Re: [PATCH 4.14 49/62] Bluetooth: btusb: Only check needs_reset_resume DMI table for QCA rome chipsets

From: gregkh@xxxxxxxxxxxxxxxxxxx
Date: Wed May 16 2018 - 03:25:00 EST


On Tue, May 15, 2018 at 12:42:36PM -0700, Guenter Roeck wrote:
> On Tue, May 15, 2018 at 10:07:02AM -0700, Brian Norris wrote:
> > + Guenter
> >
> > On Mon, May 14, 2018 at 08:49:05AM +0200, Greg Kroah-Hartman wrote:
> > > 4.14-stable review patch. If anyone has any objections, please let me know.
> >
> > FYI, this backport is wrong. See below.
> >
> > > ------------------
> > >
> > > From: Hans de Goede <hdegoede@xxxxxxxxxx>
> > >
> > > commit fc54910280eb38bde923cdf0898e74687d8e6989 upstream.
> > >
> > > Jeremy Cline correctly points out in rhbz#1514836 that a device where the
> > > QCA rome chipset needs the USB_QUIRK_RESET_RESUME quirk, may also ship
> > > with a different wifi/bt chipset in some configurations.
> > >
> > > If that is the case then we are needlessly penalizing those other chipsets
> > > with a reset-resume quirk, typically causing 0.4W extra power use because
> > > this disables runtime-pm.
> > >
> > > This commit moves the DMI table check to a btusb_check_needs_reset_resume()
> > > helper (so that we can easily also call it for other chipsets) and calls
> > > this new helper only for QCA_ROME chipsets for now.
> > >
> > > BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1514836
> > > Cc: stable@xxxxxxxxxxxxxxx
> > > Cc: Jeremy Cline <jcline@xxxxxxxxxx>
> > > Suggested-by: Jeremy Cline <jcline@xxxxxxxxxx>
> > > Signed-off-by: Hans de Goede <hdegoede@xxxxxxxxxx>
> > > Signed-off-by: Marcel Holtmann <marcel@xxxxxxxxxxxx>
> > > Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> > > ---
> > > drivers/bluetooth/btusb.c | 10 +++++++---
> > > 1 file changed, 7 insertions(+), 3 deletions(-)
> > >
> > > --- a/drivers/bluetooth/btusb.c
> > > +++ b/drivers/bluetooth/btusb.c
> > > @@ -2902,6 +2902,12 @@ static int btusb_config_oob_wake(struct
> > > }
> > > #endif
> > >
> > > +static void btusb_check_needs_reset_resume(struct usb_interface *intf)
> > > +{
> > > + if (dmi_check_system(btusb_needs_reset_resume_table))
> > > + interface_to_usbdev(intf)->quirks |= USB_QUIRK_RESET_RESUME;
> > > +}
> > > +
> > > static int btusb_probe(struct usb_interface *intf,
> > > const struct usb_device_id *id)
> > > {
> > > @@ -3037,9 +3043,6 @@ static int btusb_probe(struct usb_interf
> > > hdev->send = btusb_send_frame;
> > > hdev->notify = btusb_notify;
> > >
> > > - if (dmi_check_system(btusb_needs_reset_resume_table))
> > > - interface_to_usbdev(intf)->quirks |= USB_QUIRK_RESET_RESUME;
> > > -
> > > #ifdef CONFIG_PM
> > > err = btusb_config_oob_wake(hdev);
> > > if (err)
> > > @@ -3177,6 +3180,7 @@ static int btusb_probe(struct usb_interf
> > > hdev->setup = btusb_setup_csr;
> > >
> > > set_bit(HCI_QUIRK_SIMULTANEOUS_DISCOVERY, &hdev->quirks);
> > > + btusb_check_needs_reset_resume(intf);
> >
> > The original code puts this under the BTUSB_QCA_ROME section, but this
> > is getting placed under the BTUSB_CSR section.
> >
> Turns out 4.16.y has the same problem. Maybe git got confused because
> mainline also has
> set_bit(HCI_QUIRK_SIMULTANEOUS_DISCOVERY, &hdev->quirks);
> in the BTUSB_QCA_ROME section. This is missing in 4.14.y and 4.16.y.

Ah, thanks for catching this, I'll go fix this up. Looks like patch got
confused by the use of set_bit() here and fuzzed it into this location
instead.

greg k-h