Commit graph

673 commits

Author SHA1 Message Date
Wolfgang Bumiller
2c59385633 fixing double-free in initialzied string arrays, using them in the testcase 2013-06-12 15:53:07 +02:00
Wolfgang Bumiller
8ce331b563 allow array size to be inferred from the initializer 2013-06-12 15:47:11 +02:00
Wolfgang Bumiller
633d2ba8a4 make the ast use the initlist 2013-06-12 14:36:59 +02:00
Wolfgang Bumiller
7a36a8bdd8 parsing array initializers 2013-06-12 14:32:12 +02:00
Dale Weiler
11179a2a71 Major header reworking, this respects the namespaces properly. Makes object dependency more obvious, allows for better make caches, and prevents misuse of library features, i.e use con_* instead of printf. 2013-06-04 02:47:07 +00:00
Wolfgang Bumiller
b2348e1de0 remove the ast tree earlier - saves over 3% peak memory usage 2013-06-02 12:35:53 +02:00
Wolfgang Bumiller
d85e86141c fix a leak on a parse-error 2013-06-01 23:36:05 +02:00
Dale Weiler
062180e9a8 Fix some possible memleaks. 2013-06-01 20:18:53 +00:00
Wolfgang Bumiller
9f2b9e1b46 Merge branch 'master' into cooking 2013-06-01 20:12:27 +02:00
Wolfgang Bumiller
ff63e5bd73 fix the INCLUDE_DEF flag getting applied regardless of dotranslate... 2013-06-01 20:12:14 +02:00
Wolfgang Bumiller
1dce501b70 Merge branch 'master' into cooking 2013-06-01 18:16:56 +02:00
Wolfgang Bumiller
ed585f8c04 fix: dotranslate now sets AST_FLAG_INCLUDE_DEF to not get removed by -Ostrip-constant-names 2013-06-01 18:16:50 +02:00
Wolfgang Bumiller
5aba29006b return assignment factorial test 2013-05-29 16:58:18 +02:00
Wolfgang Bumiller
9167de1631 fixing that comment... 2013-05-29 16:56:39 +02:00
Wolfgang Bumiller
afdc0c9dc8 assignable return value now lives in ast_function, as globals can get overwritten randomly; removed parser_find_retval; updated parse_return 2013-05-29 16:51:59 +02:00
Dale Weiler
6d6a2efada Experimental support for implicit return assignments. This closes #107. To enable return assignment support use -freturn-assignments. This allows you to do the following in QC code. T name() { return = expr_eval_T; return; /* returns expr_eval_T */ }. It allows for concise code and to rid locals in functions. It also saves a tremendous amount of space since only types of certian globals need to be allocated for returns. 2013-05-29 11:13:42 +00:00
Dale Weiler
2923b718e1 Merge branch 'union-replacement' into cooking
Conflicts:
	ast.c
2013-05-29 03:43:49 +00:00
Dale Weiler
d45956f55e Fix that memory leak that I spent hours trying to find months ago and didn't succeed at. 2013-05-29 03:35:51 +00:00
Dale Weiler
f892b32335 Major export cleanup. Anything that was exported but wasn't used outside where it was implemented has been turned into static, and their exports have been removed. This actually makes the compiler compile slightly faster. 2013-05-29 03:29:04 +00:00
Dale Weiler
e02ebfe486 Result of shift operation on signed integers is bad. 2013-05-29 02:24:12 +00:00
Wolfgang Bumiller
d70b571769 killing the union types 2013-05-11 22:34:01 +02:00
Wolfgang Bumiller
a0fa90ddd5 less confusing 2013-05-07 20:35:20 +02:00
Wolfgang Bumiller
802005f571 be more strict here 2013-05-07 20:13:19 +02:00
Wolfgang Bumiller
c5225b2fa1 fixing a few leaks - code_write doesn't delete the code object anymore, code_cleanup has to be called 2013-05-07 19:56:41 +02:00
Wolfgang Bumiller
8c8ae71d65 changing a few ast_delete to ast_unref in places where the to-be-deleted maybe coming from some other place 2013-05-07 19:39:10 +02:00
Wolfgang Bumiller
9f8bee4bf1 ast_value now has an initializer list array, still unused but it's there for later 2013-04-26 10:47:55 +02:00
Wolfgang Bumiller
b3e9ef3ad9 an = before an { is also always required when the declared variable is not a function; adding because of planned array initializers 2013-04-26 10:31:38 +02:00
Wolfgang Bumiller
d4f8e4a0dd solve it, not hide it 2013-04-25 19:18:50 +02:00
Wolfgang Bumiller
35692c0b57 Revert "Just fucking initialize everything"
This reverts commit aed2b1031c.
2013-04-25 19:18:11 +02:00
Dale Weiler
aed2b1031c Just fucking initialize everything 2013-04-25 17:11:15 +00:00
Dale Weiler
68ca2c4962 Pushing the -fexpressions-for-builtins stuff and the modff for catching fractional-part builtin numbers. 2013-04-25 17:08:02 +00:00
Wolfgang Bumiller
ce73074d51 this allows builtin numbers to be constant expressions, not just constants 2013-04-25 18:47:54 +02:00
Dale Weiler
9fee84f250 less globals 2013-04-25 12:08:13 +00:00
Dale Weiler
785ab7c072 No more globals for codegen 2013-04-25 09:35:30 +00:00
Wolfgang Bumiller
3a4aba0b31 Grrrr 2013-04-24 17:47:42 +02:00
Wolfgang Bumiller
66305c676a Allow string concatenation in the parser as well, for now only for immediate strings and __FUNC__, maybe later -fpermissive or somethign could allow it on const-delcared string variables 2013-04-24 17:27:04 +02:00
Wolfgang Bumiller
dcd65f18f8 recognize the simple case of __FUNC__ in the parser 2013-04-24 11:21:10 +02:00
Dale Weiler
8bee9a6819 fix the intrinsic fail 2013-04-24 09:18:24 +00:00
Dale Weiler
dc8523c650 Merge branch 'master' into cooking 2013-04-24 01:46:53 +00:00
Dale Weiler
489ad486bc Get it compiling in visual studio again. 2013-04-24 01:43:53 +00:00
Dale Weiler
10738da2fb Make it compile as C++ code too 2013-04-23 22:50:47 +00:00
Wolfgang Bumiller
f6d554874b initialize the vector member array so that 'cleanup' won't delete uninitialized pointers 2013-04-23 16:54:05 +02:00
Dale Weiler
160e7cf7ee Remove trailing whitespace 2013-04-21 10:24:55 +00:00
Wolfgang Bumiller
46752af74b parser_const_string now uses hashtables; hashtables may want to dup an empty string without it becoming NULL - also replacing ir_strdup with the new util_ one 2013-04-17 18:23:30 +02:00
Dale Weiler
1f0a1b8140 Perliminary parser reworking, required for the proposed multithreading. 2013-04-17 00:52:52 +00:00
Wolfgang Bumiller
1f667c4b35 ast_expression_delete: delete varparam if it is set - fixes another leak 2013-04-15 21:05:59 +02:00
Dale Weiler
ef528d6710 Added tracing to strdup for allocations, fixed some memleaks, this isn't pretty (trying to track down some weird memory leak issues) 2013-04-14 23:51:16 +00:00
Dale Weiler
cbeac3e5f2 Implemented smart intrinsic / builtin system. When you use trivial math functions like "pow", if they don't exist as a builtin, the compiler will implement its own versions, likewise, if a compiler builtin depends on a function that exists, it will use it, likewise if it doesn't it will implement it. If you explicitally use __builtin_ (prefixed) versions, the compiler again will select the best option it can, be it a combination of both builtins and compiler builtins, all compiler builtins, OR, all builtins (most performant). 2013-03-09 08:53:39 +00:00
Dale Weiler
d8d78f61a7 Implemented "reverse" enum attribute, e.g enum : reverse { A, B, C, D } -> A,B,C,D = 3,2,1,0 2013-03-09 05:07:14 +00:00
Dale Weiler
662612d66f Better diagnostic when "floor" cannot be found. 2013-03-08 09:23:35 +00:00
Dale Weiler
f19adcd1b3 Implemented __builtin_mod, and % operator. Added floor builtin to the standalone executor. Mod works so long as the compiler can find a suitable definition of "float floor(float)", otherwise it prints a diagnostic and gives up (original id1 Quake had floor so we can depend on it). 2013-03-08 09:17:54 +00:00
Dale Weiler
b971ddec3a Implemented concept of enumeration attributes (can be further extended, but currently only "flag" is implemented as an attribute). An enumeration with a flag attribute will act as a "flagged enumeration", one that automatically handles exponentiation of the constants defined inside it, i.e enum : flag { A, B, C }, A,B,C will equal 2, 4, 8. 2013-03-08 08:01:45 +00:00
Dale Weiler
479f3b9343 Implemented __builtin_exp 2013-03-08 03:46:25 +00:00
Dale Weiler
12340a6bd0 Actually generate an ast_call for __builtin_pow for the ** operator, otherwise the operator yeilds a ast_function, making "a ** b" not work, but since it's a function, allows **(a, b). Also added tests for exponentiation operator. 2013-03-07 23:05:40 +00:00
Dale Weiler
d3c7b6fb39 Implement exponentiation operator ** as well as __builtin_pow (used for exponentiation operator). Use of exponentiation operator with constants results in const folded (precomputed at compile time exponentiation), otherwise runtime exponentiation with some clever loops (slow!). 2013-03-07 21:31:19 +00:00
Dale Weiler
eb0e3c2898 Make it compile as C++ code again. 2013-02-06 10:36:27 +00:00
Wolfgang Bumiller
9bc774dcd3 find aliases in parser_find_global instead of parser_sya_operand 2013-02-06 10:55:06 +01:00
Dale Weiler
cb1a6a6a6b Store the ast nodes for aliases 2013-02-06 09:44:23 +00:00
Dale Weiler
1e60bafcdc Aliases to aliases is a bad idea, revert. Added better tests to show unused aliases 2013-02-06 09:28:42 +00:00
Dale Weiler
21e890602d Fix alias bug. Implemented support for aliases of vectors (x, y, z components). Also made aliases corrector resident (e.g alias to vector foo, named bop, indexing bol_x [instead of bop_x] will result in a correction suggestion of bop_x now). 2013-02-06 09:09:47 +00:00
Dale Weiler
6fc141733f Support for aliases to aliases too. 2013-02-06 08:43:11 +00:00
Dale Weiler
060bc0be10 Some sanity checks for aliases (undeclared variable checks, and incompatible types in alias checks.) 2013-02-06 05:51:05 +00:00
Dale Weiler
3dc8e9bb9c add todo 2013-02-06 05:38:34 +00:00
Dale Weiler
e464cc1ea6 Implemented generalized attribute aliases. You can now alias functions, locals, or globals with [[alias("old_name")]] type new_name; 2013-02-06 05:30:09 +00:00
Dale Weiler
52ffc6db10 Merge branch 'cooking' of github.com:graphitemaster/gmqcc into cooking 2013-02-05 17:16:09 +00:00
Dale Weiler
b3e87c3280 ast referencing 2013-02-05 17:14:56 +00:00
Wolfgang Bumiller
4b5afe3456 ast_array_index_codegen: never regenerate code 2013-02-05 17:54:14 +01:00
Dale Weiler
e2c424d607 Merge branch 'cooking' of github.com:graphitemaster/gmqcc into cooking 2013-02-05 16:35:12 +00:00
Dale Weiler
8e87082e1d Merge branch 'cooking' of github.com:graphitemaster/gmqcc into cooking
Conflicts:
	parser.c
2013-02-05 16:34:55 +00:00
Wolfgang Bumiller
72106b0982 removing some left-over alloc checks 2013-02-05 17:34:40 +01:00
Dale Weiler
fed0f8b6b4 Work in progress <=> operator. 2013-02-05 16:34:03 +00:00
Wolfgang Bumiller
f860759758 fixing that comment 2013-02-05 17:24:02 +01:00
Wolfgang Bumiller
15141897ae insert an error when an operator doesn't get found - note the comment... 2013-02-05 17:17:38 +01:00
Dale Weiler
31ce4c2afa Better handling of multiple body functions. 2013-02-05 03:28:20 +00:00
Wolfgang Bumiller
e662efae61 fix: rotate_entfield_array_index_nodes deleting the old array caused expressions to be deleted recursively 2013-02-03 23:38:06 +01:00
Dale Weiler
319783e873 Better printing of token errors. 2013-02-02 01:03:39 +00:00
Dale Weiler
ccdfd4c65a oops 2013-02-01 23:12:48 +00:00
Dale Weiler
a3559548a8 Fix ~ unary operator (can now const-fold). Also things like b &= ~1 work now. We also cache ast_value of float (-1) for -1-x (which is how ~ is implemented). 2013-02-01 23:10:23 +00:00
Dale Weiler
03b933bc7a Work in progress ~ operator implemented as -1-x. 2013-02-01 12:56:01 +00:00
Dale Weiler
b9f46f4ef7 Not an assembler instruction, but virtual instruction used by the IR. 2013-01-30 06:31:24 +00:00
Dale Weiler
9f5f2b6d74 Fix pragma parsing that I broke. 2013-01-30 05:51:28 +00:00
Dale Weiler
e12d492d67 Removing #error/#warning/#message from parser, and making it part of preprocessor as directives again. 2013-01-30 05:49:08 +00:00
Dale Weiler
36c5722273 Rename 2013-01-30 05:35:07 +00:00
Dale Weiler
d201cfe6b4 Work in progress options cleanup. 2013-01-30 05:24:30 +00:00
Dale Weiler
f3c5c40104 Fix splint warning 2013-01-30 01:38:37 +00:00
Dale Weiler
3d35804a7d Removed #warning and #error from ftepp, and made it part of the parser routine. These are "directives" that should exist regardless if -fftepp is enabled, like #pragma. Implemented #warning, #message and #error as directives in parser instead. 2013-01-30 01:24:58 +00:00
Wolfgang Bumiller
edfe340d09 Fix a warning 2013-01-26 14:43:49 +01:00
Wolfgang Bumiller
b39b61e514 store the actual parsing context before parsing into an expression and use that for the effectless-statement warning 2013-01-25 23:37:06 +01:00
Wolfgang Bumiller
9afe41471b error when using a type not usable as boolean in a condition, ie an if statement 2013-01-25 19:19:23 +01:00
Wolfgang Bumiller
dca9dd56d1 adding missing wantop flag and some related parentheses code 2013-01-18 15:22:03 +01:00
Wolfgang Bumiller
ba207cc04c Major expression parsing code refactoring, getting rid of lots of duplicated code and in the process also fixing a bug. Still thie will require more testing. xonotic code works with it though 2013-01-18 14:50:37 +01:00
Wolfgang Bumiller
1410840ef6 Make the varargs counter more stable, it'll now work with a function pointer with a different number of fixed params 2013-01-17 10:30:32 +01:00
Wolfgang Bumiller
00a9d2a9e3 fixing a possible NULL deref 2013-01-16 20:32:37 +01:00
Wolfgang Bumiller
e311dffcb3 fixing a warning 2013-01-16 16:16:52 +01:00
Wolfgang Bumiller
0ba8455f6a initialize max_param_count to 1 so vararg functions don't try to create an accessor for an array of 0 elements... 2013-01-16 09:37:32 +01:00
Wolfgang Bumiller
e695a5919d reduce the hashtable size a bit 2013-01-15 20:35:30 +01:00
Wolfgang Bumiller
73afc98699 fix: on error it was possible that leaveblock() is called after the function's already been deleted 2013-01-15 14:28:15 +01:00
Wolfgang Bumiller
150c6e21e7 fix a goto to a wrong error-cleanup part 2013-01-15 14:23:17 +01:00
Wolfgang Bumiller
c7ca5683b9 fix a possible leak 2013-01-15 14:20:08 +01:00
Wolfgang Bumiller
0f190026e8 Error on an empty paren expression 2013-01-15 14:18:03 +01:00
Wolfgang Bumiller
ee7d45f492 add a hint about -fvariadic-args to -Wvariadic-function 2013-01-15 11:35:18 +01:00
Wolfgang Bumiller
b208c1a49f Fix a crash... if (!x) delete(x) yeah right... 2013-01-15 10:50:14 +01:00
Wolfgang Bumiller
ab8cc64dfd constant-fold the 1/N division from a_vector/N 2013-01-13 20:53:07 +01:00
Wolfgang Bumiller
3d62cb37f1 allow vec/float 2013-01-13 20:49:16 +01:00
Wolfgang Bumiller
12ffd83d4f parsewarning->compile_warning 2013-01-13 17:32:54 +01:00
Wolfgang Bumiller
8d4e395092 Replacing lots of parseerror calls with compile_error calls and a more correct context information 2013-01-13 17:31:31 +01:00
Wolfgang Bumiller
3652a122ed Removed -fenhanced-diagnostics - it's now --correct, which makes sense since it doesn't affect the compilation process itself 2013-01-12 17:10:07 +01:00
Wolfgang Bumiller
4df6d1e028 guarding vararg code by -fvariadic-args 2013-01-12 16:28:04 +01:00
Wolfgang Bumiller
7dd31ccf77 entity setter array subscript needs to get a correct field type 2013-01-12 15:49:31 +01:00
Wolfgang Bumiller
a01388ea7d Copying arg counter into the local 2013-01-12 15:06:19 +01:00
Wolfgang Bumiller
57c0a09800 Creating the argument counter if requested 2013-01-12 14:59:30 +01:00
Wolfgang Bumiller
e149551744 Added '...(idx,type)' to access varargs 2013-01-12 14:53:48 +01:00
Wolfgang Bumiller
1861660585 builtins are generally not declared const, so don't expect CV_CONST, otherwise we write out va_count for builtin calls as well... 2013-01-12 13:52:42 +01:00
Wolfgang Bumiller
c10d127fd7 While at it: moving generation of reserved:version to just before the functions so no global/immedate is generated _after_ it - at least from the parser's point of view 2013-01-12 13:49:06 +01:00
Wolfgang Bumiller
d78997599a create store to reserved:va_count; fix: don't build vararg accessor ast tree after generating immediates as they create new ones 2013-01-12 13:47:56 +01:00
Wolfgang Bumiller
6dfdf69a8e vararg accessor generation 2013-01-12 13:01:20 +01:00
Wolfgang Bumiller
c69ba2c734 parsing of vararg-counter 2013-01-12 11:29:03 +01:00
Wolfgang Bumiller
ca947d782c Declaration of type-restricted varargs 2013-01-12 11:03:17 +01:00
Wolfgang Bumiller
2f9db8972e keep track of the highest parameter count of calls for later for varargs 2013-01-12 10:58:01 +01:00
Wolfgang Bumiller
3429e7d0cb 'static' shouldn't be accepted in the global scope for now 2013-01-12 10:52:18 +01:00
Wolfgang Bumiller
3436fa7d89 added the --add-info compile switch: it adds a global const string named version to the 'reserved' namespace, ie ends up named 'reserved:version', containing the --version string 2013-01-11 19:18:51 +01:00
Wolfgang Bumiller
88cfa53dc2 replacing an error message in the enum parsing 2013-01-10 22:08:49 +01:00
Wolfgang Bumiller
9c2bc7f381 Allow accessing vector components via the . operator, including components
of expressions, like: (3 * a_vector).y.
2013-01-10 21:03:35 +01:00
Wolfgang Bumiller
aa8b9eb1ad parse_expression_leave: also end at a closing } - enum: check for } and , after an expression again 2013-01-10 18:52:43 +01:00
Wolfgang Bumiller
a468d0b478 simple 'enum' implementation 2013-01-10 18:38:57 +01:00
Wolfgang Bumiller
81cc4194dc fixed a bug which allowed some statements to end in tokens other than semicolons... (ie closing parens) 2013-01-10 15:23:04 +01:00
Wolfgang Bumiller
428453a132 for compile-time constants << and >> are now available 2013-01-10 15:12:19 +01:00
Wolfgang Bumiller
10c7f4f838 don't call correct_edit anew for each scope 2013-01-09 21:16:22 +01:00
Wolfgang Bumiller
6b0a522bb9 Factor out the adding of globals in various places into parser_addglobal; Add globals to the correct_variables lists 2013-01-09 20:42:33 +01:00
Wolfgang Bumiller
be6942d972 fixing a crash caused by the correction: setting correct=NULL after freeing so the later if(correct) check properly fails 2013-01-09 19:12:27 +01:00
Wolfgang Bumiller
57c3d48711 error when trying to break/continue outside of loops/switches instead of segfaulting at codegen 2013-01-07 15:19:53 +01:00
Wolfgang Bumiller
f4d1ef4740 Fix some possible NULL access in an error case 2013-01-07 00:22:16 +01:00
Wolfgang Bumiller
8bc9d4b427 Committing an evil allocator and a trie to speed up the correction stuff 2013-01-05 19:09:36 +01:00
Dale Weiler
793547a132 Implemented -f[no]enhanced-diagnostics, to enable/disable the usage of enhanced diagnostics. 2013-01-04 12:07:42 +00:00
Dale Weiler
5377835f1e Remove constant string literal for correction check, use parse_tokval of the current parser state instead. Accidently left it behind from debugging. 2013-01-04 11:58:27 +00:00
Dale Weiler
2d96b2a3ec Merge branch 'correct' 2013-01-04 11:56:07 +00:00
Dale Weiler
36d02d010e Got rid of all the memleaks. We can now merge with master. 2013-01-04 11:53:40 +00:00
Dale Weiler
18b9473cf8 Itegration of corrector. Seems to be some leaks in the score keeping for the probability system. 2013-01-04 11:44:25 +00:00
Wolfgang Bumiller
cf293bc669 handling the ast-destruction in computed goto 2013-01-03 13:21:11 +01:00
Dale Weiler
7d14fdf530 Merge branch 'master' of github.com:graphitemaster/gmqcc 2013-01-03 12:05:02 +00:00
Dale Weiler
9bda8f61f4 Add specialized diagnostics for when predefined macros are used and ftepp predefined macros are not enabled. 2013-01-03 12:04:32 +00:00
Wolfgang Bumiller
614fcb12f5 check for TYPE_NOEXPR in general when applying an operator 2013-01-03 12:56:26 +01:00
Wolfgang Bumiller
e146fd121c parse_expression now takes a boolean flag on whether or not it should be creating ast_labels for undefined identifiers; creating the label later will pick it up and set its undefined flag to false 2013-01-03 12:46:33 +01:00
Dale Weiler
a168c5efbe More error checking and proper unreferencing of ast nodes if expressions for computed goto fails. 2013-01-02 22:39:24 +00:00
Dale Weiler
c1734d1b27 Some error checking for computed goto. 2013-01-02 22:28:26 +00:00
Dale Weiler
e08e4a9ce0 Remove "got" left behind from debugging. 2013-01-02 21:35:18 +00:00
Dale Weiler
a421d9a33b Implemented computed goto + added goto test for testsuite (tests both normal and computed goto statements). 2013-01-02 21:32:57 +00:00
Wolfgang Bumiller
558a091c08 remove ast.h-include from lexer.h to parser.c... 2013-01-02 15:08:02 +01:00
Dale Weiler
29db4a44ed Happy new years! 2013-01-01 05:08:55 +00:00
Wolfgang Bumiller
0920cb6ec4 Another hardcoded -std=gmqcc option removed and added to the implied -std=gmqcc flags 2012-12-31 12:43:15 +01:00