Age | Commit message (Collapse) | Author |
|
|
|
Optimize tail-recursive calls of BIFs
OTP-15674
|
|
|
|
If a test case timed out, all processes created by it
should die. But because the 'evaluator' processes where
created with spawn_monitor, that was not the case, and
therefor, these processes could linger.
|
|
BEAM currently does not call BIFs at the end of a function in a
tail-recursive way. That is, when calling a BIF at the end of a
function, the BIF is first called, and then the stack frame
is deallocated, and then control is transferred to the caller.
If there is no stack frame when a BIF is called in the tail position,
the loader will emit a sequence of three instructions: first an
instruction that allocates a stack frame and saves the continuation
pointer (`allocate`), then an instruction that calls the BIF
(`call_bif`), and lastly an instruction that deallocates the stack
frame and returns to the caller (`deallocate_return`).
The old compiler would essentially allocate a stack frame for each
clause in a function, so it would not be that common that a BIF was
called in the tail position when there was no stack frame, so the
three-instruction sequence was deemed acceptable.
The new compiler only allocates stack frames when truly needed, so
the three-instruction BIF call sequence has become much more common.
This commit introduces a new `call_bif_only` instruction so that only
one instruction will be needed when calling a BIF in the tail position
when there is no stack frame. This instruction is also used when there
is a stack frame to make it possible to deallocate the stack frame
**before** calling the BIF, which may make a subsequent garbage
collection at the end of the BIF call cheaper (copying less garbage).
The one downside of this change is that the function that called the
BIF will not be included in the stack backtrace (similar to how a
tail-recursive call to an Erlang function will not be included in the
backtrace).
That was the quick summary of the commit. Here comes a detailed look
at how BIF calls are translated by the loader. The first example is a
function that calls `setelement/3` in the tail position:
update_no_stackframe(X) ->
setelement(5, X, new_value).
Here is the BEAM code:
{function, update_no_stackframe, 1, 12}.
{label,11}.
{line,[...]}.
{func_info,{atom,t},{atom,update_no_stackframe},1}.
{label,12}.
{move,{x,0},{x,1}}.
{move,{atom,new_value},{x,2}}.
{move,{integer,5},{x,0}}.
{line,[...]}.
{call_ext_only,3,{extfunc,erlang,setelement,3}}.
Because there is no stack frame, the `call_ext_only` instruction will
be used to call `setelement/3`:
{call_ext_only,3,{extfunc,erlang,setelement,3}}.
The loader will transform this instruction to a three-instruction
sequence:
0000000020BD8130: allocate_tt 0 3
0000000020BD8138: call_bif_e erlang:setelement/3
0000000020BD8148: deallocate_return_Q 0
Using the `call_bif_only` instruction introduced in this commit,
only one instruction is needed:
000000005DC377F0: call_bif_only_e erlang:setelement/3
`call_bif_only` calls the BIF and returns to the caller.
Now let's look at a function that already has a stack frame when
`setelement/3` is called:
update_with_stackframe(X) ->
foobar(X),
setelement(5, X, new_value).
Here is the BEAM code:
{function, update_with_stackframe, 1, 14}.
{label,13}.
{line,[...]}.
{func_info,{atom,t},{atom,update_with_stackframe},1}.
{label,14}.
{allocate,1,1}.
{move,{x,0},{y,0}}.
{line,[...]}.
{call,1,{f,16}}.
{move,{y,0},{x,1}}.
{move,{atom,new_value},{x,2}}.
{move,{integer,5},{x,0}}.
{line,[...]}.
{call_ext_last,3,{extfunc,erlang,setelement,3},1}.
Since there is a stack frame, the `call_ext_last` instruction will be used
to deallocate the stack frame and call the function:
{call_ext_last,3,{extfunc,erlang,setelement,3},1}.
Before this commit, the loader would translate this instruction to:
0000000020BD81B8: call_bif_e erlang:setelement/3
0000000020BD81C8: deallocate_return_Q 1
That is, the BIF is called before deallocating the stack frame and returning
to the calling function.
After this commit, the loader will translate the `call_ext_last` like this:
000000005DC37868: deallocate_Q 1
000000005DC37870: call_bif_only_e erlang:setelement/3
There are still two instructions, but now the stack frame will be
deallocated before calling the BIF, which could make the potential
garbage collection after the BIF call slightly more efficient (copying
less garbage).
We could have introduced a `call_bif_last` instruction, but the code
for calling a BIF is relatively large and there does not seem be a
practical way to share the code between `call_bif` and `call_bif_only`
(since the difference is at the end, after the BIF call). Therefore,
we did not want to clone the BIF calling code yet another time to
make a `call_bif_last` instruction.
|
|
This helps avoid long sequences of enif_is_xxx in code that
serializes terms (such as JSON encoders) by letting the user
switch on the type.
|
|
Make use of macro concat magic to simplify declarations.
OTP-15565
|
|
Add a quiet logger mode (default) which limits the
printouts in the (erlang) shell (the web log will
be as verbose as usual).
Also added a way to configure the runtime of the ttest
cases.
|
|
Moved the old socket test modules into
its own (temporary?) directory.
|
|
The test case did not consider that the machine might have both
IPv4 (inet) and IPv6 (inet6) domain interfaces.
OTP-15635
|
|
Add a (basically) placeholder test suite for the net module.
OTP-15635
|
|
Make iolist_size/1 yield
OTP-15631
|
|
The iolist_size/1 function did not yield even if the input list was
very long and a call to the function did only consume a single
reduction. This commit fixes these problems.
|
|
|
|
Add a quiet mode for ttest which is used when running in
a (terminal) shell.
Moved the esock-ttest script(s) into their own directory.
Minor improvements to the (client) result printout.
Also fixed copyright (end) dates.
OTP-14831
|
|
Fixed how to figure out local address on macOS Mojave.
OTP-14831
|
|
Add enif_set_pid_undefined & enif_is_pid_undefined
|
|
'lukas/erts/fragment-dist-messages/OTP-13397/OTP-15610/OTP-15611/OTP-15612/OTP-15613'
* lukas/erts/fragment-dist-messages/OTP-13397/OTP-15610/OTP-15611/OTP-15612/OTP-15613:
erts: Add debug dist obuf memory leak check
win32: Fix ./otp_build debuginfo_win32
Make ld.sh on windows print better error reason
erts: Fix so that externals with creation 0 compare equal to all
erts: Expand etp to look for free processes
erts: Implement trapping while sending distr exit/down
erts: Add ERL_NODE_BOOKKEEP to node tables refc
erts: Refactor ErtsSendContext to be ErtsDSigSendContext
erts: Add distr testcases for fragmentation
erts: Make remote send of exit/2 trap
erts: Implement fragmentation of distrubution messages
erts: Expand distribution protocol documentation
erts: Move reason in dist messages to payload
erts: Remove a copy of distribution data payload
erts: Yield later during process exit and allow free procs to run
erts: Refactor rbt _yielding to use reductions
erts: Limit binary printout for %.XT in erts_print
|
|
|
|
|
|
as a macro wrappper around enif_compare
|
|
|
|
* maint:
Add persistent_term:get(Key, DefaultValue)
Make dialyzer faster for left-associative andalso/orelse expressions
|
|
https://bugs.erlang.org/browse/ERL-843
|
|
|
|
* sverker/map-from-ks-vs-bug/OTP-15567:
erts: Add test for bug in enif_make_maps_from_arrays
erts: Fix bug in erts_map_from_ks_and_vs
|
|
|
|
bmk/20190204/socket_as_nif/OTP-14831
|
|
|
|
Added preliminary, and temporary, windows adaptions.
Basically they amount to letting all nif-callback functions
returning badarg for all calls if on windows (this has
been accomplished by if-defing the nif-code; if win32 then
badarg else do-something).
OTP-15526
|
|
Its now possible to set a rcvbuf (otp) option value
of {N :: pos_integer(), BufSz :: pos_integer()}.
This value is used for type stream and protocol tcp,
when calling the function recv with length = 0 (zero).
The second value, BufSz, is the actual size of the receive
buffer used when calling the socket recv function, and
the first value, N, is the max number of possible reads
that will be performed (at most), even if there is more
data to read. This is limit the effect of DoS attacks.
OTP-15497
|
|
The read and write functions now *only* use the isReadable and
isWritable fields, respectively, to decide if the "socket" can be read
from and written to (previously the state field was used outside of
the mutex lock).
Also made it possible to enable / disable test suite groups individually,
via environment variables (ESOCK_TEST_...). Specifically, the ttest
group has been excluded by default (takes to long a time).
OTP-15549
|
|
Apply type optimizations across local function calls
|
|
This commit lets the type optimization pass work across functions,
tracking return and argument types to eliminate redundant tests.
|
|
* bjorn/erts/optimize-is_function2:
Optimize the is_function/2 guard test
|
|
|
|
* lukas/erts/scheduler-pollset-fixes/OTP-15538:
erts: Fix getting of poll events on linux >= 4.15.0
erts: Use reduction based polling for starved poll-set
erts: Fix pollset test cases
|
|
Add a way to *get* the file descriptor (fd) of a socket.
Useful mostly for debugging.
OTP-15528
|
|
The is_function2 instruction is executed surprisingly
frequently when running dialyzer or the compiler. It
cannot hurt to optimize it a little.
|
|
josevalim/jv-no-bin-alloc-on-empty-append/OTP-15535
Do not allocate new bitstring/binary on empty append
|
|
|
|
* lukas/erts/spawn_opt_max_heap_tc_fix:
erts: Fix process_SUITE:max_heap tests
|
|
Conflicts:
erts/emulator/beam/bif.c
erts/preloaded/ebin/erlang.beam
erts/preloaded/ebin/erts_internal.beam
erts/preloaded/ebin/prim_file.beam
|
|
This flag allows logger and other components to set the
process which log messages from ERTS are to be sent.
|
|
* maint:
Implement integer_to_list/2 and integer_to_binary/2 as CIFs
Accept base in all integer-printing functions
Document cleanup semantics for atomics and counters
|
|
|
|
|
|
Custom message format for enif_select
|
|
Make sure to flush all queued emulator logger messages
before starting the test. The trap_exit_badarg could sometimes
leave messages and since the handler and messages now are
handled in different processes this becomes a problem.
|
|
|