On 16:41 Wed 09 Feb , Sergei Shtylyov wrote:
Hello.
On 09-02-2011 15:43, Jean-Christophe PLAGNIOL-VILLARD wrote:
quoted
From: Paul Chavent<redacted>
quoted
When listing processes on a system with SOFT/HARD_IRQ enabled,
the name of the ethernet device is [irq/eth%d] (instead of [irq/eth0] for example).
quoted
This patch call the request_irq function after having initialized the name of the device.
quoted
Signed-off-by: Paul Chavent<redacted>
Signed-off-by: Nicolas Ferre<redacted>
Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD<redacted>
[...]
quoted
diff --git a/drivers/net/macb.c b/drivers/net/macb.c
index f69e73e..d642e08 100644
--- a/drivers/net/macb.c
+++ b/drivers/net/macb.c
[...]
quoted
@@ -1219,13 +1209,23 @@ static int __init macb_probe(struct platform_device *pdev)
err = register_netdev(dev);
if (err) {
dev_err(&pdev->dev, "Cannot register net device, aborting.\n");
- goto err_out_free_irq;
+ goto err_out_iounmap;
}
- if (macb_mii_init(bp) != 0) {
+ dev->irq = platform_get_irq(pdev, 0);
platform_get_irq() can fail...
request_irq will fail too so do we really need to check it?
quoted
+ err = request_irq(dev->irq, macb_interrupt, IRQF_SAMPLE_RANDOM,
+ dev->name, dev);
Best Regards,
J.