paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-05-31 07:03:21 +02:00
Code Issues Packages Projects Releases Wiki Activity
45,818 Commits 100 Branches 75 Tags 300 MiB
68a25e154a
Commit Graph

17863 Commits

Author SHA1 Message Date
Jonas Maebe
3d9713b9c8 * has_single_field() now looks recursively into fields that are aggregate 
types until it finds a non-aggregate type, and then returns "true" if
    that aggregate (record or array) contains one element
  * it now also returns the type of that element directly rather than a
    tfieldvarsym, since we only care about the type and in case of an array
    there is no field

git-svn-id: trunk@30205 -
 2015-03-14 18:35:53 +00:00
Jonas Maebe
2e126bb723 - removed darwin/i386 special case code from i8086 code generator 
git-svn-id: trunk@30204 -
 2015-03-14 18:35:50 +00:00
Jonas Maebe
9788b01d31 * rewrote most of the special case handling of parameter passing on ppc64, 
documenting the pecularities of the various calling conventions and
    marking what we do and don't support currently
  * also handle arrays for the ELFv2 ABI when determining whether an aggregate
    only contains floating values of a single type

git-svn-id: trunk@30203 -
 2015-03-14 18:35:47 +00:00
Jonas Maebe
165aaea8a4 * made a separate abi entry for powerpc_darwin, as it's not exactly the same 
as the AIX abi (especially on ppc64)
   o replaced (for now) all checks for the AIX abi with a check for either the
     AIX or Darwin abi
   o changed the ABI of Darwin/ppc, Darwin/ppc64 and Mac OS/ppc to
     abi_powerpc_darwin

git-svn-id: trunk@30202 -
 2015-03-14 18:35:44 +00:00
Jonas Maebe
25340480b0 * factored out creating a single paraloc, so we can reuse it for creating 
function return locations

git-svn-id: trunk@30201 -
 2015-03-14 18:35:41 +00:00
Jonas Maebe
f0ee1a9ee3 * pass records that (recursively) only contain floating point values of the 
same type (even in other records or in arrays), with 8 or less such
    values, as if those values were passed individually on ppc64/ELFv2

git-svn-id: trunk@30200 -
 2015-03-14 18:35:38 +00:00
Jonas Maebe
9a73c9a460 * avoid string truncation for ait_symbolpair in assembler writer 
git-svn-id: trunk@30199 -
 2015-03-14 18:35:34 +00:00
Jonas Maebe
fb27dff638 * generate ".abiversion 2" directive when targeting ppc64/ELFv2 
* generate code to load the TOC register if it's required, and
    emit the the ".localentry" directive to indicate to the linker
    where the actual function body starts
  * support the ".localentry" directive in the ppc64 assembler reader
   o disable the fpc_qword_to_double() assembler implementation for
     ELFv2, because we can't manually insert the .localentry directive
     before the stack allocation code
  * perform indirect calls on ppc64 via R12 in ncgcal, as R12 needs to
    contain the function address on entry on ppc64/ELFv2 (and it's
    a volatile register, so there's no problem with always using
    it)

git-svn-id: trunk@30198 -
 2015-03-14 18:35:31 +00:00
Jonas Maebe
8445381929 * merged ait_set and ait_thumb_set into a single tai class 
(tai_symbolpair)

git-svn-id: trunk@30197 -
 2015-03-14 18:35:28 +00:00
Jonas Maebe
1e227f78fe * factored out the check for whether we have to generate dotted function 
names, and disabled them for ppc64 ELFv2
  * disabling those also disables the function descriptors, so use the
    _CALL_ELF compiler variable to check whether there is in fact a
    function descriptor

git-svn-id: trunk@30196 -
 2015-03-14 18:35:25 +00:00
Jonas Maebe
cc071ded4b * define _CALL_ELF=1 or _CALL_ELF=2 symbol when targeting ppc(64) 
ELFv1/ELFv2, like GCC

git-svn-id: trunk@30195 -
 2015-03-14 18:35:22 +00:00
Jonas Maebe
05f6660a25 * set the correct dynamic linker for the ppc64 ELFv2 abi 
* specify the correct linker emulation to use for ppc64 ELFv2 little endian
  * specify the correct default library directories to search for ELFv2 little
    endian

git-svn-id: trunk@30194 -
 2015-03-14 18:35:19 +00:00
Jonas Maebe
3946971c73 * default to ppc64le/elfv2 when compiling on a ppc64le platform 
* default to elfv2 abi when targeting ppc64le and no abi has been specified

git-svn-id: trunk@30193 -
 2015-03-14 18:35:16 +00:00
Jonas Maebe
0e76a84e04 + "ELFv2" abi identifier for ppc64 
git-svn-id: trunk@30192 -
 2015-03-14 18:35:13 +00:00
Károly Balogh
918c4add06 m68k: in do_spill_replace, exclude the case where the source register would be an address and the destination is a reference 
git-svn-id: trunk@30191 -
 2015-03-14 17:19:14 +00:00
Károly Balogh
935820293c m68k: in tm68ktypeconvnode.int_to_real, fix the reference before using it, and if the source is a register, make sure it's a data register 
git-svn-id: trunk@30190 -
 2015-03-14 17:14:10 +00:00
Károly Balogh
1121c2e6ce m68k: before doing a reg->ref operation, make sure the source is a datareg 
git-svn-id: trunk@30189 -
 2015-03-14 16:36:02 +00:00
Jeppe Johansen
52e505bff7 Fixed internal error in GOT related code for ARM internal assembler. 
git-svn-id: trunk@30188 -
 2015-03-14 16:03:16 +00:00
Jeppe Johansen
439027a8de Add most pre-UAL VFP instruction forms. 
Add fused mac instructions for VFPv4.

git-svn-id: trunk@30187 -
 2015-03-14 14:59:13 +00:00
Jeppe Johansen
64f127141f Add VFPv4 FPU type for ARM. 
Fix assembler reader so it can read instructions longer than 5 characters.

git-svn-id: trunk@30186 -
 2015-03-14 12:49:07 +00:00
Jeppe Johansen
73abf5e630 Merge from armiw branch. 
Update ARM internal assembler to support most ARM, Thumb and Thumb-2 instructions.
Changed generation of VFP instructions to use UAL mnemonics.
Added divided and unified assembler syntax support to ARM assembly reader.

git-svn-id: trunk@30181 -
 2015-03-13 19:22:27 +00:00
Jeppe Johansen
b75330760d Merged from trunk. 
git-svn-id: branches/laksen/armiw@30180 -
 2015-03-13 18:58:57 +00:00
Jeppe Johansen
5f622752ce Disable internal assembler 
git-svn-id: branches/laksen/armiw@30179 -
 2015-03-13 18:56:59 +00:00
Jonas Maebe
d4bc74ecc0 * support for -WM10.XX (for 10.10) and defining the correct associated 
version macro

git-svn-id: trunk@30178 -
 2015-03-13 08:35:30 +00:00
Jonas Maebe
5b3c511467 * don't crash when "return exitvalue" is used in a procedure 
git-svn-id: trunk@30177 -
 2015-03-13 08:35:26 +00:00
Jonas Maebe
2dcd42d4f1 * fixed detection of an objcprotocols descending from a forward-defined 
objcprotocol

git-svn-id: trunk@30169 -
 2015-03-12 19:37:43 +00:00
florian
f0984fb628 * abstractrecorddef.contains_float_field should search recursively, resolves #27580 
git-svn-id: trunk@30168 -
 2015-03-11 22:05:14 +00:00
pierre
b0d8f487c1 Makefile regenerated after previous commit: Fix fullinstall for win64 and disable 8086 and i386 compiler for win64 os source only for fullcycle and fullinstall 
git-svn-id: trunk@30167 -
 2015-03-11 22:03:52 +00:00
pierre
999436c5e6 Fix fullinstall for win64 and disable 8086 and i386 compiler for win64 os source only for fullcycle and fullinstall 
git-svn-id: trunk@30166 -
 2015-03-11 21:57:47 +00:00
Jonas Maebe
fc21845686 * fixed the allocation of R12 on ppc32/ppc64 when used to hold the value of 
the stack pointer during the prolog. This was done previously in
    tcg.g_proc_entry(), but that routine is called after register allocation
    and hence has no influence. Also cleaned up the deallocation of that
    register by moving the previously ifdef'd code to thlcgppcgen
    (mantis #27634)

git-svn-id: trunk@30164 -
 2015-03-11 17:49:44 +00:00
pierre
21fe7616a4 Avoid line longer than 255 characters, to be able to edit inside FP IDE 
git-svn-id: trunk@30158 -
 2015-03-10 07:56:18 +00:00
Jonas Maebe
6a70c84258 + internal bsr/bsf support for ppc32 and ppc64 
git-svn-id: trunk@30152 -
 2015-03-09 16:48:34 +00:00
Jeppe Johansen
3d7dce81fe Make MRS and MSR use the right encoding on Thumb architectures. 
Set regnumber_count_bsstart to 128 to be able to search all registers.

git-svn-id: branches/laksen/armiw@30150 -
 2015-03-08 17:30:38 +00:00
Jeppe Johansen
03b9e5117f Fix off by 1 error in assembler reader which prevented B instructions from being parsed. 
git-svn-id: branches/laksen/armiw@30147 -
 2015-03-08 12:52:43 +00:00
Jeppe Johansen
914e9e7b49 Merged from trunk 
git-svn-id: branches/laksen/armiw@30146 -
 2015-03-08 12:33:46 +00:00
Jeppe Johansen
d12b913e70 Add initial support in ARM-embedded for Allwinner A20. 
git-svn-id: trunk@30144 -
 2015-03-08 12:22:57 +00:00
Károly Balogh
0021ad714f * fixed some ancient code behind cpu_uses_separate_address_registers define to at least compile again, for some m68k experiment 
git-svn-id: trunk@30130 -
 2015-03-07 22:11:43 +00:00
Jonas Maebe
b931f78b4b * insert a type conversion from ordinal to pointer for pointerconstn 
git-svn-id: branches/hlcgllvm@30129 -
 2015-03-07 21:44:01 +00:00
Jonas Maebe
c1e7c9de02 * support for flushing a ttai_typedconstbuilder queue with an ordinal 
constant at the end (in case of e.g. ordinal -> pointer type
    conversions)

git-svn-id: branches/hlcgllvm@30128 -
 2015-03-07 21:43:57 +00:00
Jonas Maebe
eba3f534de * correctly handle empty ansistring constants for llvm 
git-svn-id: branches/hlcgllvm@30127 -
 2015-03-07 21:43:52 +00:00
Jonas Maebe
935c0c86aa * write all ait_const's of value 0 as 'zeroinitializer' so they also work 
for pointer constants

git-svn-id: branches/hlcgllvm@30126 -
 2015-03-07 21:43:47 +00:00
Legolas
ffd03563d0 * Updated Wii rtl (unicode support broken .dol executables) and libogcfpc to ver.1.8.12. Fixed a dependency to Mad library too 
git-svn-id: trunk@30125 -
 2015-03-07 21:34:51 +00:00
Jonas Maebe
9c42437326 * moved tasmlisttypedconstbuilder.tc_emit_string_offset() to 
ttai_typedconstbuilder.emit_string_offset() so it can also be used outside
    the context of parsing a Pascal-level typed constant

git-svn-id: branches/hlcgllvm@30111 -
 2015-03-06 19:45:04 +00:00
sergei
a709a9b637 * MIPS peephole: check that operand is present before accessing its fields, also check that it's not a branch target. Mantis #27608. 
git-svn-id: trunk@30110 -
 2015-03-06 00:04:06 +00:00
Jonas Maebe
511e878606 * fixed pointer plus/minus integer on llvm (since a_op_* only gets a single 
def, we have to ensure that both arguments have in fact the same type)

git-svn-id: branches/hlcgllvm@30108 -
 2015-03-05 22:45:32 +00:00
Jonas Maebe
f9ae2aadac * fixed llvm unsigned greater/lower than condition mnemoniics 
git-svn-id: branches/hlcgllvm@30107 -
 2015-03-05 22:20:50 +00:00
Jonas Maebe
ea00e559f8 * fixed compilation after previous merge 
git-svn-id: branches/hlcgllvm@30106 -
 2015-03-05 22:20:45 +00:00
Jonas Maebe
06dd396a2f * fixed writing float values using the internal assembler after r28105 
git-svn-id: branches/hlcgllvm@30105 -
 2015-03-05 21:41:56 +00:00
Jonas Maebe
67b8aceaee * synchronized with privatetrunk till r30095 
git-svn-id: branches/hlcgllvm@30101 -
 2015-03-05 20:32:15 +00:00
Károly Balogh
a71fcd290b * allocate a temp addressreg instead of a temp intreg in callnode. improves code generation on m68k for these cases 
git-svn-id: trunk@30080 -
 2015-03-02 22:21:53 +00:00
Károly Balogh
106056f462 m68k: more FMOVEM.X store/load size fixes 
git-svn-id: trunk@30062 -
 2015-03-02 02:55:23 +00:00
Károly Balogh
acaf382ea0 m68k: FMOVEM.X stores/loads each reg as 12 bytes (96 bits) not 10 bytes (80 bits), see 68k PRM, page 5-86 
git-svn-id: trunk@30061 -
 2015-03-02 02:48:35 +00:00
Jonas Maebe
6dd2fe0320 * converted widestr.charlength() to use system.Utf8CodePointLen() 
git-svn-id: trunk@30050 -
 2015-03-01 17:12:27 +00:00
Károly Balogh
5f66d63335 m68k: disabled comment generation 
git-svn-id: trunk@30045 -
 2015-03-01 15:49:57 +00:00
Károly Balogh
997ec578e0 m68k: added a simple unaryminusnode which can utilize FNEG instruction for floats on 68881 
git-svn-id: trunk@30044 -
 2015-03-01 15:00:07 +00:00
Tomas Hajny
9f9ab082da * fix typo (space instead of underscore resulting in broken help display) 
git-svn-id: trunk@30041 -
 2015-02-28 23:30:24 +00:00
Tomas Hajny
2be0cc8aae * avoid crashes when displaying help with an unsupported system codepage 
git-svn-id: trunk@30040 -
 2015-02-28 23:15:56 +00:00
Tomas Hajny
dbbce18f56 * fix determination of help line length with codepage set to UTF-8 (the string conversion routine in widestr.pas doesn't work for UTF-8 :-((( ) 
git-svn-id: trunk@30039 -
 2015-02-28 22:54:48 +00:00
florian
f0cf88533e * applying const_align when generating structures accessed by the rtl is not a good idea because their 
alignment must be immutable, should resolve #27525

git-svn-id: trunk@30038 -
 2015-02-28 22:44:58 +00:00
florian
99635658ec * corrects change flags for VSQRTSD 
git-svn-id: trunk@30037 -
 2015-02-28 22:42:25 +00:00
Jonas Maebe
dd2aa17de1 * also allow 64 bit platforms have have a 32x32->64 bit multiplication to 
take advantage of this transormation (enables this optimization for
    AArch64)

git-svn-id: trunk@30035 -
 2015-02-28 22:31:06 +00:00
Jonas Maebe
b821e31442 * force constants into a registers in the 32x32->64 optimized case 
git-svn-id: trunk@30034 -
 2015-02-28 22:31:03 +00:00
Jonas Maebe
86bde57b0f * do not switch around node subtrees if they have different resultdefs, e.g. 
because a subtree has been optimized for a 32x32->64 multiplication (so
   some subtrees have a 32 bit resultdef and others a 64 bit resultdef)

git-svn-id: trunk@30033 -
 2015-02-28 22:31:00 +00:00
Jonas Maebe
ada5060a34 * set pi_do_call for AArch64 mod/div nodes, as they call FPC_DIVBYZERO 
git-svn-id: trunk@30032 -
 2015-02-28 22:30:57 +00:00
Jonas Maebe
33b1732ca2 * avoid unnecessary sign/zero-extensions in genlinearlist() when generating 
code on a platform that doesn't have subregisters of the same size as
    the "case" value

git-svn-id: trunk@30031 -
 2015-02-28 16:46:54 +00:00
Jonas Maebe
b2427d04ed * ensure that data pools are not inserted right after add/tbb/tbh-based 
jumptable dispatches

git-svn-id: trunk@30027 -
 2015-02-27 20:52:12 +00:00
Tomas Hajny
1ecf8dcfa1 * improved paging of help pages so that lines are not missed with 80x25 windows due to lines overflowing the window width 
git-svn-id: trunk@30026 -
 2015-02-27 19:26:23 +00:00
Károly Balogh
7940806fc2 fixed ifneq vs. ifndef 
git-svn-id: trunk@30023 -
 2015-02-27 11:33:38 +00:00
Jonas Maebe
37c2d0e60d * use Object_getClass(obj) instead of obj._class to get the class of an 
Objective-C instance on AArch64, as the latter helper from the runtime
    is a plain function call rather than an expensive message dispatch,
    and moreover it also works for protocols (the "class" method is part
    of NSObject, and protocols do not support it since there is no single
    root class and hence they may represent an instance of a class that
    does not derive from NSObject)
   o also call it to get the "self" pointer when calling class methods for
     protocols on AArch64 (fixes tobjc22a and tobjc22b on Darwin/AArch64)
   o also call it to get the metaclass of a class on Darwin/AArch64 (fixes
     tobjc25)
  - removed special handling of getting the Objective-C (meta)class for
    non-type nodes on other Darwin platforms, as this is also correctly
    handled by the generic pass_generate_code

git-svn-id: trunk@30018 -
 2015-02-26 21:49:08 +00:00
florian
b7422716b2 * turn off compiler and utils building for nds 
* use NoNativeBinaries to detect if a native compiler shall be build

git-svn-id: trunk@30017 -
 2015-02-26 20:03:29 +00:00
Károly Balogh
56eeabd01b fixed a warning with DFA 
git-svn-id: trunk@30015 -
 2015-02-26 09:36:03 +00:00
sergei
657ac78304 * Improved handling of registers used in assembler blocks. Assembler nodes no longer have lists of used registers attached to them. Instead, each specified register creates an allocation/deallocation pair and an additional tai_regalloc.markused item directly in the asmlist. This way, register lists are no longer limited to integer registers, and parsing no longer depends on paramanager to know which registers are volatile. 
If assembler block has no modified register list, it is still handled in pass2, by allocating all volatile registers (not only the integer ones as before).

git-svn-id: trunk@30011 -
 2015-02-25 21:53:15 +00:00
sergei
31fddaafe8 + New subtype of tai_regalloc, "ra_markused". It marks a physical register as used in procedure, triggering saving/restoring that register if it is non-volatile. 
git-svn-id: trunk@30010 -
 2015-02-25 21:38:23 +00:00
florian
91db2a1698 + report when compilation of a unit is stopped because it has been already compiled during the recursive search in uses clauses 
git-svn-id: trunk@30001 -
 2015-02-24 21:46:09 +00:00
florian
0a17a08b28 - remove accidently committed (commented though) debug statement 
git-svn-id: trunk@30000 -
 2015-02-24 20:58:49 +00:00
pierre
da55d9ded2 Also disable overflow where range check is disabled in aasmcpu unit 
git-svn-id: trunk@29989 -
 2015-02-24 15:58:49 +00:00
marco
422df4bc77 * fixed typo removing ppca64, mantis 27533 
git-svn-id: trunk@29988 -
 2015-02-24 09:35:27 +00:00
Jonas Maebe
879afbb7be * removed -Oodfa warnings 
git-svn-id: trunk@29986 -
 2015-02-23 22:57:24 +00:00
Jonas Maebe
c41759a026 - removed a lot of unused copied internal assembler code 
git-svn-id: trunk@29985 -
 2015-02-23 22:57:21 +00:00
Jonas Maebe
2ab7f5c35d * moved x86-specific requirements from the generic bsr/bsf code to the 
x86 code generator (register size constraints)

git-svn-id: trunk@29984 -
 2015-02-23 22:57:18 +00:00
Jonas Maebe
3b5adcb3de + print iphonesim/x86_86-specific command line options in the help 
git-svn-id: trunk@29982 -
 2015-02-23 22:57:12 +00:00
Jonas Maebe
dcb6f2baae + print Darwin/AArch64-specific command line options in the help 
git-svn-id: trunk@29981 -
 2015-02-23 22:57:09 +00:00
Jonas Maebe
5e339e89e4 * source can also be a LOC_CREGISTER for bsr/bsf 
git-svn-id: trunk@29980 -
 2015-02-23 22:57:05 +00:00
Jonas Maebe
a8d05b66e2 + AArch64 internal bsf/bsr support 
git-svn-id: trunk@29979 -
 2015-02-23 22:57:02 +00:00
Jonas Maebe
b40a4d1b5a + support for "xor mmreg,mmreg" for initialising global floating point 
regvars

git-svn-id: trunk@29978 -
 2015-02-23 22:56:59 +00:00
Jonas Maebe
0a0e74342f * AArch64 doesn't need the software int64 to double code 
git-svn-id: trunk@29976 -
 2015-02-23 22:56:53 +00:00
Jonas Maebe
b7fb862f4a * enable internal handling of abs(int64) on AArch64 
git-svn-id: trunk@29975 -
 2015-02-23 22:56:50 +00:00
Jonas Maebe
fa328436ab * regenerated for iphonesim/x86-64 
git-svn-id: trunk@29974 -
 2015-02-23 22:56:41 +00:00
Jonas Maebe
749f393095 * fixed platform names for Darwin in ppudump 
git-svn-id: trunk@29973 -
 2015-02-23 22:56:24 +00:00
Jonas Maebe
79a06b1514 + iphonesim/x86_64 target (64 bit iOS simulator) 
git-svn-id: trunk@29970 -
 2015-02-23 22:56:09 +00:00
Jonas Maebe
f421e83c9a - don't allow setting the OS X version (-WM) for Darwin/ARM or Darwin/AArch64 
for now (it's all iOS)

git-svn-id: trunk@29969 -
 2015-02-23 22:56:06 +00:00
Jonas Maebe
07f31d560c + also perform sign/zero-extensions of the index in vecn using extended 
registers in references

git-svn-id: trunk@29968 -
 2015-02-23 22:56:03 +00:00
Jonas Maebe
d6de2c03cb * generic part of r26050 from the hlcgllvm branch: made tcgvecnode hlcg-safe 
o in particular, add tdef size information to the update_reference*()
     methods, and factored out offset adjustments into its own method
   o also make sure the passed size to update_reference*() corresponds to the
     actual size of the index, as it's no longer guaranteed to be ptruint
     since the previous commit

git-svn-id: trunk@29967 -
 2015-02-23 22:56:00 +00:00
Jonas Maebe
97df25cc29 + use reference/add "extended register" scale modes for vecnodes 
git-svn-id: trunk@29966 -
 2015-02-23 22:55:57 +00:00
Jonas Maebe
b5b7e0f338 * ensure that 64->32 bit truncations cannot be optimized away by the 
register allocator on AArch64 + test (did not get caught by existing
    tests)

git-svn-id: trunk@29965 -
 2015-02-23 22:55:54 +00:00
Jonas Maebe
1de8e53edd + AArch64 jump table support 
git-svn-id: trunk@29964 -
 2015-02-23 22:55:42 +00:00
Jonas Maebe
d4d9e361fc + AArch64 optimized version of a_load_regconst_subsetreg_intern() 
git-svn-id: trunk@29963 -
 2015-02-23 22:55:39 +00:00
Jonas Maebe
555194a67b + AArch64 optimized version of a_load_subsetreg_subsetreg() 
git-svn-id: trunk@29962 -
 2015-02-23 22:55:36 +00:00
Jonas Maebe
602853d40b + AArch64 optimized version of a_load_subsetreg_reg() 
git-svn-id: trunk@29961 -
 2015-02-23 22:55:33 +00:00
Jonas Maebe
c9de3b2ecd * use the zero register for a_load_const_ref(0) 
git-svn-id: trunk@29960 -
 2015-02-23 22:55:29 +00:00
Jonas Maebe
07455fb889 + inlined versions of concatcopy 
git-svn-id: trunk@29959 -
 2015-02-23 22:55:26 +00:00
Jonas Maebe
e6ab39765d * add restrictions for loaded/stored registers in case of references 
with write-back

git-svn-id: trunk@29958 -
 2015-02-23 22:55:23 +00:00
Jonas Maebe
65c2c68b5f * call the AArch64 compiler ppca64 instead of ppcaarch64 (less typing) 
git-svn-id: trunk@29957 -
 2015-02-23 22:55:20 +00:00
Jonas Maebe
14bd77f11c * removed ARM copy/paste stuff 
git-svn-id: trunk@29956 -
 2015-02-23 22:55:17 +00:00
Jonas Maebe
ff73d97e7f * temporary workaround for writing INS/UMOV operands, as these require 
vector subregisters that we cannot represent yet. We only have to
    load/store lower single/double precision contents via them though,
    so reuse SUBMMS/SUBMMD for now.

git-svn-id: trunk@29955 -
 2015-02-23 22:55:14 +00:00
Jonas Maebe
bc5a33ffac * fixed flags_to_cond() and inverse_cond() for C_GE 
git-svn-id: trunk@29954 -
 2015-02-23 22:55:11 +00:00
Jonas Maebe
42aca4db46 * pass managed function result addresses as a hidden first parameter instead 
of in X8, as the ABI allows this deviation (see added comments) and it's
    required for some internal RTL code to call interface methods via RTTI

git-svn-id: trunk@29953 -
 2015-02-23 22:55:08 +00:00
Jonas Maebe
bb1274a8ee * make DoPipe() a method of TExternalAssembler and use it in 
TAArch64AppleAssembler to determine whether we have to add extra
    parameters to support assembling via a pipe ("as" is a symlink to
    clang for Arch64 on OS X, and needs to be explicitly told its going to
    receive assembler code from standard input)

git-svn-id: trunk@29952 -
 2015-02-23 22:55:05 +00:00
Jonas Maebe
3d01d4ceb1 * regenerated with Darwin/AArch64 support 
git-svn-id: trunk@29950 -
 2015-02-23 22:54:53 +00:00
Jonas Maebe
2c6e16e49e * accesses to the ISA pointer are guaranteed to be aligned -> ensure the 
code generator knows this

git-svn-id: trunk@29944 -
 2015-02-23 22:54:21 +00:00
Jonas Maebe
2340da410c * we cannot directly load the ISA pointer on AArch64, we have to call the 
class instance method
   (http://www.sealiesoftware.com/blog/archive/2013/09/24/objc_explain_Non-pointer_isa.html )

git-svn-id: trunk@29943 -
 2015-02-23 22:54:18 +00:00
Jonas Maebe
f1fb880f18 * fixed debug register values for vector registers 
git-svn-id: trunk@29942 -
 2015-02-23 22:54:15 +00:00
Jonas Maebe
e2d21a5716 * reuse simple memory paralocs, since we always setup a stack frame and 
we can address them via the frame pointer

git-svn-id: trunk@29941 -
 2015-02-23 22:54:12 +00:00
Jonas Maebe
a81e81c775 + override second_int_to_bool(), because the generic version assumes that 
OP_OR sets the flags

git-svn-id: trunk@29940 -
 2015-02-23 22:54:09 +00:00
Jonas Maebe
966a851997 + a_loadmm_intreg_reg() and a_loadmm_reg_intreg() implementations 
git-svn-id: trunk@29939 -
 2015-02-23 22:54:06 +00:00
Jonas Maebe
41fba0c4f7 * switched to using the stack pointer as base register for the temp allocator 
instead of the frame pointer register:
      1) we exactly know the offsets of the temps from the stack pointer
         after pass 1 (based on the require parameter stack size for called
         routines), while we don't know it for the frame pointer (it depends
         on the number of saved registers)
      2) temp offsets from the stack pointer are positive while those from
         the frame pointer are negative, and we can directly encode much
         bigger positive offsets in the instructions
   o move the stack pointer register to a virtual register in
     loadparentfpn, because many instructions cannot directly operate
     on/with the stack pointer
   o add the necessary register interference edges for the stack pointer
     register

git-svn-id: trunk@29938 -
 2015-02-23 22:54:03 +00:00
Jonas Maebe
5bb89cc2f0 * keep track of the reason why a loadparentfpnode has been created: to 
load a value from a nested context, or to pass a context to a nested
    routine. In case a target uses both the stack and frame pointer, this
    difference can matter.

git-svn-id: trunk@29937 -
 2015-02-23 22:53:59 +00:00
Jonas Maebe
f69d2fe61d * AArch64 doesn't use *stret routines for Objective-C dispatching, because it 
doesn't need special code for returning a struct by reference

git-svn-id: trunk@29936 -
 2015-02-23 22:53:56 +00:00
Jonas Maebe
f6e90ae9bf * include ncgobjc in cpunode for aarch64 
git-svn-id: trunk@29935 -
 2015-02-23 22:53:53 +00:00
Jonas Maebe
f54ea490dd * fixed writing extended registers in references 
git-svn-id: trunk@29933 -
 2015-02-23 22:53:47 +00:00
Jonas Maebe
7fc9d775df + support for @page and @pageoffs addressing on AArch64: these are PIC 
references that directly take the address of a symbol, rather than
    of its GOT entry
   o use these addressing modes to access local symbols

git-svn-id: trunk@29932 -
 2015-02-23 22:53:43 +00:00
Jonas Maebe
70fc5dcee3 * don't pass TP-style objects as HFA or similar, because the calling 
convention code is triggered before the object structure has been
    finalised, so the result can be different in the interface and
    implementation. To solve this, something like r20161 has to be
    implemented for TP-style objects

git-svn-id: trunk@29930 -
 2015-02-23 22:53:29 +00:00
Jonas Maebe
620af1732a * support for AArch64 branch forwarding 
o don't forward cb(n)z, as we don't check the maxiumum offsets
   o don't forward conditional branches to global labels, as they also have a
     limited range and at least the OS X toolchain doesn't support this

git-svn-id: trunk@29929 -
 2015-02-23 22:53:26 +00:00
Jonas Maebe
55bc5d7972 * completed TAoptBaseCpu.RegModifiedByInstruction() 
git-svn-id: trunk@29928 -
 2015-02-23 22:53:23 +00:00
Jonas Maebe
28a713494c * fixed/completed spilling_get_operation_type 
+ implemented spilling_get_operation_type_ref

git-svn-id: trunk@29926 -
 2015-02-23 22:53:17 +00:00
Jonas Maebe
5041354b8e * recognise AArch64 PIC references 
git-svn-id: trunk@29925 -
 2015-02-23 22:53:14 +00:00
Jonas Maebe
5da379ab3f + inlined support for round/trunc on AArch64 
git-svn-id: trunk@29923 -
 2015-02-23 22:53:07 +00:00
Jonas Maebe
0203f5e32f * search iphoneall rather than cocoaall for Cocoa types on Darwin/AArch64 
git-svn-id: trunk@29922 -
 2015-02-23 22:53:04 +00:00
Jonas Maebe
1aee714c36 * updated PIC comment for Darwin/AArch64 
git-svn-id: trunk@29921 -
 2015-02-23 22:53:01 +00:00
Jonas Maebe
572742e546 * use Darwin-style section syntax for Darwin/AArch64 
git-svn-id: trunk@29920 -
 2015-02-23 22:52:58 +00:00
Jonas Maebe
30b0f830c3 * fixed std_param_align 
o also updated copyright and architecture name

git-svn-id: trunk@29919 -
 2015-02-23 22:52:55 +00:00
Jonas Maebe
851b3499b4 * aarch64 only/always uses its vector fpu 
git-svn-id: trunk@29917 -
 2015-02-23 22:52:48 +00:00
Jonas Maebe
5fff543c73 - removed unused taicpu.insoffset field 
git-svn-id: trunk@29916 -
 2015-02-23 22:52:45 +00:00
Jonas Maebe
71bd4e23d9 + taicpu.op_reg_reg_const_const() and taicpu.op_reg_reg_reg_cond() methods 
git-svn-id: trunk@29915 -
 2015-02-23 22:52:42 +00:00
Jonas Maebe
5bdd14e252 + AArch64 cputarg 
git-svn-id: trunk@29914 -
 2015-02-23 22:52:39 +00:00
Jonas Maebe
558b8967b6 + Aarch64 assembler reader 
git-svn-id: trunk@29913 -
 2015-02-23 22:52:36 +00:00
Jonas Maebe
96fcf6a12d * added BL and CB(N)Z to is_calljmp() 
git-svn-id: trunk@29912 -
 2015-02-23 22:52:33 +00:00
Jonas Maebe
aa0e2e9170 * fixed cgsize2subreg and cgsize2subreg for mm subreg sizes 
git-svn-id: trunk@29911 -
 2015-02-23 22:52:30 +00:00
Jonas Maebe
c2b1ff41d5 - removed ARM leftover tspecialregflag type 
git-svn-id: trunk@29910 -
 2015-02-23 22:52:27 +00:00
Jonas Maebe
4c504098ca + C_CS/C_CC condition and F_HS/F_LO flag aliases 
git-svn-id: trunk@29909 -
 2015-02-23 22:52:23 +00:00
Jonas Maebe
6e55e8356e + IP0/IP1 register aliases 
git-svn-id: trunk@29908 -
 2015-02-23 22:52:20 +00:00
Jonas Maebe
bc9b870943 * cpunode: added ncgcal, corrected ncpuinln -> ncpuinl 
git-svn-id: trunk@29907 -
 2015-02-23 22:52:17 +00:00
Jonas Maebe
1c8059e30c + dummy AArch64 hlcgcpu 
git-svn-id: trunk@29906 -
 2015-02-23 22:52:14 +00:00
Jonas Maebe
cfd7bfc8a5 + dummy AArch64 cpupi unit 
git-svn-id: trunk@29905 -
 2015-02-23 22:52:11 +00:00
Jonas Maebe
d51c87e77b * add Darwin/AArch64 to systems_darwin 
git-svn-id: trunk@29904 -
 2015-02-23 22:52:07 +00:00
Jonas Maebe
b0112fa63f + initial AArch64 support in systemh.inc, options.pas, version.pas and 
psystem.pas

git-svn-id: trunk@29903 -
 2015-02-23 22:52:04 +00:00
Jonas Maebe
f2c0b88e4f * AArch64 compiler building support to top level and compiler Makefile 
git-svn-id: trunk@29901 -
 2015-02-23 22:51:58 +00:00
Jonas Maebe
60dd2d1dd5 + register Darwin/AArch64 target 
* pass "-arch arm64" to the linker for Darwin/AArch64

git-svn-id: trunk@29897 -
 2015-02-23 22:51:44 +00:00
Jonas Maebe
bee9c4185b + default target for AArch64 
git-svn-id: trunk@29896 -
 2015-02-23 22:51:41 +00:00
Jonas Maebe
00b819307a + condition and shifterop operands for AArch64 assembler reader 
git-svn-id: trunk@29895 -
 2015-02-23 22:51:38 +00:00
Jonas Maebe
665a5e509e * default iOS deployment version for AArch64 is iOS 7 
git-svn-id: trunk@29894 -
 2015-02-23 22:51:34 +00:00
Jonas Maebe
8c89f7ee8f * AArch64 also uses the non-fragile Objective-C ABI 
git-svn-id: trunk@29893 -
 2015-02-23 22:51:31 +00:00
Jonas Maebe
de2dd592ab + shiftedregmodes and extendedregmodes set constants 
git-svn-id: trunk@29892 -
 2015-02-23 22:51:28 +00:00
Jonas Maebe
17bcd207af * fixed lowercase entry in uppercond2str 
git-svn-id: trunk@29891 -
 2015-02-23 22:51:25 +00:00
Jonas Maebe
53548663e3 + AArch64 assembler writer 
git-svn-id: trunk@29890 -
 2015-02-23 22:51:22 +00:00
Jonas Maebe
d246ababff + condition code operand for aarch64 
+ taicpu.op_reg_cond() constructor
  * use this operand for cset

git-svn-id: trunk@29889 -
 2015-02-23 22:51:19 +00:00
Jonas Maebe
9c55fa6f6c + FPCR, FPSR and TPIDR registers 
git-svn-id: trunk@29878 -
 2015-02-23 22:50:44 +00:00
Jonas Maebe
493b11da06 + Darwin/aarch64 system definition 
git-svn-id: trunk@29877 -
 2015-02-23 22:50:41 +00:00
Jonas Maebe
b5643f2a27 + aarch64 cpu identifier 
git-svn-id: trunk@29876 -
 2015-02-23 22:50:38 +00:00
Jonas Maebe
6995474ea0 + aarch64 cpunode 
git-svn-id: trunk@29875 -
 2015-02-23 22:50:35 +00:00
Jonas Maebe
820de6a17b * pass "const" arrays always by reference, because there is a lot of hacky 
code out there that relies on this, even though it's not supposed to be
    guaranteed in any way :(

git-svn-id: trunk@29874 -
 2015-02-23 22:50:32 +00:00
Jonas Maebe
edb3124ae8 * copy arrays passed by value to cdecl routines on the callee side even on 
platforms that normally must copy value parameters on the caller side,
    because
   o if we copy them on the caller side, then the behaviour will be different
     compared to other platforms if the called routine is implemented in C
     (since they are passed by reference, changes to the array contents will
      be visible on other platforms, while they wouldn't be visible if the
      copy is made on the caller side)
   o if we don't copy them on the caller nor on the callee side, then behaviour
     will be different compared to other platforms if the called routine is
     implemented in Pascal and modifies the array (since then changes won't be
     visible on the caller side, except on platforms that normally perform
     the copy on the caller side)

git-svn-id: trunk@29873 -
 2015-02-23 22:50:29 +00:00
Jonas Maebe
2dbfbd40ab * unified hlcg.g_copyvalueparas and ncgutil.copyvalueparas, they have the 
same functionality in the mean time

git-svn-id: trunk@29872 -
 2015-02-23 22:50:26 +00:00
Jonas Maebe
6e12ad817f * don't make copies of arrays passed as value parameters to cdecl routines 
(since C compilers don't copy such arrays either, as they're implciitly
     pointers)

git-svn-id: trunk@29871 -
 2015-02-23 22:50:23 +00:00
Jonas Maebe
c6ba0bb6fb + support for targets that require by-reference value parameters to be 
be copied on the caller instead of callee side
   o mark Darwin/Aarch64 as such a target (any AArch64 target will be like
     that normally, as its ABI specifies this behaviour)
   o don't mark by-reference value parameters on such targets as
     vo_has_local_copy, since a) they don't have one (the copy is on the
     caller side), and b) this ensures that all code handling such
     parameters automatically knows that they are still by reference
     after the init code has run
   o when making the copies on the caller side, don't increase the
     reference count for managed types except for variants, just like
     is done when making the copies on the callee side. This is because
     the reference count increasing code on the callee side only runs
     for non-assembler functions, and we cannot know 100% certain on the
     caller side whether the called function is assembler or not (e.g. in
     case of externally declared functions)
   o maybe over time we can reuse the Pascal code in
     tcallparanode.copy_value_by_ref_para to replace the equivalent code
     in hlcgobj and ncgutil also on the caller side for other targets

git-svn-id: trunk@29870 -
 2015-02-23 22:50:20 +00:00
Jonas Maebe
bedcc5cbb1 + Darwin/AArch64 system identifier 
git-svn-id: trunk@29869 -
 2015-02-23 22:50:16 +00:00
Jonas Maebe
0af4f3e3fb + aarch64 math nodes implementation 
git-svn-id: trunk@29868 -
 2015-02-23 22:50:13 +00:00
Jonas Maebe
4aeb22f7f4 + initial implementation of aarch64 type conversion nodes 
git-svn-id: trunk@29867 -
 2015-02-23 22:50:10 +00:00
Jonas Maebe
71ae8015da + initial implementation of aarch64 add nodes 
git-svn-id: trunk@29866 -
 2015-02-23 22:50:07 +00:00
Jonas Maebe
8a444ed96a + initial implementation of aarch64 inline nodes 
git-svn-id: trunk@29865 -
 2015-02-23 22:50:04 +00:00
Jonas Maebe
7fa414f7f1 + initial implementation of aarch64 register allocator 
git-svn-id: trunk@29864 -
 2015-02-23 22:50:01 +00:00
Jonas Maebe
1c8eb58792 * adapted copyright statement 
git-svn-id: trunk@29863 -
 2015-02-23 22:49:58 +00:00
Jonas Maebe
2214966f26 * implemented parameter passing and function result locations 
git-svn-id: trunk@29862 -
 2015-02-23 22:49:55 +00:00
Jonas Maebe
edef6b2a0a + AArch64 Darwin abi declaration 
git-svn-id: trunk@29861 -
 2015-02-23 22:49:52 +00:00
Jonas Maebe
7fbc44471b * made getparaloc ABI-compliant 
git-svn-id: trunk@29860 -
 2015-02-23 22:49:49 +00:00
Jonas Maebe
232599b429 * made push_addr_param() ABI-compliant 
git-svn-id: trunk@29859 -
 2015-02-23 22:49:46 +00:00
Jonas Maebe
e2c30eb0a1 * return base type from is_hfa() 
git-svn-id: trunk@29858 -
 2015-02-23 22:49:43 +00:00
Jonas Maebe
e7bbb31080 * implemented ret_in_param() according to the ABI 
git-svn-id: trunk@29857 -
 2015-02-23 22:49:40 +00:00
Jonas Maebe
4562731ad9 - removed custom getintparaloc() 
git-svn-id: trunk@29856 -
 2015-02-23 22:49:37 +00:00
Jonas Maebe
60768fcffe + implemented is_hfa() 
git-svn-id: trunk@29855 -
 2015-02-23 22:49:34 +00:00
Jonas Maebe
c244daaafb * don't initialize the fpu register allocator, it is/must not be used on 
AArch64

git-svn-id: trunk@29854 -
 2015-02-23 22:49:31 +00:00
Jonas Maebe
ab186f7892 + initial implementation of aarch64 code generator 
git-svn-id: trunk@29853 -
 2015-02-23 22:49:28 +00:00
Jonas Maebe
01a6777530 * simplified flag_2_cond array range 
git-svn-id: trunk@29852 -
 2015-02-23 22:49:25 +00:00
Jonas Maebe
51a094a917 + FP/LR register aliases 
git-svn-id: trunk@29851 -
 2015-02-23 22:49:22 +00:00
Jonas Maebe
e6d7c6a62a + is_shifter_const() function to determine whether a constant can be encoded 
as an immediate operand for a logical operation

git-svn-id: trunk@29850 -
 2015-02-23 22:49:19 +00:00
Jonas Maebe
d23299af1e * fixed cgsize2subreg() for integer registers (we can use 32 and 64 bit 
registers -> 64 bit values go into 64 bit registers and everything else into
    32 bit ones)

git-svn-id: trunk@29849 -
 2015-02-23 22:49:15 +00:00
Jonas Maebe
f4ada6cda0 + aarch64-specific treference fields 
git-svn-id: trunk@29848 -
 2015-02-23 22:49:13 +00:00
Jonas Maebe
123503a4f7 + GOT-related (Darwin) address modes 
git-svn-id: trunk@29847 -
 2015-02-23 22:49:10 +00:00
Jonas Maebe
0ce5a1de7f + IsSimpleRef() function that can be used to determine whether a reference 
is a valid AArch64 memory reference or not. It returns an enum that says
    it's either valid, invalid, or invalid in a way that should trigger an
    internal error in case we created the reference ourselves (as opposed to
    it coming in via the assembler reader)

git-svn-id: trunk@29846 -
 2015-02-23 22:49:07 +00:00
Jonas Maebe
84cea76f7a * ARM64 -> AArch64 
git-svn-id: trunk@29845 -
 2015-02-23 22:49:04 +00:00
Jonas Maebe
19d47610ad * added error checking to spilling_create_load/store 
git-svn-id: trunk@29844 -
 2015-02-23 22:49:01 +00:00
Jonas Maebe
92b644a19f + taicpu.op_reg_reg_const_shifterop() 
* made shifterop parameter of taicpu.op_reg_reg_reg_shifterop() const

git-svn-id: trunk@29843 -
 2015-02-23 22:48:58 +00:00
Jonas Maebe
f5950ac4cd + added remaining aarch64 shift/extension modes 
git-svn-id: trunk@29842 -
 2015-02-23 22:48:55 +00:00
Jonas Maebe
24d3b1cdf6 + tcgsizep2size[] to convert a tcgsize to its power-of-2 bytesize 
git-svn-id: trunk@29841 -
 2015-02-23 22:48:52 +00:00
Jonas Maebe
a49d386541 + SW postfix for sign extending a 32 bit integer 
git-svn-id: trunk@29840 -
 2015-02-23 22:48:49 +00:00
Jonas Maebe
ff61aba3fe * there is no ROR shiftmode on AArch64 
git-svn-id: trunk@29839 -
 2015-02-23 22:48:46 +00:00
Jonas Maebe
cd6e3d5622 * fixed another compilation error 
git-svn-id: trunk@29838 -
 2015-02-23 22:48:43 +00:00
Jonas Maebe
544c8d34ec + taicpu.op_reg_const_shifterop() constructor 
git-svn-id: trunk@29837 -
 2015-02-23 22:48:40 +00:00
Jonas Maebe
3cfa91a249 * enable cgsize2subreg() and cgsize2subreg() to differentiate between 32 and 
64 bit integer registers

git-svn-id: trunk@29836 -
 2015-02-23 22:48:37 +00:00
Jonas Maebe
df8b644190 - removed some ARM leftovers 
git-svn-id: trunk@29835 -
 2015-02-23 22:48:33 +00:00
Jonas Maebe
fc235c9400 * fixed warnings 
git-svn-id: trunk@29834 -
 2015-02-23 22:48:30 +00:00
Jonas Maebe
76d779b648 * added SW suffix 
git-svn-id: trunk@29833 -
 2015-02-23 22:48:27 +00:00
Jonas Maebe
c0548cadb0 * added some missing instructions and aliases, reordered them according 
to how they appear in the ARM ARM
   o in particular, added the "unscaled offset" variants of instructions as
     separate opcodes (seems simpler than handling them as postfixes)

git-svn-id: trunk@29832 -
 2015-02-23 22:48:24 +00:00
Jonas Maebe
f1b619a942 * made (X|W)ZR and (W)SP separate registers, because a number of 
instructions can use either depending on the encoding

git-svn-id: trunk@29831 -
 2015-02-23 22:48:21 +00:00
Jonas Maebe
e8c9b94147 * X29 is callee-saved 
git-svn-id: trunk@29830 -
 2015-02-23 22:48:18 +00:00
Jonas Maebe
67c9d60b72 * factored out the check regarding whether the index of a vecn needs to be 
loaded into a register because of its size
   o by default, also allow if the size is the same as OS_ADDR but with a
     different sign

git-svn-id: trunk@29827 -
 2015-02-23 22:48:08 +00:00
Jonas Maebe
73a9dc8077 * fixed formatting 
git-svn-id: trunk@29826 -
 2015-02-23 22:48:05 +00:00
Jonas Maebe
f886fe49d6 * only prepend the sysroot path to -Fl paths if these paths are absolute; 
if they are not, they should be interpreted relative to the current
    directory instead (solves compiling tests that depend on a library
    when cross-testing for a platform with a sysroot)

git-svn-id: trunk@29825 -
 2015-02-23 22:48:02 +00:00
Jonas Maebe
bc3d495285 * extra error checking in gen_load_cgpara_loc 
git-svn-id: trunk@29824 -
 2015-02-23 22:47:59 +00:00
Jonas Maebe
3fe0bd065e * ARM assembler reader: don't check for postfixes beyond the length 
of the opcode

git-svn-id: trunk@29823 -
 2015-02-23 22:47:56 +00:00
Jonas Maebe
1dd5f579e6 * support record fields with the same name as registers in subscripts in 
the assembler reader

git-svn-id: trunk@29822 -
 2015-02-23 22:47:53 +00:00
Jonas Maebe
504a9d1594 * fixed register sizes in a_load_ref_reg_unaligned() 
git-svn-id: trunk@29821 -
 2015-02-23 22:47:50 +00:00
Jonas Maebe
290e3f5ff8 * fixed assigning one float loc_reference to another of a different size if a 
platform does not support (hl)cg.a_loadfpu_*, but only (hl)cg.a_loadmm_*

git-svn-id: trunk@29820 -
 2015-02-23 22:47:47 +00:00
Jonas Maebe
c764826bb3 * check whether we are explicitly typecasting untypedpointer^ before other 
cases (that depend on the sizes of the typecast matching) so that we
    can (internally) generate typecasts to open array/array of const

git-svn-id: trunk@29819 -
 2015-02-23 22:47:44 +00:00
Jonas Maebe
aae879d28a * also set aktcallnode during tcallnode.pass_1, so that the callparanode 
pass_1 code can make use of it

git-svn-id: trunk@29818 -
 2015-02-23 22:47:41 +00:00
Jonas Maebe
863e81315e * changed goto usage in tcallnode.pass_typecheck into try/finally (and at 
the same time solved some issues where "exit" instead of "goto errorexit"
    was used)

git-svn-id: trunk@29817 -
 2015-02-23 22:47:38 +00:00
Jonas Maebe
68303b8df1 * add explicit check that argument of sizeof(x) is a simple load node (this 
was already assumed later on)

git-svn-id: trunk@29816 -
 2015-02-23 22:47:35 +00:00
Jonas Maebe
b8c8ef8489 * support transfering a cgpara that is stored in multiple paralocs into a 
single register, by first storing everything consecutively to memory and
    then loading it. Required for "homogeneous float aggregates" on AArch64

git-svn-id: trunk@29815 -
 2015-02-23 22:47:32 +00:00
Jonas Maebe
e6511bcb33 * support multiple paralocs in a_load_reg_cgpara() by taking a round trip 
via a_load_ref_cgpara()

git-svn-id: trunk@29814 -
 2015-02-23 22:47:29 +00:00
florian
35d550c9c6 + commented directive for easier enabling 
git-svn-id: trunk@29812 -
 2015-02-23 22:11:26 +00:00
florian
a5bef2a2b9 * prevent new definitions being added to a global unit symtable after compilation of the implementation part started 
git-svn-id: trunk@29811 -
 2015-02-23 22:10:46 +00:00
florian
bc672e8fb2 * do not mess with the global symbol when compiling the implementation part of a unit 
git-svn-id: trunk@29810 -
 2015-02-23 22:09:46 +00:00
Jonas Maebe
1a949eae1f * fixed i8086 compilation 
git-svn-id: trunk@29807 -
 2015-02-23 08:51:59 +00:00
Károly Balogh
d85b49755c m68k: some ideas for a future second_abs_long implementation. no functional changes. 
git-svn-id: trunk@29806 -
 2015-02-23 03:37:51 +00:00
Károly Balogh
c72f58bcc5 m68k: implemented sqrt_real and abs_real inlines 
git-svn-id: trunk@29805 -
 2015-02-23 02:41:33 +00:00
Károly Balogh
db50666b3b n68kinl: added inline nodes for 68k. implemented first_sqr_real and second_sqr_real for 68881 FPU 
git-svn-id: trunk@29799 -
 2015-02-22 23:11:16 +00:00
florian
78742c3c87 * use armv4 as default, FPC never generated valid armv3 code 
git-svn-id: trunk@29794 -
 2015-02-22 20:53:48 +00:00
florian
00d41dc9fa * return false, if tdosinputfile.fileopen cannot open a file for reading 
git-svn-id: trunk@29792 -
 2015-02-22 15:52:54 +00:00
florian
cab98a58d0 + parameters of nested procedure variable parameters are taken into account during name mangling, resolves #27515 
* de-duplicated parameter name mangling code

git-svn-id: trunk@29791 -
 2015-02-22 09:16:20 +00:00
florian
9eab90d8c4 * always pass the architecture to the arm assembler 
* encode pld/ldrd in arm.inc using .long, so it causes no errors with older architectures settings of the assembler

git-svn-id: trunk@29780 -
 2015-02-21 21:58:30 +00:00
florian
80cc09e350 o fix wince compilation: 
* apparently, wince (or GNU AS for it) does not support blx imm
  * set FPC_IN_SYSTEM directive, so assembler code in divide.inc is compiled right

git-svn-id: trunk@29779 -
 2015-02-21 21:55:56 +00:00
florian
5946328ed6 * CPUs not having CMOV apparently do not support the newly introduced Multibyte NOPs (Agner, Optimizing subroutines in assembly 
language, An optimization guide for x86 platforms, Page 87), so restored the 32 Bit part of the old alignment 
  bytes for use on those old CPUs and use it depending on the CPU switches

git-svn-id: trunk@29777 -
 2015-02-21 20:50:42 +00:00
florian
de17e9fa1c + cpu capability CPUX86_HAS_CMOV 
git-svn-id: trunk@29776 -
 2015-02-21 20:47:40 +00:00
florian
255c4feef6 * new code alignment fillings based on the discussion at http://www.lazarusforum.de/viewtopic.php?f=10&t=8487 
git-svn-id: trunk@29772 -
 2015-02-21 10:09:39 +00:00
florian
ba3f0281b0 * fix compilation errors with -O2 due to dfa based warnings 
git-svn-id: trunk@29771 -
 2015-02-21 10:09:37 +00:00
svenbarth
d849734acd Specialize the correct generic if it has the same name as another generic that is located at another place of the type hierarchy. 
pgenutils.pas:
  + generate_generic_name: new parameter owner_hierarchy to further influence the specialization name
  * generate_specialization: pass the ownerhierarchy name to generate_generic_name
pdecsub.pas, parse_proc_head:
  * parse_generic_interface: for now pass '' to the hierarchy parameter; it needs to be checked whether Delphi allows nested interfaces there as well

git-svn-id: trunk@29769 -
 2015-02-20 16:28:44 +00:00
svenbarth
5a344ee263 Rework the way how "specialize" is handled. Instead of initializing the specialization of a full type declaration (including unit name and parent types) it is now considered part of the specialized type itself. This means that for example the following code: 
type
  TTestLongInt = specialize SomeOtherUnit.TTest<LongInt>;

will now have to read as

type
  TTestLongInt = SomeOtherUnit.specialize TTest<LongInt>;

While this is not backwards compatible this situation should arise seldomly enough and the benefits especially in context with generic functions/procedures/methods outway the drawbacks.

pbase.pas:
  * try_consume_unitsym: add a allow_specialize parameter that allows to parse "specialize" in front of a non-unit symbol; whether it was a specialization or not is reported using a new is_specialize parameter
  + add a new overload try_consume_unitsym_no_specialize that calls try_consume_unit sym with allow_specialize=false and a dummy is_specialize parameter
  * switch calls to try_consume_unitsym to try_consume_unitsym_no_specialize
pstatmnt.pas, try_statement:
  * switch call to try_consume_unitsym to try_consume_unitsym_no_specialize
  * adjust call to parse_nested_types
ptype.pas:
  + extend id_type with the possibility to disallow unit symbols (needed if a specialize was already parsed) and to report whether a specialize was parsed
  + extend parse_nested_types with the possibility to tell it whether specializations are allowed
  * have parse_nested_types specialize generic defs if one is encountered and local type defs are allowed
  * id_type: only allow "unitsym.specialize sym" or "specialize sym", but not "specialize unitsym.sym"
  * single_type: correctly handle specializations with "specialize" keyword
  * read_named_type.expr_type: there is no longer a need to check for "specialize" keyword
pexpr.pas:
  + new function handle_specialize_inline_specialization which tries to specialize a type symbol
  * handle_factor_typenode: handle specializations after a point that follows a record or object (why isn't this part of postfixoperators anyway? O.o)
  * postfixoperators: handle "specialize" after records and objectdefs
  * factor_read_id: handle "specialize" in front of an identifier (and after unit symbols)

+ added tests
* adjusted test webtbs/tw16090.pp

git-svn-id: trunk@29768 -
 2015-02-20 16:23:40 +00:00
Károly Balogh
c421f4d91e do not include outputexedir in linker resname. it causes the path to be double-included when cross-compiling with -st and -FE on Liunuxes. also, be consistent between linking on host and target. 
git-svn-id: trunk@29753 -
 2015-02-18 14:09:15 +00:00
pierre
6ce8aeed27 + mips/mipsel libraries need PIC 
git-svn-id: trunk@29752 -
 2015-02-18 13:53:20 +00:00
florian
d6e4af8279 + applied remaining patches of Torsten Grundke: adds gather instructions of avx2 
git-svn-id: trunk@29745 -
 2015-02-17 21:43:46 +00:00
florian
726a78be08 * fix possible range check errors, resolves #27461 
git-svn-id: trunk@29743 -
 2015-02-17 21:14:47 +00:00
florian
0dfbba86f3 * properly firstpass nodes when inc(...) is converted into ...:=...+x, resolves #27471 
git-svn-id: trunk@29742 -
 2015-02-17 21:14:45 +00:00
florian
aafec52a52 * allow multiple passes on as nodes 
git-svn-id: trunk@29741 -
 2015-02-17 21:14:43 +00:00
Károly Balogh
3650991d90 m68k: support LOC_REFERENCE/LOC_CREFERENCE for floating point instructions where possible 
git-svn-id: trunk@29714 -
 2015-02-15 20:15:21 +00:00
pierre
af883cdcb7 Avoid warning for function not set if noreturn modifier is used 
git-svn-id: trunk@29710 -
 2015-02-15 18:01:22 +00:00
Károly Balogh
fdb09365ba m68k: instead of trying to mess with LOC_FLAGS, use LOC_REGISTER to return a boolean result of the compare. idea by Florian. fixes several issues with optimizer enabled and also results in better code in some cases. 
git-svn-id: trunk@29709 -
 2015-02-15 17:15:15 +00:00
Jeppe Johansen
47dbec3111 Rebase to trunk revision 
git-svn-id: branches/laksen/armiw@29708 -
 2015-02-15 16:08:18 +00:00
Jeppe Johansen
55fe3b4e77 Disable internal linker in arm-linux for now. 
git-svn-id: branches/laksen/armiw@29707 -
 2015-02-15 15:44:10 +00:00
Károly Balogh
9693b37c1a m68k: actually,overwrite the resflags with F_E, because only that would be valid after the FSxx/TST combination 
git-svn-id: trunk@29705 -
 2015-02-15 14:04:11 +00:00
Károly Balogh
c062e55aa2 m68k: after a compare on the FPU, move the condition flags back to the CPU. this should make floating point compare actually working 
git-svn-id: trunk@29704 -
 2015-02-15 13:41:40 +00:00