Ros-dev August 2008

ros-dev@reactos.org

20 participants
35 discussions

Re: [ros-dev] [ros-diffs] [sginsberg] 35600: - Remove KEBUGCHECK and KEBUGCHECKEX macros - Replace "KeBugCheck(0)" by ASSERT(FALSE) - Replace deprecated "CPRINT" by DRINT1

by Hervé Poussineau

sginsberg(a)svn.reactos.org a écrit : > Author: sginsberg > Date: Sun Aug 24 10:48:05 2008 > New Revision: 35600 > > URL: http://svn.reactos.org/svn/reactos?rev=35600&view=rev > Log: > - Remove KEBUGCHECK and KEBUGCHECKEX macros > - Replace "KeBugCheck(0)" by ASSERT(FALSE) > - Replace deprecated "CPRINT" by DRINT1 > You probably understand that KEBUGCHECK(0) is *NOT* the same as ASSERT(FALSE) ASSERT(FALSE) is valid only in debug mode, and lets the code continue in release mode. KEBUGCHECK stops the execution also in release mode. For example, in ntoskrnl/dbgk/dbgkobj.c, code now continues even if DbgkpQueueMessage failed. Are you sure it is safe to continue in that case? (same question applies for other locations) Hervé

16 years

- In IoAcquire/ReleaseCancelSpinLock, use the queued "IopCancelSpinLock" instead of a ros-specific "CancelSpinLock" standard spinlock - Misc fixes

by Stefan Ginsberg

In the kernel, stdcall IoCallDriver is macroed to the fastcall IofCallDriver. However, for clarification I think that IoCallDriver should be used, and this is what we do everywhere else. >Why this? >> ====================================================================== >> --- trunk/reactos/ntoskrnl/io/iomgr/iofunc.c [iso-8859-1] (original) >> +++ trunk/reactos/ntoskrnl/io/iomgr/iofunc.c [iso-8859-1] Sat Aug 23 >> 11:30:14 2008 >> @@ -682,7 +682,7 @@ >> StackPtr->FileObject = FileObject; >> >> /* Call the Driver */ >> - return IofCallDriver(DeviceObject, Irp); >> + return IoCallDriver(DeviceObject, Irp); >> } >> >> /* >> @@ -732,7 +732,7 @@ >> StackPtr->FileObject = FileObject; >> >> /* Call the Driver */ >> - return IofCallDriver(DeviceObject, Irp); >> + return IoCallDriver(DeviceObject, Irp); >> } >> >> _________________________________________________________________ Discover the new Windows Vista http://search.msn.com/results.aspx?q=windows+vista&mkt=en-US&form=QBRE

16 years

Re: [ros-dev] [ros-diffs] [arty] 35581: Catch failure to release the cancel spinlock. Would've spotted my error in tcpip.sys.

by Alex Ionescu

--- Please, please, don't think I'm trying to say "revert this!" --- There are dozens (if not more) places in the kernel where such checks could/should be made. I believe the more correct approach instead of adding them to every function, is to use a similar framework to Microsoft's Driver Verifier, where such functions are hooked and wrapped through similar verifications. I'm not saying this particular assert should be removed, I'm just letting people know that if there's going to be major work on this -- there's a better way. Just my 2c. On 23-Aug-08, at 4:04 PM, arty(a)svn.reactos.org wrote: > Author: arty > Date: Sat Aug 23 18:04:15 2008 > New Revision: 35581 > > URL: http://svn.reactos.org/svn/reactos?rev=35581&view=rev > Log: > Catch failure to release the cancel spinlock. Would've spotted my > error in > tcpip.sys. > > Modified: > trunk/reactos/ntoskrnl/io/iomgr/irp.c > > Modified: trunk/reactos/ntoskrnl/io/iomgr/irp.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/io/iomgr/irp.c?re… > = > = > = > = > = > = > = > = > ====================================================================== > --- trunk/reactos/ntoskrnl/io/iomgr/irp.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/io/iomgr/irp.c [iso-8859-1] Sat Aug 23 > 18:04:15 2008 > @@ -970,12 +970,14 @@ > IoCancelIrp(IN PIRP Irp) > { > KIRQL OldIrql; > + KIRQL IrqlAtEntry; > PDRIVER_CANCEL CancelRoutine; > IOTRACE(IO_IRP_DEBUG, > "%s - Canceling IRP %p\n", > __FUNCTION__, > Irp); > ASSERT(Irp->Type == IO_TYPE_IRP); > + IrqlAtEntry = KeGetCurrentIrql(); > > /* Acquire the cancel lock and cancel the IRP */ > IoAcquireCancelSpinLock(&OldIrql); > @@ -999,6 +1001,7 @@ > /* Set the cancel IRQL And call the routine */ > Irp->CancelIrql = OldIrql; > CancelRoutine(IoGetCurrentIrpStackLocation(Irp)- > >DeviceObject, Irp); > + ASSERT(IrqlAtEntry == KeGetCurrentIrql()); > return TRUE; > } > > Best regards, Alex Ionescu

16 years

Re: [ros-dev] [ros-diffs] [sginsberg] 35565: - In IoAcquire/ReleaseCancelSpinLock, use the queued "IopCancelSpinLock" instead of a ros-specific "CancelSpinLock" standard spinlock - Misc fixes

by Alex Ionescu

Please don't write formatting fixes as "Misc Fixes". It's a pain for reviewers who expect actual functionality fixes, and also for changelog writers, which will try to understand what exactly you fixed. Furthermore, such formatting patches should have [FORMATTING] attached to them. And finally, formatting patches should not be mixed with code patches. I know you're new to trunk, but perhaps you should read over the rules first. I appreciate your patches, but you have a bit of a cowboy attitude with your commits, which is bound to hurt someone in the end. On 23-Aug-08, at 9:30 AM, sginsberg(a)svn.reactos.org wrote: > Author: sginsberg > Date: Sat Aug 23 11:30:14 2008 > New Revision: 35565 > > URL: http://svn.reactos.org/svn/reactos?rev=35565&view=rev > Log: > - In IoAcquire/ReleaseCancelSpinLock, use the queued > "IopCancelSpinLock" instead of a ros-specific "CancelSpinLock" > standard spinlock > - Misc fixes > > Modified: > trunk/reactos/ntoskrnl/io/iomgr/error.c > trunk/reactos/ntoskrnl/io/iomgr/iofunc.c > trunk/reactos/ntoskrnl/io/iomgr/iomgr.c > trunk/reactos/ntoskrnl/io/iomgr/irp.c > trunk/reactos/ntoskrnl/io/iomgr/util.c > > Modified: trunk/reactos/ntoskrnl/io/iomgr/error.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/io/iomgr/error.c?… > = > = > = > = > = > = > = > = > ====================================================================== > --- trunk/reactos/ntoskrnl/io/iomgr/error.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/io/iomgr/error.c [iso-8859-1] Sat Aug 23 > 11:30:14 2008 > @@ -122,7 +122,7 @@ > NULL); > if (NT_SUCCESS(Status)) > { > - /* Remmeber we're connected */ > + /* Remember we're connected */ > IopLogPortConnected = TRUE; > return TRUE; > } > @@ -583,7 +583,7 @@ > KeAcquireSpinLock(&IopLogListLock, &Irql); > InsertHeadList(&IopErrorLogListHead, &LogEntry->ListEntry); > > - /* Check if the worker is runnign */ > + /* Check if the worker is running */ > if (!IopLogWorkerRunning) > { > #if 0 > @@ -648,7 +648,7 @@ > IN PKTHREAD Thread) > { > UNIMPLEMENTED; > - return(FALSE); > + return FALSE; > } > > /* > @@ -659,14 +659,12 @@ > IoSetThreadHardErrorMode(IN BOOLEAN HardErrorEnabled) > { > PETHREAD Thread = PsGetCurrentThread(); > - BOOLEAN Old; > + BOOLEAN OldMode; > > /* Get the current value */ > - Old = !Thread->HardErrorsAreDisabled; > + OldMode = !Thread->HardErrorsAreDisabled; > > /* Set the new one and return the old */ > Thread->HardErrorsAreDisabled = !HardErrorEnabled; > - return Old; > -} > - > -/* EOF */ > + return OldMode; > +} > > Modified: trunk/reactos/ntoskrnl/io/iomgr/iofunc.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/io/iomgr/iofunc.c… > = > = > = > = > = > = > = > = > ====================================================================== > --- trunk/reactos/ntoskrnl/io/iomgr/iofunc.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/io/iomgr/iofunc.c [iso-8859-1] Sat Aug 23 > 11:30:14 2008 > @@ -682,7 +682,7 @@ > StackPtr->FileObject = FileObject; > > /* Call the Driver */ > - return IofCallDriver(DeviceObject, Irp); > + return IoCallDriver(DeviceObject, Irp); > } > > /* > @@ -732,7 +732,7 @@ > StackPtr->FileObject = FileObject; > > /* Call the Driver */ > - return IofCallDriver(DeviceObject, Irp); > + return IoCallDriver(DeviceObject, Irp); > } > > /* > @@ -776,7 +776,7 @@ > } > > /* > - * @unimplemented > + * @implemented > */ > NTSTATUS > NTAPI > @@ -2045,7 +2045,7 @@ > CapturedByteOffset = FileObject->CurrentByteOffset; > } > > - /* Rememer we are sync */ > + /* Remember we are sync */ > Synchronous = TRUE; > } > else if (!(ByteOffset) && > @@ -2904,7 +2904,7 @@ > CapturedByteOffset = FileObject->CurrentByteOffset; > } > > - /* Rememer we are sync */ > + /* Remember we are sync */ > Synchronous = TRUE; > } > else if (!(ByteOffset) && > > Modified: trunk/reactos/ntoskrnl/io/iomgr/iomgr.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/io/iomgr/iomgr.c?… > = > = > = > = > = > = > = > = > ====================================================================== > --- trunk/reactos/ntoskrnl/io/iomgr/iomgr.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/io/iomgr/iomgr.c [iso-8859-1] Sat Aug 23 > 11:30:14 2008 > @@ -68,7 +68,6 @@ > extern LIST_ENTRY IopTimerQueueHead; > extern KDPC IopTimerDpc; > extern KTIMER IopTimer; > -extern KSPIN_LOCK CancelSpinLock; > extern KSPIN_LOCK IoVpbLock; > extern KSPIN_LOCK IoStatisticsLock; > extern KSPIN_LOCK DriverReinitListLock; > @@ -463,7 +462,6 @@ > InitializeListHead(&LastChanceShutdownListHead); > InitializeListHead(&FsChangeNotifyListHead); > InitializeListHead(&IopErrorLogListHead); > - KeInitializeSpinLock(&CancelSpinLock); > KeInitializeSpinLock(&IoVpbLock); > KeInitializeSpinLock(&IoStatisticsLock); > KeInitializeSpinLock(&DriverReinitListLock); > > Modified: trunk/reactos/ntoskrnl/io/iomgr/irp.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/io/iomgr/irp.c?re… > = > = > = > = > = > = > = > = > ====================================================================== > --- trunk/reactos/ntoskrnl/io/iomgr/irp.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/io/iomgr/irp.c [iso-8859-1] Sat Aug 23 > 11:30:14 2008 > @@ -1,8 +1,7 @@ > - > /* > * PROJECT: ReactOS Kernel > * LICENSE: GPL - See COPYING in the top level directory > - * FILE: ntoskrnl/io/irp.c > + * FILE: ntoskrnl/io/iomgr/irp.c > * PURPOSE: IRP Handling Functions > * PROGRAMMERS: Alex Ionescu (alex.ionescu(a)reactos.org) > * Gunnar Dalsnes > @@ -627,7 +626,7 @@ > > /* Allocate IRP */ > Irp = IoAllocateIrp(DeviceObject->StackSize, FALSE); > - if (!Irp) return Irp; > + if (!Irp) return NULL; > > /* Get the Stack */ > StackPtr = IoGetNextIrpStackLocation(Irp); > @@ -731,7 +730,7 @@ > Irp->UserIosb = IoStatusBlock; > Irp->Tail.Overlay.Thread = PsGetCurrentThread(); > > - /* Set the Status Block after all is done */ > + /* Return the IRP */ > IOTRACE(IO_IRP_DEBUG, > "%s - Built IRP %p with Major, Buffer, DO %lx %p %p\n", > __FUNCTION__, > @@ -763,7 +762,7 @@ > > /* Allocate IRP */ > Irp = IoAllocateIrp(DeviceObject->StackSize, FALSE); > - if (!Irp) return Irp; > + if (!Irp) return NULL; > > /* Get the Stack */ > StackPtr = IoGetNextIrpStackLocation(Irp); > @@ -857,6 +856,7 @@ > } > else > { > + /* Clear the flags */ > Irp->Flags = 0; > } > > @@ -1058,7 +1058,12 @@ > * Don't stay here forever if some broken driver doesn't > complete > * the IRP. > */ > - if (!(Retries--)) IopRemoveThreadIrp(); > + if (!(Retries--)) > + { > + /* Print out a message and remove the IRP */ > + DPRINT1("Broken driver did not complete!\n"); > + IopRemoveThreadIrp(); > + } > > /* Raise the IRQL Again */ > KeRaiseIrql(APC_LEVEL, &OldIrql); > @@ -1076,7 +1081,7 @@ > IoCallDriver(IN PDEVICE_OBJECT DeviceObject, > IN PIRP Irp) > { > - /* Call fast call */ > + /* Call fastcall */ > return IofCallDriver(DeviceObject, Irp); > } > > @@ -1112,7 +1117,7 @@ > IN PIRP Irp) > { > PDRIVER_OBJECT DriverObject; > - PIO_STACK_LOCATION Param; > + PIO_STACK_LOCATION StackPtr; > > /* Get the Driver Object */ > DriverObject = DeviceObject->DriverObject; > @@ -1126,15 +1131,15 @@ > } > > /* Now update the stack location */ > - Param = IoGetNextIrpStackLocation(Irp); > - Irp->Tail.Overlay.CurrentStackLocation = Param; > + StackPtr = IoGetNextIrpStackLocation(Irp); > + Irp->Tail.Overlay.CurrentStackLocation = StackPtr; > > /* Get the Device Object */ > - Param->DeviceObject = DeviceObject; > + StackPtr->DeviceObject = DeviceObject; > > /* Call it */ > - return DriverObject->MajorFunction[Param->MajorFunction] > (DeviceObject, > - Irp); > + return DriverObject->MajorFunction[StackPtr->MajorFunction] > (DeviceObject, > + Irp); > } > > FORCEINLINE > @@ -1417,10 +1422,9 @@ > > NTSTATUS > NTAPI > -IopSynchronousCompletion( > - IN PDEVICE_OBJECT DeviceObject, > - IN PIRP Irp, > - IN PVOID Context) > +IopSynchronousCompletion(IN PDEVICE_OBJECT DeviceObject, > + IN PIRP Irp, > + IN PVOID Context) > { > if (Irp->PendingReturned) > KeSetEvent((PKEVENT)Context, IO_NO_INCREMENT, FALSE); > @@ -1464,6 +1468,7 @@ > KeWaitForSingleObject(&Event, Suspended, KernelMode, FALSE, > NULL); > } > > + /* Return success */ > return TRUE; > } > > @@ -1550,10 +1555,12 @@ > /* > * @implemented > */ > -PEPROCESS NTAPI > +PEPROCESS > +NTAPI > IoGetRequestorProcess(IN PIRP Irp) > { > - return(Irp->Tail.Overlay.Thread->ThreadsProcess); > + /* Return the requestor process */ > + return Irp->Tail.Overlay.Thread->ThreadsProcess; > } > > /* > @@ -1563,6 +1570,7 @@ > NTAPI > IoGetRequestorProcessId(IN PIRP Irp) > { > + /* Return the requestor process' id */ > return (ULONG)(IoGetRequestorProcess(Irp)->UniqueProcessId); > } > > @@ -1586,6 +1594,7 @@ > NTAPI > IoGetTopLevelIrp(VOID) > { > + /* Return the IRP */ > return (PIRP)PsGetCurrentThread()->TopLevelIrp; > } > > @@ -1736,5 +1745,3 @@ > /* Set the IRP */ > PsGetCurrentThread()->TopLevelIrp = (ULONG)Irp; > } > - > -/* EOF */ > > Modified: trunk/reactos/ntoskrnl/io/iomgr/util.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/io/iomgr/util.c?r… > = > = > = > = > = > = > = > = > ====================================================================== > --- trunk/reactos/ntoskrnl/io/iomgr/util.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/io/iomgr/util.c [iso-8859-1] Sat Aug 23 > 11:30:14 2008 > @@ -1,9 +1,11 @@ > /* > * PROJECT: ReactOS Kernel > * LICENSE: GPL - See COPYING in the top level directory > - * FILE: ntoskrnl/io/util.c > + * FILE: ntoskrnl/io/iomgr/util.c > * PURPOSE: I/O Utility Functions > - * PROGRAMMERS: <UNKNOWN> > + * PROGRAMMERS: Alex Ionescu (alex.ionescu(a)reactos.org) > + * Aleksey Bragin (aleksey(a)reactos.org) > + * Daniel Zimmerman (netzimme(a)aim.com) > */ > > /* INCLUDES > *****************************************************************/ > @@ -17,10 +19,6 @@ > RtlpGetStackLimits(PULONG_PTR StackBase, > PULONG_PTR StackLimit); > > -/* DATA > **********************************************************************/ > - > -KSPIN_LOCK CancelSpinLock; > - > /* FUNCTIONS > *****************************************************************/ > > /* > @@ -28,10 +26,10 @@ > */ > VOID > NTAPI > -IoAcquireCancelSpinLock(PKIRQL Irql) > +IoAcquireCancelSpinLock(OUT PKIRQL Irql) > { > /* Just acquire the internal lock */ > - KeAcquireSpinLock(&CancelSpinLock,Irql); > + *Irql = KeAcquireQueuedSpinLock(LockQueueIoCancelLock); > } > > /* > @@ -113,6 +111,7 @@ > NTAPI > IoGetCurrentProcess(VOID) > { > + /* Return the current thread's process */ > return (PEPROCESS)PsGetCurrentThread()->Tcb.ApcState.Process; > } > > @@ -121,10 +120,10 @@ > */ > VOID > NTAPI > -IoReleaseCancelSpinLock(KIRQL Irql) > +IoReleaseCancelSpinLock(IN KIRQL Irql) > { > /* Release the internal lock */ > - KeReleaseSpinLock(&CancelSpinLock,Irql); > + KeReleaseQueuedSpinLock(LockQueueIoCancelLock, Irql); > } > > /* > @@ -315,4 +314,3 @@ > UNIMPLEMENTED; > return STATUS_NOT_IMPLEMENTED; > } > -/* EOF */ > Best regards, Alex Ionescu

16 years

Re: [ros-dev] [ros-diffs] [tkreuzer] 35518: Implement KiSystemStartup in C instead of asm, implement KiSwapStack intrinsic, get rid of KiSetupStackandInitializeKernel, instead do the work in KiSystemStartupReal in C. Move Stack definitions into trap.S and get rid of the whole boot.S file. Comment out the sync loop for the moment and add a comment why it doesn't work (InterlockedBitTestAndSet64 doesn't work correctly).

by Thomas Bluemel

Calling from KiSwapStack from C seems quite scary. C doesn't know anything about you messing with the stack, so any subsequent code may still refer to a now invalid stack. There's a reason this was implemented in asm... tkreuzer(a)svn.reactos.org wrote: > Author: tkreuzer > Date: Thu Aug 21 16:16:57 2008 > New Revision: 35518 > > URL: http://svn.reactos.org/svn/reactos?rev=35518&view=rev > Log: > Implement KiSystemStartup in C instead of asm, implement KiSwapStack intrinsic, get rid of KiSetupStackandInitializeKernel, instead do the work in KiSystemStartupReal in C. Move Stack definitions into trap.S and get rid of the whole boot.S file. Comment out the sync loop for the moment and add a comment why it doesn't work (InterlockedBitTestAndSet64 doesn't work correctly). > > Removed: > branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/boot.S > Modified: > branches/ros-amd64-bringup/reactos/ntoskrnl/include/internal/amd64/intrin_i.h > branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/kiinit.c > branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/trap.S > branches/ros-amd64-bringup/reactos/ntoskrnl/ntoskrnl-amd64hack.rbuild > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/include/internal/amd64/intrin_i.h > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > ============================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/include/internal/amd64/intrin_i.h [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/include/internal/amd64/intrin_i.h [iso-8859-1] Thu Aug 21 16:16:57 2008 > @@ -10,6 +10,17 @@ > #define KeSetCurrentIrql(x) __writecr8(x) > > #if defined(__GNUC__) > + > +ULONG64 > +FORCEINLINE > +KiSwapStack(ULONG64 NewStack) > +{ > + ULONG64 OldStack; > + asm volatile ("movq %%rsp, %[oldstack]\n movq %[newstack], %%rsp\n" > + : [oldstack] "=rm" (OldStack) > + : [newstack] "rm" (NewStack)); > + return OldStack; > +} > > #define Ke386SetInterruptDescriptorTable(X) \ > __asm__("lidt %0\n\t" \ > > Removed: branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/boot.S > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > ============================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/boot.S [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/boot.S (removed) > @@ -1,86 +1,0 @@ > -/* > - * FILE: ntoskrnl/ke/i386/boot.S > - * COPYRIGHT: See COPYING in the top level directory > - * PURPOSE: FreeLDR Wrapper Bootstrap Code and Bootstrap Trampoline > - * PROGRAMMERs: Alex Ionescu (alex(a)relsoft.net) > - * Thomas Weidenmueller <w3seek(a)reactos.org> > - */ > - > -/* INCLUDES ******************************************************************/ > - > -#include <asm.h> > -.intel_syntax noprefix > -.code64 > - > -/* GLOBALS *******************************************************************/ > - > -.bss > -.align 16 > - > -/* Kernel Boot Stack */ > -.globl _P0BootStack > -.space KERNEL_STACK_SIZE > -_P0BootStack: > - > -/* Kernel Double-Fault and Temporary DPC Stack */ > -.globl _KiDoubleFaultStack > -.space KERNEL_STACK_SIZE > -_KiDoubleFaultStack: > - > -/* FUNCTIONS *****************************************************************/ > -.global _KiSystemStartup > -.text > -.func KiSystemStartup > -_KiSystemStartup: > - > - /* NTLDR Boot: Call the main kernel initialization */ > - test rcx, 0x80000000 > - jnz _KiSystemStartupReal > - > - /* FREELDR Boot: Cal the FreeLDR wrapper */ > - jmp _KiRosPrepareForSystemStartup > -.endfunc > - > -/** > - * VOID > - * KiSetupStackAndInitializeKernel( > - * esp+4 = ? -> rcx > - * esp+8 = ? -> rdx > - * PVOID pNewstack // esp+12 = new stack -> r8 > - * esp+16 -> r9 > - * esp+20 -> rsp + 8 > - * esp+24 -> rsp + 16? > - */ > -.globl _KiSetupStackAndInitializeKernel > -.func KiSetupStackAndInitializeKernel > -_KiSetupStackAndInitializeKernel: > - > - /* Save current stack */ > - mov rsi, rsp > - > - /* Setup the new stack */ > - mov rsp, r8 > - sub rsp, NPX_FRAME_LENGTH + KTRAP_FRAME_ALIGN + KTRAP_FRAME_LENGTH > - push CR0_EM + CR0_TS + CR0_MP > - > - /* Copy stack parameters to the new stack */ > - push [rsi + 16] > - push [rsi + 8] > - xor rbp, rbp > - call _KiInitializeKernel > - > - /* Set the priority of this thread to 0 */ > - mov rbx, PCR[KPCR_CURRENT_THREAD] > - mov byte ptr [rbx+KTHREAD_PRIORITY], 0 > - > - /* Force interrupts enabled and lower IRQL back to DISPATCH_LEVEL */ > - sti > - mov rcx, DISPATCH_LEVEL > - mov cr8, rcx > - > - /* Set the right wait IRQL */ > - mov byte ptr [rbx+KTHREAD_WAIT_IRQL], DISPATCH_LEVEL; > - > - /* Jump into the idle loop */ > - jmp _KiIdleLoop > -.endfunc > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/kiinit.c > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > ============================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/kiinit.c [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/kiinit.c [iso-8859-1] Thu Aug 21 16:16:57 2008 > @@ -366,6 +366,7 @@ > > /* Start us out at PASSIVE_LEVEL */ > // Pcr->Irql = PASSIVE_LEVEL; > + KeSetCurrentIrql(PASSIVE_LEVEL); > > /* Set the GDI, IDT, TSS and DPC Stack */ > Pcr->GdtBase = (PVOID)Gdt; > @@ -386,6 +387,7 @@ > IN CCHAR Number, > IN PLOADER_PARAMETER_BLOCK LoaderBlock) > { > + FrLdrDbgPrint("Enter KiInitializeKernel\n"); > #if 0 > BOOLEAN NpxPresent; > ULONG FeatureBits; > @@ -642,6 +644,17 @@ > (ULONG64)TssSelector.BaseUpper << 32); > } > > +// Hack > +VOID KiRosPrepareForSystemStartup(ULONG, PROS_LOADER_PARAMETER_BLOCK); > + > +VOID > +NTAPI > +KiSystemStartup(IN ULONG_PTR Dummy, > + IN PROS_LOADER_PARAMETER_BLOCK LoaderBlock) > +{ > + KiRosPrepareForSystemStartup(Dummy, LoaderBlock); > +} > + > VOID > NTAPI > KiSystemStartupReal(IN PLOADER_PARAMETER_BLOCK LoaderBlock) > @@ -740,13 +753,15 @@ > // RtlCopyMemory(&Idt[8], &DoubleFaultEntry, sizeof(KIDTENTRY)); > } > > - > +#if 0 // FIXME: InterlockedBitTestAndSet64 is broken! It needs to be specified > + // that it should reference an absolute address! > /* Loop until we can release the freeze lock */ > do > { > /* Loop until execution can continue */ > while (*(volatile PKSPIN_LOCK*)&KiFreezeExecutionLock == (PVOID)1); > - } while(InterlockedBitTestAndSet((PLONG)&KiFreezeExecutionLock, 0)); > + } while(InterlockedBitTestAndSet64((PLONG64)&KiFreezeExecutionLock, 0)); > +#endif > > /* Setup CPU-related fields */ > Pcr->Prcb.Number = Cpu; > @@ -773,16 +788,29 @@ > KfRaiseIrql(HIGH_LEVEL); > > /* Align stack and make space for the trap frame and NPX frame */ > - InitialStack &= ~(KTRAP_FRAME_ALIGN - 1); > - > -FrLdrDbgPrint("Before KiSetupStackAndInitializeKernel\n"); > -for(;;); > - > - /* Switch to new kernel stack and start kernel bootstrapping */ > - KiSetupStackAndInitializeKernel(&KiInitialProcess.Pcb, > - InitialThread, > - (PVOID)InitialStack, > - &Pcr->Prcb, > - (CCHAR)Cpu, > - KeLoaderBlock); > + InitialStack &= ~(16 - 1); > + > + /* Switch to new kernel Stack */ > + KiSwapStack(InitialStack); > + > + /* Initialize kernel */ > + KiInitializeKernel(&KiInitialProcess.Pcb, > + InitialThread, > + (PVOID)InitialStack, > + &Pcr->Prcb, > + (CCHAR)Cpu, > + KeLoaderBlock); > + > + /* Set the priority of this thread to 0 */ > + InitialThread->Priority = 0; > + > + /* Force interrupts enabled and lower IRQL back to DISPATCH_LEVEL */ > + _enable(); > + KeLowerIrql(DISPATCH_LEVEL); > + > + /* Set the right wait IRQL */ > + InitialThread->WaitIrql = DISPATCH_LEVEL; > + > + /* Jump into the idle loop */ > + KiIdleLoop(); > } > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/trap.S > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > ============================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/trap.S [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ke/amd64/trap.S [iso-8859-1] Thu Aug 21 16:16:57 2008 > @@ -13,7 +13,22 @@ > .intel_syntax noprefix > .code64 > > +#define KERNEL_STACK_SIZE 0x6000 > + > /* GLOBALS *******************************************************************/ > + > +.bss > +.align 16 > + > +/* Kernel Boot Stack */ > +.globl _P0BootStack > +.space KERNEL_STACK_SIZE > +_P0BootStack: > + > +/* Kernel Double-Fault and Temporary DPC Stack */ > +.globl _KiDoubleFaultStack > +.space KERNEL_STACK_SIZE > +_KiDoubleFaultStack: > > .data > > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/ntoskrnl-amd64hack.rbuild > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > ============================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ntoskrnl-amd64hack.rbuild [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ntoskrnl-amd64hack.rbuild [iso-8859-1] Thu Aug 21 16:16:57 2008 > @@ -90,7 +90,6 @@ > </if> > <if property="ARCH" value="amd64"> > <directory name="amd64"> > - <file first="true">boot.S</file> > <file>cpu.c</file> > <file>except.c</file> > <file>irql.c</file> > >

16 years

Re : Ros-dev Digest, Vol 48, Issue 14

by Sylvain Petreolle

Feel free to open a bug. ----- Message d'origine ---- De : Olaf Siejka <caemyr(a)gmail.com> À : ros-dev(a)reactos.org Envoyé le : Jeudi, 21 Août 2008, 22h47mn 20s Objet : Re: [ros-dev] Ros-dev Digest, Vol 48, Issue 14 Not to mention that makex is not 100% functional, for example i cannot perform makex depends or even makex bootcd on a clean tree. I have to first do make depends, then makex bootcd. 2008/8/21 <ros-dev-request(a)reactos.org> Send Ros-dev mailing list submissions to ros-dev(a)reactos.org To subscribe or unsubscribe via the World Wide Web, visit http://www.reactos.org/mailman/listinfo/ros-dev or, via email, send a message with subject or body 'help' to ros-dev-request(a)reactos.org You can reach the person managing the list at ros-dev-owner(a)reactos.org When replying, please edit your Subject line so it is more specific than "Re: Contents of Ros-dev digest..." Today's Topics: 1. Re: [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. (gedmurphy) 2. Re: [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. (Timo Kreuzer) 3. Re: [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. (Colin Finck) 4. Has anyone given any thought to coding so that the OS can use the GPU (Matthew Therault) 5. Re: Has anyone given any thought to coding so that the OS can use the GPU (Steven Edwards) 6. Re: [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. (Aleksey Bragin) ---------------------------------------------------------------------- Message: 1 Date: Wed, 20 Aug 2008 12:59:42 +0100 From: "gedmurphy" <gedmurphy(a)gmail.com> Subject: Re: [ros-dev] [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. To: <ros-dev(a)reactos.org> Message-ID: <000f01c902bc$3ea137f0$bbe3a7d0$@com> Content-Type: text/plain; charset="utf-8" I don't want make to consume both CPU's. Ged. -----Original Message----- From: ros-diffs-bounces(a)reactos.org [mailto:ros-diffs-bounces@reactos.org] On Behalf Of dreimer(a)svn.reactos.org Sent: 20 August 2008 10:10 To: ros-diffs(a)reactos.org Subject: [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. Author: dreimer Date: Wed Aug 20 04:10:07 2008 New Revision: 35473 URL: http://svn.reactos.org/svn/reactos?rev=35473&view=rev Log: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. Modified: trunk/tools/RosBE/RosBE-Windows/Root/Build.cmd trunk/tools/RosBE/RosBE-Windows/Root/RosBE.mac

16 years

Re: [ros-dev] Ros-dev Digest, Vol 48, Issue 14

by Olaf Siejka

Not to mention that makex is not 100% functional, for example i cannot perform makex depends or even makex bootcd on a clean tree. I have to first do make depends, then makex bootcd. 2008/8/21 <ros-dev-request(a)reactos.org> > Send Ros-dev mailing list submissions to > ros-dev(a)reactos.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://www.reactos.org/mailman/listinfo/ros-dev > or, via email, send a message with subject or body 'help' to > ros-dev-request(a)reactos.org > > You can reach the person managing the list at > ros-dev-owner(a)reactos.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Ros-dev digest..." > > > Today's Topics: > > 1. Re: [ros-diffs] [dreimer] 35473: make automatically behaves > like makex when there is more then one CPU Core now. makex kept > for now to keep compatible. (gedmurphy) > 2. Re: [ros-diffs] [dreimer] 35473: make automatically behaves > like makex when there is more then one CPU Core now. makex kept > for now to keep compatible. (Timo Kreuzer) > 3. Re: [ros-diffs] [dreimer] 35473: make automatically behaves > like makex when there is more then one CPU Core now. makex kept > for now to keep compatible. (Colin Finck) > 4. Has anyone given any thought to coding so that the OS can use > the GPU (Matthew Therault) > 5. Re: Has anyone given any thought to coding so that the OS can > use the GPU (Steven Edwards) > 6. Re: [ros-diffs] [dreimer] 35473: make automatically behaves > like makex when there is more then one CPU Core now. makex kept > for now to keep compatible. (Aleksey Bragin) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 20 Aug 2008 12:59:42 +0100 > From: "gedmurphy" <gedmurphy(a)gmail.com> > Subject: Re: [ros-dev] [ros-diffs] [dreimer] 35473: make automatically > behaves like makex when there is more then one CPU Core now. > makex > kept for now to keep compatible. > To: <ros-dev(a)reactos.org> > Message-ID: <000f01c902bc$3ea137f0$bbe3a7d0$@com> > Content-Type: text/plain; charset="utf-8" > > I don't want make to consume both CPU's. > > Ged. > > -----Original Message----- > From: ros-diffs-bounces(a)reactos.org [mailto:ros-diffs-bounces@reactos.org] > On Behalf Of dreimer(a)svn.reactos.org > Sent: 20 August 2008 10:10 > To: ros-diffs(a)reactos.org > Subject: [ros-diffs] [dreimer] 35473: make automatically behaves like makex > when there is more then one CPU Core now. makex kept for now to keep > compatible. > > Author: dreimer > Date: Wed Aug 20 04:10:07 2008 > New Revision: 35473 > > URL: http://svn.reactos.org/svn/reactos?rev=35473&view=rev > Log: > make automatically behaves like makex when there is more then one CPU Core > now. makex kept for now to keep compatible. > > Modified: > trunk/tools/RosBE/RosBE-Windows/Root/Build.cmd > trunk/tools/RosBE/RosBE-Windows/Root/RosBE.mac

16 years

Re: [ros-dev] [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible.

by gedmurphy

I don't want make to consume both CPU's. Ged. -----Original Message----- From: ros-diffs-bounces(a)reactos.org [mailto:ros-diffs-bounces@reactos.org] On Behalf Of dreimer(a)svn.reactos.org Sent: 20 August 2008 10:10 To: ros-diffs(a)reactos.org Subject: [ros-diffs] [dreimer] 35473: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. Author: dreimer Date: Wed Aug 20 04:10:07 2008 New Revision: 35473 URL: http://svn.reactos.org/svn/reactos?rev=35473&view=rev Log: make automatically behaves like makex when there is more then one CPU Core now. makex kept for now to keep compatible. Modified: trunk/tools/RosBE/RosBE-Windows/Root/Build.cmd trunk/tools/RosBE/RosBE-Windows/Root/RosBE.mac Modified: trunk/tools/RosBE/RosBE-Windows/Root/Build.cmd URL: http://svn.reactos.org/svn/reactos/trunk/tools/RosBE/RosBE-Windows/Root/Bui… ============================================================================== --- trunk/tools/RosBE/RosBE-Windows/Root/Build.cmd [iso-8859-1] (original) +++ trunk/tools/RosBE/RosBE-Windows/Root/Build.cmd [iso-8859-1] Wed Aug 20 04:10:07 2008 @@ -103,11 +103,20 @@ ) :: +:: Get the number of CPUs in the system so we know how many jobs to execute. +:: To modify the number used alter the options used with cpucount: +:: No Option - Number of CPUs. +:: -x1 - Number of CPUs, plus 1. +:: -x2 - Number of CPUs, doubled. +:: -a - Determine the cpu count based on the inherited process affinity mask. +:: +for /f "usebackq" %%i in (`"%_ROSBE_BASEDIR%\Tools\cpucount.exe" -x1`) do set CPUCOUNT=%%i +:: :: Check if we are using -j or not. :: -if "%1" == "multi" ( - if not "%2" == "" ( - title 'makex %2' parallel build started: %TIMERAW% +if %CPUCOUNT% GTR 2 ( + if not "%1" == "" ( + title 'makex %1' parallel build started: %TIMERAW% ) else ( title 'makex' parallel build started: %TIMERAW% ) @@ -139,27 +148,17 @@ goto :EOF :BUILDMULTI - :: - :: Get the number of CPUs in the system so we know how many jobs to execute. - :: To modify the number used alter the options used with cpucount: - :: No Option - Number of CPUs. - :: -x1 - Number of CPUs, plus 1. - :: -x2 - Number of CPUs, doubled. - :: -a - Determine the cpu count based on the inherited process affinity mask. - :: - for /f "usebackq" %%i in (`"%_ROSBE_BASEDIR%\Tools\cpucount.exe" -x1`) do set CPUCOUNT=%%i - if %_ROSBE_SHOWTIME% == 1 ( if %_ROSBE_WRITELOG% == 1 ( - "%_ROSBE_BASEDIR%\Tools\buildtime.exe" "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %2 %3 %4 %5 %6 %7 %8 %9 2>&1 | "%_ROSBE_BASEDIR%\Tools\tee.exe" "%_ROSBE_LOGDIR%\BuildLog-%_ROSBE_GCCVERSION%-%DATENAME%-%TIMENAME%.txt" + "%_ROSBE_BASEDIR%\Tools\buildtime.exe" "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %* 2>&1 | "%_ROSBE_BASEDIR%\Tools\tee.exe" "%_ROSBE_LOGDIR%\BuildLog-%_ROSBE_GCCVERSION%-%DATENAME%-%TIMENAME%.txt" ) else ( - "%_ROSBE_BASEDIR%\Tools\buildtime.exe" "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %2 %3 %4 %5 %6 %7 %8 %9 + "%_ROSBE_BASEDIR%\Tools\buildtime.exe" "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %* ) ) else ( if %_ROSBE_WRITELOG% == 1 ( - "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %2 %3 %4 %5 %6 %7 %8 %9 2>&1 | "%_ROSBE_BASEDIR%\Tools\tee.exe" "%_ROSBE_LOGDIR%\BuildLog-%_ROSBE_GCCVERSION%-%DATENAME%-%TIMENAME%.txt" + "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %* 2>&1 | "%_ROSBE_BASEDIR%\Tools\tee.exe" "%_ROSBE_LOGDIR%\BuildLog-%_ROSBE_GCCVERSION%-%DATENAME%-%TIMENAME%.txt" ) else ( - "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %2 %3 %4 %5 %6 %7 %8 %9 + "%_ROSBE_MINGWMAKE%" -j %CPUCOUNT% %* ) ) goto :EOF Modified: trunk/tools/RosBE/RosBE-Windows/Root/RosBE.mac URL: http://svn.reactos.org/svn/reactos/trunk/tools/RosBE/RosBE-Windows/Root/Ros… ============================================================================== --- trunk/tools/RosBE/RosBE-Windows/Root/RosBE.mac [iso-8859-1] (original) +++ trunk/tools/RosBE/RosBE-Windows/Root/RosBE.mac [iso-8859-1] Wed Aug 20 04:10:07 2008 @@ -7,7 +7,7 @@ ENV = set HELP = "%_ROSBE_BASEDIR%\Help.cmd" $* MAKE = "%_ROSBE_BASEDIR%\Build.cmd" $* -MAKEX = "%_ROSBE_BASEDIR%\Build.cmd" multi $* +MAKEX = "%_ROSBE_BASEDIR%\Build.cmd" $* RADDR2LINE = "%_ROSBE_BASEDIR%\reladdr2line.cmd" $* RENV = for /f "usebackq tokens=*" %i in (`set _ROSBE_`) do @echo %i SCUT = "%_ROSBE_BASEDIR%\scut.cmd" $*

16 years

Has anyone given any thought to coding so that the OS can use the GPU

by Matthew Therault

Just a random thought wondering if anyone else has thought about it. They are making great strides into how the GPU can now be used as a processor. -- Two men walk into a bar....... Third one ducks.... Think about it, it will come to you. :P

16 years

Re: [ros-dev] [ros-diffs] [tkreuzer] 35458: Move some architecture specific code from KiRosPrepareForSystemStartup into KiArchInitSystem, defined for each architecture. Enable KiRosBuildReservedMemoryMap for _M_AMD64

by Alex Ionescu

Please rename the function to KiFreeLdrHackKiArchHackHack or something like that, and move it away from cpu.c, somewhere like i386/freeldr.c to make clear that this function has NOTHING TO DO with how NT should boot and is a hack because FreeLDR is a piece of shit that nobody seems to want to fix. On 19-Aug-08, at 12:07 PM, tkreuzer(a)svn.reactos.org wrote: > Author: tkreuzer > Date: Tue Aug 19 14:07:13 2008 > New Revision: 35458 > > URL: http://svn.reactos.org/svn/reactos?rev=35458&view=rev > Log: > Move some architecture specific code from > KiRosPrepareForSystemStartup into KiArchInitSystem, defined for each > architecture. Enable KiRosBuildReservedMemoryMap for _M_AMD64 > > Modified: > branches/ros-amd64-bringup/reactos/ntoskrnl/ke/freeldr.c > branches/ros-amd64-bringup/reactos/ntoskrnl/ke/i386/cpu.c > branches/ros-amd64-bringup/reactos/ntoskrnl/ke/powerpc/cpu.c > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/ke/freeldr.c > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > = > = > = > = > = > = > = > = > ====================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ke/freeldr.c > [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ke/freeldr.c > [iso-8859-1] Tue Aug 19 14:07:13 2008 > @@ -949,7 +949,7 @@ > /* Build entries for ReactOS memory ranges, which uses ARC > Descriptors */ > KiRosBuildOsMemoryMap(); > > -#ifdef _M_IX86 > +#if defined(_M_IX86) || defined(_M_AMD64) > /* Build entries for the reserved map, which uses ARC > Descriptors */ > KiRosBuildReservedMemoryMap(); > #endif > @@ -1266,42 +1266,15 @@ > { > PLOADER_PARAMETER_BLOCK NtLoaderBlock; > ULONG size, i = 0, *ent; > -#if defined(_M_IX86) > - PKTSS Tss; > - PKGDTENTRY TssEntry; > - > - /* Load the GDT and IDT */ > - > Ke386SetGlobalDescriptorTable(*(PKDESCRIPTOR)&KiGdtDescriptor.Limit); > - > Ke386SetInterruptDescriptorTable > (*(PKDESCRIPTOR)&KiIdtDescriptor.Limit); > - > - /* Initialize the boot TSS */ > - Tss = &KiBootTss; > - TssEntry = &KiBootGdt[KGDT_TSS / sizeof(KGDTENTRY)]; > - TssEntry->HighWord.Bits.Type = I386_TSS; > - TssEntry->HighWord.Bits.Pres = 1; > - TssEntry->HighWord.Bits.Dpl = 0; > - TssEntry->BaseLow = (USHORT)((ULONG_PTR)Tss & 0xFFFF); > - TssEntry->HighWord.Bytes.BaseMid = (UCHAR)((ULONG_PTR)Tss >> 16); > - TssEntry->HighWord.Bytes.BaseHi = (UCHAR)((ULONG_PTR)Tss >> 24); > -#endif > - > -#if defined(_M_PPC) > - // Zero bats. We might have residual bats set that will > interfere with > - // our mapping of ofwldr. > - for (i = 0; i < 4; i++) > - { > - SetBat(i, 0, 0, 0); SetBat(i, 1, 0, 0); > - } > - KiSetupSyscallHandler(); > - DbgPrint("Kernel Power (%08x)\n", LoaderBlock); > - DbgPrint("ArchExtra (%08x)!\n", LoaderBlock->ArchExtra); > -#endif > > /* Save pointer to ROS Block */ > KeRosLoaderBlock = LoaderBlock; > > /* Get debugging function */ > FrLdrDbgPrint = LoaderBlock->FrLdrDbgPrint; > + > + /* Per architecture initialisazion code */ > + KiArchInitSystem(); > > /* Save memory manager data */ > KeMemoryMapRangeCount = 0; > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/ke/i386/cpu.c > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > = > = > = > = > = > = > = > = > ====================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ke/i386/cpu.c > [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ke/i386/cpu.c > [iso-8859-1] Tue Aug 19 14:07:13 2008 > @@ -1036,3 +1036,24 @@ > /* Capture the control state */ > KiSaveProcessorControlState(State); > } > + > +VOID > +KiArchInitSystem() > +{ > + PKTSS Tss; > + PKGDTENTRY TssEntry; > + > + /* Load the GDT and IDT */ > + > Ke386SetGlobalDescriptorTable(*(PKDESCRIPTOR)&KiGdtDescriptor.Limit); > + > Ke386SetInterruptDescriptorTable > (*(PKDESCRIPTOR)&KiIdtDescriptor.Limit); > + > + /* Initialize the boot TSS */ > + Tss = &KiBootTss; > + TssEntry = &KiBootGdt[KGDT_TSS / sizeof(KGDTENTRY)]; > + TssEntry->HighWord.Bits.Type = I386_TSS; > + TssEntry->HighWord.Bits.Pres = 1; > + TssEntry->HighWord.Bits.Dpl = 0; > + TssEntry->BaseLow = (USHORT)((ULONG_PTR)Tss & 0xFFFF); > + TssEntry->HighWord.Bytes.BaseMid = (UCHAR)((ULONG_PTR)Tss >> 16); > + TssEntry->HighWord.Bytes.BaseHi = (UCHAR)((ULONG_PTR)Tss >> 24); > +} > > Modified: branches/ros-amd64-bringup/reactos/ntoskrnl/ke/powerpc/cpu.c > URL: http://svn.reactos.org/svn/reactos/branches/ros-amd64-bringup/reactos/ntosk… > = > = > = > = > = > = > = > = > ====================================================================== > --- branches/ros-amd64-bringup/reactos/ntoskrnl/ke/powerpc/cpu.c > [iso-8859-1] (original) > +++ branches/ros-amd64-bringup/reactos/ntoskrnl/ke/powerpc/cpu.c > [iso-8859-1] Tue Aug 19 14:07:13 2008 > @@ -275,3 +275,19 @@ > /* Capture the control state */ > KiSaveProcessorControlState(State); > } > + > +VOID > +KiArchInitSystem() > +{ > + ULONG i; > + > + // Zero bats. We might have residual bats set that will > interfere with > + // our mapping of ofwldr. > + for (i = 0; i < 4; i++) > + { > + SetBat(i, 0, 0, 0); SetBat(i, 1, 0, 0); > + } > + KiSetupSyscallHandler(); > + DbgPrint("Kernel Power (%08x)\n", KeRosLoaderBlock); > + DbgPrint("ArchExtra (%08x)!\n", KeRosLoaderBlock->ArchExtra); > +} > Best regards, Alex Ionescu

16 years

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Ros-dev August 2008