Ros-dev

ros-dev@reactos.org

1 participants
5790 discussions

by Ros Arm

Hello, First I was asked if a certain developer could talk to me regarding ARM3. I said yes, never heard back, and then saw commits done to pfnlist.c. While interesting and useful code, some communication would've been nice in order to talk about the goals, since much of that code is in flux (I feel bad when someone improves a function that I will later remove, as two of the improved functions are only temporary -- the changes to the other functions will remain, obviously). Normally, this would not have been a problem, if not for the strange formatting and the fact this code was now broken and causing a boot regression. I showed up in the #reactos-dev channel and instructed developers on how to fix the problem: remove the entire function and use MiInsertPageInFreeList instead, or fix MiInitializePagingLists to perform its work at DISPATCH_LEVEL instead, by acquiring the PFN lock as it should do. None of this was done, instead "janderwald", in 47514, "Fixed" the ASSERT by saying "Any IRQL lower than DISPATCH_LEVEL" is okay. Janderwald, whoever you are, do you even understand this code? Are you even a kernel-mode developer? Does it make to you that the code would need to assert for low IRQL? The whole point is to validate that the PFN lock is being held. Even if you are not a kernel-mode dev, isn't it clear that all the other assertions are also checking for DISPATCH_LEVEL+? And even if that's not clear, I had written down the solution in the #reactos-dev channel, expecting that probably not many people would figure it out. So I guess some lessons to learn for this project: 1) If you modify someone's code, make sure you test your changes (and preferably don't mess up the formatting) so their code is not now blamed for a boot regression. 2) If someone tells you how to fix their code after it's been messed up, apply the fix. 3) If you choose to roll your own fix, write something useful instead of breaking the function (logic-wise) even more. Again, I'm afraid the problem is most of you are students and don't have work ethics of real employees -- this seems to be the root of many problems this project has had since the last 2 years I've been watching and working on the ARM port. Next steps: - Janderwald: remove your ASSERT hack. Either 1) Remove the ASSERT (stop checking for an invariant that doesn't happen) or 2) Fix the caller (so the invariant holds) - Arthur: changes are appreciated, but let's talk more so you don't end up improving dead code. Carry on! -r

15 years, 2 months

Some proposals for roadmap

by Javier Agustìn Fernàndez Arroyo

hello, i have thinking about an easy way to show people the ReactOS roadmap, and there came to my mind two projects where i have submitted some bug reports before joining ReactOS. They use the "milestone" field in their bug reports for the roadmaps, and its a quite clean way to show it, IMO Here are the links Is there any way to do something similar using Bugzilla? (none of these 2 projects use this Bug Management System) http://www.kdenlive.org/mantis/roadmap_page.php http://developer.pidgin.im/roadmap

15 years, 2 months

Testbots fail to boot ReactOS since 47510

by Olaf Siejka

Since arty's commit in rev 47510, one of his asserts is being triggered in bot qemu and vbox. This practically lock down both testbots as they cannot cont it. I have two requests: - Could you all refrain from commiting until this problem is resolved? - If this issue is not to be fixed soon enough, can this particular assert be commented out for the time being? Debug log below: WARNING: HalStopProfileInterrupt at hal\halx86\generic\profil.c:24 is UNIMPLEMENTED! Assertion 'KeGetCurrentIrql() == DISPATCH_LEVEL' failed at ARMÂł::PFNLIST line 61 Entered debugger on embedded INT3 at 0x0008:0x808e9b7e. kdb:> bt Eip: <808e9b7f (DbgBreakPoint)> Frames: <8089840f (ARM┬│::PFNLIST:61 (MiInsertInListTail@8))> <808a26cd (ntoskrnl/mm/freelist.c:499 (MmInitializePageList@0))> <8089088b (ARM┬│::INIT:X86:521 (MiInitMachineDependent@4))> <80896a61 (ARM┬│::INIT:1846 (MmArmInitSystem@8))> <808a51f7 (ntoskrnl/mm/mminit.c:388 (MmInitSystem@8))> <8084404b (ntoskrnl/ex/init.c:1008 (ExpInitializeExecutive@8))> <8080450b (ntoskrnl/ke/i386/kiinit.c:564 (KiInitializeKernel@24))> <8080467e (ntoskrnl/include//internal/arch/../i386/ke.h:235 (KiSystemStartupBootStack@0))> <0000000e> Entered debugger on last-chance exception (Exception Code: 0xc0000005) (Page Fault) Memory at 0x0007767C could not be read: Page not present. *** Fatal System Error: 0x0000001e (0xC0000005,0x80877985,0x80967DC0,0x00000000) Entered debugger on embedded INT3 at 0x0008:0x808e9b84.

15 years, 2 months

Re: [ros-dev] [ros-diffs] [janderwald] 47509: [NPFS] - Add more buffer checks - Clear DO_DEVICE_INITIALIZING flag

by Ged Murphy

janderwald(a)svn.reactos.org wrote: > + DeviceObject->Flags &= ~DO_DEVICE_INITIALIZING; This is done by the IO manager for any device objects created in DriverEntry If it's not then we have a bug, but I'm fairly sure it does otherwise we wouldn't be able to open devices from umode :) Ged.

15 years, 2 months

Re: [ros-dev] [ros-diffs] [tkreuzer] 47484: - add explicit braces to avoid ambiguous 'else' - fix deprecated conversion from string constant to 'char*'

by Ged Murphy

This is Wine code, it'll get lost in the next sync On 31 May 2010 05:13, <tkreuzer(a)svn.reactos.org> wrote: > Author: tkreuzer > Date: Mon May 31 06:13:40 2010 > New Revision: 47484 > > URL: http://svn.reactos.org/svn/reactos?rev=47484&view=rev > Log: > - add explicit braces to avoid ambiguous 'else' > - fix deprecated conversion from string constant to 'char*' > > Modified: > trunk/reactos/dll/win32/browseui/internettoolbar.cpp > > Modified: trunk/reactos/dll/win32/browseui/internettoolbar.cpp > URL: > http://svn.reactos.org/svn/reactos/trunk/reactos/dll/win32/browseui/interne… > > ============================================================================== > --- trunk/reactos/dll/win32/browseui/internettoolbar.cpp [iso-8859-1] > (original) > +++ trunk/reactos/dll/win32/browseui/internettoolbar.cpp [iso-8859-1] Mon > May 31 06:13:40 2010 > @@ -414,10 +414,16 @@ > { > SMINFO *infoPtr = (SMINFO *)lParam; > if ((infoPtr->dwMask & SMIM_FLAGS) != 0) > + { > if (psmd->uId == FCIDM_MENU_FAVORITES) > + { > infoPtr->dwFlags |= > SMIF_DROPCASCADE; > - else{ > - infoPtr->dwFlags |= > SMIF_TRACKPOPUP;} > + } > + else > + { > + infoPtr->dwFlags |= > SMIF_TRACKPOPUP; > + } > + } > if ((infoPtr->dwMask & SMIM_ICON) != 0) > infoPtr->iIcon = -1; > } > @@ -1213,7 +1219,7 @@ > CComPtr<IObjectWithSite> > objectWithSite; > CComPtr<IContextMenu> contextMenu; > CMINVOKECOMMANDINFO > commandInfo; > - char > *searchGUID = "{169A0691-8DF9-11d1-A1C4-00C04FD75D13}"; > + const char > *searchGUID = "{169A0691-8DF9-11d1-A1C4-00C04FD75D13}"; > HRESULT > hResult; > > // TODO: Query shell if this command is enabled first > > >

15 years, 3 months

Re: [ros-dev] [ros-diffs] [mjmartin] 47393: [win32k] - The timer is created usingUserCreateObject. It may be a good idea to save the handle in the timer object so that it can be deleted later. - Dereference the object before attempting to delete it.

by Ged Murphy

There's no lock on the list access. On 29 May 2010 07:51, <mjmartin(a)svn.reactos.org> wrote: > Author: mjmartin > Date: Sat May 29 08:51:03 2010 > New Revision: 47393 > > URL: http://svn.reactos.org/svn/reactos?rev=47393&view=rev > Log: > [win32k] > - The timer is created usingUserCreateObject. It may be a good idea to save > the handle in the timer object so that it can be deleted later. > - Dereference the object before attempting to delete it. > > Modified: > trunk/reactos/subsystems/win32/win32k/ntuser/timer.c > > Modified: trunk/reactos/subsystems/win32/win32k/ntuser/timer.c > URL: > http://svn.reactos.org/svn/reactos/trunk/reactos/subsystems/win32/win32k/nt… > > ============================================================================== > --- trunk/reactos/subsystems/win32/win32k/ntuser/timer.c [iso-8859-1] > (original) > +++ trunk/reactos/subsystems/win32/win32k/ntuser/timer.c [iso-8859-1] Sat > May 29 08:51:03 2010 > @@ -50,13 +50,21 @@ > if (!FirstpTmr) > { > FirstpTmr = UserCreateObject(gHandleTable, NULL, &Handle, otTimer, > sizeof(TIMER)); > - if (FirstpTmr) InitializeListHead(&FirstpTmr->ptmrList); > + if (FirstpTmr) > + { > + FirstpTmr->head.h = Handle; > + InitializeListHead(&FirstpTmr->ptmrList); > + } > Ret = FirstpTmr; > } > else > { > Ret = UserCreateObject(gHandleTable, NULL, &Handle, otTimer, > sizeof(TIMER)); > - if (Ret) InsertTailList(&FirstpTmr->ptmrList, &Ret->ptmrList); > + if (Ret) > + { > + Ret->head.h = Handle; > + InsertTailList(&FirstpTmr->ptmrList, &Ret->ptmrList); > + } > } > return Ret; > } > @@ -66,14 +74,17 @@ > FASTCALL > RemoveTimer(PTIMER pTmr) > { > + BOOL Ret = FALSE; > if (pTmr) > { > /* Set the flag, it will be removed when ready */ > RemoveEntryList(&pTmr->ptmrList); > - UserDeleteObject( UserHMGetHandle(pTmr), otTimer); > - return TRUE; > - } > - return FALSE; > + UserDereferenceObject(pTmr); > + Ret = UserDeleteObject( UserHMGetHandle(pTmr), otTimer); > + } > + if (!Ret) DPRINT1("Warning unable to delete timer\n"); > + > + return Ret; > } > > PTIMER > @@ -528,9 +539,7 @@ > { > if ((pTmr) && (pTmr->pti == pti) && (pTmr->pWnd == Window)) > { > - RemoveEntryList(&pTmr->ptmrList); > - UserDeleteObject( UserHMGetHandle(pTmr), otTimer); > - TimersRemoved = TRUE; > + TimersRemoved = RemoveTimer(pTmr); > } > pLE = pTmr->ptmrList.Flink; > pTmr = CONTAINING_RECORD(pLE, TIMER, ptmrList); > @@ -557,9 +566,7 @@ > { > if ((pTmr) && (pTmr->pti == pti)) > { > - RemoveEntryList(&pTmr->ptmrList); > - UserDeleteObject( UserHMGetHandle(pTmr), otTimer); > - TimersRemoved = TRUE; > + TimersRemoved = RemoveTimer(pTmr); > } > pLE = pTmr->ptmrList.Flink; > pTmr = CONTAINING_RECORD(pLE, TIMER, ptmrList); > > >

15 years, 3 months

Recent Bootvid Commit

by Ros Arm

Hello! While the state of the original code is horrendous, and removing it was a step in the good direction, this simplified implementation omits many of the optimizations and specific VGA tricks that the original "author" of the code had intended to duplicate. If you'd like, I would be open to talking to eVb (we both have some experience with VGA programming) to create an efficient 4BitBlt routine similar to the original one that has been removed, but one that actually has basis in actual fact and isn't of the nature this original implementation was. I can see the original code was converting packed data into per-plane (planar) data and had special edge handling, but it looks like the original author did not realize this, and most of the modulo-and-shifts appear as bitwise ANDs (compiler optimizations), which is just plain confusing. Since the rest of bootvid is probably the same "quality", we could take a good look at it, both from a coding and legal standpoint, if you'd like. -r Author: gschneider Date: Sun May 30 01:54:47 2010 New Revision: 47431 URL: http://svn.reactos.org/svn/reactos?rev=47431&view=rev Log: [BOOTVID] Dramatically simplify 4bpp blitting routine See issue #5103 for more details. Modified: trunk/reactos/drivers/base/bootvid/i386/vga.c Modified: trunk/reactos/drivers/base/bootvid/i386/vga.c URL: http://svn.reactos.org/svn/reactos/trunk/reactos/drivers/base/bootvid/i386/… ============================================================================== --- trunk/reactos/drivers/base/bootvid/i386/vga.c [iso-8859-1] (original) +++ trunk/reactos/drivers/base/bootvid/i386/vga.c [iso-8859-1] Sun May 30 01:54:47 2010 @@ -402,33 +402,20 @@ IN ULONG BitsPerPixel, IN ULONG Delta) { - ULONG LeftAnd, LeftShifted, LeftPlusOne, LeftPos; - ULONG lMask, rMask; - UCHAR NotlMask; - ULONG Distance; - ULONG DistanceMinusLeftBpp; - ULONG SomeYesNoFlag, SomeYesNoFlag2; - PUCHAR PixelPosition, m; - PUCHAR i, k; - ULONG j; - ULONG x; - ULONG Plane; - UCHAR LeftArray[84]; - PUCHAR CurrentLeft; - PUCHAR l; - ULONG LoopCount; - UCHAR pMask, PlaneShift; - BOOLEAN Odd; - UCHAR Value; + ULONG sx, dx, dy; + UCHAR color; + ULONG offset = 0; + const ULONG Bottom = Top + Height; + const ULONG Right = Left + Width; /* Check if the buffer isn't 4bpp */ if (BitsPerPixel != 4) { /* FIXME: TODO */ DbgPrint("Unhandled BitBlt\n" - "%lxx%lx @ (%lx,%lx)\n" - "Bits Per Pixel %lx\n" - "Buffer: %p. Delta: %lx\n", + "%lux%lu @ (%lu|%lu)\n" + "Bits Per Pixel %lu\n" + "Buffer: %p. Delta: %lu\n", Width, Height, Left, @@ -439,181 +426,28 @@ return; } - /* Get the masks and other values */ - LeftAnd = Left & 0x7; - lMask = lMaskTable[LeftAnd]; - Distance = Width + Left; - rMask = rMaskTable[(Distance - 1) & 0x7]; - Left >>= 3; - - /* Set some values */ - SomeYesNoFlag = FALSE; - SomeYesNoFlag2 = FALSE; - Distance = (Distance - 1) >> 3; - DistanceMinusLeftBpp = Distance - Left; - - /* Check if the distance is equal to the left position and add the masks */ - if (Left == Distance) lMask += rMask; - - /* Check if there's no distance offset */ - if (DistanceMinusLeftBpp) - { - /* Set the first flag on */ - SomeYesNoFlag = TRUE; - - /* Decrease offset and check if we still have one */ - if (--DistanceMinusLeftBpp) - { - /* Still have a distance offset */ - SomeYesNoFlag2 = TRUE; - } - } - - /* Calculate initial pixel position */ - PixelPosition = (PUCHAR)VgaBase + (Top * 80) + Left; - - /* Set loop buffer variable */ - i = Buffer; - - /* Switch to mode 0 */ - ReadWriteMode(0); - - /* Leave now if the height is 0 */ - if (Height <= 0) return; - - /* Set more weird values */ - CurrentLeft = &LeftArray[Left]; - NotlMask = ~(UCHAR)lMask; - LeftPlusOne = Left + 1; - LeftShifted = (lMask << 8) | 8; - j = Height; - - /* Start the height loop */ + /* 4bpp blitting */ + dy = Top; do { - /* Start the plane loop */ - Plane = 0; + sx = 0; do { - /* Clear the current value */ - *CurrentLeft = 0; - LoopCount = 0; - - /* Set the buffer loop variable for this loop */ - k = i; - - /* Calculate plane shift and pixel mask */ - PlaneShift = 1 << Plane; - pMask = PixelMask[LeftAnd]; - - /* Check if we have a width */ - if (Width > 0) - { - /* Loop it */ - l = CurrentLeft; - x = Width; - do - { - /* Check if we're odd and increase the loop count */ - Odd = LoopCount & 1 ? TRUE : FALSE; - LoopCount++; - if (Odd) - { - /* Check for the plane shift */ - if (*k & PlaneShift) - { - /* Write the pixel mask */ - *l |= pMask; - } - - /* Increase buffer position */ - k++; - } - else - { - /* Check for plane shift */ - if ((*k >> 4) & PlaneShift) - { - /* Write the pixel mask */ - *l |= pMask; - } - } - - /* Shift the pixel mask */ - pMask >>= 1; - if (!pMask) - { - /* Move to the next current left position and clear it */ - l++; - *l = 0; - - /* Set the pixel mask to 0x80 */ - pMask = 0x80; - } - } while (--x); - } - - /* Set the plane value */ - __outpw(0x3C4, (1 << (Plane + 8) | 2)); - - /* Select the bitmask register and write the mask */ - __outpw(0x3CE, (USHORT)LeftShifted); - - /* Read the current Pixel value */ - Value = READ_REGISTER_UCHAR(PixelPosition); - - /* Add our mask */ - Value = (Value & NotlMask) | *CurrentLeft; - - /* Set current left for the loop, and write new pixel value */ - LeftPos = LeftPlusOne; - WRITE_REGISTER_UCHAR(PixelPosition, Value); - - /* Set loop pixel position and check if we should loop */ - m = PixelPosition + 1; - if (SomeYesNoFlag2) - { - /* Set the bitmask to 0xFF for all 4 planes */ - __outpw(0x3CE, 0xFF08); - - /* Check if we have any distance left */ - if (DistanceMinusLeftBpp > 0) - { - /* Start looping it */ - x = DistanceMinusLeftBpp; - do - { - /* Write the value */ - WRITE_REGISTER_UCHAR(m, LeftArray[LeftPos]); - - /* Go to the next position */ - m++; - LeftPos++; - } while (--x); - } - } - - /* Check if the first flag is on */ - if (SomeYesNoFlag) - { - /* Set the mask value */ - __outpw(0x3CE, (rMask << 8) | 8); - - /* Read the current Pixel value */ - Value = READ_REGISTER_UCHAR(m); - - /* Add our mask */ - Value = (Value & ~(UCHAR)rMask) | LeftArray[LeftPos]; - - /* Set current left for the loop, and write new pixel value */ - WRITE_REGISTER_UCHAR(m, Value); - } - } while (++Plane < 4); - - /* Update pixel position, buffer and height */ - PixelPosition += 80; - i += Delta; - } while (--j); + /* Extract color */ + color = Buffer[offset + sx]; + + /* Calc destination x */ + dx = Left + (sx << 1); + + /* Set two pixels */ + SetPixel(dx, dy, color >> 4); + SetPixel(dx + 1, dy, color & 0x0F); + + sx++; + } while (dx < Right); + offset += Delta; + dy++; + } while (dy < Bottom); } VOID

15 years, 3 months

Translations of usrmgr cpl

by Eric Kohl

Hi! It is really funny to see that translators don't seem to understand the simple meaning of: /* * Attention Translators: * DO NOT TRANSLATE THESE RESOURCES YET! */ It means: DO NOT TRANSLATE THESE RESOURCES YET!!!!!!! Got it??? Yes, but why? Because usrmgr is not finished yet from a programmers point of view. Most of the underlying infrastructure ist entirely missing. Right now, usrmgr is completely useless and its look and feel might change without prior notice. So please don't waste your precious time by adding more translations! Please translate other missing texts instead, but leave usrmgr alone. At least as long as the little note at the top is still in place. Thank you very much for your attention! Regards, Eric

15 years, 3 months

"- Setup can not handle more than one primary partition per disk."

by Javier Agustìn Fernàndez Arroyo

hello, isnt this string at usetup obsolete? AFAIK usetup does allow more partitions from a long time now....

15 years, 3 months

Kernel Question

by ahmet alper parker

Hi all, I am new to this mail group and have a question that probably needs your kind of technical knowledge to answer. I want to use reactos, however, up to its completion, can I use some libraries on windows to replace them? In example, wine libraries can be superseded by original windows libraries if you own them. I want to make the inverse of this simply for security reasons to use a kernel fully opensource and which have a documented api. So can I use ntkrnlos.exe and hal.dll or similar in original windows installations to replace propriety kernel and libraries for security reasons? If yes, which of them should I use, and are they complete to replace them? Also a second question may arise as "the priority of reactos development can be focused on firstly these parts of the project or not?"... Best Regards... Ahmet Alper Parker

15 years, 3 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