paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-12-04 14:47:17 +01:00
Code Issues Packages Projects Releases Wiki Activity
53,744 Commits 102 Branches 76 Tags 307 MiB
84b5265312
Commit Graph

21143 Commits

Author SHA1 Message Date
nickysn
84b5265312 + append a TIS trailer after the debug ELF (this helps the watcom debugger find 
the beginning of the ELF debug image)

git-svn-id: trunk@39229 -
 2018-06-15 13:41:40 +00:00
nickysn
480eedac9f * fix/pretty print the segment and group list with too long segment/group/class 
names in the linker map file, produced by the i8086 internal linker

git-svn-id: trunk@39228 -
 2018-06-15 00:03:58 +00:00
nickysn
8af56ad6c8 * fixed silly (caused by copy/paste) bug when writing elf section contents in the dwarf debug binary, produced by TMZExeOutput.writeDebugElf 
git-svn-id: trunk@39227 -
 2018-06-14 16:29:25 +00:00
nickysn
11cf2a6996 * emit a warning, when encountering an omf reference to a section, that has been 
removed due to smartlinking

git-svn-id: trunk@39226 -
 2018-06-14 16:22:47 +00:00
nickysn
c41f736958 + allow MemPos_EndExeSection to be called with an empty section name 
git-svn-id: trunk@39225 -
 2018-06-14 16:17:00 +00:00
nickysn
97365abbcf + set the oso_debug flag to dwarf MZ exe sections 
git-svn-id: trunk@39224 -
 2018-06-14 16:15:11 +00:00
nickysn
239fb25179 + assign an MZExeUnifiedLogicalSegment object to the OMF dwarf sections, so that 
OMF relocations work on them too

git-svn-id: trunk@39223 -
 2018-06-14 16:03:05 +00:00
nickysn
7103e3cb5c + set the oso_debug section flags when reading an OMF section, which belongs to 
the 'DWARF' class

git-svn-id: trunk@39222 -
 2018-06-14 15:27:08 +00:00
nickysn
b7bfac11cb + enable using the i8086 internal linker with debug info, if the compiler is 
compiled with -di8086_link_intern_debuginfo

git-svn-id: trunk@39221 -
 2018-06-14 15:24:39 +00:00
nickysn
5b47bad32a * use uppercase object section names in the OMF linker script for matching the 
dwarf debug sections. This is because they are converted to upper case when
  reading (so that we treat omf section names as case-insensitive)

git-svn-id: trunk@39220 -
 2018-06-14 15:19:11 +00:00
nickysn
c56ebad183 + implemented writing of dwarf debug info (by calling writeDebugElf) in 
TMZExeOutput.writeData. Note that writing i8086 debug info with the internal
  linker is still disabled (i.e. it switches to external linking) in 
  options.TOption.checkoptionscompatibility

git-svn-id: trunk@39219 -
 2018-06-14 14:30:59 +00:00
nickysn
29b3ec3fdb * call MaybeSwapHeader and MaybeSwapSecHeader in TMZExeOutput.writeDebugElf, so 
the debug ELF file is generated with the correct endianness

git-svn-id: trunk@39218 -
 2018-06-14 13:02:54 +00:00
nickysn
b34ab6b279 * The MaybeSwapXXXHeader functions moved from ogelf to the elfbase unit, so they 
can be used by TMZExeOutput.writeDebugElf as well

git-svn-id: trunk@39217 -
 2018-06-14 13:01:08 +00:00
nickysn
daa47f4b7f + completed the internal debug ELF writer in TMZExeOutput.writeDebugElf (not 
enabled yet)

git-svn-id: trunk@39216 -
 2018-06-13 16:08:54 +00:00
svenbarth
be0d51d64c * allow $MinEnumSize, $PackSet and $PackRecords to be used with $Push and $Pop 
+ added tests

git-svn-id: trunk@39215 -
 2018-06-12 19:40:45 +00:00
svenbarth
da672d1344 * instead of using separate booleans use a set to keep track of pending changes for popped settings 
git-svn-id: trunk@39214 -
 2018-06-12 19:40:41 +00:00
nickysn
b3e4ace4e3 * fixed compilation of .com files in the i8086 tiny memory model, when used with 
section based smartlinking

git-svn-id: trunk@39213 -
 2018-06-12 00:38:51 +00:00
nickysn
420a57ddc6 * use upper case OMF section names when using i8086 section-based smart linking 
git-svn-id: trunk@39212 -
 2018-06-12 00:14:08 +00:00
nickysn
b9ab9e5782 + initial (experimental) implementation of section-based smartlinking for i8086. 
Use -dI8086_SMARTLINK_SECTIONS to enable.

git-svn-id: trunk@39211 -
 2018-06-11 16:56:33 +00:00
nickysn
bfdd75eabf + support AB_LOCAL symbols in OMF object files, by implementing support for 
LPUBDEF/LPUBDEF32 OMF records

git-svn-id: trunk@39210 -
 2018-06-11 15:29:23 +00:00
nickysn
188e9ff64a * only read the first OMF record in TOmfLibObjectWriter.closefile, since we no 
longer copy the entire file there, so we parse the file only to get the module
  name, and the THEADR record should always be the first record in the OMF file.

git-svn-id: trunk@39198 -
 2018-06-08 14:21:18 +00:00
nickysn
3f5577077a * automatically create omf libraries with the smallest page size possible. This 
greatly reduces the size of i8086 snapshots by keeping the amount of padding
  in the library files to the minimum.

git-svn-id: trunk@39195 -
 2018-06-07 22:27:47 +00:00
nickysn
345713c1fa * refactored omf library writing, so that the whole library is written in 
TOmfLibObjectWrite.WriteLib, after determining the page size (currently still
  fixed to 512 bytes). Prerequisite for OMF lib page size optimization.

git-svn-id: trunk@39194 -
 2018-06-07 22:11:03 +00:00
nickysn
00c6826896 * moved the PageNum property of TOmfLibObjectWriter.TOmfLibDictionaryEntry to 
TOmfLibObjectWriter.TOmfLibObjectModule. Instead, store the module index (in
  FObjectModules) in TOmfLibDictionaryEntry. This is a prerequisite for
  implementing OMF library page size optimization.

git-svn-id: trunk@39193 -
 2018-06-07 18:32:50 +00:00
nickysn
4bcfe8dbb2 * refactored TOmfLibObjectWriter, so it keeps all object modules in a 
TFPObjectList. This will allow implementing omf library page size optimization
  later.

git-svn-id: trunk@39192 -
 2018-06-07 15:03:05 +00:00
nickysn
af5e6e182d * the TOmfLibDictionaryEntry class in unit owomflib converted to two identical 
nested classes inside TOmfLibObjectWriter and TOmfLibObjectReader. The reason
  is I'm planning to change the class, used in TOmfLibObjectWriter, but not the
  one in TOmfLibObjectReader. And using nested classes keeps things clean and
  avoids exposing implementation details in the interface of the owomflib unit.
* also changed 'private' to 'strict private' in the classes of the owomflib to
  reduce the risk of interference between TOmfLibObjectWriter and TOmfLibObjectReader.

git-svn-id: trunk@39191 -
 2018-06-07 12:24:36 +00:00
Károly Balogh
bbbaa39d79 ncginl: another approach to fix the same issue #39184 supposed to address. keep left.resultdef, but try to make the target def always ordinal of an equal size, so it ends up in a data register on m68k 
git-svn-id: trunk@39188 -
 2018-06-07 06:43:02 +00:00
Károly Balogh
3286623c3c ncginl: revert the previous commit, it seems to cause some issues on other CPUs 
git-svn-id: trunk@39186 -
 2018-06-07 06:21:46 +00:00
Károly Balogh
42efddc481 ncginl: in inc/dec, do not use left.resultdef as the dest to force a register location, because if left is a pointer, then an ordinal value might get forced into an address register on m68k, which results in a lot of unnecessary register shuffling to complete the operation 
git-svn-id: trunk@39184 -
 2018-06-07 06:04:19 +00:00
Károly Balogh
ddf8788c7a m68k: try to generate better code from some pointermath - when the left is an address register and we do add/sub, also allocate an address register as temp, also only require that left and right sizes are same, when right is not a const 
git-svn-id: trunk@39174 -
 2018-06-05 16:45:02 +00:00
nickysn
e6a8c058e0 * moved the handling of LIDATA OMF records (which currently only produces a 
'LIDATA records are not supported' error) to TOmfObjInput.ReadLeOrLiDataAndFixups

git-svn-id: trunk@39171 -
 2018-06-04 14:20:07 +00:00
nickysn
02fc2827e0 * TOmfObjInput.ReadLEDataAndFixups renamed ReadLeOrLiDataAndFixups 
git-svn-id: trunk@39170 -
 2018-06-04 14:10:11 +00:00
nickysn
e410a0aefb + support OMF FIXUPP records, containing only THREAD subrecords to occur without 
a preceding LEDATA or LIDATA record (this is allowed by the OMF spec)

git-svn-id: trunk@39169 -
 2018-06-04 14:00:36 +00:00
nickysn
002b116322 * use a case statement, instead of 'if', based on the OMF record type in 
TOmfObjInput.ReadLEDataAndFixups. This is to allow (in the future) handling of
  LIDATA records and FIXUPP records without preceding LEDATA/LIDATA. No
  functional changes in this commit.

git-svn-id: trunk@39166 -
 2018-06-04 13:11:55 +00:00
florian
21d785e41b * do not unroll loops if the counter variable is accessed outside the current scope 
git-svn-id: trunk@39163 -
 2018-06-03 15:12:48 +00:00
florian
3b3e33d863 + correctly set var state of addr and frame in raise statements, resolves #33818 
git-svn-id: trunk@39162 -
 2018-06-03 09:31:19 +00:00
florian
f03a396757 * target classic MacOS builds again 
git-svn-id: trunk@39156 -
 2018-05-31 19:31:36 +00:00
nickysn
98634899f2 + support OMF fixups, determined by thread 
git-svn-id: trunk@39151 -
 2018-05-30 15:49:27 +00:00
nickysn
2fb73de404 + support reading OMF files with THREAD FIXUPP subrecords; note that fixups, 
determined by thread are still not yet implemented

git-svn-id: trunk@39150 -
 2018-05-30 14:30:54 +00:00
nickysn
ebecfb1ce9 + added a FFixupThreads property with a TOmfThreads object to TOmfObjInput 
git-svn-id: trunk@39149 -
 2018-05-30 14:10:12 +00:00
nickysn
189cba817e + added method TOmfSubRecord_THREAD.ApplyTo, which applies the data in the 
current thread subrecord to a TOmfThreads object

git-svn-id: trunk@39148 -
 2018-05-30 13:52:34 +00:00
nickysn
26e808e155 + added classes for maintaining the OMF fixup thread state 
git-svn-id: trunk@39147 -
 2018-05-30 13:35:52 +00:00
nickysn
49b414ffee + support ugly constructs like 'DD BYTE PTR 5' in the x86 intel syntax inline 
assembler; these are TP7 and Delphi compatible

git-svn-id: trunk@39143 -
 2018-05-29 15:43:08 +00:00
nickysn
54a8d8de58 + added class for decoding/encoding THREAD OMF subrecords 
git-svn-id: trunk@39142 -
 2018-05-29 15:05:07 +00:00
pierre
b1537a7f40 Add some more debug information to temp-generation when compiled with -dEXTDEBUG 
git-svn-id: trunk@39141 -
 2018-05-28 22:39:21 +00:00
nickysn
052da5c921 * use DW_OP_reg0..DW_OP_reg31 instead of DW_OP_regx in TDebugInfoDwarf.append_seg_reg 
in case the dwarf register number is 31 or smaller

git-svn-id: trunk@39131 -
 2018-05-28 14:21:53 +00:00
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
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