paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-09-10 22:09:18 +02:00
Code Issues Packages Projects Releases Wiki Activity
53,640 Commits 101 Branches 76 Tags 313 MiB
f077c7d950
Commit Graph

53640 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
svenbarth
f077c7d950 + add support for Unicode code point constants > $FFFF; they are converted to a surrogate pair so they are in 
fact a UnicodeString constant
+ added tests

git-svn-id: trunk@39123 -
 2018-05-27 14:06:19 +00:00
pierre
c64455cfd4 Do not assume that size of integer type is 4 (it's 2 for msdos) 
git-svn-id: trunk@39122 -
 2018-05-26 06:04:57 +00:00
pierre
a1bbd4c02d Try to avoid inserting all RTTI symbols into executables on targets not supporting smartlinked sections 
git-svn-id: trunk@39121 -
 2018-05-25 20:44:24 +00:00
nickysn
ad96cf758b + emit a DW_AT_address_class dwarf attribute on i8086 for procdefs, indicating 
their call model (near or far)

git-svn-id: trunk@39120 -
 2018-05-25 20:03:15 +00:00
svenbarth
199b5809a3 * optimize "dynarr := dynarr + [elem]" to "Insert(elem, dynarr, High(SizeInt))" and "dynarr := [elem] + dynarr" to "Insert(elem, dynarr, 0)" (we need to do this in the typecheck of taddnode as otherwise the array constructor is already converted) 
+ added test
* adjusted test for Mantis #30463 as p1 triggers the optimization as well

git-svn-id: trunk@39119 -
 2018-05-25 16:03:56 +00:00
svenbarth
85439a0fa0 * also check for nf_assign_done_in_right flag after typechecking the right side 
git-svn-id: trunk@39118 -
 2018-05-25 15:54:41 +00:00
nickysn
22e8763100 + generate script code for creating the dwarf exe sections when using the 
internal linker with dwarf debug info enabled on i8086-msdos

git-svn-id: trunk@39117 -
 2018-05-25 15:35:17 +00:00
pierre
4efb978858 Fix wrong preprocessor: $undefine is not accepted, must be $undef 
git-svn-id: trunk@39116 -
 2018-05-25 14:51:13 +00:00
nickysn
f505718e6c + started implementing TMZExeOutput.writeDebugElf; nothing is written yet, only 
the elf header fields are filled with the values, generated by wlink

git-svn-id: trunk@39115 -
 2018-05-25 14:20:13 +00:00
nickysn
b18602485a * fixed base segment calculation on i8086 (code moved after call to OpenDwarf, 
because it needs to have e.processsegment initialized)

git-svn-id: trunk@39114 -
 2018-05-25 00:52:14 +00:00
nickysn
45bcbca71b * fixed reading of DW_FORM_ref_addr on i8086 
git-svn-id: trunk@39113 -
 2018-05-25 00:21:53 +00:00
nickysn
6bd99b65bf + emit DW_AT_frame_base dwarf attributes on i8086. Even though the Watcom 
Debugger doesn't require them (it uses BP as the frame register by default),
  the dwarf spec requires that we emit them on i8086, because we also use
  DW_OP_fbreg on i8086.

git-svn-id: trunk@39112 -
 2018-05-24 22:49:12 +00:00
nickysn
c5fe78201c + emit DW_OP_bregx dwarf ops for local variables and parameters, relative to BP 
on i8086. This fixes Watcom Debugger not showing them right in memory models,
  where DS<>SS (Watcom seems to ignore the DW_AT_segment attribute for register
  relative variables, even though we emit it correctly :/ )

git-svn-id: trunk@39111 -
 2018-05-24 22:27:11 +00:00
nickysn
855c810c09 + emit a DW_AT_WATCOM_memory_model dwarf attribute, when compiling for i8086 
git-svn-id: trunk@39110 -
 2018-05-24 21:59:50 +00:00
nickysn
87e1fa8fc7 + added enum with the values for DW_AT_WATCOM_memory_model 
git-svn-id: trunk@39109 -
 2018-05-24 21:50:55 +00:00
nickysn
dc12d4d789 + added constants with the watcom dwarf attribute extensions 
git-svn-id: trunk@39108 -
 2018-05-24 21:49:28 +00:00
nickysn
749872aaf7 + generate DW_AT_segment dwarf attribute for local variables and parameters on i8086 
git-svn-id: trunk@39107 -
 2018-05-24 21:30:37 +00:00
nickysn
217c75f84e + generate dwarf DW_AT_address_class attributes for pointer types on i8086 
git-svn-id: trunk@39106 -
 2018-05-24 20:22:59 +00:00
svenbarth
b94a8c93a5 * test (for Mantis #5516; 3531 is the old bug ID) is valid now that we have dynamic array initializers 
git-svn-id: trunk@39105 -
 2018-05-24 19:56:35 +00:00
nickysn
76fb8c3afd * fix for dwarf debug info generation of localsym vars with offset (thx Pierre for spotting this) 
git-svn-id: trunk@39104 -
 2018-05-24 16:31:01 +00:00
nickysn
69fbf0e0d8 * use unaligned constants, when writing the final dwarf arange record on i8086 
git-svn-id: trunk@39103 -
 2018-05-24 15:43:02 +00:00
pierre
a5a6b70658 Add tf_no_pic_supported to go32v2 target (was generating wrong code) 
git-svn-id: trunk@39102 -
 2018-05-23 21:12:56 +00:00
pierre
f80e73eac2 Remove references to sym variable when not set (revealed warning generated when compiling with -O4 option) 
git-svn-id: trunk@39101 -
 2018-05-23 20:45:21 +00:00
pierre
019da9517f Set return value when calling inherited in ti8086inlinenode.pass_typecheck_cpu 
git-svn-id: trunk@39100 -
 2018-05-23 20:43:40 +00:00
nickysn
5417eb043d * fix for the encoding of dwarf debug info of local variables, whose base is a 
register that has a dwarf register number higher than 31

git-svn-id: trunk@39099 -
 2018-05-23 15:58:51 +00:00
nickysn
e5de7d0eac * set all 16/32-bit specific stabs/dwarf register numbers of x86_64-only 
registers to -1 in x86reg.dat. The values that used to be there weren't used
  at all (most were just copies of the 32-bit version of the register). This can
  be easily demonstrated by the fact that running 'make regdat' in the compiler
  directory doesn't change any of the generated files for i8086/i386/x86_64.

git-svn-id: trunk@39098 -
 2018-05-23 15:29:33 +00:00
nickysn
a275d15da2 + added an i8086-specific column in x86reg.dat for the dwarf register numbers 
and filled it with the dwarf register mapping, used by Open Watcom (Watcom
  also uses this mapping on i386, but we don't need to support their debugger on
  i386 for now)

git-svn-id: trunk@39097 -
 2018-05-23 15:12:58 +00:00
nickysn
6f44deb557 + implemented the -Xg option for putting the debug information in a separate 
file on i8086-msdos and i8086-win16

git-svn-id: trunk@39096 -
 2018-05-22 15:33:26 +00:00
nickysn
0a01c0c76f * autoswitch to writing 32-bit SEGDEF OMF records if there's a need to encode a 
32-bit segment length

git-svn-id: trunk@39095 -
 2018-05-22 14:03:59 +00:00
nickysn
70be0dc6e9 * changed internal error number to make it unique 
git-svn-id: trunk@39094 -
 2018-05-22 13:49:40 +00:00
nickysn
8c842c1337 * changed the type of the 'size' parameter of TOmfObjOutput.AddSegment to 
TObjSectionOfs instead of aword

git-svn-id: trunk@39093 -
 2018-05-22 13:48:55 +00:00
nickysn
f5862efcd4 + support writing 32-bit LEDATA OMF records 
git-svn-id: trunk@39092 -
 2018-05-22 13:23:56 +00:00
nickysn
99f2f3d4b6 * changed the TObjRelocation.DataOffset and orgsize types to TObjSectionOfs instead of aword 
git-svn-id: trunk@39091 -
 2018-05-22 13:17:35 +00:00
nickysn
dcdfecbf35 * changed internal error number to make it unique 
git-svn-id: trunk@39090 -
 2018-05-22 12:55:53 +00:00
nickysn
a656d98657 * autoswitch to writing 32-bit PUBDEF OMF records if there's a need to encode a 
32-bit offset

git-svn-id: trunk@39089 -
 2018-05-22 12:54:11 +00:00
nickysn
ca50b14e21 * set the omf section size limit to 32-bit for the dwarf debug sections 
git-svn-id: trunk@39088 -
 2018-05-22 12:41:51 +00:00
nickysn
065a7b63d9 * changed the TObjSectionOfs type to be longword on i8086 
+ introduced the TObjSection.SizeLimit property. This will allow setting
  different size limits for different sections (e.g. 32-bit for the dwarf
  sections, 16-bit for the regular code and data sections on i8086).

git-svn-id: trunk@39087 -
 2018-05-22 09:48:28 +00:00
nickysn
584635c591 + introduce align_objsecofs; use it to align integers of type TObjSectionOfs 
git-svn-id: trunk@39086 -
 2018-05-22 00:20:21 +00:00
nickysn
f9748e928b * also changed the type of the 'DataPos' parameter of TObjData.layoutsections to TObjSectionOfs 
git-svn-id: trunk@39085 -
 2018-05-21 16:48:56 +00:00
nickysn
04129b8c6e + introduce the TObjSectionOfs type - an unsigned integer, representing an 
offset in an obj file section on the current platform; equal to PUInt for now,
  but will probably change to LongWord on i8086, in order to support 32-bit
  sections (needed for dwarf debug info or when linking with 32-bit object files
  in a mixed 16-bit/32-bit application, while using the internal linker)

git-svn-id: trunk@39084 -
 2018-05-21 16:31:50 +00:00
florian
8f472d5212 * check for more controll flow statements before doing loop unrolling 
git-svn-id: trunk@39083 -
 2018-05-21 15:46:02 +00:00
nickysn
308e439d5d + emit segment information in the dwarf debug info for static variables on i8086 
git-svn-id: trunk@39082 -
 2018-05-21 15:30:58 +00:00
nickysn
76ff01beda + also emit the non-standard DW_LNE_set_segment watcom extended opcode in the 
dwarf line info on i8086

git-svn-id: trunk@39081 -
 2018-05-21 14:54:11 +00:00
nickysn
9d167dbd2a + emit DW_AT_segment attributes in the dwarf debug info for subroutines 
(DW_TAG_subprogram) on i8086

git-svn-id: trunk@39080 -
 2018-05-21 14:40:39 +00:00
svenbarth
4354821f05 * regenerate msg* files after r39078 
git-svn-id: trunk@39079 -
 2018-05-21 08:14:21 +00:00
svenbarth
b7cc225149 * revert r39043 and r39045 which disallow static array constants inside dynamic constants for Delphi modes 
git-svn-id: trunk@39078 -
 2018-05-21 08:13:17 +00:00
nickysn
4aa66d9033 * emit a 32-bit offset for DW_FORM_ref_addr on i8086 (Watcom compatibility) 
git-svn-id: trunk@39077 -
 2018-05-21 01:36:23 +00:00
nickysn
2a5fbd4598 * continue reading the segment when walking the dwarf arange tuples on i8086 
git-svn-id: trunk@39076 -
 2018-05-20 22:51:52 +00:00
nickysn
d1cdd8be7c * do not emit padding bytes after the dwarf arange header on i8086 for 
compatibility with Open Watcom (also, the dwarf spec is not very clear about
  the alignment requirements on segmented architectures, so that's why we do
  this by default, instead of putting it behind a -go debug switch)

git-svn-id: trunk@39075 -
 2018-05-20 22:50:46 +00:00
nickysn
1a2f364e68 * use 32-bit address_size in the dwarf arange sections on i8086 (for Open Watcom 
compatibility)

git-svn-id: trunk@39074 -
 2018-05-20 21:59:32 +00:00