otp.git - Mirror of Erlang/OTP repository.

Age	Commit message (Collapse)	Author
2018-06-18	Update copyright year	Henrik Nord

2018-01-26	Eliminate get_list/3 internally in the compiler	Björn Gustavsson
	Instructions that produce more than one result complicate optimizations. get_list/3 is one of two instructions that produce multiple results (get_map_elements/3 is the other). Introduce the get_hd/2 and get_tl/2 instructions that return the head and tail of a cons cell, respectively, and use it internally in all optimization passes. For efficiency, we still want to use get_list/3 if both head and tail are used, so we will translate matching pairs of get_hd and get_tl back to get_list instructions.
2018-01-22	Don't build a stacktrace if it's only passed to erlang:raise/3	Björn Gustavsson
	Consider the following function: function({function,Name,Arity,CLabel,Is0}, Lc0) -> try %% Optimize the code for the function. catch Class:Error:Stack -> io:format("Function: ~w/~w\n", [Name,Arity]), erlang:raise(Class, Error, Stack) end. The stacktrace is retrieved, but it is only used in the call to erlang:raise/3. There is no need to build a stacktrace in this function. We can avoid the building if we introduce an instruction called raw_raise/3 that works exactly like the erlang:raise/3 BIF except that its third argument must be a raw stacktrace.
2017-12-08	Merge pull request #1634 from bjorng/bjorn/get_stacktrace-syntax/OTP-14692	Björn Gustavsson
	Add syntax in try/catch to retrieve the stacktrace directly
2017-11-30	Add syntax in try/catch to retrieve the stacktrace directly	Björn Gustavsson
	This commit adds a new syntax for retrieving the stacktrace without calling erlang:get_stacktrace/0. That allow us to deprecate erlang:get_stacktrace/0 and ultimately remove it. The problem with erlang:get_stacktrace/0 is that it can keep huge terms in a process for an indefinite time after an exception. The stacktrace can be huge after a 'function_clause' exception or a failed call to a BIF or operator, because the arguments for the call will be included in the stacktrace. For example: 1> catch abs(lists:seq(1, 1000)). {'EXIT',{badarg,[{erlang,abs, [[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20\|...]], []}, {erl_eval,do_apply,6,[{file,"erl_eval.erl"},{line,674}]}, {erl_eval,expr,5,[{file,"erl_eval.erl"},{line,431}]}, {shell,exprs,7,[{file,"shell.erl"},{line,687}]}, {shell,eval_exprs,7,[{file,"shell.erl"},{line,642}]}, {shell,eval_loop,3,[{file,"shell.erl"},{line,627}]}]}} 2> erlang:get_stacktrace(). [{erlang,abs, [[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22, 23,24\|...]], []}, {erl_eval,do_apply,6,[{file,"erl_eval.erl"},{line,674}]}, {erl_eval,expr,5,[{file,"erl_eval.erl"},{line,431}]}, {shell,exprs,7,[{file,"shell.erl"},{line,687}]}, {shell,eval_exprs,7,[{file,"shell.erl"},{line,642}]}, {shell,eval_loop,3,[{file,"shell.erl"},{line,627}]}] 3> We can extend the syntax for clauses in try/catch to optionally bind the stacktrace to a variable. Here is an example using the current syntax: try Expr catch C:E -> Stk = erlang:get_stacktrace(), . . . In the new syntax, it would look like: try Expr catch C:E:Stk -> . . . Only a variable (not a pattern) is allowed in the stacktrace position, to discourage matching of the stacktrace. (Matching would also be expensive, because the raw format of the stacktrace would have to be converted to the cooked form before matching.) Note that: try Expr catch E -> . . . is a shorthand for: try Expr catch throw:E -> . . . If the stacktrace is to be retrieved for a throw, the 'throw:' prefix must be explicitly included: try Expr catch throw:E:Stk -> . . .
2017-11-30	Stop trying to maximize the use of x(0)	Björn Gustavsson
	X register 0 used to be mapped to a hardware register, and therefore faster than the other registers. Because of that, the compiler tried to use x(0) as much as possible as a temporary register. That was changed a few releases ago. X register 0 is now placed in the array of all X registers and has no special speed advantage compared to the other registers. Remove the code in the compiler that attempts to use x(0) as much as possible. As a result, the following type of instruction will be much less frequent: {put_list,Src,{x,0},{x,0}} Instead, the following type of instruction will be more frequent: {put_list,Src,{x,X},{x,X}} (Where X is an arbitrary X register.) Update the runtime system to specialize that kind of put_list instruction.
2017-10-21	Optimize instructions for comparing a register with a literal	Björn Gustavsson
	We can avoid calling eq() from the is_eq_exact_literal/3 and is_ne_exact_literal/3 instructions if the source operand is an immediate (since a literal is either a boxed or a list, never an immediate).
2017-10-09	Slightly speed up try/catch	Björn Gustavsson
	The try_end and try_case instructions are implemented the same way (try_case is translated to try_end by the loader). We can do better than that. We know that try_case will only be executed when an exception has been caught. Therefore, we know that x(0) is the non-value and that x(1) through x(3) need to be shifted down to x(0) through x(2). There is no need to test x(0) before shifting down. try_end does not need the register shifting code at all.
2017-10-01	Move out variables from the head of combined instructions	Björn Gustavsson
	Move out from the head the variables that are only used in the excute phase.
2017-10-01	instrs.tab: Add missing -no_next directives	Björn Gustavsson

2017-09-13	Refactor instructions to support relative jumps	Björn Gustavsson
	Introduce new macros that can be used for relative jumps and use them consistently. Test that everything works by using a non-zero constant JUMP_OFFSET. The loader subtracts JUMP_OFFSET from loaded labels, and all instructions that use 'f' operands add it back.
2017-09-11	Avoid using $Src more than once	Björn Gustavsson
	The C compiler will probably optimize this, but just to be sure...
2017-09-04	Don't allow macros to assign to hard-coded variables	Björn Gustavsson
	It's bad style. Pass the name of the variable as an extra argument to the macro.
2017-09-04	Eliminate unecessary instruction macro i_move_call_only()	Björn Gustavsson

2017-08-31	Add missing -no_next directives	Björn Gustavsson

2017-08-31	Eliminate three arguments for the apply() helper	Björn Gustavsson
	We don't need to pass x(0), x(1), and x(2) because they can already be found in the register array.
2017-08-31	Add annotations for likely/unlikely	Björn Gustavsson
	In a correct Erlang programs, we can expect that: * A GC test instruction (such as test_heap) is more likely not to do the GC. * A BIF is more likely to succeed than to fail. * A BIF is more likely to fail in a guard than in a body. * An apply or fun call is likely to succeed. Annotate conditions accordingly.
2017-08-22	Add missing -no_next for badarg instruction	Björn Gustavsson

2017-08-11	Break out most instructions from beam_emu.c	Björn Gustavsson

2017-08-08	Simplify specifying implementation of instructions	Björn Gustavsson
	Eliminate the need to write pre-processor macros for each instruction. Instead allow the implementation of instruction to be written in C directly in the .tab files. Rewrite all existing macros in this way and remove the %macro directive.