| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
We can't just leave it in queue with dist_ext=NULL.
Two symptoms seen:
1. 'receive' trying to deref dist_ext as NULL.
2. GC think it's a term and put THE_NON_VALUE in root set.
|
|
* bjorn/erts/improve-beam-ops:
Move out variables from the head of combined instructions
Change operand from 'P' to 'Q' for i_apply_last and i_apply_fun_last
Add CHECK_ALIGNED() for testing storage destinations
instrs.tab: Add missing -no_next directives
beam_load.c: Generalize the 'P' operator in the packing engine
Break out most of the initialization from process_main()
Eliminate the OpCode() macro
Eliminate unnecessary and inconsistent casts
Refactor macros for accessing Beam instructions
beam_emu: Make order of macros consistent
beam_SUITE: Strengthen test of packed registers
|
|
Point out the correct line number in stack traces
|
|
* maint:
Don't allow null in filenames
|
|
* lukas/erts/poll-thread/OTP-14346: (25 commits)
erts: Trigger ready events when erts_io_control fails
erts: enif_select steal test
kernel: Rewrite gen_udp_SUITE:read_packet tc
erts: disable kernel-poll on OS X vsn < 16
erts: Fix msacc testcase with new poll-thread
erts: Add testcases to test IOp and IOt options
erts: get_internal_state(check_io_debug) now prints to error_logger
erts: Remove eager check io
erts: Move all I/O polling to a seperate thread
erts: Fix smp_select testcase to use ERL_DRV_USE
erts: Fix msacc unmanaged state counter
erts: Optimize port_task quick allocator
erts: Add ERTS_THR_PREF_QUICK_ALLOC_IMPL
erts: Update suspend of scheduler to handle multiple pollsets
erts: Add multiple poll sets
erts: Some code cleanup for gdb to work better
erts: temp_alloc can no longer be disabled
erts: Refactor check_io to use one static struct
erts: Replace check_io spinlock with lock-less list insertion
erts: Add number of enif_select's to check_io_debug
...
|
|
|
|
It is not longer relevant when using the poll thread
|
|
|
|
OTP-14652
|
|
for non scheduler threads by using ERTS_THR_PREF_QUICK_ALLOC_IMPL.
|
|
usable from any (managed?) thread.
|
|
|
|
|
|
|
|
temp_alloc is used in such a way that if it ever results
in a malloc/free sequence it will slow down the system
alot. So it will no longer be possible to disable it and
it will not be disabled when using +Mea min.
OTP-14651
|
|
Move out from the head the variables that are only used in the excute
phase.
|
|
All other instructions that increment the stack pointer takes a 'Q'
operand.
|
|
Add the CHECK_ALIGNED() macro that can be used for testing that
the storage destination is word-aligned.
|
|
|
|
In the 'P' operator, don't assume that a packed target label ('f'
or 'j') is always the leftmost argument. Instead, transfer the
patch position from the accumulator to the stack.
|
|
process_main() is already too big.
|
|
Introduce the IsOpCode() macro that can be used to compare
instructions.
|
|
Consider the types in the code below:
BeamInstr* I;
.
.
.
BeamInstr* next;
next = (BeamInstr *) *I;
Goto(next);
This is illogical. If 'I' points to a BeamInstr, then 'next' should
be a BeamInstr, not a pointer to a BeamInstr. The Goto() macros does
not require a pointer, because it will cast its argument to a void*
anyway.
Therefore, this code example can be simplified to:
BeamInstr* I;
.
.
.
BeamInstr next;
next = *I;
Goto(next);
Similarly, we can remove the casts in the macros when NO_JUMP_TABLE
is defined.
|
|
The BeamOp() macro in erl_vm.h is clumsy to use. All users
cast the return value to BeamInstr.
Define new macros that are easier to use. In the future,
we might want to pack an operand into the same word as
the pointer to the instruction, so we will define two macros.
BeamIsOpCode() is used to rewrite code like this:
if (Instr == (BeamInstr) BeamOp(op_i_func_info_IaaI) {
...
}
to:
if (BeamIsOpCode(Instr, op_i_func_info_IaaI)) {
...
}
BeamOpCodeAddr(op_apply_bif) is used when we need the address
for an instruction.
Also elimiminate the global variables em_* in beam_emu.c.
They are not really needed. Use the BeamOpCodeAddr() macro
instead.
|
|
The inconsistent order has annoyed me for a long time.
While at it, also remove the unecessary definition of LabelAddr() if
NO_JUMP_TABLE is defined.
|
|
Sometimes the line number in a stack trace could be wrong,
for example for this code:
t() ->
Res = id(x), %<== Wrong line number.
Res + 1.
id(I) -> I.
The line number pointed out in the stack trace would be the
line before the line where the exception occurred.
The reason is the way the increment instruction instruction
is implemented:
OpCase(i_increment_rWtd):
{
increment_reg_val = r(0);
}
I -= 1;
goto increment__execute;
OpCase(i_increment_xWtd):
{
increment_reg_val = xb(I[1]);
}
goto increment__execute;
increment__execute:
/* Common code for increment */
.
.
.
(The implementation in OTP 20 is similar, but hand-coded directly
in beam_emu.c instead of generated.)
The instruction i_increment_rWtd decrements the instruction pointer (I)
before jumping to the common code. That means that I points *before*
the 'increment' instruction. If there is a 'line' instruction directly
before the 'increment' instruction (as there is in this example), the
instruction pointer will point before that line. Thus the previous line
will be picked up instead.
To eliminate this bug, we must never decrement the instruction pointer.
Instead, we can increment the other (longer) instructions in the
same group of combined instructions:
OpCase(i_increment_rWtd):
{
increment_reg_val = r(0);
}
goto increment__execute;
OpCase(i_increment_xWtd):
{
increment_reg_val = xb(I[1]);
}
I += 1;
goto increment__execute;
increment__execute:
/* Common code for increment */
.
.
.
Also fix a bug that was only a potential bug when ddaed7774eb0a
introduced relative jumps, but is now a real bug. See the added
comment for SET_I_REL() in macros.tab.
|
|
For a long time, there has been the two macros IS_SSMALL() and
MY_IS_SSMALL() that do exactly the same thing.
There should only be one, and it should be called IS_SSMALL().
However, we must decide which implementation to use. When
MY_IS_SSMALL() was introduced a long time ago, it was the most
efficient. In a modern C compiler, there might not be any
difference.
To find out, I used the following small C program to examine
the code generation:
#include <stdio.h>
typedef unsigned int Uint32;
typedef unsigned long Uint64;
typedef long Sint;
#define SWORD_CONSTANT(Const) Const##L
#define SMALL_BITS (64-4)
#define MAX_SMALL ((SWORD_CONSTANT(1) << (SMALL_BITS-1))-1)
#define MIN_SMALL (-(SWORD_CONSTANT(1) << (SMALL_BITS-1)))
#define MY_IS_SSMALL32(x) (((Uint32) ((((x)) >> (SMALL_BITS-1)) + 1)) < 2)
#define MY_IS_SSMALL64(x) (((Uint64) ((((x)) >> (SMALL_BITS-1)) + 1)) < 2)
#define MY_IS_SSMALL(x) (sizeof(x) == sizeof(Uint32) ? MY_IS_SSMALL32(x) : MY_IS_SSMALL64(x))
#define IS_SSMALL(x) (((x) >= MIN_SMALL) && ((x) <= MAX_SMALL))
void original(Sint n)
{
if (IS_SSMALL(n)) {
printf("yes\n");
}
}
void enhanced(Sint n)
{
if (MY_IS_SSMALL(n)) {
printf("yes\n");
}
}
gcc 7.2 produced the following code for the original() function:
.LC0:
.string "yes"
original(long):
movabs rax, 576460752303423488
add rdi, rax
movabs rax, 1152921504606846975
cmp rdi, rax
jbe .L4
rep ret
.L4:
mov edi, OFFSET FLAT:.LC0
jmp puts
clang 5.0.0 produced the following code which is slightly better:
original(long):
movabs rax, 576460752303423488
add rax, rdi
shr rax, 60
jne .LBB0_1
mov edi, .Lstr
jmp puts # TAILCALL
.LBB0_1:
ret
.Lstr:
.asciz "yes"
However, in the context of beam_emu.c, clang could produce
similar to what gcc produced.
gcc 7.2 produced the following code when MY_IS_SSMALL() was used:
.LC0:
.string "yes"
enhanced(long):
sar rdi, 59
add rdi, 1
cmp rdi, 1
jbe .L4
rep ret
.L4:
mov edi, OFFSET FLAT:.LC0
jmp puts
clang produced similar code.
This code seems to be the cheapest. There are four instructions, and
there is no loading of huge integer constants.
|
|
|
|
* lukas/erts/fix_threads_error_printout:
erts: Print the error reason when threads fail to start
|
|
|
|
The byte_offset of sub-binaries wasn't taken into account for
ProcBins, subtly ruining the results. The test suite didn't catch
it since it didn't check for sub-binaries in particular, and only
checked for equality between variations -- not whether the output
was equal to the input.
|
|
|
|
|
|
* bjorn/erts/pack-combined:
Pack combined instructions
beam_makeops: Refactor code generation
Correct disassembly of select instructions
|
|
|
|
|
|
|
|
# Conflicts:
# erts/emulator/beam/erl_process.c
# erts/emulator/beam/sys.h
# erts/emulator/sys/common/erl_check_io.c
# erts/emulator/sys/common/erl_check_io.h
# erts/emulator/sys/unix/sys.c
|
|
* lukas/erts/remove-dirty-scheduler-defines/OTP-14613:
erts: Remove possibility to disable dirty schedulers
|
|
Make sure to use the 'unpacked[-1]' when accessing the unpacked
arguments.
|
|
* bjorn/erts/relative-jumps:
Pack failure labels in i_select_val2 and i_select_tuple_arity2
Optimize i_select_tuple_arity2 and is_select_lins
Rewrite select_val_bins so that its labels can be packed
Pack sequences of trailing 'f' operands
Implement packing of 'f' and 'j'
Make sure that mask literals are 64 bits
Use relative failure labels
Add information about offset to common group start position
Remove JUMP_OFFSET
Refactor instructions to support relative jumps
Introduce a new trace_jump/1 instruction for tracing
Avoid using $Src more than once
|
|
|
|
Don't save a pointer to the default failure label. That could
relieve register pressure. Instead, if we'll need to signal an
error in i_select_tuple_arity2, delay to the execution phase.
That should be a clear win because i_select_tuple_arity2 very
rarely fails because of the term being selected is not a tuple.
|
|
|
|
Pack sequences of trailing 'f' operands for instructions
such at jump_on_val or i_select_val_lins.
|
|
|
|
Relative failure in itself is not an optimization, but we plan to
pack failure labels in the future to save memory.
|
|
|
