Age | Commit message (Collapse) | Author |
|
* rickard/ethread-rewrite/OTP-8544:
Rewrite ethread library
|
|
Large parts of the ethread library have been rewritten. The
ethread library is an Erlang runtime system internal, portable
thread library used by the runtime system itself.
Most notable improvement is a reader optimized rwlock
implementation which dramatically improve the performance of
read-lock/read-unlock operations on multi processor systems by
avoiding ping-ponging of the rwlock cache lines. The reader
optimized rwlock implementation is used by miscellaneous
rwlocks in the runtime system that are known to be read-locked
frequently, and can be enabled on ETS tables by passing the
`{read_concurrency, true}' option upon table creation. See the
documentation of `ets:new/2' for more information.
The ethread library can now also use the libatomic_ops library
for atomic memory accesses. This makes it possible for the
Erlang runtime system to utilize optimized atomic operations
on more platforms than before. Use the
`--with-libatomic_ops=PATH' configure command line argument
when specifying where the libatomic_ops installation is
located. The libatomic_ops library can be downloaded from:
http://www.hpl.hp.com/research/linux/atomic_ops/
The changed API of the ethread library has also caused
modifications in the Erlang runtime system. Preparations for
the to come "delayed deallocation" feature has also been done
since it depends on the ethread library.
Note: When building for x86, the ethread library will now use
instructions that first appeared on the pentium 4 processor. If
you want the runtime system to be compatible with older
processors (back to 486) you need to pass the
`--enable-ethread-pre-pentium4-compatibility' configure command
line argument when configuring the system.
|
|
|
|
* sverker/one_offheap_list/OTP-8737:
One off-heap list, to eliminate two words per ETS object.
|
|
Merging the three off-heap lists (binaries, funs and externals) into
one list. This reduces memory consumption by two words (pointers) per
ETS object.
|
|
* sverker/ets_select_hang/OTP-8732:
Fix ets:select hanging on ordered_set with empty list as key.
|
|
* sverker/full_INTEGER_EXT/OTP-8540:
term_to_binary use all 32 bits of INTEGER_EXT
|
|
* sverker/driver_int64_bug/OTP-8716:
Fix bug that caused faulty 64-bit integers from drivers.
|
|
erl_db_tree.c incorrectly used NIL (empty list) as "lastkey" to mark
start of the iteration. A real NIL key could then cause a
select or match iteration to be restarted over and over again if the
last key before a trap happended to be NIL.
Changed NIL to THE_NON_VALUE. Should be ok as the initial key value can
never be put into any continuation tuple.
|
|
* se/http_response_empty_phrase:
{packet,http} allow empty response phrase
|
|
Earlier, external format INTEGER_EXT was only produced for 28-bit
signed integers. Now full 32-bit signed integers are produced as
INTEGER_EXT to avoid the more costly SMALL_BIG_EXT format. Both old
and new code can read 32-bit INTEGER_EXT.
Also fixed integer encoding bugs in erl_interface erl_encode/erl_decode.
(Thanks to Alexander Demidenko for reporting)
|
|
|
|
A user defined CPU topology set via a call to <seealso
marker="erlang#system_flag_cpu_topology">erlang:system_flag(cpu_topology,
CPUTopology)</seealso> was not properly verified, and could in worst case
cause an emulator crash. The emulator crash could only occur when a user
defined CPU topology already existed and was redefined.
|
|
Packet type 'http' for erlang:deocode:packet and gen_tcp allow a
response line with an empty phrase string, like
"HTTP/1.1 200\r\n".
Earlier, an empty phrase was only accepted if the status code
was followed by space or tab, like
"HTTP/1.1 200 \r\n".
|
|
* commit 'bg/nif_error':
crypto: Add type specs for all documented functions
crypto: Use erlang:nif_error/1 to squelch false Dialyzer warnings
Add erlang:nif_error/1,2
|
|
* pg/fix_erts_bld_string_n:
Fix erts_bld_string_n (called by enif_make_string and erlang:decode_packet/3) to work with 8 bits characters
OTP-8685 pg/fix_erts_bld_string_n
Fix bug causing erlang:decode_packet and enif_make_string to generate
faulty strings with negative character values for ascii values larger than
127. (Thanks to Paul Guyot)
|
|
erlang:decode_packet/3) to work with 8 bits characters
|
|
|
|
Without the instruction defined in ops the interpreter will not
compile when using NO_JUMPTABLE.
|
|
If a rehash was done the old hash value was used. This was incorrect.
|
|
|
|
erlang:system_info(snifs) lists all static native implemented
functions. The function presents the lists with three tuple
values containing MFAs [{Module, Function, Arity}, ...].
|
|
|
|
|
|
Call count previously used a global lock for accessing and writing
its counter in the breakpoint. This is now changed to atomics instead.
The change will let call count tracing and cprof to scale better
when increasing the number of schedulers.
|
|
To solve the issue of multiple schedulers constantly updating the
head pointer to the bp data wheel, each scheduler now has its own
entrypoint to the wheel. This head pointer can be updated without
a locking being taken. Previously there were no lock ...
|
|
|
|
|
|
call_time trace will use instruction pointers instead of
breakpoint data pointers. More costly lookup but the bdt
structure might be deallocated, we do not want that.
Remove unnecessary pattern lock.
|
|
|
|
|
|
op_i_time_breakpoint is now used
|
|
Call time breakpoint tracing traces per call trace per process.
- Add hashes to support the extra dimension.
- Teach trace_info/2 to handle the extra information dimension.
|
|
|
|
Initial commit with a new breakpoint instruction and PSD areas
for temporary time storage during tracing.
|
|
erlang:system_info(multi_scheduling) sometimes erroneously returned enabled
when it should have returned blocked.
|
|
Alignment of scheduler data and run queues were adjusted.
|
|
The runtime system will by default bind schedulers to logical processors
using the default_bind bind type if the amount of schedulers are at least
equal to the amount of logical processors configured, binding of schedulers
is supported, and a CPU topology is available at startup.
NOTE: If the Erlang runtime system is the only operating system process
that binds threads to logical processors, this improves the performance of
the runtime system. However, if other operating system processes (as for
example another Erlang runtime system) also bind threads to logical
processors, there might be a performance penalty instead. If this is the
case you, are are advised to unbind the schedulers using the <seealso
marker="erl#+sbt">+sbtu</seealso> command line argument, or by invoking
<seealso
marker="erlang#system_flag_scheduler_bind_type">erlang:system_flag(schedule
r_bind_type, unbound)</seealso>.
|
|
New NIF features:
Send messages from a NIF, or from thread created by NIF, to any local
process (enif_send)
Store terms between NIF calls (enif_alloc_env, enif_make_copy)
Create binary terms with user defined memory management
(enif_make_resource_binary)
|
|
* pan/unregister_error:
Add possibly missing process lock before unregistering oneself
Keep process lock over trace of unregister
OTP-8663 Unregister may crash VM
Extreme combinations of register/unregister in a highly parallell SMP
application could crash the VM. The error is corrected.
|
|
Port locks could be prematurely destroyed.
|
|
Missing memory barriers in erts_poll() could cause the runtime system to
hang indefinitely.
|
|
This BIF was only used by the now broken SAE support.
|
|
The experimental Standalone Erlang (SAE) support based on
Joe Armstrong's work has long been broken. Remove the remaining
code and Makefile rules.
|
|
|
|
|
|
A stub function that is supposed to be replaced by a NIF usually
calls erlang:error/1 to cause an exception if the NIF library
is not loaded. For example:
foo() ->
erlang:error(nif_not_loaded).
The problem is that although erlang:error/1 will normally never be
called, Dialyzer will think that any call to the function will fail
and thus generate false warnings. Adding a spec for the function
will not help because Dialyzer will not believe the spec.
Add erlang:nif_error/1,2 that work exactly like erlang:error/1,2.
Define the return types for both BIFs to be t_any().
erlang:nif_error is used like this:
-spec foo() -> binary().
foo() ->
erlang:nif_error(nif_not_loaded).
(The -spec is optional but highly recommended, since Dialyzer
otherwise has no chance to figure out the types.)
|
|
* ta/extend-nif-api:
erl_nif: add make_atom_len, make_existing_atom_len and make_string_len
erl_nif: add enif_get_atom_length and enif_get_list_length
erl_nif: add enif_is_list and enif_is_tuple
OTP-8640 ta/extend-nif-api
New NIF API functions: enif_make_atom_len, enif_make_existing_atom_len,
enif_make_string_len, enif_get_atom_length, enif_get_list_length,
enif_is_list, enif_is_tuple (by Tuncer Ayaz)
|
|
* fm/file-operations:
Update preloaded modules
Add file:advise/4 - a wrapper to the POSIX syscall posix_fadvise
Add file:datasync/1 for syncing file contents only
sys.h: Correct the get_int64() macro
OTP-8637 fm/file-operations
The functions file:advise/4 and file:datasync/1 have been added. (Thanks to
Filipe David Manana.)
|
|
Add new NIF API functions
- enif_make_atom_len
- enif_make_existing_atom_len
- enif_make_string_len
These are basically the same as enif_make_atom,
enif_make_existing_atom and enif_make_string except that the
new functions require a length parameter instead of a
null-terminated C-string.
Signed-off-by: Tuncer Ayaz <[email protected]>
|