Ros-dev

ros-dev@reactos.org

5789 discussions

by Filip Navara

Hi, I think I found a race condition in KQUEUE code, but I'm not sure my solution is optimal. If no items are queued, KeRemoveQueue calls KeWaitForSingleObject and then after it finishes it acquires dispatcher lock and removes a list item. The problem arises when someone calls KeRemoveQueue just after the KeWaitForSingleObject call finishes and before the dispatcher lock is acquired. In this case he can remove the last item in the list and this will cause the first KeRemoveQueue (the one that waited) to return garbage. Can anybody think of better solution than the attached patch? Regards, Filip Index: ntoskrnl/ke/queue.c =================================================================== RCS file: /CVS/ReactOS/reactos/ntoskrnl/ke/queue.c,v retrieving revision 1.11 diff -u -r1.11 queue.c --- ntoskrnl/ke/queue.c 15 Aug 2004 16:39:05 -0000 1.11 +++ ntoskrnl/ke/queue.c 21 Nov 2004 00:39:52 -0000 @@ -159,27 +159,38 @@ return ListEntry; } - //need to wait for it... - KeReleaseDispatcherDatabaseLock (OldIrql); - - Status = KeWaitForSingleObject(Queue, - WrQueue, - WaitMode, - TRUE,//Alertable, - Timeout); - - if (Status == STATUS_TIMEOUT || Status == STATUS_USER_APC) - { - return (PVOID)Status; - } - else + for (;;) { - OldIrql = KeAcquireDispatcherDatabaseLock (); - ListEntry = RemoveHeadList(&Queue->EntryListHead); + //need to wait for it... KeReleaseDispatcherDatabaseLock (OldIrql); - return ListEntry; - } + Status = KeWaitForSingleObject(Queue, + WrQueue, + WaitMode, + TRUE,//Alertable, + Timeout); + + if (Status == STATUS_TIMEOUT || Status == STATUS_USER_APC) + { + return (PVOID)Status; + } + else + { + OldIrql = KeAcquireDispatcherDatabaseLock (); + + /* + * Prevent a race condition. If someone calls KeRemoveQueue + * just after we end up waiting and before acquiring the spin + * lock he can get the last item in the list... + */ + if (IsListEmpty(&Queue->EntryListHead)) + continue; + + ListEntry = RemoveHeadList(&Queue->EntryListHead); + KeReleaseDispatcherDatabaseLock (OldIrql); + return ListEntry; + } + } }

20 years, 7 months

Keyboard layout DLLs

by Casper Hornstrup

These are a big mystery to me. I'd like to document how to create them, but I need to know how to create them first. Please send me any information you have on them. Casper

20 years, 7 months

A call for help getting putty working

by art yerkes

I'm hoping to get putty working to use as another test case for the network code. I'm running into problems right now, however and could really use a hand. There are two versions of putty in here, built with symbols as well as a diff against the original sources and a mingw-able makefile named makefile.mgw I haven't tried the mingw makefile on windows but it shouldn't take too much effort. putty-noshow differs from the release putty only in that it contains debug messages and symbols. The putty dialog is created and the caret timer even fires, but the window is never shown. w3seek thinks that the dialog functions in user32 and win32k are correct, but the dialog shows on windows and wine so something is wrong. If I add a ShowWindow as in putty-show, the dialog shows, so the problem is clearly in reactos. I'm seeing some random crashes as well accessing the registry that will be the subject of another mail. my putty musings: http://64.81.145.152/~arty/putty.zip Putty page: http://www.chiark.greenend.org.uk/~sgtatham/putty/ -- Here's a simple experiment. Stand on a train track between two locomotives which are pushing on you with equal force in opposite directions. You will exhibit no net motion. None the less, you may soon begin to notice that something important is happening. -- Robert Stirniman

20 years, 7 months

headers in reactos/w32api/include and reactos/include/wine

by Steven Edwards

Does anyone care if I merge the duplicate headers from reactos/include/wine in to the headers in the reactos/w32api/include folder? There are a few headers we will want to keep in reactos/include/wine but now that we have our own copy of the w32api we dont need most of these headers that do the #include_next thing. Thanks Steven __________________________________ Do you Yahoo!? Meet the all-new My Yahoo! - Try it today! http://my.yahoo.com

20 years, 7 months

Request for use of Logo on items at CafePress

by Wierd Wierd

I have set up an account at CafePress, and would like to make some ReactOS merchanise, for advertisment purposes. I do not intend on adding a retail markup to the base price of the items in question- the prices will be solely the price mandated by CafePress for fabrication and shipment. (this means I make nothing) Some of the items that can be made, include Mugs, buttons, mousepads, T-shirts, hoodies... ...Thongs... and other misc. items. I think it could have utility for us as a project, on occasions where we are promoting our project-- Things like "ReactOS Developer" buttons for our booth sitters, etc... I would like a "Yeah"/"Nay" vote, and then I may require a written letter of permission allowing me to use the logo. (The letter is required by the CafePress legal terms of service contract) Please let me know what you think of the idea. __________________________________ Do you Yahoo!? The all-new My Yahoo! - Get yours free! http://my.yahoo.com

20 years, 7 months

RE: [ros-dev] question about some old changes (hartmut)

by Hartmut Birr

Hi, >What was the reason for these changes? It looks like a workaround for some other bug. it is a nice question after more than one year. >http://cvs.reactos.com/cgi-bin/cvsweb.cgi/reactos/ntoskrnl/ke/wait.c.diff?r 1=1.51&r2=1.52 KeRemoveAllWaitsThread was called in two situations. One was a real unblocking of the thread, if the thread was waiting on a waitable object which was signaled. The other was the terminating of the thread by an other thread. In this case KeRemoveAllWaitsThread has worked if the thread was waiting on a waitable object. If the thread was simply suspended, RemoveEntryList has bug checked because Waiter->WaitListEntry wasn't in any list. The reason for these changes was the implementation of the CTRL-C handlers in csrss. - Hartmut

20 years, 7 months

fix for regedit

by Christian Koinig

This fixes the problem with regedit (regedit freezes after opening a child window). Regards Christian ------------------------------------------------------ Index: framewnd.c =================================================================== RCS file: /CVS/ReactOS/reactos/subsys/system/regedit/framewnd.c,v retrieving revision 1.11 diff -u -p -r1.11 framewnd.c --- framewnd.c 11 Oct 2004 21:08:05 -0000 1.11 +++ framewnd.c 18 Nov 2004 21:56:01 -0000 @@ -670,8 +670,10 @@ FrameWndProc(HWND hWnd, UINT message, WP OnMenuSelect(hWnd, LOWORD(wParam), HIWORD(wParam), (HMENU)lParam); break; case WM_ACTIVATE: - if (LOWORD(hWnd)) - SetFocus(g_pChildWnd->hWnd); + // I don't know what this is supposed to do (except for creating an infinite "setfocus loop"). + // If someone ever figures it out put it back in plz. + //if (LOWORD(hWnd)) + // SetFocus(g_pChildWnd->hWnd); break; case WM_DESTROY: WinHelp(hWnd, _T("regedit"), HELP_QUIT, 0); ------------------------------------------------------

20 years, 7 months

Greenville: The story of Freetype VS the windows rasterizer---

by Wierd Wierd

Hey everybody- I seem to be consistantly running into troubles with Freetype and Greenville. I suspect the problem stems from Freetype trying to render the glyph with too many shades of grey, and I *SERIOUSLY NEED* somebody to fix it. Here is a more technical view of the situation, and the problem: The TrueType hinting system allows each pixel to be subdivided into sub-units. There are 81 subunits per pixel, in a 9x9 grid, with the center being the origin point. So, it looks something like this: 87654321012345678 7XXXXXXXXXXXXXXXX 6XXXXXXXXXXXXXXXX 5XXXXXXXXXXXXXXXX 4XXXXXXXXXXXXXXXX 3XXXXXXXXXXXXXXXX 2XXXXXXXXXXXXXXXX 1XXXXXXXXXXXXXXXX 0XXXXXXXXXXXXXXXX 1XXXXXXXXXXXXXXXX 2XXXXXXXXXXXXXXXX 3XXXXXXXXXXXXXXXX 4XXXXXXXXXXXXXXXX 5XXXXXXXXXXXXXXXX 6XXXXXXXXXXXXXXXX 7XXXXXXXXXXXXXXXX 8XXXXXXXXXXXXXXXX That represents the total number of subunits, and their coordinate placements within 1 pixel. The windows rasterizer determines the 'shade' the pixel will get, based on % of coverage this pixel gets from the outline, as does freetype. The problem shows up, because The windows type rasterizer only renders 16 shades, including Black and White, while the Freetype one renders 256 shades. There are only 81 units, however. What does this mean? It means that under freetype as it currently stands, if even *ONE* pixel subunit is covered by an outline, it will render as a shade of grey--- While the exact same pixel, with the windows rasterizer, will render as 'white' instead. This is because "white" is 1/256th coverage, and there are only 81 subunits-- Essentially meaning that in order to get 'white' out of freetype, the pixel cannot be touched *AT ALL* by the glyph outline. On the windows rasterizer, there are only 16 shades of grey, so up to 1/16th of the area can be covered, and still rendered as 'white'-- or 5 subunits covered. This is why the windows rasterizer produces crisper images than Freetype. I don't know how to fix this behavior in Freetype, and *SERIOUSLY* need said behavior changed, As soon as possible. Thank you. __________________________________ Do you Yahoo!? The all-new My Yahoo! - Get yours free! http://my.yahoo.com

20 years, 7 months

RE: [ros-dev] ReactOS Versions

by Waldo Alvarez Cañizares

Hi Rick: I think you mean something like the make xconfig or make gconfig you can do when building the Linux kernel. Hmm I think if that´s what you mean ROS could have a little GUI application to configure the system. I mean to make life easier to non-programmers that wish to build a custom system. That's one of the things Linux users enjoy very much and I think that would be good for ROS too and won't take very long to do. Also if it is done general enough won't have to be maintained in the future. Well at least not that much. Regards Waldo ________________________________ De: ros-dev-bounces(a)reactos.com en nombre de Rick Langschultz Enviado el: Jue 11/11/2004 6:53 p.m. Para: ros-dev(a)reactos.com Asunto: [ros-dev] ReactOS Versions When ReactOS is built and runs windows executables stabily I think we should have a program for windows that will let you customize the platform like windows ce and compile it with all the features you want. I think that it would be a great idea because of different types of systems such as PBX/VOIP servers, application servers, and domain name servers. This would allow security to be integrated into ReactOS by limiting programs allowing open ports, and Denial of service attacks. Maybe even support for multiple architectures like SH, ARM, X86, AMD64, PPC??? With FAT, FAT32, NTFS, XFS, EXT3, EXT2, and many others. Maybe even a rom based os... Please tell me what you all think.

20 years, 7 months

"no current process" problem

by Gunnar Dalsnes

Hi. I'm was fixing some code, se\semgr.c->SeLockSubjectContext/SeUnlockSubjectContext, where apc's were not disabled before/after ExAcquireResourceExclusiveLite/ExReleaseResourceLite using KeEnterCriticalRegion/KeLeaveCriticalRegion. KeEnterCriticalRegion/KeLeaveCriticalRegion access current thread and this triggers bsod->"no current process". call chain: CmInitializeRegistry->ObCreateObject->SeCaptureSubjectContext->SeLockSubject Context CmInitializeRegistry is called in ke\main.c before the initial process is created. What should i do? Add checks for if the current process/thread exist or not in KeEnterCriticalRegion/KeLeaveCriticalRegion or is there some other way to fix this? Gunnar

20 years, 7 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Ros-dev