Enter an SEH2_TRY one time and do all testing within it instead of entering the SEH_TRY multiple times.
The commit is an addendum to 0.4.15-dev-2734-g 514147776a
Thanks to patches author Doug-Lyons
This is 1:1 the final approved content of #3758, but I committed
by hand and closed the PR to avoid squash+rebase.
in CreateDIBPalette() when passing invalid arguments to CreateDIBSection.
This could be triggered by using the broken test-application "GDIProg".
After this patch not only the BSOD is fixed but also the app does
properly start up, like it is the case on 2k3sp2.
Thanks to the patches author Doug Lyons.
It allows to properly pass the surface handle (and palette entry pointer) from MS ddraw into win32k.
Otherwise, they are passing into the wrong parameters of actual NtGdi* function, and due to this,
since they're detected as invalid, they become NULL, and that function does not work correctly.
See https://docs.microsoft.com/en-us/windows/win32/devnotes/-dxgkernel-ntgdiddgetdc for the reference
(and confirmed by our headers).
Required by MS DirectDraw stack (ddraw.dll & dxg.sys).
CORE-17561
flip_fix_9.patch fixes:
CORE-16984 " 'SPINA Thrulg' / 'SPINA Thyr' / 'Sim Thyr' have images flipped"
CORE-17194 "StretchDIBits test" isn't work correctly"
"Output of 'Project 3 Test'"
CORE-14701 "DVDStyler 3.0.4 transparent toolbars"
CORE-14701 "DVDStyler 3.0.4 erroneously black around icons of welcome-dlg"
CORE-14671 "'Peazip' shows icons in buttons and menubar vertically flipped"
"Double Commander shoes icons flipped in buttons, menubar, listview and the treeview"
CORE-13273 "Welcome to Lazarus" icon shows flipped
CORE-13026 "'CudaText app' icon shows flipped"
Not all of those are duplicates, although they appear to be at first glance.
It affects different controls and some of those tickets do have different 'guilty revs' than others.
The patch does consist of 3 parts:
1.) win32ss/gdi/ntgdi/dibobj.c
This one is the most clean part of it, that addresses most of the flipping issues now.
2.) The hack in comctl32.h redefining the version:
We used that in the past to appease some, but not all of the issues listed above.
But it does hide additional issues, e.g. in DvDStyler, therefore we seem to still need that appeasement even today.
Most likely it would make sense to aim to avoid this part in the future.
part 2.) was committed as first appeasement on its own already into
0.4.14-RC-24-g 198b61e
0.4.13-RC-7-g 67211fa
0.4.12-RC-5-g 8449527
0.4.11-RC-16-g b906163
0.4.10-RC-7-g f1e80fe
0.4.9-RC-34-g 9d758ae
3.) toolbar.c change
That part fixes at least the toolbar case for DvDStyler
without relying on the comctl32.h hack any longer,
but it was still not enough to completely get rid of part 2.) yet.
Many thanks to all contributors: 'I_kill_Bugs', Doug Lyons and also 'Julenuri' for testing.
The patch gave nice testbot results:
KVM: https://reactos.org/testman/compare.php?ids=75704,75714
VBox: https://reactos.org/testman/compare.php?ids=75705,75715
and we also created a summary of manual test-results:
https://jira.reactos.org/browse/CORE-17415?focusedCommentId=126668&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-126668
If you read this comment via git blame and your goal is to get rid of the hack in comctl32.h, I would highly recommend
to redo the tests of that testing table, and if that is asked too much, then please test at least the following cases:
CORE-14701 "DVDStyler 3.0.4 transparent toolbars"
CORE-14701 "DVDStyler 3.0.4 erroneously black around icons of welcome-dlg"
"DVDStyler 3.0.4 erroneously black/transparent within comboboxes of properties of VMGM menu"
and try to add what might be needed to fix them, and double-check again:
"DoubleCommander optionsDlg the most complex testcase, contains flip-prone icons in treeview, listview, menubar, buttons, statics". Only some of its flipped icons were impacted by the comctl32.h change before.
- Change INIT_FUNCTION and INIT_SECTION to CODE_SEG("INIT") and DATA_SEG("INIT") respectively
- Remove INIT_FUNCTION from function prototypes
- Remove alloc_text pragma calls as they are not needed anymore
ReactOS has no Raster fonts today, so compensate for this affecting the editbox margins.
CORE-1091
This patch fixes the inability to see all of the characters in the CD-KEY box of Starcraft and Visual Basic 5 and Visual Basic 6 Installs. It supersedes #2656
In return of GetGlyphOutline function call, gm.gmBlackBoxX and gm.gmBlackBoxY must be non-zero to avoid Division by Zero. At epilogue of ftGdiGetGlyphOutline, we adjust the values. CORE-15949
By Reverting beginnings of raster-fonts-works (*.fnt and *.fon)
Thanks Katayama Hirofumi MZ for helping with this revert.
The regression was introduced by 0.4.13-dev-681-g
ae99df1675
I will also port this revert back into 0.4.13-RC
CORE-12149
VIDEOPRT:
=========
Improve interfacing with INBV, so as to detect when an external module
acquired INBV display ownership, and whether ownership is being released
later on. (This does NOT rely on hooking!)
For this purpose we improve the IntVideoPortResetDisplayParameters(Ex)
callback that gets registered with an InbvNotifyDisplayOwnershipLost()
call during initialization, and we add a monitoring thread.
The callback is called whenever an external module calls
InbvAcquireDisplayOwnership(), for example the bugcheck code or the KDBG
debugger in SCREEN mode. When this happens, a flag that tells the
monitoring thread to start monitoring INBV is set (ReactOS-specific),
and the display adapters get reset with HwResetHw() (as done on Windows).
Due to the fact that this INBV callback can be called at *ANY* IRQL, we
cannot use dispatcher synchronization mechanisms such as events to tell
the INBV monitoring thread to start its operations, so we need to rely
instead on a flag to be set. And, since INBV doesn't provide with any
proper callback/notification system either, we need to actively monitor
its state by pooling. To reduce the load on the system the monitoring
thread performs 1-second waits between each check for the flag set by
the INBV callback, and during checking the INBV ownership status.
When the INBV ownership is detected to be released by an external module,
the INBV callback is re-registered (this is *MANDATORY* since the
external module has called InbvNotifyDisplayOwnershipLost() with a
different callback parameter!), and then we callout to Win32k for
re-enabling the display.
This has the virtue of correctly resetting the display once the KDBG
debugger in SCREEN mode is being exited, and fixes CORE-12149 .
The following additional fixes were needed:
VIDEOPRT & WIN32K:
==================
Remove the registration with INBV that was previously done in a ReactOS-
specific hacked IRP_MJ_WRITE call; it is now done correctly during the
video device opening done by EngpRegisterGraphicsDevice() in the VIDEOPRT's
IRP_MJ_CREATE handler, as done on Windows.
WIN32K:
=======
- Stub the VideoPortCallout() support, for VIDEOPRT -> WIN32 callbacks.
This function gets registered with VIDEOPRT through an
IOCTL_VIDEO_INIT_WIN32K_CALLBACKS call in EngpRegisterGraphicsDevice().
- Only partially implement the 'VideoFindAdapterCallout' case, that just
re-enables the primary display by refreshing it (using the new function
UserRefreshDisplay()).
VIDEOPRT:
=========
- PVIDEO_WIN32K_CALLOUT is an NTAPI (stdcall) callback.
- In the IntVideoPortResetDisplayParameters(Ex) callback, reset all the
"resettable" adapters registered in the HwResetAdaptersList list.
We thus get rid of the global ResetDisplayParametersDeviceExtension.
- Make the IntVideoPortResetDisplayParameters(Ex) callback slightly more
robust (using SEH) against potential HwResetListEntry list corruption
or invalid DriverExtension->InitializationData.HwResetHw() that would
otherwise trigger a BSOD, and this would be disastrous since that
callback is precisely called when INBV is acquired, typically when the
BSOD code initializes the display for displaying its information...
Extras:
- Validate the IrpStack->MajorFunction in IntVideoPortDispatchDeviceControl()
and implement IRP_MJ_SHUTDOWN handling. Stub out the other IOCTLs that
are handled by VIDEOPRT only (and not by the miniports).
- VIDEOPRT doesn't require IRP_MJ_INTERNAL_DEVICE_CONTROL (unused).
- Implement IOCTL_VIDEO_PREPARE_FOR_EARECOVERY that just resets the
display to standard VGA 80x25 text mode.
- Delete ItalicFromStyle and WeightFromStyle functions.
- Don't use FT_FaceRec.style_name for font style but pOS2->fsSelection, pOS2->usWeightClass, WinFNT.italic and WinFNT.weight.
- Don't use the bitmap glyph as possible if the glyph will be oblique'd.
- Tahoma Italic is working!
CORE-9614
CORE-16269
- Add IntGdiAddFontResourceEx function that is extended from IntGdiAddFontResource, in order to add dwFlags parameter.
- Add IntLoadFontsInRegistry function that will load the fonts from registry info.
- If loading from registry failed, IntLoadSystemFonts will be called.
- Use IntLoadFontsInRegistry rather than IntLoadSystemFonts in the OS startup.
- Add NameFromCharSet function.
- Append " (CharSetName)" to registry value name if not TrueType.
Fix and improve font dumping. CORE-16245
- FontGDI->FontObj is not a pointer.
- "%s" does null check, so we remove redundant null checks.
- Add FaceName and StyleName dumping.
addendum to 0.4.13-dev-707-g
0e4db883e5
Jim Tabor intended to structurally use flTextAlign over
lTextAlign internally within IntExtTextOutW().
He initially committed these changes in 0.4.12-dev-805-g
3377fe184e
But we unintentionally lost these changes via 0.4.13-dev-370-g
8f482af0a8
Many thanks to Doug Lyons who helped a lot with fixing CORE-16133.
We assume pso->sizlBitmap.cy is positive. CORE-15995
Because pso->sizlBitmap.cy is certificated as a positive value after initialization in SURFACE_AllocSurface, so we can safely omit abs and labs (absolute) function calls for it.
- Use an auxiliary list where we add the new enumerated faces, for easy
roll-back in case a failure happens during font loading. Only at the
very end when font loading has been successfully done, the auxiliary
list is appended to the corresponding global or per-proceess font list.
- Perform actual cleanup in case of failures.
- Introduce CleanupFontEntryEx() helper to simplify cleanup in
IntGdiLoadFontsFromMemory() in failure paths; implement
CleanupFontEntry() around it.
Additions:
- Perform string buffer NULL check in IntGdiAddFontResource().
- Reorder memory unmapping and section object dereferencing in
IntGdiAddFontResource().
- Move DuplicateUnicodeString() upwards.
- Mark explicitly the Marlett font handling hacks.
Add raster font (*.fnt and *.fon files) support. CORE-16165
- Add IntGetCharSet() function to get the charset by index and/or the number of charsets.
- Make IntGdiLoadFontsFromMemory() a non-recursive function.
- IntGetOutlineTextMetrics() accepts raster fonts.
- Improve CharMap handling (especially TT_PLATFORM_APPLE_UNICODE).
- Write the raster font file info to the registry.
- Don't request font size for raster fonts in IntRequestFontSize() function.
by reverting essential parts of "Font rotation at arbitrary angles"
which was once committed in several steps:
Initial commit (PR#1207)
0.4.12-dev-243-g
f757a13519
broke font rendering for GTK apps CORE-15623
Second commit (PR#1238)
0.4.12-dev-558-g
64987cf273
restored font rendering in GTK apps but introduced slowdown CORE-16020
Katayama agreed to the temporal revert and will try to come up later
with a performance optimized version.
We revert in 0.4.12-RC and 0.4.13-dev.
The revert will cause failure-increases in gdi32:TextTransform
https://reactos.org/testman/compare.php?ids=67898,67901
In NtGdiGetGlyphIndicesW function, allocation of zero size had caused fatal failures.
Avoid allocation of zero size in ExAllocatePoolWithTag calls. Optimize for cwc == 0. CORE-12825
Thanks to the patches author Doug Lyons
and the reviewers hbelusca and Katayama Hirofumi MZ.
The regression was introduced by
0.4.13-dev-24-g
61475b9fed
Fix results from tests, add (last one) gdi batch support for
ExtSelectClipRgn. Left commented out test code in tree this time.
Pass Katayama Hirofumi MZ SelectClipRgn tests. After commit will plug in
the last batch. After 12 years.
See CORE-13817 and CORE-15906.
Add WORLD_XFORM_CHANGED flag to some positioning functions.
Fix IntExtTextOutW function for TA_UPDATECP alignment, so that it can correctly update the current position. CORE-14994
Eliminate some bugs about font enumeration. CORE-15755
- Add "Microsoft Sans Serif" font substitution.
- Fix and improve the treatment of the nominal font names.
- Split IntGetFontFamilyInfo function from NtGdiGetFontFamilyInfo.
- Add DoFontSystemUnittest function for font system unittest to GDI32.
- Call DoFontSystemUnittest function at CreateFontIndirectW first call.
There is no need to compile our DLLs as shared libraries since we are
managing symbols exports and imports through spec files.
On my system, this reduces the configure-time by a factor of two.
- Assume that GetTextExtentExPointWPri() uses the same kind of
arguments as GetTextExtentExPointW().
- Add GetTextExtentExPointWPri() declaration to undocgdi.h.
There was a missing break statement in ftGdiGetGlyphOutline().
Found by JIRA contributor 'I_Kill_Bugs' :-)
Add also brackets around 'case's to make code navigation easier.
Fixes regression CORE-15785 (Zim Desktop Wiki 0.67 crashed) and
Fixes regression CORE-15755 (NLite 1.4.9.3 used wrong font)
without reintroducing regression CORE-15558 (AbiWord 2.6.8 font enumeration)
This is achieved by partial revert of
0.4.12-dev-320-g
6e4e5a004c
and got ack of Katayama Hirofumi MZ.
Thanks to patches author Doug Lyons.
Test-results: https://reactos.org/testman/compare.php?ids=66264,66267
Add fix ups to PolyPatBlt and freetype.
TextOut supports 580 characters w/o Dx and Dx at a max of 193 characters
with Dx*1, both if offset is 0.
Noticed a lot of over drawing with start menu from ComCtl32 SB_SETTEXT
going through User32 DrawText/Worker ExtTextOutW. Explain why it might
be slowing down.
If issues arise, commenting out lines in win32ss/include/gdi32p.h can
help.
Example:
else if (Cmd == GdiBCTextOut) cjSize = 0;//sizeof(GDIBSTEXTOUT); <----
this one most of all.
else if (Cmd == GdiBCExtTextOut) cjSize = 0;//sizeof(GDIBSEXTTEXTOUT);
Test results:
https://reactos.org/testman/compare.php?ids=66260,66262
Separate TextOut for internal use.
Use flTextAlign internally.
Not sure why GetDeviceCap was used for determining rendering. SVN commit
logs are unavailable. Include code from wine that might make more sense.
Fix EnumFonts, EnumFontFamilies and EnumFontFamiliesEx functions.
The charsets of the font substitutes are also enumerated.
Delete meaningless codes. CORE-15558
* [WIN32K] Fix handle calculation in DbgGdiHTIntegrityCheck
* [NOTEPAD] Fix MSVC warnings
* [PSDK] Simplify *PROC definitions in windef.h
* [VIDEOPRT] Don't try to use NtVdmControl on x64
* [FREELDR] Fix some macros
* [CRT] Make qsort 64 bit compatible
* [NTOS] Use #ifndef _WIN64 instead of #ifdef _M_IX86 around C_ASSERTs
* [FAST486] Fix 64 bit warnings and change DWORD to ULONG, so it can be used in kernel mode
* [APPHELP_APITEST] Fix 64 bit issue
This helps when e.g. changing the resolution on the Dell Latitude D531,
which reports that it supports large resolutions (e.g. 1920x1440x32 and
others larger than 1024x768x32) but fails to apply these.
This usually happens because PDEVOBJ_pSurface(), and more precisely
ppdev->pldev->pfn.EnableSurface(), fails for these resolutions.
- PDEVOBJ_bSwitchMode(): Set the new video mode, or restore the original
one in case of failure + release the allocated ppdevTmp if previous
calls fail. Also unlock in reverse order of locking order.
- UserChangeDisplaySettings(): In case PDEVOBJ_pSurface() fails (but has
reverted the original video mode), we still need to refresh the
display since the display may have been messed up.
Partially implement font/text rotation by LOGFONT.lfEscapement and improve the rendering image. If the angle was not a multiple of 90 degrees, then the background won't be filled. CORE-11848, CORE-15319
* Add an NDK header to define INIT_FUNCTION/INIT_SECTION globally
* Use _declspec(allocate(x)) and _declspec(code_seg(x)) on MSVC versions that support it
* Use INIT_FUNCTION on functions only and INIT_SECTION on data only (required by MSVC)
* Place INIT_FUNCTION before the return type (required by MSVC)
* Make sure declarations and implementations share the same modifiers (required by MSVC)
* Add a global linker option to suppress warnings about defined but unused INIT section
* Merge INIT section into .text in freeldr
- Define IntUnicodeStringToBuffer function to copy the contents of a UNICODE_STRING to a string buffer with a UNICODE_NULL correctly.
- Use it.
JIRA issue: N/A
Preparation to support pen width.
Split IntGdiWidenPath function from PATH_WidenPath function.
IntGdiWidenPath is independent from DC.
CORE-2527, CORE-13534
Add a mechanism that detects via the version resource whether it's needed to mirror
the application layout or not. This is a "follow up" to commit 5b14b6f.
This code is taken from the GetProcessDefaultLayout() function from Wine
(called there "WIN_CreateWindowEx").
In Windows the version-resource mirroring happens when the GDI is loading the LPK.
- Implement mirroring via version resource. GDI linking on startup is required.
- GDI32: Add LPK loading on GDI startup.
To enable mirroring via version resource the lpk.dll must be loaded before any windows are created.
This is done through GdiInitializeLanguagePack() function which is called in GdiProcessSetup().
- LPK: Use Wine debug messaging.
