On Fri, Sep 06, 2024 at 10:58:40AM +0900, Damien Le Moal wrote:
> A struct ata_device flags are always set iand cleared with the device
s/iand/and/
I can see many cases function that currently set or clear dev->flags
withouth holding ap->lock, done by functions that run in EH context,
which will thus not hold ap->lock.
Perhaps rephrase this to:
struct ata_device flags should always be set and cleared with the device
> port locked. Testing for a flag should thus also be done while holding
> the device port lock. In accordance to this, modify
> ata_scsi_handle_link_detach() to both test and clear the
> ATA_DFLAG_DETACHED flag while holding the device port lock.
>
> Signed-off-by: Damien Le Moal <dlemoal@xxxxxxxxxx>
> ---
> drivers/ata/libata-scsi.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c
> index a3ffce4b218d..1a135d44286c 100644
> --- a/drivers/ata/libata-scsi.c
> +++ b/drivers/ata/libata-scsi.c
> @@ -4616,10 +4616,12 @@ static void ata_scsi_handle_link_detach(struct ata_link *link)
> ata_for_each_dev(dev, link, ALL) {
> unsigned long flags;
>
> - if (!(dev->flags & ATA_DFLAG_DETACHED))
> + spin_lock_irqsave(ap->lock, flags);
> + if (!(dev->flags & ATA_DFLAG_DETACHED)) {
> + spin_unlock_irqrestore(ap->lock, flags);
> continue;
> + }
>
> - spin_lock_irqsave(ap->lock, flags);
> dev->flags &= ~ATA_DFLAG_DETACHED;
> spin_unlock_irqrestore(ap->lock, flags);
>
> --
> 2.46.0
>
Reviewed-by: Niklas Cassel <cassel@xxxxxxxxxx>
