summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* compiler: implement tree-bind special op.Kaz Kylheku2018-03-221-0/+134
| | | | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Wire in tree-bind case via comp-tree-bind method. (compiler comp-tree-bind): New method. (expand-bind-mac-parse-params, expand-bind-mac-params): New functions.
* vm: initialize vd->funvec to nil.Kaz Kylheku2018-03-221-0/+1
| | | | | | | | * vm.c (vm_make_desc): Initialize vd->funvec to nil so that when the object is created, it doesn't have a garbage field. However, there is no risk here that the field will be traversed by the garbage collector, since immediately after allocating the objct, we initialize the fields.
* vm: bugfix: cannot access vec during gc.Kaz Kylheku2018-03-221-6/+8
| | | | | | | | | | | | | | Let's have a field in struct vm_desc that keeps track of the ftab size. The problem is when we do length_vec(vd->funvec) in vm_desc_mark, we are in the middle of garbage collection. The object's tag has a reachable bit which blows up the type check. * vm.c (struct vm_desc): New member, ftsz. (vm_make_desc): Store the length of the function table into ftsz. (vm_desc_mark): Use the stored length as the loop bound; do not access vd->funvec.
* gc bugs: incorrect struct init function assignments.Kaz Kylheku2018-03-221-2/+2
| | | | | | | | | | * struct.c (struct_set_initfun, struct_set_postinitfun): Replace incorrect direct assignments with set macro. This manifested itself as corruption. I ran into a situation in which the postinitfun of a struct type was prematurely reclaimed and the heap object was re-used for something else wreaking havoc on the postinit call when the struct was instantiated.
* compiler: implement return-from and return.Kaz Kylheku2018-03-221-0/+18
| | | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Handle return and return-from special forms. (compiler comp-return-from, compiler comp-return): New methods.
* vm: ret instructions: throw error if block not found.Kaz Kylheku2018-03-221-0/+14
| | | | | | * vm.c (vm_no_block_err): New static function. (vm_retsr, vm_retrs, vm_retrr): Call vm_no_block_err if uw_block_return returns.
* compiler: sys:setq bug: wrong way mov.Kaz Kylheku2018-03-221-1/+1
| | | | | * share/txr/stdlib/compiler.tl (compiler comp-setq): Fix reversed mov instruction arguments in lexical case.
* new special op sys:upenvKaz Kylheku2018-03-212-0/+10
| | | | | | | | | | | | | | | | | | | | | | | This is something that will be useful in compiling some forms. At first I added it to the compiler only, but it seems wise to have it in the interpreter also. (sys:upenv form) causes form to be treated as if it were not in the immediately surrounding lexical environment but its parent. Thus (let ((a 1)) (let ((a 2)) (list a (sys:upenv a)))) yields (2 1). This operator needs no special treatment in the expander; it is expanded as a function call. This is not 100% correct in the face of all conceivable use. For instance given (symacrolet ((a 1)) (let ((a 2)) (sys:upenv a))), we probably want sys:upenv to skip the inner environment at expansion time too so that a is replaced by 1. However, it is not documented for application use, and will never be used in such a situation in the compiler. * eval.c (op_upenv): New static function. (eval_init): Register sys:upenv special operator. * compiler.tl (compiler compile): Implement compiled version of sys:upenv.
* compiler: append-each: bugfix for non-lists.Kaz Kylheku2018-03-211-1/+1
| | | | | | | | | * share/txr/stdlib/compiler.tl: If a non-list sequence like a string or vector is being accumulated, we must have an append call, because the accumulation cell is never extended with new cells. After we replace the cdr of ,accum with a string or vector atom, (last ,accum) just finds ,accum and so then the cdr will be clobbered on the next iteration.
* compiler: let* bugfix: not extending env.Kaz Kylheku2018-03-211-2/+8
| | | | | | | | | | * share/txr/stdlib/compiler.tl (sys:env extend-var*): New method. (compiler comp-let): When doing sequential bindings, we must still add the variable to the environment even if there is no init-form. Oops! Since we're not doing any gensym renaming trick in this case but binding the real symbol directly, we need a binding method that allows duplicates.
* compiler: disallow duplicate symbols in same env.Kaz Kylheku2018-03-211-3/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We want to diagnose (let (a a)) and (lambda (a a)), as well as the analogous situations for function bindings. This is certainly ambiguous: if the code accesses a, which one was intended? The choice is undocumented and so this is a likely programming error; nobody will write code this deliberately, investigate what the implementation actually does, and then rely on that. However, multiple bindings of special variables are allowed. Duplicates are also allowed in sequential binding. Handling duplicates in let* binding creates the following problem: (a a) means that a should be initialized from the previous a. We would like to compile the init form in the environment in which the second a doesn't exist, yet indicate the destination location of the new a. Since we are clobbering one environment as we go by adding bindings to it, this is not possible. The solution is a rename hack: we introduce the new a as a gensym, so that it is invisible. We indicate the gensym's location as the destination register for the init-form code fragment. Then immediately after compiling the init-form, we rename the gensym variable. * share/txr/stdlib/compiler.tl (sys:env extend-var, sys:env extend-fun): Diagnose duplicates. (sys:env rename-var): New methods. * share/txr/stdlib/compiler.tl (compile comp-let): Bind to a gensym instead of the real symbol when binding sequentially. Then rename the symbol.
* compiler: handle sys:each-op special form.Kaz Kylheku2018-03-211-0/+21
| | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Handle sys:each-op with help of expand-each. (expand-each): New function.
* compiler: handle special forms and or.Kaz Kylheku2018-03-201-0/+27
| | | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Handle and and or cases via comp-and-or method. (compiler comp-and-or): New method. This is based on comp-progn.
* expander: bugfix: sys:for-op init forms.Kaz Kylheku2018-03-201-5/+8
| | | | | | | * eval.c (do_expand): The first argument of the sys:for-op special operator isn't "vars" but a sequence of initialization forms. Name the variable appropriately. The neglected expansion of these forms is now performed.
* compiler: handle dohash special formKaz Kylheku2018-03-201-0/+12
| | | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Handle dohash by rewriting the form via expand-dohash, then compiling result. (expand-dohash): New function.
* vm: funcall wrappers need to check arg count.Kaz Kylheku2018-03-201-0/+16
| | | | | | | | | | | | | | There is the possibility that funcall1 through funcall3 will pass the wrong number of arguments to vm_execute_closure, which doesn't do any checks. This is only for the code paths which don't go through generic_funcall, when the function has no optional arguments. * lib.c (funcall1, funcall2, funcall3, funcall4): Check that the closure doesn't require more fixed parameters than we're giving it in the variadic case, or that it doesn't require a different number of fixed parameters we're giving it in the non-variadic case.
* gc: tail recurse on env rather than code.Kaz Kylheku2018-03-201-4/+5
| | | | | | | * gc.c (mark_obj): When marking functions, tail recurse on the environment. That's the pointer more likely to be a gateway to serious depth, rather than the vm description or interpreted function's source.
* vm: allocate display in same object as closure.Kaz Kylheku2018-03-201-14/+9
| | | | | | | | | | * vm.c (struct vm_closure): Redeclare display member as array of 1: old-fashioned version of C struct hack. (vm_make_closure): Make the necessary adjustments to the allocation code. (vm_closure_destroy): Static function removed. (vm_closure_mark): Wire in cobj_destroy_free_op once again, since there is just one object to free.
* compiler: lambda bug: wrong reg in defaulting.Kaz Kylheku2018-03-201-1/+1
| | | | | | | | | | * share/txr/stdlib/compiler.tl (compiler comp-lambda): To determine whether the argument is missing, we must test it for equivalence to the : symbol. What we're testing here is the wrong thing: the register which will hold the output of the default initform. That fragment would even be executed yet at this spot in the code, never mind being the wrong value to test.
* compiler: improve progn.Kaz Kylheku2018-03-201-6/+15
| | | | | | | | | | * share/txr/stdlib/compiler.tl (compiler comp-progn): If the suggested output register is a variable, progn should not use it for the discarded values of the leading forms, only for the last form. Writes to the variable space of the display are costly because in closures, the underlying vector of a given level has to be passed to gc_mutated. We use a new temp register for the discarded forms.
* compiler: bugfix: arg passing in progn dispatch.Kaz Kylheku2018-03-201-1/+1
| | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Fix nonsensical fourth argument expression in comp-progn call.
* lib: new ldiff function.Kaz Kylheku2018-03-203-2/+61
| | | | | | | | | | * eval.c (eval_init): Use the old ldiff function under compatibility with 190 or lower. * lib.c (ldiff): Rewritten. (ldiff_old): New function, copy of previous version of ldiff. * lib.h (ldiff_old): Declared.
* compiler: let bugfix: wrong output register designated.Kaz Kylheku2018-03-201-1/+1
| | | | | | * share/txr/stdlib/compiler.tl (compiler comp-let): The returned fragment must specify boreg not oreg, because the output is either in oreg or bfrag.oreg.
* compiler: factor recurring conditional mov.Kaz Kylheku2018-03-201-29/+19
| | | | | | | | | | | | | Anyway, this kind of thing should really be (and will be) taken care of peephole optimization. But for now, it's nice to get somewhat better code without doing too much work. * share/txr/stdlib/compiler.tl (compiler comp-setq, compiler comp-cond, compiler cond-if, compiler cond-let, compiler comp-lambda, compiler comp-for): Replace occurrences of (if (nequal reg1 reg2) ^((mov ,reg ,reg2))) with call to maybe-mov. (maybe-mov): New function.
* compiler: handle if special formKaz Kylheku2018-03-201-0/+52
| | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Handle if case via comp-if method. (compiler comp-if): New method.
* compiler: handle cond special form.Kaz Kylheku2018-03-201-0/+54
| | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Handle cond case via comp-cond method. (compiler comp-cond): New method.
* vm: support deferred resolution of ftab bindings.Kaz Kylheku2018-03-191-4/+20
| | | | | | | | | | | | | | | | This allows a virtual machine's funvec to refer to functions that are not yet defined. We resolve all the ones that are defined when the virtual machine description is constructed. The others are resolved at vm execution time when accessed. * vm.c (vm_make_desc): When filling ftab with resolved function bindings, check that lookup_fun has returned nil; in that case, initialize the binding location to zero, instead of blowing up calling cdr_l(nil). (vm_ftab): New static function. Implements the null check and late resolution. (vm_gcall, vm_gapply): Acces ftab through vm_ftab instead of open-coded expression.
* compiler: dump warnings from form expansion.Kaz Kylheku2018-03-191-1/+3
| | | | | | | * share/txr/stdlib/compiler.tl (usr:compile-toplevel): Call release-deferred-warnings to dump accumulated warnings. Otherwise when we are working in the listener, the pent-up warnings get dumped in relation to the next input line.
* vm: bug: vm-desc created with incorrect display depth.Kaz Kylheku2018-03-192-3/+5
| | | | | | | | | | | | | | | | | | * share/txr/stdlib/compiler.tl (sys:env :postinit): The call to register the environment with the compiler must be outside of the unless form. Otherwise it never takes place, and so the compiler doesn't find the maximum number of environment levels, keeping the value at 2. The executing vm then accesses out of bounds memory when setting up display frames. (usr:compile-toplevel): Give the root environment the compiler. Not strictly necessary since we are constent in doing this elsewhere, so we are not relying on inheritance of the compiler from parent environment to child. * vm.c (vm_make_closure): assert added for the environment levels of the closure not exceeding the display depth given in the machine description. This was added during debugging and going off; I'm keeping it.
* vm: variadic arg closures bug 3/3.Kaz Kylheku2018-03-191-1/+1
| | | | | | | | | * share/txr/stdlib/asm.tl (op-close asm): Fix argument count check being incorrect in the variadic argument case, causing the assembler to blow up on a correct (close ...) instruction. The integer indicating the number of fixed args is one *less* than the number of arguments given to the instruction not one *more*.
* vm: variadic arg closures bug 2/3.Kaz Kylheku2018-03-191-6/+5
| | | | | | | | | | | | | | * vm.c (vm_execute_closure): Remove bogus nargs local variable, and replace its uses by fixparam which is the correct one to use for getting the list of trailing arguments using args_get_rest and for the loop which extracts the fixed arguments. The "if (variadic)" statement near the end specifically depends on the variadic destination register not having been extracted yet from the instruction space, so the argument count used in the previousl loop cannot include that register. This bug was causing that statement to extract a zero, thus register t0, which then blew up in the vm execution as an attempt to modify t0.
* vm: variadic arg closures bug 1/3.Kaz Kylheku2018-03-191-1/+1
| | | | | | | * lib.c (func_vm): fix incorrect assignment of int boolean to one-bit-wide bitfield. The argument isn't zero or one; in fact the value 256 gets passed down representing true. This gets truncated to one bit which is zero.
* compiler: bug: inappropriate clobber in block.Kaz Kylheku2018-03-191-2/+2
| | | | | | | | | | | * share/txr/stdlib/compiler.tl (compiler comp-block): Don't use bfrag.oreg as the output register of block. Firstly, bfrag.oreg might be (t 0) which is read-only. Worse, bfrag.oreg could just be a variable from which the block obtains its return value; we mustn't clobber that location with block's dynamic return value. Not only mustn't but possibly we cannot: the location could be out of scope!
* compiler: bug: nil entered into data tab by block.Kaz Kylheku2018-03-191-1/+1
| | | | | | * share/txr/stdlib/compiler.tl (compiler comp-block): If the block is anonymous, just refer to (t 0) as the name; don't intern nil via get-dreg.
* compiler: use VM's function table for global calls.Kaz Kylheku2018-03-191-4/+23
| | | | | | | | | | | | | * share/txr/stdlib/compiler.tl (compiler): New slots fidx-cntr, fidx and ftab. (compiler get-fidx, compiler get-funvec): New methods. (compiler comp-call-impl): New method. (compiler comp-call): Look up the symbol to determine whether the function lexical or global. Call comp-call-impl appropriately to generate a gcall or call. (usr:compile-toplevel): Obtain the funvec from the compiler and pass to vm-make-desc.
* asm: return original object from disassemble function.Kaz Kylheku2018-03-191-1/+2
| | | | | | | | * share/txr/stdlib/asm.tl (usr:disassemble): Return the object that was disassembled, rather than nil. This is useful in the listener: we can compile and disassemble something in one step, then have access to the compiled object via a listener variable.
* compiler: bug: lambda body uses inappropriate output reg.Kaz Kylheku2018-03-191-3/+5
| | | | | | | | | | * share/txr/stdlib/compiler.tl (compiler comp-lambda): The incoming oreg, which indicates where the surrounding context would like to put the closure, cannot be used for the output location of the lambda body. Because then when the closure is called, its return value will overwrite the location where the closure was placed. We must allocate a a new temporary register for this, and be sure to free it.
* compiler: bugfix: misused assoc in function lookup.Kaz Kylheku2018-03-191-1/+1
| | | | | * share/txr/stdlib/compiler.tl (sys:env lookup-fun): Fix swapped args in assoc call.
* vm: function table for faster calls.Kaz Kylheku2018-03-185-34/+176
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The vm now supports gcall and gapply opcodes which index numerically (using an immediate integer field in the instruction word) into a table of pre-resolved global function bindings. * share/txr/stdlib/asm.tl (op-gcall, op-gapply): New opcodes. (disassemble-c-d): Take the function vector as an argument and dump it too. (usr:disassemble): Extract function vector from VM description and pass it to disassemble-c-d. * share/txr/stdlib/compiler.tl (usr:compile-toplevel): Pass empty function symbol vector to vm-make-desc; it's now a required argument. * vm.c (struct vm_desc): New members funvec and ftab. (struct vm_ftent): New struct type. (vm_make_desc): New argument, funvec. Store funvec in the descriptor. Allocate a table of vm_ftent structures equal in number of elements to the function table, and populate it with resolved bindings. (vm_desc_funvec, vm_desc_destroy): New static functions. (vm_desc_mark): Mark the captured bindings in vd->ftab. (vm_gcall, vm_gapply): New static functions. (vm_execute): Handle GCALL and GAPPLY opcodes. (vm_desc_ops): Wire vm_desc_destroy in place of cobj_destroy_free_op. (vm_init): Add argument to vm-make-desc intrinsic. Register vm-desc-funvec intrinsic. * vm.h (vm_make_desc): Declaration updated. * vmop.h: Regenerated
* vm: free display memory when closure reclaimed.Kaz Kylheku2018-03-181-1/+8
| | | | | | * vm.c (vm_closure_destroy): New static function. (vm_closure_ops): Use vm_closure_destroy rather than generic cobj_destroy_free_op.
* compiler: change in output register protocol.Kaz Kylheku2018-03-181-108/+107
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | With this change we get better code, as well fewer situations in which tregs are manually released. There are some subtleties. Previously, a compiled fragment of code would dictate the identity of the location into which it has placed output; then the higher level compile had to work with that location. The new protocol is that the caller specifies a suggested output register to the recursive compile call. If the recursive compile requires a temporary register for the output, it uses the suggested one. If it doesn't require a temporary register, it specifies its own output (for instance a (v x y) entry in the display). The caller has to deal with the output not being in the suggested register. The binding constructs lambda and let also have to deal with the possibility that the body specifies an output which is going out of scope, rather than the suggested register, and in that case must generate a move instruction to transfer from there to the suggested register before the (end ...). * share/txr/stdlib/compiler.tl (sys:env out-of-scope): New method. (compiler compile): Restructured handling of atom forms. Function takes output register argument and passes it down to the special form compile methods. (compiler comp-atom): Bugfix: this must handle nil by returning (t 0) rather than wastefully interning nil into the data table. (quote nil) triggered this. We no longer need to allocate a temporary register in this function; we just use oreg when we need it. (compiler comp-var): Don't alloc temporary, use oreg. (compiler comp-setq): We don't have to free the temporary register from compiling the value expression. When the target is a lexical variable, we pass its location to the sub-compile, so the result is placed in the variable directly. Thus we don't have to generate an additional mov instruction any more, except if that compile wasn't able to use the suggested location. (compiler comp-block): Pass oreg to compile of block bodies. Then use whatever register that block specifies as the block output. (compiler comp-let): Binding of variables is streamlined with oreg in a manner similar to assignment in comp-setq. The body is compiled with the suggested oreg as its output. Because it may put the output in a location that is going out of scope, we have to do the scope check and insert a mov. (compiler comp-lambda): Similar changes as in comp-let. (compiler comp-progn): Just use the incoming oreg as the destination for every form. Use the actual output register of the last frag as the output register, unless there were no frags in which case the nil register is specified as the output. (compiler comp-prog1): Allocate a register for the ignored values of the second and subsequent forms. Use oreg for compiling the first form. (compiler comp-quasi): Take oreg param and pass down. (compiler comp-call): Now allocates a batch of suggested registers for each of the argument compiles, and passes each down to its corresponding argument compile. These registers are freed, and the actual output are used in generating the call. The output of the call goes into oreg; no need to allocate. (compiler comp-for): Mostly the obvious changes here. (usr:compile-toplevel): Here we need to allocate an output register for the top-level call into the compiler.
* compiler: use mac-param-bindKaz Kylheku2018-03-171-15/+15
| | | | | | | | | | | | It's better to use mac-param-bind than tree-bind because it provides diagnostics related to the form being destructured. * share/txr/stdlib/compiler.tl (compiler compile): Pass the whole form rather than (cdr form) to a number of special form handlers. (compiler comp-seq, compiler comp-block, compiler comp-let, compiler comp-lambda, compiler comp-for): Destructure arguments with mac-param-bind.
* listener: fix poor regex handling in balance check.Kaz Kylheku2018-03-171-1/+33
| | | | | | | * parser.c (is_balanced_line): Introduce the ST_RGXC state to which we switch when we encounter a regex character class. Also introduce ST_RGXE for regex subexpressions. In these states, do not recognize / as the regex terminator.
* compiler: replace invalid compound form message.Kaz Kylheku2018-03-171-1/+1
| | | | | | * share/txr/stdlib/compiler.tl (compiler compile): The message will show (car form) followed by a colon, so the wording should refer to that object rather than the entire form.
* compile-toplevel: expand with warnings.Kaz Kylheku2018-03-172-1/+2
| | | | | | | | * eval.c (eval_init): Expose raw expand function as sys:expand*, since sys:expand squelches warnings. * share/txr/stdlib/compiler.tl (usr:compile-toplevel): Use expand* instead of expand.
* compiler: compile string quasiliterals.Kaz Kylheku2018-03-171-0/+84
| | | | | | | | * share/txr/stdlib/compiler.tl (compiler compile): Add sys:quasi case, dispatching new comp-quasi method. (compiler comp-quasi): New method. (expand-quasi-mods, expand-quasi-args, expand-quasi): New functions.
* vm: handle FVM function type thorughout run-time.Kaz Kylheku2018-03-162-2/+51
| | | | | | | | | | | | | | | | * gc.c (mark_obj): Recognize FVM functions and mark their vm_desc. * lib.c (equal): Handle equality for FVM. If the environment pointers are equal, consider the functions equal. (funcall, funcall1, funcall2, funcall3, funcall4): Recognize and call FVM functions. However, there is a lack of robustness here that needs to be addressed: vm_execute_closure doesn't check whether there are too many or not enough arguments. Interpreted functions have a run-time check inside bind_args. (obj_print_impl): Don't print VM functions as #<intrinsic fun...> but rather #<vm fun>.
* quasi: factor out repeated code.Kaz Kylheku2018-03-161-19/+24
| | | | | | | | * eval.c (fmt_tostring, fmt_cat): New static functions. (do_format_field): Replace code block with call to fmt_cat. (fmt_simple, fmt_flex): Insert needed call to fmt_tostring. (subst_vars): Replace blocks of code with calls to fmt_tostring and fmt_cat.
* quasilit: expose access to field formatting.Kaz Kylheku2018-03-161-0/+35
| | | | | | | | | | | | | | | | | | | | | | | | | This is the second round of changes in perparation for compiling the string quasiliterals special form. Having split format_fields into a lower level do_format_field and format_field, we now provide two functions which allow Lisp code to call do_format_field. The sys:fmt-simple function has fixed arguments for the field width/alignment, indexing/slicing, separator and plist. Any of them can be defaulted with nil. This function will be useful when the quasiliteral specifies modifiers that are simple literals or constant expressions. The sys:fmt-flex function takes a variable number of arguments, after the object and the plist. This function has to scan through the arguments and classify them by type: a string is a separator, an integer is a width and left/right alignment and so on. The compiler will use sys:flex when format field modifiers are present whose arguments contain expressions that get evaluated. * eval.c (fmt_simple, fmt_flex): New static functions. (eval_init): sys:fmt-simple and sys:fmt-flex registered.
* compiler: closure bug: (dframe ...) without (end ...).Kaz Kylheku2018-03-161-0/+2
| | | | | | * share/txr/stdlib/compiler.tl (compiler comp-lambda): When we have specials we must generate an extra (end ...), to terminate the (dframe ...) that we inserted.
generated by cgit v1.2.3 (git 2.25.1) at 2025-03-25 22:19:56 +0000