[ros-diffs] [tfaber] 60231: [NPFS] - Release cancel spinlock in NpCancelListeningQueueIrp

[Hermès BÉLUSCA - MAÏTO]

I think a call to IoReleaseCancelSpinLock should be balanced with a call to
IoAcquireCancelSpinLock somewhere in the npfs_new code, however, looking at:
http://git.reactos.org/?p=reactos.git
<http://git.reactos.org/?p=reactos.git&a=search&h=HEAD&st=grep&s=IoAcquireCa
ncelSpinLock> &a=search&h=HEAD&st=grep&s=IoAcquireCancelSpinLock
I don’t see any call to IoAcquireCancelSpinLock in npfs_new. Can someone
explain to me, why ?
Hermès
Author: tfaber
Date: Fri Sep 20 06:49:21 2013
New Revision: 60231
URL: http://svn.reactos.org/svn/reactos?rev=60231&view=rev
Log:
[NPFS]
- Release cancel spinlock in NpCancelListeningQueueIrp
Modified:
    trunk/reactos/drivers/filesystems/npfs_new/statesup.c
Modified: trunk/reactos/drivers/filesystems/npfs_new/statesup.c
URL:
http://svn.reactos.org/svn/reactos/trunk/reactos/drivers/filesystems/npfs_ne
w/statesup.c?rev=60231&r1=60230&r2=60231&view=diff
============================================================================
==
--- trunk/reactos/drivers/filesystems/npfs_new/statesup.c     [iso-8859-1]
(original)
+++ trunk/reactos/drivers/filesystems/npfs_new/statesup.c  [iso-8859-1] Fri
Sep 20 06:49:21 2013
@@ -20,6 +20,8 @@
NpCancelListeningQueueIrp(IN PDEVICE_OBJECT DeviceObject,
                           IN PIRP Irp)
{
+    IoReleaseCancelSpinLock(Irp->CancelIrql);
+
     FsRtlEnterFileSystem();
     NpAcquireExclusiveVcb();

[Thomas Faber]

Yay someone actually reviewing code :D
Cancel routines are called with the cancel spinlock held and must
release it at some point during their execution.
http://msdn.microsoft.com/en-us/library/windows/hardware/ff540742(v=vs.85).…
On 2013-09-20 11:53, Hermès BÉLUSCA - MAÏTO wrote:
>   I think a call to IoReleaseCancelSpinLock should be
> balanced with a call to
>  IoAcquireCancelSpinLock somewhere in the npfs_new code, however, looking at:
>  http://git.reactos.org/?p=reactos.git
> <http://git.reactos.org/?p=reactos.git&a=search&h=HEAD&st=grep&s=IoAcquireCa
>  ncelSpinLock> &a=search&h=HEAD&st=grep&s=IoAcquireCancelSpinLock
>  I don’t see any call to IoAcquireCancelSpinLock in npfs_new. Can someone
>  explain to me, why ?
>  Hermès
>  Author: tfaber
>  Date: Fri Sep 20 06:49:21 2013
>  New Revision: 60231
>  URL: http://svn.reactos.org/svn/reactos?rev=60231&view=rev
>  Log:
>  [NPFS]
>  - Release cancel spinlock in NpCancelListeningQueueIrp
>  Modified:
>      trunk/reactos/drivers/filesystems/npfs_new/statesup.c
>  Modified: trunk/reactos/drivers/filesystems/npfs_new/statesup.c
>  URL:
>  http://svn.reactos.org/svn/reactos/trunk/reactos/drivers/filesystems/npfs_ne
>  w/statesup.c?rev=60231&r1=60230&r2=60231&view=diff
>  ============================================================================
>  ==
>  --- trunk/reactos/drivers/filesystems/npfs_new/statesup.c     [iso-8859-1]
>  (original)
>  +++ trunk/reactos/drivers/filesystems/npfs_new/statesup.c  [iso-8859-1] Fri
>  Sep 20 06:49:21 2013
>  @@ -20,6 +20,8 @@
>  NpCancelListeningQueueIrp(IN PDEVICE_OBJECT DeviceObject,
>                             IN PIRP Irp)
>  {
>  +    IoReleaseCancelSpinLock(Irp->CancelIrql);
>  +
>       FsRtlEnterFileSystem();
>       NpAcquireExclusiveVcb();

[Hermès BÉLUSCA - MAÏTO]

>   Cancel routines are called with the cancel spinlock
> held and must release it at some point during their execution.
>  http://msdn.microsoft.com/en-us/library/windows/hardware/ff540742(v=vs.85).… 
Ok, nice to know, thanks Thomas :)
Hermès
-----Message d'origine-----
De : ros-dev-bounces(a)reactos.org [mailto:[e-mail]] De la part de Thomas
Faber
Envoyé : vendredi 20 septembre 2013 11:53
À : ReactOS Development List
Objet : [SPAM]Re: [ros-dev] [ros-diffs] [tfaber] 60231: [NPFS] - Release cancel spinlock
in NpCancelListeningQueueIrp
Yay someone actually reviewing code :D
Cancel routines are called with the cancel spinlock held and must release it at some point
during their execution.
http://msdn.microsoft.com/en-us/library/windows/hardware/ff540742(v=vs.85).…
On 2013-09-20 11:53, Hermès BÉLUSCA - MAÏTO wrote:
>   I think a call to IoReleaseCancelSpinLock should be
> balanced with a
>  call to IoAcquireCancelSpinLock somewhere in the npfs_new code, however, looking at:
>  http://git.reactos.org/?p=reactos.git
>  <http://git.reactos.org/?p=reactos.git&a=search&h=HEAD&st=grep&s=IoAcq
>  uireCa
>  ncelSpinLock> &a=search&h=HEAD&st=grep&s=IoAcquireCancelSpinLock
>  I don’t see any call to IoAcquireCancelSpinLock in npfs_new. Can
>  someone explain to me, why ?
>  Hermès
>  Author: tfaber
>  Date: Fri Sep 20 06:49:21 2013
>  New Revision: 60231
>  URL: http://svn.reactos.org/svn/reactos?rev=60231&view=rev
>  Log:
>  [NPFS]
>  - Release cancel spinlock in NpCancelListeningQueueIrp
>  Modified:
>      trunk/reactos/drivers/filesystems/npfs_new/statesup.c
>  Modified: trunk/reactos/drivers/filesystems/npfs_new/statesup.c
>  URL:
>  http://svn.reactos.org/svn/reactos/trunk/reactos/drivers/filesystems/n
>  pfs_ne w/statesup.c?rev=60231&r1=60230&r2=60231&view=diff
>  ======================================================================
>  ======
>  ==
>  --- trunk/reactos/drivers/filesystems/npfs_new/statesup.c     [iso-8859-1]
>  (original)
>  +++ trunk/reactos/drivers/filesystems/npfs_new/statesup.c
>  +++ [iso-8859-1] Fri
>  Sep 20 06:49:21 2013
>  @@ -20,6 +20,8 @@
>  NpCancelListeningQueueIrp(IN PDEVICE_OBJECT DeviceObject,
>                             IN PIRP Irp)
>  {
>  +    IoReleaseCancelSpinLock(Irp->CancelIrql);
>  +
>       FsRtlEnterFileSystem();
>       NpAcquireExclusiveVcb(); 
_______________________________________________
Ros-dev mailing list
Ros-dev(a)reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev