2022-09-27 02:20:42

by Jason Yan

Subject: [PATCH v3 4/8] scsi: mvsas: use sas_find_attached_phy() instead of open coded

The attached phy finding is open coded. Now we can replace it with
sas_find_attached_phy().

Signed-off-by: Jason Yan <[email protected]>
Reviewed-by: Jack Wang <[email protected]>
---
drivers/scsi/mvsas/mv_sas.c | 15 ++++-----------
1 file changed, 4 insertions(+), 11 deletions(-)

diff --git a/drivers/scsi/mvsas/mv_sas.c b/drivers/scsi/mvsas/mv_sas.c
index a6867dae0e7c..f6576ffabe9f 100644
--- a/drivers/scsi/mvsas/mv_sas.c
+++ b/drivers/scsi/mvsas/mv_sas.c
@@ -1190,23 +1190,16 @@ static int mvs_dev_found_notify(struct domain_device *dev, int lock)
mvi_device->sas_device = dev;
if (parent_dev && dev_is_expander(parent_dev->dev_type)) {
int phy_id;
- u8 phy_num = parent_dev->ex_dev.num_phys;
- struct ex_phy *phy;
- for (phy_id = 0; phy_id < phy_num; phy_id++) {
- phy = &parent_dev->ex_dev.ex_phy[phy_id];
- if (SAS_ADDR(phy->attached_sas_addr) ==
- SAS_ADDR(dev->sas_addr)) {
- mvi_device->attached_phy = phy_id;
- break;
- }
- }

- if (phy_id == phy_num) {
+ phy_id = sas_find_attached_phy(&parent_dev->ex_dev, dev);
+ if (phy_id == -ENODEV) {
mv_printk("Error: no attached dev:%016llx"
"at ex:%016llx.\n",
SAS_ADDR(dev->sas_addr),
SAS_ADDR(parent_dev->sas_addr));
res = -1;
+ } else {
+ mvi_device->attached_phy = phy_id;
}
}

--
2.31.1

2022-09-27 02:43:16

by Damien Le Moal

Subject: Re: [PATCH v3 4/8] scsi: mvsas: use sas_find_attached_phy() instead of open coded

On 9/27/22 11:29, Jason Yan wrote:
> The attached phy finding is open coded. Now we can replace it with
> sas_find_attached_phy().
>
> Signed-off-by: Jason Yan <[email protected]>
> Reviewed-by: Jack Wang <[email protected]>

Reviewed-by: Damien Le Moal <[email protected]>

> ---
> drivers/scsi/mvsas/mv_sas.c | 15 ++++-----------
> 1 file changed, 4 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/scsi/mvsas/mv_sas.c b/drivers/scsi/mvsas/mv_sas.c
> index a6867dae0e7c..f6576ffabe9f 100644
> --- a/drivers/scsi/mvsas/mv_sas.c
> +++ b/drivers/scsi/mvsas/mv_sas.c
> @@ -1190,23 +1190,16 @@ static int mvs_dev_found_notify(struct domain_device *dev, int lock)
> mvi_device->sas_device = dev;
> if (parent_dev && dev_is_expander(parent_dev->dev_type)) {
> int phy_id;
> - u8 phy_num = parent_dev->ex_dev.num_phys;
> - struct ex_phy *phy;
> - for (phy_id = 0; phy_id < phy_num; phy_id++) {
> - phy = &parent_dev->ex_dev.ex_phy[phy_id];
> - if (SAS_ADDR(phy->attached_sas_addr) ==
> - SAS_ADDR(dev->sas_addr)) {
> - mvi_device->attached_phy = phy_id;
> - break;
> - }
> - }
>
> - if (phy_id == phy_num) {
> + phy_id = sas_find_attached_phy(&parent_dev->ex_dev, dev);
> + if (phy_id == -ENODEV) {
> mv_printk("Error: no attached dev:%016llx"
> "at ex:%016llx.\n",
> SAS_ADDR(dev->sas_addr),
> SAS_ADDR(parent_dev->sas_addr));
> res = -1;

Would be nice to keep the -ENODEV error here instead of this generic, and
meaningless, "-1" code. Same for the previous patch too.

> + } else {
> + mvi_device->attached_phy = phy_id;
> }
> }
>

--
Damien Le Moal
Western Digital Research