanding I/O exist when checking for I/O in debug build
|
|
|
|
* pg/fix-system_info-cpu_topology-segfault:
Fix crash with erlang:system_info({cpu_topology,junk})
OTP-8914
|
|
* rickard/sched-misc/OTP-8932:
Fix erts_schedule_misc_op() so it can be called from arbitrary threads
|
|
* rickard/halfword-fix/OTP-8910:
Fix ERL_DRV_INT and ERL_DRV_UINT in halfword emulator
|
|
* rickard/dist/OTP-8901:
Be less eager to set dist entry in busy state
|
|
* rickard/tsd-get/OTP-8889:
Fix erl_drv_tsd_get() and enif_tsd_get()
|
|
Tracing to port could cause an emulator crash when unloading the
trace driver.
|
|
The runtime system is now less eager to suspend
processes sending messages over the distribution. The
default value of the distribution buffer busy limit
has also been increased from 128 KB to 1 MB. This in
order to improve throughput.
|
|
Id: OTP-8912
This patch creates a new family of flags with the "+z" prefix. It
further creates a new configuration option called "dbbl" (which is the
first letter of the name dist_buf_busy_limit). Example usage of this
flag would be "+zdbbl 1048576".
This patch creates an adjustable buffer limit for the amount of data
that may be buffered by the erlang distribution code (in dist.c
specifically). Before this patch, this hard-coded constant was used:
#define ERTS_DE_BUSY_LIMIT (128*1024)
When large binaries are transmitted between nodes (or simply a lot of
medium-sized binaries), it is very easy to hit the old 128KB limit.
Processes that use the erlang:system_monitor() BIF to monitor system
events can be spammed by {monitor, busy_dist_port, ...} message tuples
at rates of tens to even hundreds of messages/second.
A larger buffer limit will allow processes to buffer more outgoing
messages over the distribution. When the buffer limit has been
reached, sending processes will be suspended until the buffer size has
shrunk. The buffer limit is per distribution channel. A higher limit
will give lower latency and higher throughput at the expense of
higher memory usage.
A variation of this patch has been in commercial production use in at
least two companies that the author is aware of. Larger buffer values
can reduce the number of {monitor, busy_dist_port, ...} system
messages drastically, lower overall messaging latencies, and prevent
false timeouts and 'nodedown' messages in extremely busy Mnesia systems.
Test suite: there are two tests:
a. In erlexec_SUITE.erl to test basic set & get of the value
b. In distribution_SUITE.erl, to verify that setting +zdbbl very
low will actually change behavior.
|
|
|
|
A bug in erl_drv_tsd_get() and enif_tsd_get() could cause an
emulator crash. These functions are currently not used in OTP.
That is, the crash only occur on systems with user implemented
NIF libraries or drivers that use one of these functions.
|
|
* rickard/halfword-bug:
Fix newly introduced halfword emulator bugs
|
|
|
|
There is a bug in system_info BIF causing a crash if
erts_get_cpu_topology_term fails. The fix comes with a non-regression
test.
|
|
It seems to work (at least on a little-endian architecture)
by sheer luck.
|
