Re: [patch V2 20/23] PCI/MSI: Move msi_lock to struct pci_dev

[Jason Gunthorpe <jgg@xxxxxxxxxx>]

On Mon, Dec 06, 2021 at 11:27:56PM +0100, Thomas Gleixner wrote:
> It's only required for PCI/MSI. So no point in having it in every struct
> device.
> 
> Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> ---
> V2: New patch
> ---
>  drivers/base/core.c    |    1 -
>  drivers/pci/msi/msi.c  |    2 +-
>  drivers/pci/probe.c    |    4 +++-
>  include/linux/device.h |    2 --
>  include/linux/pci.h    |    1 +
>  5 files changed, 5 insertions(+), 5 deletions(-)

Reviewed-by: Jason Gunthorpe <jgg@xxxxxxxxxx>
 
> --- a/drivers/base/core.c
> +++ b/drivers/base/core.c
> @@ -2875,7 +2875,6 @@ void device_initialize(struct device *de
>       device_pm_init(dev);
>       set_dev_node(dev, NUMA_NO_NODE);
>  #ifdef CONFIG_GENERIC_MSI_IRQ
> -     raw_spin_lock_init(&dev->msi_lock);
>       INIT_LIST_HEAD(&dev->msi_list);
>  #endif
>       INIT_LIST_HEAD(&dev->links.consumers);
> --- a/drivers/pci/msi/msi.c
> +++ b/drivers/pci/msi/msi.c
> @@ -18,7 +18,7 @@ int pci_msi_ignore_mask;
>  
>  static noinline void pci_msi_update_mask(struct msi_desc *desc, u32 clear, 
> u32 set)
>  {
> -     raw_spinlock_t *lock = &desc->dev->msi_lock;
> +     raw_spinlock_t *lock = &to_pci_dev(desc->dev)->msi_lock;
>       unsigned long flags;
>  
>       if (!desc->pci.msi_attrib.can_mask)

It looks like most of the time this is called by an irq_chip, except
for a few special cases list pci_msi_shutdown()

Is this something that should ideally go away someday and use some
lock in the irq_chip - not unlike what we've thought is needed for
IMS?

Jason