Re: [PATCH] x86/PCI: add spinlock held check to 'pcibios_fwaddrmap_lookup()'

[Jesse Barnes <jbarnes@virtuousgeek.org>]

[-- Attachment #1: Type: text/plain, Size: 2483 bytes --]

On Fri, 2 Mar 2012 12:41:36 -0800
Greg KH <gregkh@linuxfoundation.org> wrote:

> On Fri, Mar 02, 2012 at 12:24:05PM -0800, Jesse Barnes wrote:
> > On Fri, 2 Mar 2012 12:00:27 -0800
> > Greg KH <gregkh@linuxfoundation.org> wrote:
> > 
> > > On Fri, Mar 02, 2012 at 12:45:01PM -0700, Myron Stowe wrote:
> > > > 'pcibios_fwaddrmap_lookup()' is used to maintain FW-assigned BIOS BAR
> > > > values for reinstatement when normal resource assignment attempts
> > > > fail and must be called with the 'pcibios_fwaddrmap_lock' spinlock
> > > > held.
> > > > 
> > > > This patch adds a WARN_ON notification if the spinlock is not currently
> > > > held by the caller.
> > > > 
> > > > Signed-off-by: Myron Stowe <myron.stowe@redhat.com>
> > > > ---
> > > > 
> > > >  arch/x86/pci/i386.c |    2 ++
> > > >  1 files changed, 2 insertions(+), 0 deletions(-)
> > > > 
> > > > diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c
> > > > index 33e6a0b..831971e 100644
> > > > --- a/arch/x86/pci/i386.c
> > > > +++ b/arch/x86/pci/i386.c
> > > > @@ -57,6 +57,8 @@ static struct pcibios_fwaddrmap *pcibios_fwaddrmap_lookup(struct pci_dev *dev)
> > > >  {
> > > >  	struct pcibios_fwaddrmap *map;
> > > >  
> > > > +	WARN_ON(!spin_is_locked(&pcibios_fwaddrmap_lock));
> > > > +
> > > 
> > > What is this going to help with?  How can someone then recover from this
> > > issue?  Just adding a warning message isn't going to fix any problems
> > > here, why not fix the root cause?
> > 
> > It's just a self-documenting assert; doesn't trigger anything and has
> > more functionality than
> >   /* Must hold the fwaddrmap_lock here */
> 
> Don't we have sparse markups that we can use to verify this instead
> somehow?  Adding asserts isn't the nicest, as what will a user really do
> about this if it ever gets hit?  And if a user isn't supposed to do
> anything, then yes, a comment would be best I would think.

The user is supposed to report a bug, but it likely won't cause a
crash, just a worrying message in the log.

And yeah I think we have some sparse bits for this, and at one point I
thought we had an assert_spin_is_locked or somesuch.

I like the idea of self-documenting code checks better than just
comments.  Sparse annotations are a good second solution, but not as
nice since sparse isn't always run, and has a harder time with control
flow dependent stuff like this.

-- 
Jesse Barnes, Intel Open Source Technology Center

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 836 bytes --]