linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] phy_ethtool_{sset,gset} -- check phydev for NULL
@ 2007-05-10  8:57 Matvejchikov Ilya
  2007-05-10 14:35 ` Kumar Gala
  0 siblings, 1 reply; 2+ messages in thread
From: Matvejchikov Ilya @ 2007-05-10  8:57 UTC (permalink / raw)
  To: linuxppc-embedded

Good Day!

The command 'brctl addbr br0 eth0' brings the kernel oops if the eth0
has PHY, but the phydev is NULL (for ex., ifconfig eth0 0.0.0.0 was
not issued firstly)

Call Trace:
[C02CFBD0] [7FFFFFFF] 0x7fffffff (unreliable)
[C02CFBE0] [C0109634] dev_ethtool+0x1b0/0xfac
[C02CFCD0] [C0155EF0] port_cost+0x5c/0x130
[C02CFD40] [C01561FC] br_add_if+0x17c/0x308
[C02CFD70] [C01569E0] add_del_if+0x70/0xa0
[C02CFD90] [C0156A58] br_dev_ioctl+0x48/0x9fc
[C02CFE10] [C0106FCC] dev_ifsioc+0x144/0x3ac
[C02CFE30] [C01080D8] dev_ioctl+0x3b8/0x4c8
[C02CFEB0] [C00F9BC0] sock_ioctl+0x78/0x260
[C02CFED0] [C006889C] do_ioctl+0x38/0x84
[C02CFEE0] [C006896C] vfs_ioctl+0x84/0x3d8
[C02CFF10] [C0068D00] sys_ioctl+0x40/0x74
[C02CFF40] [C000EF48] ret_from_syscall+0x0/0x38

The following patch fixes the problem.

Signed-off-by: Matvejchikov Ilya <matvejchikov <at> gmail.com>

===================================================================
diff -purN linux-2.6.21-clean/drivers/net/phy/phy.c
linux-2.6.21/drivers/net/phy/phy.c
--- linux-2.6.21-clean/drivers/net/phy/phy.c	2007-04-26 07:08:32.000000000 +0400
+++ linux-2.6.21/drivers/net/phy/phy.c	2007-05-04 08:22:01.000000000 +0400
@@ -245,6 +245,9 @@ EXPORT_SYMBOL(phy_sanitize_settings);
  */
  int phy_ethtool_sset(struct phy_device *phydev, struct ethtool_cmd *cmd)
 {
+	if (unlikely(NULL == phydev))
+		return -EINVAL;
+
  	if (cmd->phy_address != phydev->addr)
  		return -EINVAL;

@@ -289,6 +292,9 @@ EXPORT_SYMBOL(phy_ethtool_sset);

  int phy_ethtool_gset(struct phy_device *phydev, struct ethtool_cmd *cmd)
 {
+	if (unlikely(NULL == phydev))
+		return -EINVAL;
+
  	cmd->supported = phydev->supported;

  	cmd->advertising = phydev->advertising;

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] phy_ethtool_{sset,gset} -- check phydev for NULL
  2007-05-10  8:57 [PATCH] phy_ethtool_{sset,gset} -- check phydev for NULL Matvejchikov Ilya
@ 2007-05-10 14:35 ` Kumar Gala
  0 siblings, 0 replies; 2+ messages in thread
From: Kumar Gala @ 2007-05-10 14:35 UTC (permalink / raw)
  To: matvejchikov; +Cc: Jeff Garzik, linux-ppc-embedded ((((E-Mail))))


On May 10, 2007, at 3:57 AM, Matvejchikov Ilya wrote:

> Good Day!
>
> The command 'brctl addbr br0 eth0' brings the kernel oops if the eth0
> has PHY, but the phydev is NULL (for ex., ifconfig eth0 0.0.0.0 was
> not issued firstly)
>
> Call Trace:
> [C02CFBD0] [7FFFFFFF] 0x7fffffff (unreliable)
> [C02CFBE0] [C0109634] dev_ethtool+0x1b0/0xfac
> [C02CFCD0] [C0155EF0] port_cost+0x5c/0x130
> [C02CFD40] [C01561FC] br_add_if+0x17c/0x308
> [C02CFD70] [C01569E0] add_del_if+0x70/0xa0
> [C02CFD90] [C0156A58] br_dev_ioctl+0x48/0x9fc
> [C02CFE10] [C0106FCC] dev_ifsioc+0x144/0x3ac
> [C02CFE30] [C01080D8] dev_ioctl+0x3b8/0x4c8
> [C02CFEB0] [C00F9BC0] sock_ioctl+0x78/0x260
> [C02CFED0] [C006889C] do_ioctl+0x38/0x84
> [C02CFEE0] [C006896C] vfs_ioctl+0x84/0x3d8
> [C02CFF10] [C0068D00] sys_ioctl+0x40/0x74
> [C02CFF40] [C000EF48] ret_from_syscall+0x0/0x38
>
> The following patch fixes the problem.
>
> Signed-off-by: Matvejchikov Ilya <matvejchikov <at> gmail.com>

while you're patch looks reasonable, this is best sent to the netdev  
list and jeff garzik as its not ppc specific at all.

- k

>
> ===================================================================
> diff -purN linux-2.6.21-clean/drivers/net/phy/phy.c
> linux-2.6.21/drivers/net/phy/phy.c
> --- linux-2.6.21-clean/drivers/net/phy/phy.c	2007-04-26  
> 07:08:32.000000000 +0400
> +++ linux-2.6.21/drivers/net/phy/phy.c	2007-05-04  
> 08:22:01.000000000 +0400
> @@ -245,6 +245,9 @@ EXPORT_SYMBOL(phy_sanitize_settings);
>   */
>   int phy_ethtool_sset(struct phy_device *phydev, struct  
> ethtool_cmd *cmd)
>  {
> +	if (unlikely(NULL == phydev))
> +		return -EINVAL;

Would -ENODEV; make more sense?

> +
>   	if (cmd->phy_address != phydev->addr)
>   		return -EINVAL;
>
> @@ -289,6 +292,9 @@ EXPORT_SYMBOL(phy_ethtool_sset);
>
>   int phy_ethtool_gset(struct phy_device *phydev, struct  
> ethtool_cmd *cmd)
>  {
> +	if (unlikely(NULL == phydev))
> +		return -EINVAL;
> +

Would -ENODEV; make more sense?


>   	cmd->supported = phydev->supported;
>
>   	cmd->advertising = phydev->advertising;
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2007-05-10 14:35 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-05-10  8:57 [PATCH] phy_ethtool_{sset,gset} -- check phydev for NULL Matvejchikov Ilya
2007-05-10 14:35 ` Kumar Gala

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).