Merge branch 'fsl-fman-next'
Christophe JAILLET says:
====================
fsl/fman: Fix some error handling code in mac_probe
Commit c6e26ea8c8 ("dpaa_eth: change device used") generated some
conflicts in my patches waiting for submission. So I took a closer look at
it.
So here is a serie of 4 patches.
The 1st one is just about a spurious call to 'dev_set_drvdata()', which is
done in only 1 error handling path in the function.
The 2nd one removes some devm_iounmap/release/kfree functions which look
useless to me.
The 3rd one fixes a missing of_node_put.
The 4th one is just cosmetic and removes a useless message.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
@@ -615,7 +615,6 @@ static int mac_probe(struct platform_device *_of_dev)
|
||||
mac_dev = devm_kzalloc(dev, sizeof(*mac_dev), GFP_KERNEL);
|
||||
if (!mac_dev) {
|
||||
err = -ENOMEM;
|
||||
dev_err(dev, "devm_kzalloc() = %d\n", err);
|
||||
goto _return;
|
||||
}
|
||||
priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
|
||||
@@ -709,12 +708,8 @@ static int mac_probe(struct platform_device *_of_dev)
|
||||
}
|
||||
|
||||
if (!of_device_is_available(mac_node)) {
|
||||
devm_iounmap(dev, priv->vaddr);
|
||||
__devm_release_region(dev, fman_get_mem_region(priv->fman),
|
||||
res.start, res.end + 1 - res.start);
|
||||
devm_kfree(dev, mac_dev);
|
||||
dev_set_drvdata(dev, NULL);
|
||||
return -ENODEV;
|
||||
err = -ENODEV;
|
||||
goto _return_of_get_parent;
|
||||
}
|
||||
|
||||
/* Get the cell-index */
|
||||
@@ -825,6 +820,7 @@ static int mac_probe(struct platform_device *_of_dev)
|
||||
phy = of_phy_find_device(mac_dev->phy_node);
|
||||
if (!phy) {
|
||||
err = -EINVAL;
|
||||
of_node_put(mac_dev->phy_node);
|
||||
goto _return_of_get_parent;
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user