On Tue, 19 Sep 2006, Jiri Kosina wrote:
> With 2.6.18-rc6-mm2 it is possible on my IBM T42p to do suspend/resume
> cycle only once. The second time, suspend fails with
>
> usb_hcd_pci_suspend(): ehci_pci_suspend+0x0/0xd0 [ehci_hcd]() returns -22
> pci_device_suspend(): usb_hcd_pci_suspend+0x0/0x280() returns -22
> suspend_device(): pci_device_suspend+0x0/0x70() returns -22
>
> or, when ehci is unloaded, uhci fails in a similar way:
>
> usb_hcd_pci_suspend(): uhci_suspend+0x0/0xf0 [uhci_hcd]() returns -16
> pci_device_suspend(): usb_hcd_pci_suspend+0x0/0x1c0() returns -16
> suspend_device(): pci_device_suspend+0x0/0x70() returns -16
>
> (by the way this looks to me like a mess, that the user could be presented
> with either -EINVAL or -EBUSY, in case of the same error, depending on the
> device driver which reports it. I will probably submit a consolidating
> patch later this evening).
>
> After some time I figured out that usb_resume_both() is not resuming
> devices properly - it calls resume_device() only for those devices which
> are USB_STATE_SUSPENDED (in which case we want to propagate the resume
> recursively up the device tree), but other cases (most notably
> USB_STATE_CONFIGURED) stay unhadled.
>
> This patch fixes my issue, now the suspend/resume can be done arbitrary
> number of times without any error.
>
> Signed-off-by: Jiri Kosina <jikos@jikos.cz>
>
> --- linux-2.6.18-rc6-mm2.orig/drivers/usb/core/driver.c 2006-09-14 16:20:48.000000000 +0200
> +++ linux-2.6.18-rc6-mm2/drivers/usb/core/driver.c 2006-09-19 03:03:12.000000000 +0200
> @@ -1083,6 +1083,8 @@ int usb_resume_both(struct usb_device *u
> status = resume_device(udev);
> if (parent)
> mutex_unlock(&parent->pm_mutex);
> + } else {
> + status = resume_device(udev);
> }
>
> /* Now the parent won't suspend until we are finished */
Oddly enough, I identified exactly the same problem last week and
submitted a nearly-identical patch just a few minutes ago. The problem
occurs only when CONFIG_USB_SUSPEND isn't set, and it stems from the fact
that without this addition, the device's dev.power.power_state.event field
doesn't get set properly during the resume.
Alan Stern
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]