2014-06-17 12:08:31

by Mikko Perttunen

[permalink] [raw]
Subject: [PATCH RESEND] libahci_platform: Fail when PHY required but PHY support disabled

ahci_platform_get_resources handles resource management for
platform AHCI drivers, including getting a possible PHY
from the device tree. Since not all drivers need a PHY, it
ignores -ENODEV and -ENOSYS from devm_get_phy. However, when
the PHY subsystem is mistakenly disabled, -ENOSYS can be
returned even when a PHY is needed.

This patch modifies the -ENOSYS case to check if a "phys"
device tree node exists. If it exists, then clearly the PHY
subsystem is mistakenly disabled and the driver cannot work,
ahci_platform_get_resources will fail and propagate the error.

Signed-off-by: Mikko Perttunen <[email protected]>
---
drivers/ata/libahci_platform.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
index 3a5b4ed..b007758 100644
--- a/drivers/ata/libahci_platform.c
+++ b/drivers/ata/libahci_platform.c
@@ -250,8 +250,13 @@ struct ahci_host_priv *ahci_platform_get_resources(struct platform_device *pdev)
if (IS_ERR(hpriv->phy)) {
rc = PTR_ERR(hpriv->phy);
switch (rc) {
- case -ENODEV:
case -ENOSYS:
+ /* No PHY support. Check if PHY is required. */
+ if (of_find_property(dev->of_node, "phys", NULL)) {
+ dev_err(dev, "couldn't get sata-phy: ENOSYS\n");
+ goto err_out;
+ }
+ case -ENODEV:
/* continue normally */
hpriv->phy = NULL;
break;
--
1.8.1.5


Subject: Re: [PATCH RESEND] libahci_platform: Fail when PHY required but PHY support disabled

On Tuesday, June 17, 2014 03:07:55 PM Mikko Perttunen wrote:
> ahci_platform_get_resources handles resource management for
> platform AHCI drivers, including getting a possible PHY
> from the device tree. Since not all drivers need a PHY, it
> ignores -ENODEV and -ENOSYS from devm_get_phy. However, when
> the PHY subsystem is mistakenly disabled, -ENOSYS can be
> returned even when a PHY is needed.
>
> This patch modifies the -ENOSYS case to check if a "phys"
> device tree node exists. If it exists, then clearly the PHY
> subsystem is mistakenly disabled and the driver cannot work,
> ahci_platform_get_resources will fail and propagate the error.
>
> Signed-off-by: Mikko Perttunen <[email protected]>

Acked-by: Bartlomiej Zolnierkiewicz <[email protected]>

> ---
> drivers/ata/libahci_platform.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
> index 3a5b4ed..b007758 100644
> --- a/drivers/ata/libahci_platform.c
> +++ b/drivers/ata/libahci_platform.c
> @@ -250,8 +250,13 @@ struct ahci_host_priv *ahci_platform_get_resources(struct platform_device *pdev)
> if (IS_ERR(hpriv->phy)) {
> rc = PTR_ERR(hpriv->phy);
> switch (rc) {
> - case -ENODEV:
> case -ENOSYS:
> + /* No PHY support. Check if PHY is required. */
> + if (of_find_property(dev->of_node, "phys", NULL)) {
> + dev_err(dev, "couldn't get sata-phy: ENOSYS\n");
> + goto err_out;
> + }
> + case -ENODEV:
> /* continue normally */
> hpriv->phy = NULL;
> break;

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

2014-06-17 14:27:32

by Tejun Heo

[permalink] [raw]
Subject: Re: [PATCH RESEND] libahci_platform: Fail when PHY required but PHY support disabled

On Tue, Jun 17, 2014 at 03:07:55PM +0300, Mikko Perttunen wrote:
> ahci_platform_get_resources handles resource management for
> platform AHCI drivers, including getting a possible PHY
> from the device tree. Since not all drivers need a PHY, it
> ignores -ENODEV and -ENOSYS from devm_get_phy. However, when
> the PHY subsystem is mistakenly disabled, -ENOSYS can be
> returned even when a PHY is needed.
>
> This patch modifies the -ENOSYS case to check if a "phys"
> device tree node exists. If it exists, then clearly the PHY
> subsystem is mistakenly disabled and the driver cannot work,
> ahci_platform_get_resources will fail and propagate the error.
>
> Signed-off-by: Mikko Perttunen <[email protected]>

Applied to libata/for-3.17.

Thanks.

--
tejun