2023-11-15 09:46:06

by Jarkko Nikula

[permalink] [raw]
Subject: Re: [PATCH v3 01/25] i2c: designware: Delete adapter before disabling in i2c_dw_pci_remove()

On 11/10/23 20:11, Andy Shevchenko wrote:
> Make i2c_dw_pci_remove() to be aligned with dw_i2c_plat_remove() on
> the sequence of the ceasing I²C operations.
>
> Fixes: 18dbdda89f5c ("i2c-designware: Add runtime power management support")

I think this is wrong. Commit 18dbdda89f5c is 12 years ago so is this
patch really fixing a real bug that was hiding all these years?

> Signed-off-by: Andy Shevchenko <[email protected]>
> ---
> drivers/i2c/busses/i2c-designware-pcidrv.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/i2c/busses/i2c-designware-pcidrv.c b/drivers/i2c/busses/i2c-designware-pcidrv.c
> index 61d7a27aa070..35d35d7c8e23 100644
> --- a/drivers/i2c/busses/i2c-designware-pcidrv.c
> +++ b/drivers/i2c/busses/i2c-designware-pcidrv.c
> @@ -354,11 +354,13 @@ static void i2c_dw_pci_remove(struct pci_dev *pdev)
> {
> struct dw_i2c_dev *dev = pci_get_drvdata(pdev);
>
> + i2c_del_adapter(&dev->adapter);
> +
> dev->disable(dev);
> +
> pm_runtime_forbid(&pdev->dev);
> pm_runtime_get_noresume(&pdev->dev);
>
> - i2c_del_adapter(&dev->adapter);
> devm_free_irq(&pdev->dev, dev->irq, dev);
> pci_free_irq_vectors(pdev);
> }

Then I believe this patch may potentially introduce a bug. As far as I
can see nothing wrong regarding adapter deletion after the commit
18dbdda89f5c:

i2c_dw_pci_probe()
{
...
i2c_add_numbered_adapter()
pm_runtime_enable_stuff()
...
}

i2c_dw_pci_remove()
{
...
pm_runtime_disable_stuff()
i2c_del_adapter()
...
}

Order is still practically the same in current i2c-designware-pcidrv.c
code but after this patch disabling/deletion is not done in reverse order.


2023-11-15 13:49:29

by Andy Shevchenko

[permalink] [raw]
Subject: Re: [PATCH v3 01/25] i2c: designware: Delete adapter before disabling in i2c_dw_pci_remove()

On Wed, Nov 15, 2023 at 11:45:10AM +0200, Jarkko Nikula wrote:
> On 11/10/23 20:11, Andy Shevchenko wrote:
> > Make i2c_dw_pci_remove() to be aligned with dw_i2c_plat_remove() on
> > the sequence of the ceasing I?C operations.
> >
> > Fixes: 18dbdda89f5c ("i2c-designware: Add runtime power management support")
>
> I think this is wrong. Commit 18dbdda89f5c is 12 years ago so is this patch
> really fixing a real bug that was hiding all these years?

Why not? But let me check more careful the ordering between two and probe.

--
With Best Regards,
Andy Shevchenko