David!
On Mon, Jan 16 2023 at 09:56, David Woodhouse wrote:
> On Fri, 2022-11-25 at 00:24 +0100, Thomas Gleixner wrote:
>> +
>> + if (ops->domain_free_irqs)
>> + ops->domain_free_irqs(domain, dev);
>
> Do you want a call to msi_free_dev_descs(dev) here? In the case where
> the core code calls ops->domain_alloc_irqs() it *has* allocated the
> descriptors first... but it's expecting the irqdomain to free them?
No. Let me stare at it.
> However, it's not quite as simple as adding msi_free_dev_descs()...
Correct.
> diff --git a/kernel/irq/msi.c b/kernel/irq/msi.c
> index 955267bbc2be..812e1ec1a633 100644
> --- a/kernel/irq/msi.c
> +++ b/kernel/irq/msi.c
> @@ -1533,9 +1533,10 @@ static void msi_domain_free_locked(struct device *dev, struct msi_ctrl *ctrl)
> info = domain->host_data;
> ops = info->ops;
>
> - if (ops->domain_free_irqs)
> + if (ops->domain_free_irqs) {
> ops->domain_free_irqs(domain, dev);
> - else
> + msi_free_msi_descs(dev);
This is just wrong. I need to taxi my grandson. Will have a look
afterwards.
Thanks,
tglx
