Ros-dev

ros-dev@reactos.org

5788 discussions

msports commit (51813) - CMakeLists.txt references nonexistent files

by Adam

When I was trying to build trunk (using Cmake) the configure.cmd did not complete. Host tools were OK but reactos directory was not. I tracked the issue to /dll/win32/msports/CMakeLists.txt which contained references to two nonexistent files (serial.c and parallel.c) and commenting them out allowed the build to continue (albeit with an error at the very end) there.

13 years, 4 months

Re: [ros-dev] [ros-diffs] [tkreuzer] 51748: [NTOSKRNL] Patch by Paolo Bonzini <bonzini [at] gnu [dot] org> Fix implementation of ExInterlockedAddLargeStatistic The old version wasn't really atomic. See issue #6223 for more ...

by Alex Ionescu

Congrats on once again breaking another key part of Windows compatibility. This function is DOCUMENTED to tear. It was DESIGNED that way on purpose. Utter failure, once more. Best regards, Alex Ionescu On Sat, May 14, 2011 at 12:29 PM, <tkreuzer(a)svn.reactos.org> wrote: > Author: tkreuzer > Date: Sat May 14 19:29:42 2011 > New Revision: 51748 > > URL: http://svn.reactos.org/svn/reactos?rev=51748&view=rev > Log: > [NTOSKRNL] > Patch by Paolo Bonzini <bonzini [at] gnu [dot] org> > Fix implementation of ExInterlockedAddLargeStatistic > The old version wasn't really atomic. > > See issue #6223 for more details. > > Modified: > trunk/reactos/ntoskrnl/ex/i386/fastinterlck_asm.S > > Modified: trunk/reactos/ntoskrnl/ex/i386/fastinterlck_asm.S > URL: > http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/ex/i386/fastinter… > > ============================================================================== > --- trunk/reactos/ntoskrnl/ex/i386/fastinterlck_asm.S [iso-8859-1] > (original) > +++ trunk/reactos/ntoskrnl/ex/i386/fastinterlck_asm.S [iso-8859-1] Sat May > 14 19:29:42 2011 > @@ -4,8 +4,9 @@ > * FILE: ntoskrnl/ex/i386/fastinterlck_asm.S > * PURPOSE: FASTCALL Interlocked Functions > * PROGRAMMERS: Alex Ionescu (alex(a)relsoft.net) > - */ > - > + * Paolo Bonzini <bonzini [at] gnu [dot] org> > + */ > + > /* INCLUDES > ******************************************************************/ > > #include <asm.inc> > @@ -31,25 +32,55 @@ > */ > PUBLIC @ExInterlockedAddLargeStatistic@8 > @ExInterlockedAddLargeStatistic@8: > - > -#ifdef CONFIG_SMP > - /* Do the addition */ > - lock add [ecx], edx > - > - /* Check for carry bit and return */ > - jb .l1 > - ret > - > -.l1: > - /* Add carry */ > - lock adc dword ptr [ecx+4], 0 > -#else > - /* Do the addition and add the carry */ > - add dword ptr [ecx], edx > - adc dword ptr [ecx+4], 0 > -#endif > - /* Return */ > - ret > + push ebp > + push ebx > + mov ebp, ecx > + > +Again: > + /* Load comparand in eax for cmpxchg */ > + mov eax, [ebp] > + > + /* Compute low word of the result in ebx */ > + mov ebx, edx > + add ebx, eax > + > + /* Carry needs cmpxchg8b */ > + jc Slow > + > + /* Fast path still needs to be atomic, so use cmpxchg and retry if it > fails > + * Hopefully it will still get through this path :) */ > + LOCK cmpxchg [ecx], ebx > + jnz Again > + > + /* Thats it */ > + pop ebx > + pop ebp > + ret > + > +Slow: > + /* Save increment across cmpxchg8b */ > + push edx > + > + /* Finish loading comparand in edx:eax */ > + mov edx, [ebp+4] > + > + /* Result in ecx:ebx (we know there's carry) */ > + lea ecx, [edx+1] > + > + /* Do a full exchange */ > + LOCK cmpxchg8b [ebp] > + > + /* restore increment */ > + pop edx > + > + /* Need to retry */ > + jnz Again > + > + /* Thats it */ > + pop ebx > + pop ebp > + ret > + > > /*ULONG > *FASTCALL > > >

13 years, 4 months

Current patch review assignment

by Olaf Siejka

Hiya Attached, you will find a rough draft of current patch assignment proposals. I`m trying to keep it in realtime, so several of those are already in place. As usual, please comment if you find you shouldn't be reviewing assigned patch. Best regards

13 years, 4 months

Compatibility List moderation

by Javier Agustìn Fernàndez Arroyo

Speaking about bug reports drama, im still watching how there are several people that are still sending totally useless compatibility test results to ROS database... I want to ask for a moderation system for that, and i would want to offer myself to moderate new incoming results, such as Wine system....

13 years, 4 months

Re: [ros-dev] [EVENTVWR/EVENTLOG] Fix a double off-by-one bug

by Love Nystrom

Hi Eric, Sounds like you got it licked, the right way. Good job ! Best Regards L. 2011/5/6 <ros-dev-request(a)reactos.org> > From: Eric Kohl <eric.kohl(a)t-online.de> > To: ReactOS Development List <ros-dev(a)reactos.org> > Date: Thu, 05 May 2011 22:41:24 +0200 > Subject: Re: [ros-dev] [EVENTVWR/EVENTLOG] Fix a double off-by-one bug > Hi Love, > > I just committed a fix for the calculation of the available number of event > records. It addresses the issues you mentioned. > > And this is how it works: > When a new log file is created, the first (or oldest) record gets the > record number 1. With each new event record the record number is incremented > by 1. To get the number of records in the log file we need to subtract the > OldestRecordNumber from the CurrentRecordNumber. If the log file is empty > the OldestRecordNumber will be 0. In this case we will return 0 for the > number of event records. > > Now we call clear a filled log file and fill it with new records. The > oldest record will not be record number 1 but we can still return the true > number of records in the log file. > > > Regards > Eric >

13 years, 4 months

Re: [ros-dev] [freeldr] Add PXE "filesystem" (Aleksey Bragin)

by Love Nystrom

Hi Aleksey, I'm sorry if my opinion came across as critique of the work as such. It was definitely not intended that way. I have great respect for all your efforts. Just a point about unfortunate choice of AT&T syntax. I would gladly volunteer to translate all old AT&T code to Intel. After all, I do have 25 years of expertise in assembly coding. It's just that I'm completely bogged down by R.L and have no time. But if it can wait, just leave it be, and I'll do it later. I'm sure Timo has enough to do with other issues already. I abbreviated the pro's and con's to just two points, of which the 2nd applies just to gcc's inline asm syntax. Translating non-inline AT&T to Intel is fairly straight forward. InvalidateHarshMode(TRUE); No offence taken. Best Regards L. On 2011-05-05 14.58, "Aleksey Bragin" <aleksey(a)reactos.org> wrote: > SetHarshMode(TRUE); > > Your opinion would be interesting if it was motivated by your knowledge and experience in writing assembly > in both AT&T and Intel syntax, so that you could provide us pros and cons of each approach by your judgement. > > Or, if you would be author of that code part in ReactOS. > > Otherwise, I’m afraid, it’s just criticizing a very good work, didn’t even noticing the great PXE milestone, > but nitpicking to a syntax. Don’t join complaints if you can’t fix it. (Timo’s point is valid because that’s him who actually > converts this stuff!) > > SetHarshMode(FALSE); > > WBR, > Aleksey.

13 years, 4 months

ReactOS YouTube Channel

by Adam

I noticed this YouTube channel here - http://www.youtube.com/reactosorg - and apparently it seems to have 3 videos about ReactOS on it. Nothing seems to be uploaded to it for a little over 1 year and I was wondering perhaps it wouldn't be a bad idea to upload some videos about various events to it. I suspect a major issue with the lack of developers is lack of publicity and I think this seems like a good opportunity. (note I am not sure if it is the official ReactOS channel or not - if not time to make one - apparently it seems the person who made the channel is displaying the Age as 16 years old)

13 years, 4 months

Apologies

by Olaf Siejka

I am sorry to Vicmarcal for outbursting on ros-dev. The reason for that was yet another pointless bugreport, after my requestes, passed several times, regarding bug reports quality. Sorry, Vicmarcal, as well as those who had to witness it. Regards

13 years, 4 months

Vicmarcal - are you mad?

by Olaf Siejka

Bugzilla is crowded with untouched bugs for so long, and you HAVE to add more crud to it? Bugs like no.6214, very usefull as we can see below: "1)Download ClickToDesktop 2)Run in ReactOS 3)Click on the desktop to hide. It doesn't work. Apps just lost the focus. Nothing appears in the DebugLog." Now that is informative... Why dont you get back doing something useful for a change?

13 years, 4 months

Re: [ros-dev] [EVENTVWR/EVENTLOG] Fix a double off-by-one bug

by Love Nystrom

Hi Gabriel, Eric, Isn't the real issue why LogFile->Header.CurrentRecordNumber is off by one ? If that is found and fixed, this patch will make the error re-emerge as an off by -1. And if CurrentRecordNumber is actually used as the next *pending* record number, the identifier is somewhat mis-leading .. ? Just my penny to the pot Best Regards L. ---------- Forwarded message ---------- > From: Gabriel ilardi <gabrielilardi(a)hotmail.it> > To: ros-dev <ros-dev(a)reactos.org> > Date: Mon, 2 May 2011 21:50:03 +0200 > Subject: Re: [ros-dev] [ros-diffs] [ekohl] 51558: [EVENTVWR/EVENTLOG] Fix a > double off-by-one bug: - The eventlog service was reporting one event more > than was available (+1). - The event viewer did not display the latest event > from... > Great work on event service and event viewer... keep it up Eric :) > > > > Date: Mon, 2 May 2011 19:38:24 +0000 > > To: ros-diffs(a)reactos.org > > From: ekohl(a)svn.reactos.org > > Subject: [ros-diffs] [ekohl] 51558: [EVENTVWR/EVENTLOG] Fix a double > off-by-one bug: - The eventlog service was reporting one event more than was > available (+1). - The event viewer did not display the latest event from... > > > > Author: ekohl > > Date: Mon May 2 19:38:23 2011 > > New Revision: 51558 > > > > URL: http://svn.reactos.org/svn/reactos?rev=51558&view=rev > > Log: > > [EVENTVWR/EVENTLOG] > > Fix a double off-by-one bug: > > - The eventlog service was reporting one event more than was available > (+1). > > - The event viewer did not display the latest event from the eventlog > service (-1). > > > > See issue #6182 for more details. > > > > Modified: > > trunk/reactos/base/applications/mscutils/eventvwr/eventvwr.c > > trunk/reactos/base/services/eventlog/rpc.c > > > > Modified: trunk/reactos/base/applications/mscutils/eventvwr/eventvwr.c > > URL: > http://svn.reactos.org/svn/reactos/trunk/reactos/base/applications/mscutils… > > > ============================================================================== > > --- trunk/reactos/base/applications/mscutils/eventvwr/eventvwr.c > [iso-8859-1] (original) > > +++ trunk/reactos/base/applications/mscutils/eventvwr/eventvwr.c > [iso-8859-1] Mon May 2 19:38:23 2011 > > @@ -503,7 +503,7 @@ > > HWND hwndDlg; > > HANDLE hEventLog; > > EVENTLOGRECORD *pevlr; > > - DWORD dwRead, dwNeeded, dwThisRecord, dwTotalRecords = 0, > dwCurrentRecord = 1, dwRecordsToRead = 0, dwFlags, dwMaxLength; > > + DWORD dwRead, dwNeeded, dwThisRecord, dwTotalRecords = 0, > dwCurrentRecord = 0, dwRecordsToRead = 0, dwFlags, dwMaxLength; > > LPWSTR lpSourceName; > > LPWSTR lpComputerName; > > LPSTR lpData; > > > > Modified: trunk/reactos/base/services/eventlog/rpc.c > > URL: > http://svn.reactos.org/svn/reactos/trunk/reactos/base/services/eventlog/rpc… > > > ============================================================================== > > --- trunk/reactos/base/services/eventlog/rpc.c [iso-8859-1] (original) > > +++ trunk/reactos/base/services/eventlog/rpc.c [iso-8859-1] Mon May 2 > 19:38:23 2011 > > @@ -199,6 +199,7 @@ > > DWORD *NumberOfRecords) > > { > > PLOGHANDLE lpLogHandle; > > + DWORD dwRecords; > > > > lpLogHandle = ElfGetLogHandleEntryByHandle(LogHandle); > > if (!lpLogHandle) > > @@ -206,7 +207,9 @@ > > return STATUS_INVALID_HANDLE; > > } > > > > - *NumberOfRecords = lpLogHandle->LogFile->Header.CurrentRecordNumber; > > + dwRecords = lpLogHandle->LogFile->Header.CurrentRecordNumber; > > + > > + *NumberOfRecords = (dwRecords > 0) ? (dwRecords - 1) : 0; > > > > return STATUS_SUCCESS; > > } > > > > >

13 years, 4 months

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Ros-dev