Commit Graph

16820 Commits

Author SHA1 Message Date
nickysn
e6c5c35d65 * tarraydef.create_from_pointer changed to require the actual pointerdef as a
parameter, instead of the element def. This will allow e.g. creating huge
  arrays from huge pointers on i8086.

git-svn-id: trunk@28268 -
2014-07-26 10:39:18 +00:00
svenbarth
d46ae01589 Fix inlining of record methods across units if the unit with the method definitions is already compiled.
symdef.pas:
  + tabstractrecorddef: implement buildderefimpl and derefimpl which works on the symtable (especially derefimpl is important here)
  - trecorddef & tobjectdef: remove buildderefimpl; it's already handled by the tabstractrecorddef one now
  - tobjectdef: remove the part of derefimpl that's already handled by tabstractrecorddef

git-svn-id: trunk@28267 -
2014-07-25 22:01:23 +00:00
sergei
0d96963459 * Do not promote integer comparison operations to native size if one side is a constant and its value is within range of opposite side. Instead, cast the constant to type of the opposite side.
Comparisons are already not promoted if sides are equally signed, but this wasn't true for "unsigned vs constant" case, because constants are preferably parsed as signed.
The change considerably improves generated code for i386 and even more considerably for m68k. Tested on i386-win32.

git-svn-id: trunk@28266 -
2014-07-25 21:40:22 +00:00
sergei
a0934e897e * i386: improvements to 64-bit comparison code, also serves as a good base for m68k target:
* Better handling of LOC_CREGISTER locations.
  + Support direct compares of constants with references.
  + Optimize away comparison of low dword when its result is constant (cases of "less" and "greater or equal" comparing against a constant with zero low dword).

git-svn-id: trunk@28263 -
2014-07-25 13:21:44 +00:00
sergei
bd5ce35130 * m68k: completely replaced 64-bit comparison code with a clone of i386 implementation. No optimizations yet, but it already makes system unit about 300 instructions shorter due to changing node location to LOC_JUMP.
* Cleaned up 32-bit comparison code and implemented one of the optimizations (TST instruction for comparison with constant zero which can act directly on references). This makes system unit shorter by about 1000 instructions.

git-svn-id: trunk@28260 -
2014-07-24 19:44:41 +00:00
nickysn
86726a5dc8 * return ptrsinttype instead of sinttype in pointerdef.pointer_arithmetic_int_type
git-svn-id: trunk@28259 -
2014-07-24 15:17:15 +00:00
nickysn
33dac14554 * defutil.get_int_type_for_pointer_arithmetic replaced with a virtual method
tpointerdef.pointer_arithmetic_int_type

git-svn-id: trunk@28258 -
2014-07-24 14:18:57 +00:00
nickysn
bb48761652 * taddnode.pass_typecheck_internal: always pass a tpointerdef to
get_int_type_for_pointer_arithmetic, even when adding a zero-based array with
  a number

git-svn-id: trunk@28257 -
2014-07-24 13:28:19 +00:00
sergei
dac52f503c * m68k: fixed extension in a_load_ref_reg. Existing code cleans only bits 16-31 when loading a 8-bit register from 16-bit reference, and leaves garbage in bits 8-15.
git-svn-id: trunk@28256 -
2014-07-24 12:17:29 +00:00
Jonas Maebe
5836478bce * add alignment for .lcomm data in AIX, required for int64 data put in bss
in case it's passed to interlocked* (requires 8 byte alignment, default
    is 4). Supported in AIX 5.3 and later
    (http://www-01.ibm.com/support/docview.wss?uid=isg1IY90802) and
    GNU binutils 2.24 or later

git-svn-id: trunk@28252 -
2014-07-23 14:06:09 +00:00
Jonas Maebe
2c7a1b4e91 * fixed optimisation from r25067 for the generic case where a signed value
is converted to a larger unsigned value (e.g. OS_S16 -> OS_32)

git-svn-id: trunk@28251 -
2014-07-23 14:06:06 +00:00
Jonas Maebe
d5895a4d98 - removed ppc-specific a_load_const_cgpara: not necessary, and in fact wrong
in case the para location size is larger than the source size

git-svn-id: trunk@28250 -
2014-07-23 14:06:03 +00:00
svenbarth
856ae9d3cf Fix for Mantis #26482.
pgenutil.pas, generate_specialization:
  * also search for the generic type in general if it was previously found in a record/class as there might be a non-generic type (or one with a different number of parameters) defined elsewhere

+ added test

git-svn-id: trunk@28244 -
2014-07-20 15:47:18 +00:00
svenbarth
f1bfb398e4 Fix for Mantis #26456.
symtable.pas, searchsym_in_helper:
  * check whether extendeddef is assigned as this function is called when parsing the extended type and it was not found (the second pass is to display a more accurate error message and was added some weeks ago)

+ added test

git-svn-id: trunk@28242 -
2014-07-20 13:40:01 +00:00
Jonas Maebe
4ff597afae * call "message" rather than "cgmessage" when throwing fatal errors, because
cgmessage won't do anything in case cgerror is already set (and in several
    places we assume that the compiler will immediately abort after throwing a
    fatal error)

git-svn-id: trunk@28229 -
2014-07-18 09:09:21 +00:00
Jonas Maebe
9e29fd424d * changed the "laststatement" parameter of internalstatements() from "var"
to "out"

git-svn-id: trunk@28228 -
2014-07-18 09:09:16 +00:00
Jonas Maebe
627c83e828 * when creating a bare procdef copy, don't copy the funcret parameter.
This is useful if you want to change the calling convention of the copy.
   o call insert_funcret_para() after creating a bare copy
  * don't copy aliasnames of copied procdefs either (can at best result
    in duplicate symbol errors)

git-svn-id: trunk@28227 -
2014-07-18 09:09:11 +00:00
Jonas Maebe
256f2fcf69 * fixed finish_copied_procdef() in case newstruct is nil after support for
generics was added to it

git-svn-id: trunk@28226 -
2014-07-18 09:09:03 +00:00
nickysn
6d54046300 * output a proper error message in case number+nil is encountered (just like
it's already done for nil+number and nil-number)

git-svn-id: trunk@28225 -
2014-07-15 22:26:58 +00:00
nickysn
9b3d41d9c7 * the default value of tpointerdef.pointer_subtraction_result_type changed to
ptrsinttype

git-svn-id: trunk@28220 -
2014-07-15 06:51:56 +00:00
nickysn
d86a8ebc99 * defutil.get_int_result_type_from_pointer_subtraction moved to a virtual
method tpointerdef.pointer_subtraction_result_type

git-svn-id: trunk@28219 -
2014-07-14 23:30:29 +00:00
nickysn
99c555cea8 + implemented huge pointer subtraction
* the result of far pointer subtraction changed to word (for TP7 compatibility)

git-svn-id: trunk@28218 -
2014-07-13 23:40:43 +00:00
nickysn
200df7a136 * set expectloc correctly for far pointer comparison as well
git-svn-id: trunk@28217 -
2014-07-13 16:25:20 +00:00
nickysn
04234c733f * set expectloc correctly (LOC_JUMP instead of LOC_FLAGS) for non-normalized
huge pointer comparison

git-svn-id: trunk@28216 -
2014-07-13 15:20:58 +00:00
nickysn
5818451bb5 * non-normalized huge pointer comparison made unsigned
git-svn-id: trunk@28214 -
2014-07-13 12:54:12 +00:00
Tomas Hajny
82d62aa883 * added description for option @ and missing bracket at the end of list for -P
git-svn-id: trunk@28188 -
2014-07-08 22:33:18 +00:00
Tomas Hajny
109d890adf * added description for option @ and missing bracket at the end of list for -P
git-svn-id: trunk@28187 -
2014-07-08 22:33:00 +00:00
Tomas Hajny
dfb44e1642 * improve identation of long options and add support for displaying parameter @
git-svn-id: trunk@28186 -
2014-07-08 22:30:43 +00:00
sergei
b7da785688 * m68k: support stack cleanup at caller side, fixed calculation of pushed parameters size and offsets and cleaned out another pile of junk.
* Parameter offsets are now calculated similar to other targets, target_info,first_parm_offset is applied only to callee side, while at caller side offsets start from 0.
* For cdecl procedures, parameters are removed by caller, for the rest it's still done by callee (resembles i386 target, except there is no 'register' calling convention).

git-svn-id: trunk@28185 -
2014-07-08 19:08:56 +00:00
sergei
cfe13734e0 * m68k: Fixed damage caused by r27573 and r28177:
* longjmp: restored loading of function result (d0) which got removed in r28177.
  * While at the point, added test for zero, because longjmp must never return zero values.
  * Assembler reader: 'fp' refers to frame pointer, not the stack pointer.

git-svn-id: trunk@28183 -
2014-07-08 15:04:54 +00:00
sergei
15ff2c1236 * m68k: Fixed name of ELF interpreter, and updated cprt0.as to somewhat working state. Now a "hello world" program using libc can be linked and launched, but anything more complex needs another round of fixing the parameter passing code, in order to get the stack properly balanced.
git-svn-id: trunk@28181 -
2014-07-07 20:08:50 +00:00
nickysn
8604372da3 + implemented far pointer comparison operating only on the offset (+ added test)
git-svn-id: trunk@28179 -
2014-07-06 22:45:19 +00:00
sergei
df60309d96 * m68k: fixed the last remaining warning and removed "$WARNINGS OFF" directive.
git-svn-id: trunk@28176 -
2014-07-06 11:36:33 +00:00
sergei
e4fea2ebc8 * Dummy implementations of a_bit_scan_reg_reg and g_stackpointer_alloc in tcg, removes the need to override these methods in every descendant code generator solely to avoid "constructing a class with abstract method" warning.
git-svn-id: trunk@28175 -
2014-07-06 11:34:04 +00:00
Károly Balogh
e5656a6f7e m68k assembler reader: fixed parsing of slash-separated movem/fmovem register lists. only the first register was stored.
git-svn-id: trunk@28174 -
2014-07-06 11:29:45 +00:00
sergei
1678fcdc2d - m68k assembler reader: removed Oper.InitRef which was causing failure of tbs/tb0142.pp and, in general, making impossible calls/jumps to non-alias procedure names.
git-svn-id: trunk@28173 -
2014-07-06 10:35:35 +00:00
svenbarth
441e6c6083 Fix for Mantis #26177. Use the correct value for the method data when using a method pointer of a type or record helper.
Please note that Delphi has the exact same bug as we have and thus code working in FPC will not work in Delphi.
Additionally taking the method address of a local variable or a local/global constant for a method pointer is dangerous as the variable (in case of constants is a temporary local variable) will go out of scope once the containing procedure/function/method exits!

ncgld.pas, tcgloadnode.pass_generate_code:
  * only use the value of Self if it is an implicit pointer object (class instance) or a class reference, but not for everything else (objects, records, primitive types)

+added test

git-svn-id: trunk@28160 -
2014-07-05 08:54:10 +00:00
nickysn
2b2224494b * call first_addpointer and first_cmppointer in pass_1 for pchars as well. Fixes
huge pointer arithmetic/comparison on huge pchars.

git-svn-id: trunk@28159 -
2014-07-04 23:58:26 +00:00
nickysn
c8130843b2 * fixed 'cannot convert type' error when comparing two voidhugepointers with
= or <>

git-svn-id: trunk@28158 -
2014-07-04 23:54:47 +00:00
nickysn
3bb730d5b9 + implemented normalized huge pointer comparison
git-svn-id: trunk@28157 -
2014-07-04 23:22:33 +00:00
nickysn
e80ecd4d5b + moved handling of pointer comparison in pass_1 to new virtual method
taddnode.first_cmppointer. To be overriden for i8086 huge pointer comparisons.

git-svn-id: trunk@28156 -
2014-07-04 22:47:21 +00:00
nickysn
7f9fa145e8 + added error message in case the HUGEPOINTERNORMALIZATION directive is used
with an invalid argument

git-svn-id: trunk@28152 -
2014-07-04 17:16:58 +00:00
nickysn
63305cc2e6 + added directive {$HUGEPOINTERARITHMETICNORMALIZATION} to control the
cs_hugeptr_arithmetic_normalization local setting
+ added the cs_hugeptr_comparison_normalization local setting and corresponding
  directive {$HUGEPOINTERCOMPARISONNORMALIZATION} to set it
* the {HUGEPOINTERNORMALIZATION} changed to set both of the huge pointer
  normalization settings, according to the specified compiler type ('BORLANDC',
  'MICROSOFTC' or 'WATCOMC')

git-svn-id: trunk@28151 -
2014-07-04 14:43:46 +00:00
nickysn
a3b48b79cb - rm the scan_w_hugepointernormalization_not_support message, replaced with
scanner_w_directive_ignored_on_target

git-svn-id: trunk@28150 -
2014-07-04 14:03:18 +00:00
nickysn
aee568e6fc * cs_hugeptr_normalization renamed to cs_hugeptr_arithmetic_normalization,
because we're going to also introduce cs_hugeptr_comparison_normalization as
  an independent option

git-svn-id: trunk@28149 -
2014-07-04 13:14:06 +00:00
nickysn
d9e94ffab7 + implemented huge pointer arithmetic
git-svn-id: trunk@28139 -
2014-07-03 16:53:22 +00:00
nickysn
da77535631 * the handling of pointer arithmetic in pass_1 moved to a new virtual method
taddnode.first_addpointer. This will be overridden in the i8086 code generator
  in order to implement huge pointer arithmetic.

git-svn-id: trunk@28138 -
2014-07-03 13:47:47 +00:00
svenbarth
30be5d8d33 As discussed on core enable the abstract class construction warning by default and also make it controllable with {$warn constructing_abstract ...}.
* scandir.pas, dir_warn: also adjust the message state for type_w_instance_abstract_class if the identifier is CONSTRUCTING_ABSTRACT
* msg/errore.msg: enable type_w_instance_abstract_class by default

* msgidx.inc, msgtxt.inc: updated

git-svn-id: trunk@28137 -
2014-07-02 19:40:22 +00:00
Jonas Maebe
43839d1bda * update of German error message files by Karl-Michael Schindler
(mantis #26282)

git-svn-id: trunk@28129 -
2014-07-02 17:17:38 +00:00
svenbarth
a7a9440692 Add a new warning message that is generated if an instance of an abstract class is created. This message is disabled by default, but can be switched on by using {$warn 4122 on} or {$warn 4122 error}.
Please note that this warning won't be triggered if an instance of that class is created using a class variable of that class type as the compiler can not know the type contained in the variable at compile time (see also the added test).

+ msg/errore.msg: added disabled message which informs about the instantiation of an abstract class
* pexpr.pas, do_member_read: generate the message if we have a constructor call for an abstract class using a loadvmtaddrnode (thus the type name is used and not a class variable)

* msg{idx,txt}.inc: updated

+ added test

git-svn-id: trunk@28127 -
2014-07-01 20:41:05 +00:00