Re: [PATCH] ata:Remove no longer valid FIXME comment in pata_ninja32.c for the function,ninja32_init_one

Re: [PATCH] ata:Remove no longer valid FIXME comment in pata_ninja32.c for the function,ninja32_init_one

[Bartlomiej Zolnierkiewicz]

Hi,

On Wednesday, February 04, 2015 06:47:41 PM Nicholas Krause wrote:
> Removes the no longer valid fix me comment in the function, ninja32_init_one for
> questioning if we are required to remove the activated device at the end of this
> function by calling the function, ata_host_activate to activate the device passed
> by the caller of the function ninja_init_one. This comment is no longer needed as
> we need to deallocate the memory used to store this device's information when the
> device is no longer in use by the system in order to avoid leaking memory. Due to
> this the driver in the file,ninja32_init_one has the function,ata_pci_remove_one
> to release the memory allocated in it's init function and therefore this comment
> can now be removed due to use correctly freeing the memory allocated  from the
> driver's init function,ninja32_init_one.
> 
> Signed-off-by: Nicholas Krause <xerofoify@gmail.com>
> ---
>  drivers/ata/pata_ninja32.c | 1 -
>  1 file changed, 1 deletion(-)
> 
> diff --git a/drivers/ata/pata_ninja32.c b/drivers/ata/pata_ninja32.c
> index efb272d..47b30d3 100644
> --- a/drivers/ata/pata_ninja32.c
> +++ b/drivers/ata/pata_ninja32.c
> @@ -147,7 +147,6 @@ static int ninja32_init_one(struct pci_dev *dev, const struct pci_device_id *id)
>  	ap->pflags = ATA_PFLAG_PIO32 | ATA_PFLAG_PIO32CHANGE;
>  
>  	ninja32_program(base);
> -	/* FIXME: Should we disable them at remove ? */

After looking at the driver history I think that the comment is about
ninja32_program() enabling interrupt lines (which don't get disabled when
the host controller is detached from the driver) not about the cleanup
done during ata_pci_remove_one().  Alan, is this correct?

>  	return ata_host_activate(host, dev->irq, ata_bmdma_interrupt,
>  				 IRQF_SHARED, &ninja32_sht);
>  }

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics

Re: [PATCH] ata:Remove no longer valid FIXME comment in pata_ninja32.c for the function,ninja32_init_one

[Bartlomiej Zolnierkiewicz]

On Thursday, February 05, 2015 08:00:24 AM Nick Krause wrote:
> On Thu, Feb 5, 2015 at 6:02 AM, Bartlomiej Zolnierkiewicz
> <b.zolnierkie@samsung.com> wrote:
> >
> > Hi,
> >
> > On Wednesday, February 04, 2015 06:47:41 PM Nicholas Krause wrote:
> >> Removes the no longer valid fix me comment in the function, ninja32_init_one for
> >> questioning if we are required to remove the activated device at the end of this
> >> function by calling the function, ata_host_activate to activate the device passed
> >> by the caller of the function ninja_init_one. This comment is no longer needed as
> >> we need to deallocate the memory used to store this device's information when the
> >> device is no longer in use by the system in order to avoid leaking memory. Due to
> >> this the driver in the file,ninja32_init_one has the function,ata_pci_remove_one
> >> to release the memory allocated in it's init function and therefore this comment
> >> can now be removed due to use correctly freeing the memory allocated  from the
> >> driver's init function,ninja32_init_one.
> >>
> >> Signed-off-by: Nicholas Krause <xerofoify@gmail.com>
> >> ---
> >>  drivers/ata/pata_ninja32.c | 1 -
> >>  1 file changed, 1 deletion(-)
> >>
> >> diff --git a/drivers/ata/pata_ninja32.c b/drivers/ata/pata_ninja32.c
> >> index efb272d..47b30d3 100644
> >> --- a/drivers/ata/pata_ninja32.c
> >> +++ b/drivers/ata/pata_ninja32.c
> >> @@ -147,7 +147,6 @@ static int ninja32_init_one(struct pci_dev *dev, const struct pci_device_id *id)
> >>       ap->pflags = ATA_PFLAG_PIO32 | ATA_PFLAG_PIO32CHANGE;
> >>
> >>       ninja32_program(base);
> >> -     /* FIXME: Should we disable them at remove ? */
> >
> > After looking at the driver history I think that the comment is about
> > ninja32_program() enabling interrupt lines (which don't get disabled when
> > the host controller is detached from the driver) not about the cleanup
> > done during ata_pci_remove_one().  Alan, is this correct?
> >
> >>       return ata_host_activate(host, dev->irq, ata_bmdma_interrupt,
> >>                                IRQF_SHARED, &ninja32_sht);
> >>  }
> >
> > Best regards,
> > --
> > Bartlomiej Zolnierkiewicz
> > Samsung R&D Institute Poland
> > Samsung Electronics
> >
> If that is the case then I will disable the interrupt lines for the
> host in this driver's
> remove function.

Unless you have hardware to test such change (to check that there are no
side-effects) just moving FIXME to ninja32_program() seems to be a better
approach for now..

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics

Re: [PATCH] ata:Remove no longer valid FIXME comment in pata_ninja32.c for the function,ninja32_init_one

[Bartlomiej Zolnierkiewicz]

On Thursday, February 05, 2015 08:49:24 AM Nick Krause wrote:
> On Thu, Feb 5, 2015 at 8:15 AM, Bartlomiej Zolnierkiewicz
> <b.zolnierkie@samsung.com> wrote:
> > On Thursday, February 05, 2015 08:00:24 AM Nick Krause wrote:
> >> On Thu, Feb 5, 2015 at 6:02 AM, Bartlomiej Zolnierkiewicz
> >> <b.zolnierkie@samsung.com> wrote:
> >> >
> >> > Hi,
> >> >
> >> > On Wednesday, February 04, 2015 06:47:41 PM Nicholas Krause wrote:
> >> >> Removes the no longer valid fix me comment in the function, ninja32_init_one for
> >> >> questioning if we are required to remove the activated device at the end of this
> >> >> function by calling the function, ata_host_activate to activate the device passed
> >> >> by the caller of the function ninja_init_one. This comment is no longer needed as
> >> >> we need to deallocate the memory used to store this device's information when the
> >> >> device is no longer in use by the system in order to avoid leaking memory. Due to
> >> >> this the driver in the file,ninja32_init_one has the function,ata_pci_remove_one
> >> >> to release the memory allocated in it's init function and therefore this comment
> >> >> can now be removed due to use correctly freeing the memory allocated  from the
> >> >> driver's init function,ninja32_init_one.
> >> >>
> >> >> Signed-off-by: Nicholas Krause <xerofoify@gmail.com>
> >> >> ---
> >> >>  drivers/ata/pata_ninja32.c | 1 -
> >> >>  1 file changed, 1 deletion(-)
> >> >>
> >> >> diff --git a/drivers/ata/pata_ninja32.c b/drivers/ata/pata_ninja32.c
> >> >> index efb272d..47b30d3 100644
> >> >> --- a/drivers/ata/pata_ninja32.c
> >> >> +++ b/drivers/ata/pata_ninja32.c
> >> >> @@ -147,7 +147,6 @@ static int ninja32_init_one(struct pci_dev *dev, const struct pci_device_id *id)
> >> >>       ap->pflags = ATA_PFLAG_PIO32 | ATA_PFLAG_PIO32CHANGE;
> >> >>
> >> >>       ninja32_program(base);
> >> >> -     /* FIXME: Should we disable them at remove ? */
> >> >
> >> > After looking at the driver history I think that the comment is about
> >> > ninja32_program() enabling interrupt lines (which don't get disabled when
> >> > the host controller is detached from the driver) not about the cleanup
> >> > done during ata_pci_remove_one().  Alan, is this correct?
> >> >
> >> >>       return ata_host_activate(host, dev->irq, ata_bmdma_interrupt,
> >> >>                                IRQF_SHARED, &ninja32_sht);
> >> >>  }
> >> >
> >> > Best regards,
> >> > --
> >> > Bartlomiej Zolnierkiewicz
> >> > Samsung R&D Institute Poland
> >> > Samsung Electronics
> >> >
> >> If that is the case then I will disable the interrupt lines for the
> >> host in this driver's
> >> remove function.
> >
> > Unless you have hardware to test such change (to check that there are no
> > side-effects) just moving FIXME to ninja32_program() seems to be a better
> > approach for now..
> >
> > Best regards,
> > --
> > Bartlomiej Zolnierkiewicz
> > Samsung R&D Institute Poland
> > Samsung Electronics
> >
> Very well I will send in a patch moving this fix me.

Please wait with this to give a chance for somebody more familiar with
pata_ninja32 driver/hardware to comment on it.  There is absolutely no need
to rush with this change..

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics