Andreas Zwinkau [Fri, 25 Mar 2011 14:03:31 +0000 (15:03 +0100)]
Remove gcse optimization
It does exactly the same as 'place',
so where 'gcse' was called by default,
the driver now calls 'place'.
Andreas Zwinkau [Thu, 17 Mar 2011 14:43:51 +0000 (15:43 +0100)]
Put parallelize-mem into safe-defaults
It was simply disabled before.
Matthias Braun [Wed, 16 Mar 2011 17:44:36 +0000 (18:44 +0100)]
redo big_endian bitfield packing in a simpler and more correct way
Matthias Braun [Tue, 15 Mar 2011 11:47:22 +0000 (12:47 +0100)]
fix whitespace errors
Matthias Braun [Tue, 15 Mar 2011 11:45:53 +0000 (12:45 +0100)]
remove broken/unused caml export
Matthias Braun [Mon, 14 Mar 2011 12:41:49 +0000 (13:41 +0100)]
do not force -m32 by default but make -m32 work correctly
Matthias Braun [Mon, 14 Mar 2011 12:01:58 +0000 (13:01 +0100)]
forward -static and -shared flag to linker (like gcc does)
Christoph Mallon [Fri, 11 Mar 2011 21:00:00 +0000 (22:00 +0100)]
Fix type semantics: &, ^ and | require integer operands.
Matthias Braun [Mon, 7 Mar 2011 12:07:34 +0000 (13:07 +0100)]
fix confusing error message when misspelling -mXXX options
Matthias Braun [Fri, 4 Mar 2011 11:53:31 +0000 (11:53 +0000)]
compound parameter lowering is done in lower_for_target now
[r28473]
Matthias Braun [Thu, 3 Mar 2011 15:54:57 +0000 (15:54 +0000)]
make the uninitialized warning depend on the -wuninitialized switch
[r28472]
Matthias Braun [Mon, 28 Feb 2011 11:46:14 +0000 (11:46 +0000)]
adapt to firm Cmp changes
[r28461]
Christoph Mallon [Thu, 24 Feb 2011 11:01:30 +0000 (11:01 +0000)]
Fix another typo in the same comment.
[r28448]
Christoph Mallon [Thu, 24 Feb 2011 10:22:09 +0000 (10:22 +0000)]
Fix typo in comment.
[r28444]
Christoph Mallon [Thu, 24 Feb 2011 10:22:08 +0000 (10:22 +0000)]
Remove duplicate assert.
[r28443]
Christoph Mallon [Thu, 24 Feb 2011 10:22:06 +0000 (10:22 +0000)]
There is no need to #ifdef an identifier before #undefing it.
[r28442]
Christoph Mallon [Thu, 24 Feb 2011 10:22:04 +0000 (10:22 +0000)]
Allow declarations after labels in C++ mode.
[r28441]
Christoph Mallon [Thu, 24 Feb 2011 10:22:02 +0000 (10:22 +0000)]
Improve error message: Show the position of the label statement instead of the } when complaining about a label at the end of a compound statement.
[r28440]
Christoph Mallon [Thu, 24 Feb 2011 10:21:59 +0000 (10:21 +0000)]
Do not check for a null pointer in the type printer, because we stopped using null pointers as error type long ago.
[r28439]
Andreas Zwinkau [Thu, 24 Feb 2011 09:59:09 +0000 (09:59 +0000)]
Handle GNU extension __REDIRECT
Used to specify a different ld_name than function name.
E.g. use to redirect fstat to fstat64 in GNU.
Fixes C/gnu99/redirect
[r28437]
Michael Beck [Sun, 20 Feb 2011 11:13:11 +0000 (11:13 +0000)]
Attemp to fix typedef_init.c testcase.
[r28425]
Michael Beck [Sun, 20 Feb 2011 11:13:02 +0000 (11:13 +0000)]
add some comments, made for loop look "more usual", no functional change
[r28424]
Christoph Mallon [Fri, 18 Feb 2011 21:05:13 +0000 (21:05 +0000)]
Change is_error_entity() to !is_entity_valid(), which is consistent with is_type_valid().
[r28419]
Christoph Mallon [Fri, 18 Feb 2011 21:05:10 +0000 (21:05 +0000)]
Correct logic in is_error_entity().
This fixes C/should_fail/typedef_enum.c.
[r28418]
Matthias Braun [Fri, 18 Feb 2011 16:45:53 +0000 (16:45 +0000)]
implement sizeof(func) and alignof(func) in gnu99 mode
[r28414]
Christoph Mallon [Fri, 18 Feb 2011 12:09:26 +0000 (12:09 +0000)]
Correct off-by-whatever errors of the source position in parse_select_expression().
[r28407]
Christoph Mallon [Fri, 18 Feb 2011 12:09:24 +0000 (12:09 +0000)]
Correct typo in comment.
[r28406]
Christoph Mallon [Fri, 18 Feb 2011 12:09:23 +0000 (12:09 +0000)]
Prefer get_mode_{minus_one,null}() over new_tarval_from_long({-1,0}).
[r28405]
Christoph Mallon [Fri, 18 Feb 2011 12:09:19 +0000 (12:09 +0000)]
Assign the proper namespace to all new entities and assert that nobody tries to retrieve entities with NAMESPACE_INVALID.
This corrects C/should_fail/conflicting_types.c.
[r28404]
Andreas Zwinkau [Fri, 18 Feb 2011 10:02:51 +0000 (10:02 +0000)]
Support -pthread argument on Linux
Other platforms require different arguments or different behavior.
[r28402]
Christoph Mallon [Thu, 17 Feb 2011 13:45:49 +0000 (13:45 +0000)]
Rate EXPR_REFERENCE (variable uses) with error type as EXPR_CLASS_ERROR instead of EXPR_CLASS_VARIABLE.
For "case unknown_var:" this prevents the subsequent error that the case label is not constant.
[r28393]
Matthias Braun [Wed, 9 Feb 2011 16:13:41 +0000 (16:13 +0000)]
fix last commit: conditional_expression might use compounds
[r28359]
Matthias Braun [Wed, 9 Feb 2011 14:50:48 +0000 (14:50 +0000)]
fix cp_error065.c
[r28357]
Andreas Zwinkau [Wed, 9 Feb 2011 14:48:45 +0000 (14:48 +0000)]
Adding some file extensions for detection
Interpretation according to gcc.
[r28355]
Matthias Braun [Wed, 9 Feb 2011 13:55:49 +0000 (13:55 +0000)]
fix warning
[r28348]
Matthias Braun [Wed, 9 Feb 2011 13:55:48 +0000 (13:55 +0000)]
adapt to remove of Quot node in latest libfirm
[r28347]
Michael Beck [Sun, 6 Feb 2011 20:39:18 +0000 (20:39 +0000)]
Turn all exits inside main into returns, replace exit(1) by exit(FAILURE).
[r28335]
Christoph Mallon [Sun, 6 Feb 2011 07:51:23 +0000 (07:51 +0000)]
Enter constant folding mode before doing so. Fixes some regression tests.
[r28326]
Christoph Mallon [Sun, 6 Feb 2011 07:51:19 +0000 (07:51 +0000)]
Prefer new_Const(get_mode_null(mode)) over new_Const_long(mode, 0).
[r28325]
Christoph Mallon [Sun, 6 Feb 2011 07:51:16 +0000 (07:51 +0000)]
Add the helper function create_Const_from_bool() to create 0/1 constant from a boolean value.
[r28324]
Christoph Mallon [Sun, 6 Feb 2011 07:51:12 +0000 (07:51 +0000)]
Let is_constant_expression() and friends return an enum so constant expressions can be differentiated from erroneous expressions. Now the latter do not cause panics when trying to constant fold them.
[r28323]
Christoph Mallon [Sun, 6 Feb 2011 07:51:08 +0000 (07:51 +0000)]
Improve error recovery: eat_until_anchor() when parse_primary_expression() finds no token to consume.
[r28322]
Christoph Mallon [Sun, 6 Feb 2011 07:51:04 +0000 (07:51 +0000)]
Fix C/should_fail/decl_without_name.c.
[r28321]
Christoph Mallon [Sun, 6 Feb 2011 07:51:00 +0000 (07:51 +0000)]
Use get_break_label() instead of duplicating it.
[r28320]
Christoph Mallon [Sun, 6 Feb 2011 07:50:57 +0000 (07:50 +0000)]
Overhaul conditional_to_firm(). Now it is a tad shorter/simpler and more friendly towards SSA construction due to maturing in dominance order.
[r28319]
Christoph Mallon [Sun, 6 Feb 2011 07:50:53 +0000 (07:50 +0000)]
Simplify produce_condition_result().
[r28318]
Christoph Mallon [Sun, 6 Feb 2011 07:50:49 +0000 (07:50 +0000)]
Add jump_if_reachable() to reduce code duplication.
[r28317]
Christoph Mallon [Sun, 6 Feb 2011 07:50:46 +0000 (07:50 +0000)]
Overhaul for_statement_to_firm() so it is more friendly towards SSA construction due to maturing in dominance order.
[r28316]
Christoph Mallon [Sun, 6 Feb 2011 07:50:42 +0000 (07:50 +0000)]
Simplify maturing of the header block in do_while_statement_to_firm(). This also is more friendly towards SSA construction due to maturing in dominance order.
[r28315]
Christoph Mallon [Sun, 6 Feb 2011 07:50:39 +0000 (07:50 +0000)]
Overhaul while_statement_to_firm(). Now it is a tad shorter/simpler and more friendly towards SSA construction due to maturing in dominance order.
[r28314]
Christoph Mallon [Sun, 6 Feb 2011 07:50:35 +0000 (07:50 +0000)]
Overhaul if_statement_to_firm(). Now it is a tad shorter/simpler and more friendly towards SSA construction due to maturing in dominance order.
[r28313]
Christoph Mallon [Sat, 5 Feb 2011 09:04:11 +0000 (09:04 +0000)]
Simplify daisy chaining local labels.
[r28309]
Christoph Mallon [Sat, 5 Feb 2011 09:04:07 +0000 (09:04 +0000)]
Do not eat() the semicolon at the end of a local label declaration, but expect() it.
[r28308]
Christoph Mallon [Sat, 5 Feb 2011 09:04:03 +0000 (09:04 +0000)]
Improve error recovery: Do not create an invalid statement if the semicolon at the end of a goto statement is missing.
[r28307]
Christoph Mallon [Sat, 5 Feb 2011 09:04:00 +0000 (09:04 +0000)]
Correct typo in label name.
[r28306]
Christoph Mallon [Sat, 5 Feb 2011 09:03:56 +0000 (09:03 +0000)]
Remove unnecessary gotos.
[r28305]
Christoph Mallon [Sat, 5 Feb 2011 09:03:51 +0000 (09:03 +0000)]
Provide the specific error message about a label at the end of a compound statement not just for goto labels but for case and default labels, too.
[r28304]
Christoph Mallon [Sat, 5 Feb 2011 09:03:47 +0000 (09:03 +0000)]
Reduce variable scope.
[r28303]
Christoph Mallon [Sat, 5 Feb 2011 09:03:43 +0000 (09:03 +0000)]
Simplify case_label_to_firm() a bit, which can be done since new_immBlock() does not set current_block anymore.
[r28302]
Christoph Mallon [Sat, 5 Feb 2011 09:03:39 +0000 (09:03 +0000)]
Simplify do_while_statement_to_firm() a bit, which can be done since new_immBlock() does not set current_block anymore.
[r28301]
Christoph Mallon [Sat, 5 Feb 2011 09:03:34 +0000 (09:03 +0000)]
Correct copy&paste-o in warning message.
[r28300]
Christoph Mallon [Fri, 4 Feb 2011 14:27:21 +0000 (14:27 +0000)]
Remove unnecessary test in ast2firm: A label-statement always has an inner statement.
[r28299]
Christoph Mallon [Fri, 4 Feb 2011 14:27:18 +0000 (14:27 +0000)]
Remove unnecessary test in ast2firm: A case-label-statement always has an inner statement.
[r28298]
Christoph Mallon [Fri, 4 Feb 2011 14:27:16 +0000 (14:27 +0000)]
Remove unnecessary test in ast2firm: A switch-statement always has a body.
[r28297]
Christoph Mallon [Fri, 4 Feb 2011 14:27:14 +0000 (14:27 +0000)]
Remove unnecessary test in ast2firm: A for-statement always has a body.
[r28296]
Christoph Mallon [Fri, 4 Feb 2011 14:27:13 +0000 (14:27 +0000)]
Remove unnecessary tests in ast2firm: An if-statement always has a true_statement.
[r28295]
Christoph Mallon [Fri, 4 Feb 2011 14:27:10 +0000 (14:27 +0000)]
Improve error recovery: Produce a valid statement and continue parsing even if the ":" after "default" is missing.
[r28294]
Christoph Mallon [Fri, 4 Feb 2011 14:27:06 +0000 (14:27 +0000)]
Whitespace fixes.
[r28293]
Christoph Mallon [Mon, 31 Jan 2011 10:02:34 +0000 (10:02 +0000)]
Remove unnecessary/wrong cast. ARR_SHRINKLEN() uses size_t now.
[r28284]
Christoph Mallon [Mon, 31 Jan 2011 10:02:31 +0000 (10:02 +0000)]
Use correct parameter name in comment.
[r28283]
Michael Beck [Fri, 28 Jan 2011 01:06:19 +0000 (01:06 +0000)]
Removed superflous computations, these are done later again.
[r28275]
Michael Beck [Fri, 28 Jan 2011 01:06:10 +0000 (01:06 +0000)]
BugFix: Implement missing void casts, this fixed cp_error064.c
[r28274]
Christoph Mallon [Wed, 19 Jan 2011 14:53:32 +0000 (14:53 +0000)]
Simplify while_statement_to_firm() a bit, which can be done since new_immBlock() does not set current_block anymore.
[r28260]
Matthias Braun [Thu, 13 Jan 2011 11:55:59 +0000 (11:55 +0000)]
workaround/fix jam dependency problems
[r28245]
Matthias Braun [Thu, 13 Jan 2011 11:55:58 +0000 (11:55 +0000)]
constructors/destructors pointers must have visibility_private
[r28244]
Matthias Braun [Wed, 5 Jan 2011 17:28:27 +0000 (17:28 +0000)]
attempt to workaround input problems on OS/X gdb
[r28207]
Michael Beck [Thu, 30 Dec 2010 21:23:22 +0000 (21:23 +0000)]
Fixed type sizes and alignment for -m64.
[r28205]
Matthias Braun [Fri, 3 Dec 2010 15:05:26 +0000 (15:05 +0000)]
only write out functions for the main file when generating JNA declarations
[r28178]
Matthias Braun [Wed, 1 Dec 2010 13:11:32 +0000 (13:11 +0000)]
adapt to latest libfirm
[r28173]
Matthias Braun [Wed, 1 Dec 2010 12:45:55 +0000 (12:45 +0000)]
we might also have array null-initializers; do NOT do anythign for union members
[r28171]
Matthias Braun [Wed, 1 Dec 2010 12:02:48 +0000 (12:02 +0000)]
fix dyanimc_initializer creation for IR_INITIALIZER_NULL on struct types
[r28170]
Matthias Braun [Wed, 1 Dec 2010 12:02:47 +0000 (12:02 +0000)]
cond may be NULL
[r28169]
Matthias Braun [Mon, 29 Nov 2010 18:01:48 +0000 (18:01 +0000)]
new_d_defaultProj fails if its argument is already a Bad. Do not use it anymore (defaultProj is odd anyway IMO)
[r28165]
Matthias Braun [Mon, 29 Nov 2010 18:01:48 +0000 (18:01 +0000)]
long double must not be exported as java double
[r28164]
Andreas Zwinkau [Thu, 25 Nov 2010 14:47:21 +0000 (14:47 +0000)]
Fix Bug 62 aka C_struct_init testcase
[r28158]
Matthias Braun [Mon, 22 Nov 2010 14:04:44 +0000 (14:04 +0000)]
adapt style of generated code to java coding conventions
[r28151]
Matthias Braun [Thu, 4 Nov 2010 12:15:45 +0000 (12:15 +0000)]
adapt to latest libfirm
[r28129]
Michael Beck [Sat, 30 Oct 2010 00:56:42 +0000 (00:56 +0000)]
Let git ignore more visual studio files.
[r28123]
Michael Beck [Thu, 28 Oct 2010 19:19:12 +0000 (19:19 +0000)]
Use EXIT_SUCCESS and EXIT_FAILURE where appropriate.
[r28122]
Matthias Braun [Wed, 13 Oct 2010 14:59:07 +0000 (14:59 +0000)]
avoid warning in generated jna code
[r28117]
Andreas Zwinkau [Wed, 13 Oct 2010 14:47:22 +0000 (14:47 +0000)]
Fix JNA direct-mode wrapper
[r28115]
Matthias Braun [Tue, 12 Oct 2010 22:17:54 +0000 (22:17 +0000)]
add more definitions for __builtin___XXX_chk
[r28113]
Matthias Braun [Tue, 12 Oct 2010 22:17:53 +0000 (22:17 +0000)]
initial implementation of builtin_object_size
[r28112]
Matthias Braun [Tue, 12 Oct 2010 21:53:20 +0000 (21:53 +0000)]
fix bug on darwn introduced in the '-mtriple' commit
[r28110]
Andreas Zwinkau [Mon, 11 Oct 2010 12:52:02 +0000 (12:52 +0000)]
JNA generator now emits for JNA direct-mode
(hopefully performance gains)
[r28094]
Matthias Braun [Mon, 11 Oct 2010 12:03:01 +0000 (12:03 +0000)]
work when strings are const char*
[r28092]
Matthias Braun [Mon, 11 Oct 2010 12:03:00 +0000 (12:03 +0000)]
perform archep lowering before the target-lowering pass
[r28091]
Matthias Braun [Fri, 8 Oct 2010 13:37:02 +0000 (13:37 +0000)]
adapt to libfirm (rename tarval to ir_tarval)
[r28063]
Matthias Braun [Wed, 6 Oct 2010 08:40:49 +0000 (08:40 +0000)]
adapt to latest libfirm
[r28045]
Matthias Braun [Wed, 6 Oct 2010 08:40:48 +0000 (08:40 +0000)]
add -mtriple= to specify target architecture; refactor target architecture specific code
[r28044]