Commit Graph

9 Commits

Author SHA1 Message Date
Matt Reimer 0ff8d1b3c8 USB: pxa27x_udc: use four bits to store endpoint addresses
Endpoint addresses on pxa27x can be programmed as 1-15, but since
only three bits were being used to store the endpoint number it
was possible to overflow.

Signed-off-by: Matt Reimer <mreimer@sdgsystems.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-05-20 13:21:36 -07:00
Robert Jarzmik 5e23e90f33 USB: pxa27x_udc: Fix deadlocks on request queueing
As reported by Antonio, there are cases where the ep->lock
can be taken twice, triggering a deadlock.
The typical sequence is :
 irq_handler
   \
    -> gadget.complete()
       \
        -> pxa27x_udc.pxa_ep_queue() : ep->lock is taken
           \
            -> gadget.complete()
               \
                -> pxa27x_udc.pxa_ep_queue() : ep->lock is taken
                                               ==> *deadlock*
The patch fixes this by :
 - releasing the lock each time gadget.complete() is called
 - adding a check in handle_ep() to detect a recursive call,
   in which case the function becomes on no-op.

The patch is still not good enough for ep0. For this unique
endpoint, another well thought over patch will be needed.

Reported-by: Antonio Ospite <ospite@studenti.unina.it>
Tested-by: Antonio Ospite <ospite@studenti.unina.it>
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-03-02 14:54:47 -08:00
Robert Jarzmik 367815eea4 USB: pxa27x_udc: single-thread setup requests
Since the PXA 27x UDC automatically ACK's some control
packets such as SET_INTERFACE, the gadgets may not get a
chance to process the request before another control packet
is received. The Linux gadgets do not expect to receive
setup callbacks out of order. The file storage gadget only
saves the "highest" priority request.

The PXA27x UDC driver must make sure it only sends one up at
a time, allowing the gadget to make changes before
continuing. In theory, the host would be NACK'd while the
gadget processes the change but the UDC has already ACK'd
the request. If another request is sent by the host that is
not automatically ACK'd by the UDC, then the throttling
happens properly to regain sync.

The observed case was the file_storage gadget timing out on
a BulkReset request because the SET_INTERFACE was being
processed by the gadget. Since SET_INTERFACE is higher
priority than BulkReset, the BulkReset was dropped.  This
was exacerbated by turning on the debug which delayed the
fsg signal processing thread.

This also fixes the "should never get in
WAIT_ACK_SET_CONF_INTERF state here!!!" warning.

Reported-by: Vernon Sauder <vernoninhand@gmail.com>
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

index 51790b0..1937d8c 100644
2009-06-15 21:44:42 -07:00
Robert Jarzmik 7fec3c25b7 USB: pxa27x_udc: add otg transceiver support
When a transceiver driver is used, no automatic udc enable
is done. The transceiver (OTG or not) should :
 - take care of VBus sensing
 - call usb_gadget_vbus_connect()
 - call usb_gadget_vbus_disconnect()

The pullup should remain within this driver's management,
either by gpio_pullup of udc_command() fields.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-24 16:20:27 -07:00
Robert Jarzmik b799a7eb68 USB: pxa27x_udc: add vbus session handling
On vbus_session() call, optionally activate D+ pullup
resistor and enable the udc, or deactivate D+ pullup
resistor and disable the udc.

It is intentional to not handle any VBus sense related irq.
An external transceiver driver (like gpio_vbus) should
catch VBus sense signal, and call usb_gadget_vbus_connect()
or usb_gadget_vbus_disconnect().

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-24 16:20:26 -07:00
Robert Jarzmik eb50702539 USB: pxa27x_udc: factor pullup code to prepare otg transceiver
Prepare pxa27x_udc to handle usb D+ pullup properly : it
should connect the pullup resistor and disconnect it only
if no external transceiver is handling it.

[ dbrownell@users.sourceforge.net: kerneldoc and gpio fixes ]

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-03-24 16:20:26 -07:00
Philipp Zabel d438ae5796 [ARM] 5080/1: touch PSSR_OTGPH only on pxa27x in ohci-pxa27x and pxa27x_udc
and include pxa2xx-regs.h as build fix since PSSR definitions
moved from pxa-regs.h into pxa2xx-regs.h.

Note: This change is temporary as pxa27x processor specific
code will be finally moved elsewhere (both drivers should
support pxa3xx, too).

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2008-06-15 19:54:23 +01:00
Robert Jarzmik 5a59bc544d USB: pxa27x_udc: minor fixes
Minor fixes to pxa27x udc driver :
 - don't clobber driver model bus_id field
 - wrong endianess fix (no functional change; cpu is little-endian)
 - double udc disable fix
 - resume/suspend fix (OTG hold bit)
 - make driver pxa27x dependant (check cpu at runtime)

Signed-off-by: Robert Jarzmik <rjarzmik@free.fr>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-05-14 10:00:29 -07:00
Robert Jarzmik d75379a538 usb: pxa27x_udc driver
Adds pxa27x udc driver to support USB peripherals on pxa27x chips.

The driver is compatible with: Gadget Zero, the File Storage
gadget, and the Ethernet gadget (only in CDC subset mode).

The driver can't properly support multiple interfaces, because
of hardware bugs without possible workaround.  That means no
RNDIS support from g_ether, and no CDC ACM support in g_serial.

Signed-off-by: Robert Jarzmik <rjarzmik@free.fr>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-05-02 10:25:53 -07:00