Correct fix was to fix the HCELL_INDEX <-> HKEY conversions, much like
is being done with UlongToHandle / HandleToUlong.
The on-disk/in-memory hive file structures are platform-independent:
their layout must not depend on whether code is compiled in 32 or 64
bits.
They are correct and are more performant.
- Rewrite WinLdrAddDriverToList() on the model of CmpAddDriverToList()
with support for CmpIsDriverInList().
- Disable RegEnumKey() as it is now unused.
This allows: (i) to shorten dramatically the lengths of the static
registry path strings being used (and thus, allow for reduced binary
file size), and (ii) to prepare future support for selecting different
control sets (and related Last-Known-Good configurations).
- Use UNICODE_STRINGs wherever possible.
- Retrieve the OEMHAL font file name. TODO: Load it!
- In WinLdrLoadNLSData(), don't open/close each file twice (and showing
we are loading them twice). Instead, open all files at once (and keep
them that way), retrieve their sizes and do the calculations, then
load their respective contents one by one -- giving loading feedback
only at that point -- and finally close all of them at once.
... as it should have always been done (and must be done for NTDLL
and NTOS kernel as well). This allows using the RTL with the correct
definitions and the reduced functionality available at boot-time.
+ Make the RTL main header compatible.
In addition, this will permit re-using existing code that already
uses the RTL (mostly string conversions).
See commits 427c90af3 (r36761) and b46e8cc18 (r36980) for some
background.
Reviewed all Chinese Simplified translation files and updated those
inappropriate or outdated ones.
Co-authored-by: Liu Wenyuan <15816141883@163.com>
Reviewed-by: Chan Chilung <eason066@gmail.com>
Reviewed-by: He Yang <1160386205@qq.com>
Reviewed-by: Zheng Jianping <robsean@126.com>
Reviewed-by: Luo Yufan <njlyf2011@hotmail.com>
Reviewed-by: Hermès Bélusca-Maïto <hermes.belusca-maito@reactos.org>
Reviewed-by: Stanislav Motylkov <x86corez@gmail.com>
- Added zh-HK translation for following files:
- [BASE/SERVICES/W32TIME]
- [BASE/SYSTEM/...] (except CMD console only applications)
- [FDEBUG]
- [DLL/CPL/...] (except Wine related applications)
- [DLL/SHELLEXT/...]
- [DLL/WIN32/...] (not all applications are translated, and Wine related applications are excluded from this part)
- [MODULES/ROSAPPS/APPLICATIONS/...] (not all applications are translated)
- [SCREENSAVERS]
- [NTVDM]
- [USERSRV]
- Translation Improvement
- Fix header for zh-TW and zh-HK translation files
Reviewed-by: Luo Yufan <njlyf2011@hotmail.com>
Reviewed-by: Hermès Bélusca-Maïto <hermes.belusca-maito@reactos.org>
Reviewed-by: Stanislav Motylkov <x86corez@gmail.com>
- Move boot options parsing early on in LoadWindowsCore().
- Add WinLdrIsPaeSupported(), move in it the PAE options parsing.
The aim is to perform other tests to determine whether or not to
enable PAE, and select an adequate kernel image.
Co-authored-by: Vadim Galyant <vgal@rambler.ru>
- Fix parsing of the NOEXECUTE options, taking their precedence
into account. Most of these are also x86-specific.
Now, default UI colors/strings (e.g. titles/timeout text) that are
dependent of the UI "theme" (mini vs. full TUI) are set first when
calling their corresponding Initialize() function.
Then, the user UI settings are read from the user's freeldr.ini file
and override the UI theme defaults.
These settings get effectively applied at the first drawing operation
(usually the UiFadeInBackdrop() call done by the main UiInitialize()
function).
For "directui" we don't care about the settings -- they are hardcoded.
This allows not having to specify the default settings in the
freeldr.ini files anymore.
+ Add support for "None" UiBackdropFillStyle: fill with whitespace,
instead of the specific patterns.
Fixes CORE-13827, where the timer countdown would otherwise display
incorrectly if there were no trailing spaces after the timer string.
Now, the countdown is always appended with one space separator after
the end of the string, **or**, it is inserted at a specific place
delimited by a %d insertion point in the string.
Also, don't repaint the date & time (in full-UI mode) many times per
second -- instead of only once per second -- while displaying a menu.
A Kernel Debugger Transport DLL is always loaded for Windows XP+ :
either the standard KDCOM.DLL (by default), or an alternative
user-provided one via the /DEBUGPORT= option. If this alternative
does not exist or fails to be loaded, fall back to the standard
KDCOM.DLL.
If no KD dll is loaded, kernel loading would fail because of the
resulting unsatisfied KDCOM dll import (tested on Windows and ReactOS).
+ Normalize kernel, HAL and KD dll file names to lowercase (needed for
case-sensitive installations).
CORE-9023, CORE-18033
- Reset the UI to a minimal one in SOS mode.
- In SOS mode, a trace of loaded files is displayed on the screen,
instead of the usual progress bar.
- Add a callback to the PE loader to notify when imported DLLs are
loaded for a main image. This allows getting an accurate SOS trace.
- If the started OS loader failed and we are back to the OS selection
menu, re-initialize the UI as the loader may have messed up the display
in the meantime.
- Tear down allocated off-screen back-buffer when uninitializing the TUI.
- Clear up the screen when initializing the direct-UI.
- Remove excessive UiDrawBackdrop() calls that caused too many
unnecessary redraws.
- ProgressBar: Clear only the portions that need to be cleared up.
This allows to not use DrawBackdrop anymore and the flickering.
- Remove duplicated code from directui.c and use the one from
TUI instead, with the latter properly #ifdef'ed for _M_ARM.
- Fix their title color.
- Add SAL annotations.
- TuiDrawCenteredText: Partly fix centering calculations (susceptible
to give negative coordinates).
- TuiDrawText2: Don't display anything if X or Y are out of the screen.
- Remove duplicated code from directui.c and use the one from
TUI instead, with the latter properly #ifdef'ed for _M_ARM.
- Provide the minimal implementations for NoUiDrawText*().
- TuiDrawText() is just a particular case of TuiDrawText2().
- Isolate the TuiPrintf() and TuiDraw*Text*() functions as separate
"Generic TUI utils".
- Fix "TAG_TAG" typo in TAG_TAG_TUI_PALETTE.
- Add SAL annotations.
- Added translation for:
- [W32TIME]
- [MODE]
- [TIMEOUT]
- [MODEMUI]
- [ARPING]
- [FRAGINATOR]
- and more
- Updated existing translation
- [WINNLS] Update cht (zh-TW) and zhh (zh-HK)
- [DOC] Update rules in Chinese translation notes.txt
- Wine translation sync
Some of the punctuation use in WIne has been converted to half-width,
so the punctuation will no longer synced.
- Don't store trailing newlines in the exception description text strings.
- Remove unused i386PrintChar().
- Display CR4 in x86.
- Use the "indentation" printf generation trick in order to get aligned
strings for (CF4 and) DR6 and DR7, without having to hardcode the tons
of alignment whitespaces (--> make the strings stored in freeldr shorter).
- Show the IP/ErrorCode/EFlags/GDTR/IDTR/LDTR values vertically aligned.
- Display the stack frames in both x86 and x64 modes.
- Adjust the instruction pointer when a BREAKPOINT or OVERFLOW exception
arises, so that the offending instruction can show up in the instruction
stream.
CORE-16748
- Display the correct TR register value.
- Ensure that the x86 segment register values displayed are really
2-byte long.
Segment registers are intrinsically 16 bits. Even if the x86
KTRAP_FRAME structure stores them as ULONG, only their lower 16 bits
are initialized. We thus cast them to USHORT before display.
These segment registers are saved in a stack-based KTRAP_FRAME by the
CPU trap mechanism (for SS), and by 'push CS' etc. instructions for
the others, and from Intel documentation, we know that:
"
If the source operand is a segment register (16 bits) and the operand
size is 64-bits, a zero-extended value is pushed on the stack; if the
operand size is 32-bits, either a zero-extended value is pushed on the
stack or the segment selector is written on the stack using a 16-bit
move. For the last case, all recent Core and Atom processors perform
a 16-bit move, leaving the upper portion of the stack location unmodified.
"
So it may happen, when using the push, that either they get zero-extended,
or garbage gets stored in the higher bits, and these need to be trimmed.
Actually, RegCloseKey() is here a dummy macro that just "returns"
success. Indeed, the internal implementation of RegOpenKey() doesn't
really allocate internal structures: it just returns a "pointer" to
already existing allocated data. Therefore nothing needs to be closed/freed
later on.
- Move a few lines around.
- Switch to RtlZeroMemory() from memset().
- Make while() more explicit.
For CORE-12802:
- Add/Fix FrLdrHeapAlloc() failure handling and related.
Especially, add/fix FrLdrHeapFree() calls.
- Add/Improve ERR() to some FrLdrHeapAlloc() failure cases.
Co-authored-by: Hermès BÉLUSCA - MAÏTO <hermes.belusca-maito@reactos.org>
objcopy bug is now solved and I can't remember how this used to work anyway.
Use strip instead.
If you need to debug freeldr
- set SEPARATE_DBG cmake var to TRUE
- use qemu GDB stub
- In GDB : add-symbol-file <build_dir>/symbols/freeldr_pe.exe
This splits MM_MAX_PAGE into the maximum addressable PFN (which is 2^20-1 on x86, 2^26-1 on x86PAE and 2^36-1 on x64) and the maximum mapped/accessible virtual memory MM_MAX_PAGE_LOADER (which is 4 GB on x86, since paging is disabled, but only 1 GB on x64, since only that much is identity-mapped).
This fixes the recursion when pci.sys attempts to get PCI bridge children
devices, which in fact are non-existing.
Note that the problem in pci.sys itself also deserves a separate fix.
CORE-16319 CORE-16216
We can already restart reading the whole file if seeking backward,
and there is currently no more use case of backward seek.
By doing so, we gain 1MB of memory when using PXE.
at runtime from the boot selection menu or from FREELDR.INI.
CORE-17350, CORE-9023
For a proper override of the options by new user options, specify the
/SIFOPTIONSOVERRIDE switch in addition. Otherwise, user options are
merged with those retrieved from TXTSETUP.SIF, with priority given to
the former ones.
- Update the documentation for the 'ReactOSSetup' OS type in the
FREELDR.INI file template.
- Use a different prompt in the custom boot options editor for the
'ReactOSSetup' OS type, with adequate explanation.
- Get rid of the ReactOS-specific TXTSETUP.SIF 'DbgOsLoadOptions' value,
and use instead the Windows-compatible 'SetupDebugOptions' value that
is added to the other load options when debugging is to be enabled.
- Introduce a set of NtLdrGet(Next)Option(Ex) helpers that allow
retrieving respectively, the "next" option in an options string,
and retrieving a given named option in such an options string,
by correctly parsing that string.
Valid syntaxes:
/OPTION1 /OPTION2/OPTION3 OPTION4 /OPTION5(=...) ...
Options separators are slashes, or whitespace (space, tab), mandatory
if no slash is used, and otherwise optional.
- Use these functions wherever NT load options are being parsed.
- Simplify the parsing of /DEBUGPORT=... using these functions.
- When parsing the /HAL=... or /KERNEL=... options, only the first
encountered one is taken into account, any other ones are discarded.
- When parsing the other load options, only their first occurrences are
taken into account, any other repetitions are discarded.
* The NOPAE option overrides any previous PAE option.
* Any NOEXECUTE(=) option should override any previous
NOEXECUTE=ALWAYSOFF (or equivalently, EXECUTE) option.
This parameter is not needed since it's possible to determine, from
within this function, whether or not we are in Setup mode, by just
looking for the presence of a non NULL LoaderBlock->SetupLdrBlock.