940 lines
31 KiB
Plaintext
940 lines
31 KiB
Plaintext
07-Jan-1996 BryanT
|
||
Has it been this long? Merge in the speed changes from SteveSm for faster UDT lookups.
|
||
Rename all the VC sourcecode to *.vc. We'll eliminate them all together shortly.
|
||
Rename the NT .cxx files to *.cpp.
|
||
Switch to the NT build process.
|
||
Move the tracking for the PDATA and FPO into the EXE struct. This allows multiple
|
||
copies of the same image for different processes.
|
||
|
||
16-Jun-1995
|
||
loadomf.cpp
|
||
Query the registry for Pdb Dirs and pass information to mspdb?0.dll, so it
|
||
can search for pdbs along the path.
|
||
|
||
01-May-1995 SanjayS
|
||
shsymlb0.c
|
||
Don't search the publics in SHdNearestSymbol if the address passed in
|
||
lies within the range of the proc that we found.
|
||
|
||
25-Feb-1995 BryanT
|
||
Nuke all 16-bit files and build rules. Use the Dolphin/Crystal codebase if it's
|
||
interesting to you.
|
||
|
||
03-Nov-1994 BryanT
|
||
All *.cxx files, sources, shinc.hxx
|
||
Add support to build in NT build environment w/o effecting VC.
|
||
*.cxx files are built in NT, *.c are built for VC (will change
|
||
once it's stable). For NT code, remove FAR/PASCAL/LOADDS/NEAR
|
||
macros. Replace _ftcsxxx macros with _tcsxxx macros. Remove
|
||
need for shwin32.c by using CRITICAL_SECTION wrappers. Format
|
||
according to OSDebug4 standards.
|
||
|
||
23-Jun-1994 SanjayS
|
||
loadomf.cpp
|
||
Add support to read NT .dbg files.
|
||
|
||
18-May-1994 v-LeoL
|
||
loadomf.cpp inc\sapi.h
|
||
When the IDE preloads a DLL it stands the risk of NT moving the
|
||
loadaddress when loaded for real. This causes all the
|
||
IMAGE_RUNTIME_FUNCTION_ENTRY's to be invalid unless relocated.
|
||
So OLLoadOmf is changed to check for that problem and LoadPdata
|
||
stores the assumed loadaddress when first loaded.
|
||
|
||
18-May-1994 MarkBro
|
||
shsymbol.c:
|
||
Corrected CheckHandles. It assumed that psym->off was never zero.
|
||
Changed all refs from
|
||
GetAddrOff(paddr) == 0
|
||
to
|
||
GetAddrOff(paddr) < psym->off
|
||
|
||
Also corrected SHIsInProlog to correctly check for the end (limit)
|
||
when validating epilog.
|
||
|
||
29-Apr-1994 MarkBro
|
||
shinit.c:
|
||
Exported SHIsEmiLoaded.
|
||
|
||
29-Apr-1994 MarkBro
|
||
sh.c:
|
||
In SHAddDllExt, when a filename isn't found in the correct
|
||
list, look in the other list (exgDll vs. exgExe). If found
|
||
put in correct list. Solves problem with calling SHLoadDll
|
||
which automatically puts filename into dll list (doesn't
|
||
take an arg to specify)
|
||
|
||
25-Apr-1994 v-LeoL
|
||
shsymbol.c
|
||
Because MIPS uses nested procedures in C++EH and SEH do an
|
||
exhaustive search in SetBlks when PC is beyond DebugEnd of
|
||
top level procedure.
|
||
|
||
20-Apr-1994 SanjayS
|
||
sh.c
|
||
Store full path name for dlls we don't find on the host machine.
|
||
This is so we don't report errors during a restart.
|
||
|
||
19-Apr-1994 v-LeoL
|
||
shsymbol.c
|
||
Protect against GetSymbols failure and prep for S_LINK32
|
||
support
|
||
sl.c
|
||
Protect against FHOpen failure
|
||
sh.c
|
||
Protect SHGetHexeName against null Hexe
|
||
|
||
14-Apr-1994 MarkBro
|
||
sh.c:
|
||
When destroying an Exs node, make sure that the
|
||
find node succeeded before calling LLRemove
|
||
|
||
13-Apr-1994 v-LeoL
|
||
shsymbol.c
|
||
return FALSE in SHIsInProlog if addr beyond end of proc
|
||
|
||
1-Apr-1994 v-LeoL
|
||
sh.c shsymbol.c
|
||
Don't stop for S_ENDARG when looking for REGREL/BPREL
|
||
Look for REGREL in SHGetSymbol sopStack
|
||
|
||
30-Mar-1994 MarkBro
|
||
sl.c:
|
||
Big changes to SLFAddrFromLineExtended. Removed cache and
|
||
now will loop through entire source table to find multiple
|
||
occurances of a line w/i the table.
|
||
|
||
SLFLineToAddr now has its own cache for the previous query.
|
||
(SLCAddrFromLine already had a cache).
|
||
|
||
SLCAddrFromLine now calls SLFAFLExt with new arg for finding
|
||
the nth entry in the specified table
|
||
|
||
24-Mar-1994
|
||
loadomf.cpp
|
||
Under OSDEBUG4X don't exit OLStart if DEBUG_DIRECTORY==NULL
|
||
until after having checked/loaded pdata
|
||
|
||
22-Mar-1994 MarkBro
|
||
shinit.c shsymbol.c shiproto.h:
|
||
Added export of SHFindSymInExe
|
||
|
||
17-Mar-1994 MarkBro
|
||
Changes below to fix Dolphin 8654.
|
||
|
||
loadomf.c:
|
||
#ifdef'd out guts of LoadOmap since it will do an allocation
|
||
which isn't saved. Basicall, lost memory
|
||
|
||
sh.c:
|
||
Reworked SHUnloadSymbolhandler. Now correctly destroy debug
|
||
data lists. Also takes an arg to deterimine if the DLL
|
||
is actually being unloaded or just the debuggee.
|
||
|
||
Various KillxxxList functions now free up corresponding
|
||
memory. Still some holes, but not so bad. The problems
|
||
were mostly lists being retained. Not anymore
|
||
|
||
shinit.c:
|
||
Change SHInit call of InitLlexg to FInitLists and now include
|
||
initializing hlliPds
|
||
|
||
inc\shiproto.h:
|
||
Changed prototype for SHUnloadSymbolHandler (now takes BOOL)
|
||
|
||
15-Mar-1994 MarkBro
|
||
sh.c:
|
||
When loading symbols, don't put cbSym into WORD. Should go
|
||
into UINT. Dolphin 8577
|
||
|
||
18-Feb-1994 MarkBro
|
||
loadomf.c:
|
||
Close the hfile (.exe/.dll) if NB10 and LoadPdb fails.
|
||
Dolphin:3386
|
||
|
||
16-Feb-1994 MarkBro
|
||
*.c inc\sapi.h:
|
||
Dbcs enabling only.
|
||
|
||
18-Jan-1994 MikeMo
|
||
shsymbol.c inc\cvproto.h:
|
||
OSDebug 4 changes: SYFIsOverlayLoaded callback is gone.
|
||
|
||
30-Nov-1993 MarkBro
|
||
ph.c:
|
||
Enabled NB09 FindNearest to search global syms also. SteveSm
|
||
vehemently argueed not to put in additional support for NB10
|
||
|
||
12-Nov-1993 MarkBro
|
||
loadomf.c dosdll.asm shinit.c sh.c inc\shiproto.h inc\sapi.h:
|
||
Handle PDB load errors like NB09 errors and added new
|
||
function void SHPdbNameFromExe( lszName, lszPdbName, cbMax )
|
||
|
||
4-Nov-1993 MarkBro
|
||
sl.c:
|
||
When calculating the CB for the last line in the source line
|
||
table, add 1 to the offEnd since GetSFBounds returns an
|
||
inclusive offset so cb = offBounds - srcstart + 1:
|
||
Fix for dolphin 3701
|
||
|
||
2-Nov-1993 MarkBro
|
||
ph.c:
|
||
Added code in FindNearest to include sstGlobals in search
|
||
ifdef'd out until M5
|
||
|
||
shsymlb0.c:
|
||
When getting nearest address, also allow for GDATAxx when
|
||
creating strings.
|
||
|
||
21-Oct-1993 MarkBro
|
||
shinit.c sh.c inc\shiproto.h dosdll.asm:
|
||
Added SHUnloadSymbolHandler for NB10 notifications. Can
|
||
also be used in the future to free up memory so symbol
|
||
handler doesn't need to be free'd and reloaded.
|
||
|
||
7-Oct-1993 MarkBro
|
||
loadomf.c:
|
||
OLLoadFileIndexes - handle NB10 pdb size instead of lpdssCur->cb
|
||
|
||
6-Oct-1993 MarkBro
|
||
loadomf.c:
|
||
OLLoadFileIndexes - correct off by 1 on cbFileName calculation
|
||
|
||
5-Oct-1993 MarkBro
|
||
inc\sapi.h:
|
||
Added cbFileNames field to _EXS struct. This will be used
|
||
to determine the end of the lpchFileNames buffer.
|
||
|
||
loadomf.c:
|
||
OLLoadFileIndexes - initialize the cbFileNames field of the _EXS
|
||
struct
|
||
|
||
sl.c:
|
||
Added SLFFileInHexe. Scans lpchFileNames for a specific file
|
||
being searched for in SLCAddrFromLine. Also changed SLCAddrFromLine
|
||
to call SLFFileInHexe to short circuit and NOT go through all of
|
||
the modules in an exe if the file isn't referenced in the exe
|
||
|
||
1-Oct-1993 MarkBro
|
||
sl.c:
|
||
cbLine was off by 1. (offNext - offPrev - 1) was the cb in
|
||
SLFLineToAddrExtended. All the way from CV400!
|
||
|
||
29-Sep-1993 v-LeoL
|
||
loadomf.c:
|
||
Handle uSig == 8 in CheckSignature
|
||
|
||
ph.c
|
||
Add UNALIGNED for MIPS
|
||
|
||
shsymbol.c shsymlb0.c shsymlb1.c
|
||
Add support for S_GPROCMIPS, S_LPROCMIPS, S_REGREG32
|
||
|
||
27-Sep-1993 MarkBro
|
||
sl.c:
|
||
Corrected bug in SLFLineToAddrExtended. Improperly caching
|
||
iSegStartSave.
|
||
|
||
21-Sep-1993 MarkBro
|
||
sh.c:
|
||
Corrected SHHexeAddNew to first check the exs list to see
|
||
if the hexg is already in the list if so, don't re-add (bug)
|
||
and return success.
|
||
|
||
shsymbol.c inc\sapi.h:
|
||
DBCS enabling. Missing call to STRLWR instead of _fstrlwr
|
||
|
||
sl.c:
|
||
Changed SLCAddrFromLine to go through each segment table for
|
||
a source file. Also SLFLineToAddr now calls SLFLineToAddrExteneded
|
||
which SLCAddrFromLine uses. This new API (not exported) takes
|
||
an index into the hsf (SL table) to start the search for the line.
|
||
|
||
08-Sep-1993 MarkBro
|
||
sl.c:
|
||
Added SLCAddrFromLine. This will walk through all of the hexe's
|
||
in our list to get ALL occurances of a source line in all modules.
|
||
Will allocate an array of SLP (source line pairs)
|
||
|
||
shinit.c:
|
||
Added export of SLCAddrFromLine to the SHF struct. Also added new
|
||
SHFree so caller can free up memory allocated by SLCAddrFromLine
|
||
|
||
inc\shiproto.h:
|
||
Added prototype for SLCAddrFromLine
|
||
|
||
26-May-1993 DanS
|
||
sh.c
|
||
cuda:4825
|
||
GetNextSym and FixAlign were not operating on the same wavelength
|
||
wrt UFOP's and what was stored in the data portion of S_ALIGN records
|
||
once we had the page loaded.
|
||
ph.c
|
||
sh.c
|
||
turn optimizations on for WIN32 builds
|
||
|
||
20-Apr-1993 RicoM
|
||
ph.c
|
||
CUDA #3513
|
||
PsymFromRef now returns NULL if the module in question doesn't
|
||
have any associated symbols. There was only one call to this
|
||
function and it now correctly checks for this case. This fixes
|
||
problems that occur if cvpack can't write out the symbols for
|
||
a module because they are too big or some other such wierdness.
|
||
|
||
23-Mar-1993 DanS
|
||
shsymbol.c
|
||
shsymlb0.c
|
||
changes required by new cvinfo.h, use bitfields instead of hard
|
||
coded bit positions.
|
||
|
||
19-Mar-1993 dans
|
||
shsymbol.c
|
||
don't count on proc symbols being ordered by offset in SHSetBlksInCXT.
|
||
this means that we have to look at every proc sym in a module before
|
||
giving up.
|
||
|
||
16-Mar-1993 dans
|
||
sh.c
|
||
a couple of off-by-1 bugs in LpvFromAlmLfo regarding cb == cbBlock
|
||
on an ending alignment record that is on the boundary, exactly.
|
||
Also fixed a couple of warnings.
|
||
|
||
05-Mar-1993 dans
|
||
makefile
|
||
sh.c
|
||
shinit.c
|
||
shnt.bld
|
||
added critical sections around SHLoadDll and SHAddDll, cause they
|
||
both dork with the list of modules.
|
||
|
||
29-Feb-1993
|
||
loadomf.c
|
||
Change cTypeIndChunk to 0x2000 to prevent wrap-around when loading
|
||
type index table in a segmented architecture.
|
||
shsymbol.c
|
||
shsymlb0.c
|
||
shsymlb1.c
|
||
Added S_[L|G]Thread cases where appropriate in order to support
|
||
thread local storage.
|
||
|
||
31-Dec-1992 RicoM
|
||
sh.c:
|
||
Check for null hpds in SHGetNextExe() and return failure code
|
||
instead of asserting
|
||
|
||
17-Dec-1992 MarkBro
|
||
shw0.bld:
|
||
Fix version stuff in windows builds.
|
||
|
||
15-Dec-1992 MarkBro
|
||
shsymbol.c:
|
||
Moved assert( pcxt->hmod ) in SHGetSymLoc to places where
|
||
hmod is referenced. There are places where we don't care if the
|
||
hmod is null (where it's not used, duh).
|
||
|
||
11-Dec-1992 MarkBro
|
||
shsymbol.c:
|
||
SHGetSymLoc wasn't handling multiple register symbols. Needed
|
||
to see if register index > 255. Second register value to be
|
||
in hiword if applicable
|
||
|
||
11-Dec-1992 MarkBro
|
||
shinit.c
|
||
Change #ifdef around version stuff so we can get update version #'s
|
||
for release builds.
|
||
|
||
07-Dec-1992 MarkBro
|
||
sh.c:
|
||
Add extra check for end of symbol info in GetNextSym. Was only
|
||
checking for ~1. Now check for ~1 or -1.
|
||
|
||
05-Dec-1992 MarkBro
|
||
sh.c:
|
||
Allow SHHexeFromHmod to accept an HMOD of NULL. If the hmod is
|
||
NULL, return (HEXE)NULL.
|
||
|
||
03-Dec-1992 GlennN
|
||
ph.c
|
||
fix bug 5859.
|
||
Add function to verify an hexe from an hpid.
|
||
The problem was that we assume that hpdsCurr is
|
||
changed and since we changed to having 1 hpds we
|
||
lost the ability to cache multiple hpids ( as well
|
||
as the setting of hpidCurr ). To get around this (
|
||
since windows is weird ) we now make sure that the
|
||
hexe is an hexe in the list.
|
||
|
||
loadomf.c
|
||
added fix from dans change to OLLoadOmf
|
||
|
||
02-Dec-1992 DavidGra
|
||
ph.c
|
||
Incorporate DanS's fix for binary searching.
|
||
inc\sapi.h
|
||
sh.c
|
||
loadomf.c
|
||
Handle mis-aligned symbols by maintaining an align bit in
|
||
the Alm structure and using the low bit of an alignment
|
||
record length to maintain the alignment bit for linear
|
||
searching of tables.
|
||
|
||
30-Nov-1992 MikeMo
|
||
loadomf.c:
|
||
(1) In default (non-LOADALL) case, skip over all sstModules
|
||
as an optimization. (This was already being done in
|
||
the LOADALL case.)
|
||
(2) Fix for NT: compare iMod to 0xFFFF not -1, since it's an
|
||
unsigned short. (Was fixed earlier, someone accidentally
|
||
changed it back on a merge with 410.)
|
||
|
||
24-Nov-1992 DavidGra
|
||
loadomf.c
|
||
ph.c
|
||
inc\sapi.h
|
||
Added support for hash functions 10 & 12 while still supporting
|
||
functions 6 & 8.
|
||
|
||
18-Nov-1992 DavidGra
|
||
sl.c:
|
||
If a source line table is unsorted, we now sort it at load time.
|
||
|
||
13-Nov-1992 DavidGra
|
||
inc\sapi.h Change type alignment to 48k.
|
||
|
||
05-Nov-1992 DavidGra
|
||
loadomf.c Forgot to include the types table in the fix for demand
|
||
load of symbols from a .dbg file.
|
||
|
||
03-Nov-1992 DavidGra
|
||
loadomf.c Added a new field to the exg structure, lszDebug which
|
||
ph.c is the full path name to the file that contains the
|
||
sh.c dedbug info and is used during demand loading of symbols.
|
||
sl.c
|
||
sapi.h
|
||
|
||
03-Nov-1992 MarkBro
|
||
loadomf.c:
|
||
Changed return she values for various signatures.
|
||
< NB08 == relink
|
||
NB08 == repack
|
||
NB09 == OK
|
||
> NB09 == future symbols.
|
||
|
||
ph.c:
|
||
Correction to fix for 4539. Conditions for checking selector
|
||
limits would always fail. Corrected if() to check boundary
|
||
condition.
|
||
|
||
02-Nov-1992 GlennN
|
||
loadomf.c
|
||
fix problem for 32bit world since we indexed on all
|
||
imods. in our case -1 was special and subtracting
|
||
1 would cause us to wrap and index off of the mod
|
||
array. In the 32 bit world this would cause them
|
||
to index off the end of the array and gpf.
|
||
|
||
ph.c
|
||
fix bug 4539. problem was that we can't always get
|
||
the right seg limit for the publics address hash (
|
||
we don't know the real logical segment limit until
|
||
later so we use the largest ref'd in the publics
|
||
table itself. Well, in cases where we are trying
|
||
to find the nearest symbol we can walk off the end
|
||
of the address hash table because of this. Now, we
|
||
check to see of the segment coming in is with range
|
||
of the segment sort table before looking it up.
|
||
|
||
30-Oct-1992 DavidGra
|
||
ph.c
|
||
Initialized some variables for publics searching to avoid
|
||
randomly doing a linear search.
|
||
sh.c
|
||
Fix bug when types table is an exactl multiple of 0x4000 bytes long.
|
||
|
||
30-Oct-1992 MarkBro
|
||
sh.c:
|
||
Fix problem with SHFAddNewPds for dos and windows. We were
|
||
returning a bad pds when we failed and we were also calling LLUnlock
|
||
with a NULL handle.
|
||
|
||
28-Oct-1992 GlennN
|
||
sh.c
|
||
Fix bug 4666. We were loading debug information
|
||
twice since ProgLoad in codeview system.c would
|
||
call SYLoadDebugInfo(). Compounding the problem
|
||
was the fact that Run.Load used to attach the debug
|
||
info on a per process basis. Now we have one and
|
||
only one hpds and all debug information tables are
|
||
associated to this and therefore exhaustive
|
||
searches don't always need the context operator
|
||
like they used to.
|
||
|
||
28-Oct-1992 MikeMo
|
||
loadomf.c:
|
||
We no longer assume IMAGE_FILE_HEADER.SizeOfOptionalHeader ==
|
||
sizeof(IMAGE_OPTIONAL_HEADER).
|
||
|
||
27-Oct-1992 MikeMo
|
||
loadomf.c:
|
||
Fix OLLoadOmf so that, when reading directory entries past
|
||
the list of module entries, we don't try to index into
|
||
rghmod [ iMod - 1 ] if iMod == 0xFFFF.
|
||
|
||
26-Oct-1992 MarkBro
|
||
shinit.c inc\cvtypes.h:
|
||
Change to AVS structure to have alphabetic version
|
||
|
||
19-Oct-1992 v-natjm
|
||
sh.c:
|
||
Made changes to FHOpen to share the static with the SHUnloadDll
|
||
function. The current file handle was locked until CVW.EXE is
|
||
unloaded. We now close this handle in SHUnloadDll if the file
|
||
name is matching.
|
||
|
||
17-Oct-1992 MikeMo
|
||
cvexefmt.h
|
||
Deleted, moved to the CVINC.420 project.
|
||
|
||
*.bld
|
||
Changed to reflect new location of CVEXEFMT.H.
|
||
|
||
16-Oct-1992 MarkBro
|
||
shw0.bld shd1.bld sho1.bld:
|
||
Changed nmake from using a tmp file (which it doesn't delete)
|
||
to a constant $(OBJDIR)\compiler.flg
|
||
|
||
16-Oct-1992 MikeMo
|
||
cvinfo.h
|
||
cvtypes.h
|
||
shapi.h
|
||
types.h
|
||
Deleted, moved to the CVINC.420 project.
|
||
|
||
makefile
|
||
*.bld
|
||
Modified to support the new CVINC.420 project.
|
||
|
||
11-Oct-1992 MikeMo
|
||
makefile:
|
||
Created CDEBUGNT macro which is distinct from CDEBUG in that
|
||
it uses -Z7 instead of -Zi, and it does not use -f.
|
||
|
||
10-Oct-1992 MikeMo
|
||
makefile shnt1.bld:
|
||
Fixes for NT build.
|
||
types.h:
|
||
SEGBASED(x) maps to _based(x) for 16-bit, nothing for 32-bit
|
||
INLINE maps to __inline (the EE uses this)
|
||
_fstrcmp maps to strcmp on NT
|
||
loadomf.c:
|
||
Fixed very minor bug in CheckSignature, where it wouldn't return
|
||
the right result if the read failed.
|
||
sl.c:
|
||
Minor fix for NT
|
||
shinit.c:
|
||
Remove shdllinit (we now set _CRT_INIT as DLL entry point in NT)
|
||
shapi.h:
|
||
Revise prototypes for use with NT.
|
||
|
||
09-Oct-1992 JimSa
|
||
ph.c
|
||
Fixes for ambiguous functions and determining when we are done.
|
||
Previously we would recurse indefinitely because we didn't pay
|
||
attention to the incoming hsym, and we assumed we could just
|
||
get the next symbol in the hash. We needed to keep fetching from
|
||
the hash until we got to the current symbol. Then we know we
|
||
can just get one more.
|
||
|
||
08-Oct-1992
|
||
sh.c: initialize static to 0 in FHOpen. Don't ask me
|
||
why this was required but it fixed the problem with
|
||
multiple runs of CV in the same Windows DOS VM.
|
||
|
||
6-Oct-1992 v-natjm
|
||
sh.c : Compiler was complaining about (lp & 1) in debug build
|
||
|
||
5-Oct-1992 v-natjm
|
||
makefile: Updated these two files for the new Visual C++ DLL names
|
||
shw0.bld:
|
||
|
||
4-Oct-1992 MikeMo
|
||
loadomf.c:
|
||
32-bit fixes: changed "far" to "FAR", and changed "int" to "WORD"
|
||
when reading OMF info off the disk.
|
||
|
||
1-Oct-1992 MarkBro
|
||
makefile:
|
||
Changed DOS sapi to use -G2 for compiles.
|
||
|
||
29-Sept-1992 v-natjm
|
||
shw0.bld:
|
||
Added /NOE to link command line to avoid the linker error on
|
||
multiple symbols.
|
||
|
||
18-Sept-1992 MarkBro
|
||
inc\newexe.h:
|
||
Updated with pcode bit in header
|
||
|
||
*.c inc\shinc.h precomp.c:
|
||
Stuff for precompiled header builds!
|
||
|
||
4-Sept-1992 Rossg
|
||
inc\cvinfo.h
|
||
Update to cvinfo with LF_VARSTRING numeric leaf
|
||
shsymlb1.c
|
||
process LF_VARSTRING in S_CONSTANT
|
||
|
||
28-Aug-1992 JimSa
|
||
sh.c
|
||
Fix for loading Dll's without symbols. We weren't registering an
|
||
emi.
|
||
loadomf.c
|
||
Fixes for PE loading with multiple debug directory entries.
|
||
|
||
14-Aug-1992 DavidGra
|
||
sh.c Correct checking of fLin32 bit.
|
||
|
||
10-Aug-1992 Davidgra
|
||
sh.c Don't throw away exe nodes which don't contain symbols.
|
||
|
||
03-Aug-1992 MarkBro
|
||
sh.c:
|
||
Bad use of ADDRSEG16. ADDRSEG16 ASSIGNS the addr packet to be
|
||
16 bits not check to see if it is a 16 bit addr. Changed check
|
||
to do fAddr32 instead.
|
||
|
||
03-Aug-1992 RossG
|
||
shsymbol.c\shsymlb1.c
|
||
Add support for S_COBOLUDT symbol type. This is a UDT that cannot
|
||
be symbol packed.
|
||
inc\cvinfo.h
|
||
Update to master copy
|
||
|
||
30-Jul-1992 Davidgra
|
||
inc\shapi.h
|
||
Add sopRegister for enregistered variable support by
|
||
the disassmbler.
|
||
|
||
27-Jul-1992 Davidgra/MarkBro
|
||
sh.c
|
||
Fix 32-bit bp-relative symbol lookup from disasm.
|
||
|
||
shsymbol.c
|
||
Fix another bug with SHGetSymLoc for negative offsets.
|
||
|
||
27-Jul-1992 MarkBro
|
||
sl.c:
|
||
Removed some dead code from SLLineToAddr.
|
||
|
||
21-Jul-1992 MarkBro
|
||
shsymbol.c:
|
||
Fix for Caviar 2411. "bp 0:0;bl" causes GPF in SHSetCxtMod.
|
||
We were using the segment (unfixed addr) as an index into
|
||
the segmap table. Problem is that index = seg-1. 0-1 = 0xffff
|
||
which gpf's its brains out. To fix, if the seg is zero, then
|
||
return NULL for the mod.
|
||
|
||
20-Jul-1992 JimSa/MarkBro
|
||
loadomf.c
|
||
Fix problem with module contributor with a seg index of 0.
|
||
|
||
20-Jul-1992 Davidgra/MarkBro
|
||
shsymbol.c
|
||
Changes SHGetSymLoc to return a flat address for 32-bit
|
||
static symbols and to return a 32-bit BP relative offset
|
||
only when the offset will not fit in a 16-bit number.
|
||
|
||
01-Jul-1992 MarkBro
|
||
shsymbol.c:
|
||
Fixed problem in CheckHandles() in hblk test. was checking
|
||
pcxt->hProc instead of pcxt->hBlk.
|
||
|
||
inc\cvtypes.h:
|
||
Updated DBG_API_VERSION
|
||
|
||
25-Jun-1992 MarkBro
|
||
|
||
*********************************************************************
|
||
*********************************************************************
|
||
** **
|
||
** **
|
||
** SPLIT OFF FROM HOBIE - old 410 project became 420 on HOBIE and **
|
||
** moved to SLUG\DEBUG for 410 **
|
||
** **
|
||
** **
|
||
*********************************************************************
|
||
*********************************************************************
|
||
|
||
18-Jun-1992 MarkBro
|
||
sh.c Corrected caching in SHHexeFromHmod. It didn't check
|
||
the hpdsCurr when determining if the cache hexe is the
|
||
correct one to return. It's possible for Windows
|
||
multiple instances to hit the same hmod when in different
|
||
hpds's. This would yield different hexes.
|
||
|
||
04-Jun-1992 MarkBro
|
||
loadomf.c Added tons of error checking for OOM failures.
|
||
|
||
19-May-1992 Jonki
|
||
sh.c when SHLoadDll was invoked twice for the same dll, and
|
||
it failed to load the omf, it would leave an hexe lying around
|
||
which would cause the second invocation to return true.
|
||
Now we remove the hexe that was created by SHLoadDll if the
|
||
load failed to load the OMF
|
||
|
||
08-Apr-1992 Davidgra
|
||
loadomf.c Added code to get debug info location from PE header
|
||
|
||
02-Apr-1992 DavidGra
|
||
loadomf.c Added code to detect PE exes for win32s and
|
||
shsymblol.c to mark addresses as 32-bit based on type
|
||
sl.c of exe when the sh cannot get this information
|
||
inc\exe386.h elsewhere.
|
||
inc\newexe.h
|
||
|
||
23-Mar-1992 DavidGra
|
||
Merged fixes from the 400 version of sapi.
|
||
|
||
23-Jan-1992 davidgra
|
||
sl.c Correct off by one error in source file lookup to fix
|
||
code in include file problem.
|
||
|
||
07-Jan-1992 GlennN
|
||
new bld files for DOS C7 builds and new STU libs
|
||
|
||
02-Jan-1992 Jonki
|
||
shsymbol.c fix call to SyProcessor()
|
||
|
||
05-Dec-1991 GlennN
|
||
|
||
dosdll.asm
|
||
ldllcew.lib new C7 lib
|
||
libw.lib new C7 lib
|
||
loadomf.c change _HUGE to _HUGE_ conflict with C7 math.h
|
||
shd1.bld change build to include oldnames
|
||
sho1.bld ""
|
||
shw0.bld ""
|
||
th.c change _HUGE to _HUGE_ conflict with C7 math.h
|
||
|
||
inc/sapi.h change _HUGE to _HUGE_
|
||
inc/shapi.h change _HUGE to _HUGE_
|
||
inc/types.h change _HUGE to _HUGE_
|
||
02-Dec-1991 Davidgra
|
||
|
||
shsymbol.c Added search on symbol type in SHFindNameInContext
|
||
shsymlb1.c and SHFindNameInGlobal.
|
||
|
||
19-Nov-1991 Davidgra
|
||
|
||
dosdll.asm Added version checking.
|
||
shd1.bld
|
||
shinit.c
|
||
shnt1.bld
|
||
sho1.bld
|
||
shw0.bld
|
||
inc\cvtypes.h
|
||
|
||
ph.c Surpress hashing when SSTR_NoHash bit is set.
|
||
inc\shapi.h
|
||
shsymbol.c
|
||
|
||
04-Nov-1991 Jonki
|
||
inc\sapi.h
|
||
increased size of _MAX_FNAME under !OS2 to 32
|
||
|
||
sho1.bld
|
||
shw0.bld
|
||
shd1.bld
|
||
shnt1.bld
|
||
Added sapi dependency
|
||
|
||
28-Oct-1991 Davidgra
|
||
ph.c Make find nearest public find only publics less than
|
||
or equal to the address requested.
|
||
|
||
28-Oct-1991 GlennN
|
||
sl.c Make sure that in SLNameFromHmod that we check the
|
||
lpsm before the derefence.
|
||
|
||
24-Oct-1991 MikeMo
|
||
sl.c Changed SLHmodFromHsf so it doesn't fail when a non-NULL
|
||
HEXE is passed in.
|
||
|
||
24-Oct-1991 MikeMo
|
||
sl.c The line number passed to SLFLineToAddr is now a WORD
|
||
rather than an SHOFF.
|
||
|
||
17-Oct-1991 Jonki
|
||
shsymbol.c Fix SHSetCxtMod to set the address even if no mod
|
||
was found
|
||
|
||
16-Oct-1991 JimSa
|
||
shsymbol.c SHFindNameInContext was passing of lpsstr to pfnCmp rather than
|
||
lpsstr->lpName.
|
||
|
||
15-Oct-1991 Jonki
|
||
shsymbol.c inc\shapi.h inc\shiproto.h
|
||
Fix SHGetSymLoc to display the location of global stuff correctly.
|
||
This required that we pass it a context, so that we could get the
|
||
emi and fixup the address
|
||
|
||
12-Oct-1991 GlennN
|
||
sl.c fix problems with SLHmodFromHsf. We would sometimes
|
||
return the wrong Hmod due to duplicate files ( includes ) in
|
||
different modules.
|
||
|
||
|
||
10-Oct-1991 RossG
|
||
loadomf.c ph.c shsymbol.c
|
||
Changed the interface to the EE to allow specification of whether
|
||
or not the local symbols of a function are searched while in the
|
||
prolog\epilog of a function. Also, formalized the hidden convention
|
||
to the interface to the symbol handler. The parameter that is passed
|
||
as the hvoid parameter passed back to the symbol compare routine is
|
||
now a structure that contains the pointer to the name, length, flags,
|
||
etc. that the symbol handler can use for global and public symbol
|
||
hashing. Also, there is provision for limiting the symbol searching
|
||
to specific types of symbols (not implemented).
|
||
|
||
07-Oct-1991 GlennN
|
||
sl.c fix problem with cb calculation for line count bytes. In
|
||
the case of the last line I was off by one.
|
||
|
||
07-Oct-1991 GlennN
|
||
sl.c initialize cb inside of SLLineToAddr so that we don't
|
||
pass in garbage.
|
||
|
||
02-Oct-1991 DavidGra
|
||
loadomf.c Removed the start/end fixups to the source lines. The
|
||
packer should now be taking care of this.
|
||
|
||
01-Oct-1991 GlennN
|
||
dosdll.asm fix build for dos due to API name change
|
||
|
||
27-Sep-1991 GlennN
|
||
|
||
makefile Add OBJDIR as part of parameters passed to shw0.bld
|
||
for shw0.dll build
|
||
|
||
24-Sep-1991 GlennN
|
||
|
||
Bonehead fixes to
|
||
loadomf.c Fix FixOfpEnd allocation of lpopt table. We now alloc
|
||
cfiles * cseg * sizeof(opt). The reason we need this
|
||
is because a file may occur mutiple times for a segment.
|
||
also fix potential problem with return of this function.
|
||
|
||
Also, make sure that we free the lpopt table.
|
||
|
||
sl.c Fix problem with GetLpslFromIndex, I was indexing too
|
||
far into the baseSrcLn table.
|
||
|
||
23-Sep-1991 GlennN
|
||
dosdll.asm
|
||
loadomf.c
|
||
sh.c
|
||
shd1.bld
|
||
shinit.c
|
||
shsymbol.c
|
||
shsymlb0.c
|
||
shsymlb1.c
|
||
shw0.bld
|
||
inc/cvexefmt.h
|
||
inc/cvtypes.h
|
||
inc/shiproto.h
|
||
inc/shtypes.h
|
||
|
||
add new source line table api.
|
||
|
||
|
||
05-Setp01991 Jimsch
|
||
Change handle sizes to be consistent across all machines -- i.e
|
||
32-bit handles and such
|
||
|
||
04-Sept-1991 MarkBro
|
||
DOS DLLs -- many small changes all over the place.
|
||
|
||
22-Aug-1991 Jonki
|
||
shsymbol.c SHSetBlksInCXT MUST compare segments. Otherwise,
|
||
we screw up when there are multiple logical segments
|
||
in a mod (ex. alloc_text)
|
||
|
||
19-Aug-1991 Jonki
|
||
loadomf.c fix to support C++ member functions. Member functions
|
||
can cause code to come from different contribution locations
|
||
within the segment. We need to calulate the end offset based
|
||
on the last mds with a matching index.
|
||
|
||
13-Aug-1991 Jonki
|
||
shsymbol.c hacked SHSetBlksInCXT until the packer is fixed.
|
||
|
||
12-Ayg-1991 Jonki
|
||
shsymbol.c Corrected SHAddrFromHsym - make sure that fAddrLI is set
|
||
after we have built the addr to be returned
|
||
|
||
12-Aug-1991 Jonki
|
||
ph.c Correct error in computation of pPubEnd in PHFindNameInPublics
|
||
|
||
07-Aug-1991 Davidgra
|
||
\sapi\ph.c Check to see that there is a valid publics or
|
||
\sapi\shsymbol.c globals table before attempting to search it.
|
||
this used to fall out with a cpubs or cglobs
|
||
of 0, now we check explicitly for the
|
||
appropriate lpssmr of NULL.
|
||
|
||
06-Aug-1991 Davidgra
|
||
\sapi\loadomf.c Rework the publics hashing code. The hash table
|
||
\sapi\ph.c now appears on disk with length prefixed chain
|
||
\sapi\shtypes.h tables. The chain tables are loaded in sections
|
||
if the total size of the tables is greater that
|
||
cbMaxAlloc.
|
||
|
||
05-Aug-1991 Jonki
|
||
loadomf.c sh.c th.c cvproto.h cvtypes.h shproto.h shtypes.h
|
||
Allow for up to 64K distinct types, by making rgitd a huge array.
|
||
Also defined _HUGE. Also removed some old ems junk from OLLoadTypes
|
||
|
||
01-Aug-1991 GlennN
|
||
|
||
\sapi\loadomf.c disabled the loading of hash table to get past of
|
||
gpf loading excel.
|
||
|
||
|
||
26-Jul-1991 Davidgra
|
||
\sapi\shsymbol.c Fix bug in SHSetCXTMod with caching. We were not
|
||
checking to see if the current process was the same as the process
|
||
when we created the cache.
|
||
|
||
26-Jul-1991 MarkBro
|
||
\sapi\shsymbol.c Fix bug in SHSetCXTMod with caching. We were not
|
||
checking to see if the emi of the addr matched the cached hexs. Now
|
||
we are.
|
||
|
||
08-Jul-1991 Davidgra
|
||
\sapi\shsymbol.c Calculate end of symbol table correctly ( size includes
|
||
the leading long ) for SHSetBlksInCXT
|
||
|
||
03-Jul-1991 MikeMo
|
||
\sapi\loadomf.c Improved return values from CheckSignature so the user
|
||
gets a better error message depending on NBxx format.
|
||
|
||
01-Jul-1991 GlennN
|
||
added hack for code in include files for RossG.
|
||
bug fix for callstack navigation.
|
||
bug fix 1062 for MarkBro a at csip other than that at main.
|
||
|
||
|
||
\sapi\LoadOmf.c hack is marked as M00KLUDGE. The change was to first
|
||
save the first file with line number information that
|
||
has .c .cxx .bas .for .asm extension as the main module
|
||
stored in the hmds structure in OLLoadSrc.
|
||
\sapi\shsymlb1.c
|
||
bug was that we weren't checking offset for files in
|
||
SHLineFromAddr(). So we put in a check for offset
|
||
verification.
|
||
|
||
\sapi\inc\shtypes.h
|
||
Added defines for STRUPR and STRSTR into sapi for calls
|
||
to the runtime. This was for hack made to loadomf.c
|