Re: eepro100 problems in 2.4.0

From: Jeff Garzik (jgarzik@mandrakesoft.com)
Date: Sat Jan 27 2001 - 11:30:40 EST


Andrey Savochkin wrote:
>
> Hi,
>
> On Thu, Jan 25, 2001 at 04:19:27PM -0500, Jeff Garzik wrote:
> > Oops, sorry guys. Thanks to DaveM for correcting me -- my patch has
> > nothing to do with the "card reports no resources" problem. My
> > apologies.
>
> No problems.
>
> However, there is a real problem with eepro100 when the system resumes
> operations after a sleep.
> May be, you could guess what's wrong in this case?

The regions shouldn't be disabled, the attached patch (against
2.4.1-pre10) adds a call to pci_enable_device to enable things in
eepro100_resume().

It also includes the patch I posted previously; the patch itself is
correct -- the pci_enable_device call should be moved up -- however it
was my description of the patch ("fixes 'card has no resources'") which
was completely incorrect.

Scott, does the attached patch help you out?

        Jeff

-- 
Jeff Garzik       | "You see, in this world there's two kinds of
Building 1024     |  people, my friend: Those with loaded guns
MandrakeSoft      |  and those who dig. You dig."  --Blondie

Index: linux_2_4/drivers/net/eepro100.c diff -u linux_2_4/drivers/net/eepro100.c:1.1.1.9 linux_2_4/drivers/net/eepro100.c:1.1.1.9.42.4 --- linux_2_4/drivers/net/eepro100.c:1.1.1.9 Sat Nov 25 15:53:20 2000 +++ linux_2_4/drivers/net/eepro100.c Sat Jan 27 08:27:00 2001 @@ -560,6 +560,9 @@ if (speedo_debug > 0 && did_version++ == 0) printk(version); + if (pci_enable_device(pdev)) + return -EIO; + if (!request_region(pci_resource_start(pdev, 1), pci_resource_len(pdev, 1), "eepro100")) { printk (KERN_ERR "eepro100: cannot reserve I/O ports\n"); @@ -598,9 +601,6 @@ acpi_idle_state = pwr_command & PCI_PM_CTRL_STATE_MASK; } - if (pci_enable_device(pdev)) - goto err_out_free_mmio_region; - pci_set_master(pdev); if (speedo_found1(pdev, ioaddr, cards_found, acpi_idle_state) == 0) @@ -2146,6 +2146,13 @@ struct net_device *dev = pdev->driver_data; struct speedo_private *sp = (struct speedo_private *)dev->priv; long ioaddr = dev->base_addr; + + /* Make sure power state is D0, a.k.a. alive, and also + * make sure PIO and MMIO are active. Apparently some + * cases exist where PCI_COMMAND_{IO,MEM} is not set when + * we return from resume. -jgarzik + */ + pci_enable_device(pdev); /* I'm absolutely uncertain if this part of code may work. The problems are:

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Wed Jan 31 2001 - 21:00:27 EST