This patch takes a reference to the wa_xfer object in wa_urb_dequeue to
prevent the urb giveback code from completing the xfer and freeing it
while wa_urb_dequeue is executing. It also checks for done at the start
to avoid a double completion scenario. Adding the check for done in
urb_dequeue means that any other place where a submitted transfer
segment is marked as done must complete the transfer if it is done.
__wa_xfer_delayed_run was not checking this case so that check was added
as well.
Signed-off-by: Thomas Pugliese <thomas.pugliese@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds error checking to the abort request callback to forcibly
clean up the dequeued transfers if the abort request failed. The
wa_complete_remaining_xfer_segs was modified so that it could be used in
this situation as well. This fixes a stranded URB/PNP hang when the HWA
is unplugged while playing audio to a wireless audio device.
Signed-off-by: Thomas Pugliese <thomas.pugliese@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds ref counting to sections of code that operate on struct
wa_xfer objects that were missing it. Specifically, error handling
cases need to be protected from freeing the xfer while it is still in
use elsewhere. This fixes a kernel panic that can occur when pulling
the HWA dongle while data is being transferred to a wireless device.
Signed-off-by: Thomas Pugliese <thomas.pugliese@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ehci_irq() and ehci_hrtimer_func() can deadlock on ehci->lock when
threadirqs option is used. To prevent the deadlock use
spin_lock_irqsave() in ehci_irq().
This change can be reverted when hrtimer callbacks become threaded.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: stable <stable@vger.kernel.org>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Hello,
the following patch adds an entry for the PID of a Cressi Leonardo
diving computer interface to kernel 3.13.0.
It is detected as FT232RL.
Works with subsurface.
Signed-off-by: Joerg Dorchain <joerg@dorchain.net>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Delete ARM's asm/system.h. It's the last holdout and should be got rid of.
This builds for defconfig, lpc32xx_defconfig, exynos_defconfig + XEN, the
previous changed to a Gemini system and an omap3 config with TI_DAVINCI_EMAC.
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
It is an error for a driver to call usb_clear_halt() or
usb_reset_endpoint() while there are URBs queued for the endpoint,
because the end result is not well defined. At the time the endpoint
gets reset, it may or may not be actively running.
As far as I know, no kernel drivers do this. But some userspace
drivers do, and it seems like a good idea to bring this error to their
attention.
This patch adds a warning to the kernel log whenever a program invokes
the USBDEVFS_CLEAR_HALT or USBDEVFS_RESETEP ioctls at an inappropriate
time, and includes the name of the program. This will make it clear
that any subsequent errors are not due to the misbehavior of a kernel
driver.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Suggested-by: Bjørn Mork <bjorn@mork.no>
CC: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
No need to return a 'fake' return value on platform_get_irq() failure.
Just return the error code itself instead.
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch makes it possible to set the chipidea udc into full-speed only mode.
It is set by the oftree property "maximum-speed = full-speed".
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use the wrapper function for retrieving the platform data instead
of accessing dev->platform_data directly. This is a cosmetic change
to make the code simpler and enhance the readability.
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
- delete the warning message at interrupt handler, and adds judgement at
ep_enable, if non-ep0 requests ctrl transfer, it will indicate an error.
- delete hw_test_and_clear_setup_status which is a broken code
- Tested with g_mass_storage, g_ncm, g_ether
Cc: matthieu.castet@parrot.com
Reported-by: Michael Grzeschik <mgr@pengutronix.de>
Acked-by: Michael Grzeschik <mgr@pengutronix.de>
Tested-by: Michael Grzeschik <mgr@pengutronix.de>
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
- Delete global_phy due to we can get the phy from phy layer now
- using devm_usb_get_phy to instead of usb_get_phy
- delete the otg_set_peripheral, which should be handled by otg layer
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ENDPTFLUSH and ENDPTPRIME registers are set by software and clear
by hardware. There is a bit for each endpoint. When we are setting
a bit for an endpoint we should make sure we do not touch other
endpoint bit. There is a race condition if the hardware clear the
bit between the read and the write in hw_write.
Cc: stable <stable@vger.kernel.org> # 3.11+
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Matthieu CASTET <matthieu.castet@parrot.com>
Tested-by: Michael Grzeschik <mgrzeschik@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Here are 10 fixes for our current -rc cycle. It's likely
the last round of fixes for this merge window. All patches
have soaked for a long time and have all been tested in real
HW.
The most interesting fixes are a fix for enumeration of superspeed
hubs when MUSB is acting as host, and a remote-wakeup fix also on
MUSB.
Signed-of-by: Felipe Balbi <balbi@ti.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=AnZJ
-----END PGP SIGNATURE-----
Merge tag 'fixes-for-v3.14-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-linus
Felipe writes:
usb: fixes for v3.14-rc4
Here are 10 fixes for our current -rc cycle. It's likely
the last round of fixes for this merge window. All patches
have soaked for a long time and have all been tested in real
HW.
The most interesting fixes are a fix for enumeration of superspeed
hubs when MUSB is acting as host, and a remote-wakeup fix also on
MUSB.
Signed-of-by: Felipe Balbi <balbi@ti.com>
debugfs files to show the contents of important dsps registers.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
schedule_delayed_work() takes the delay in jiffies, not msecs.
This bug slipped in with the recent reset logic cleanup
(8ed1fb790e: "usb: musb: finish suspend/reset work independently from
musb_hub_control()").
Signed-off-by: Daniel Mack <daniel@zonque.org>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Both the PM_RUNTIME and PM_SLEEP callbacks call into the common
msm_otg_{suspend,resume} routines, however these routines are only being
built when CONFIG_PM_SLEEP. In addition, msm_otg_{suspend,resume} also
depends on msm_hsusb_config_vddcx(), which is only built when
CONFIG_PM_SLEEP.
Fix the CONFIG_PM_RUNTIME, !CONFIG_PM_SLEEP case by changing the
preprocessor conditional, and moving msm_hsusb_config_vddcx().
While we're here, eliminate the CONFIG_PM conditional for setting
up the dev_pm_ops.
This address the following errors Russell King has hit doing randconfig
builds:
drivers/usb/phy/phy-msm-usb.c: In function 'msm_otg_runtime_suspend':
drivers/usb/phy/phy-msm-usb.c:1691:2: error: implicit declaration of function 'msm_otg_suspend'
drivers/usb/phy/phy-msm-usb.c: In function 'msm_otg_runtime_resume':
drivers/usb/phy/phy-msm-usb.c:1699:2: error: implicit declaration of function 'msm_otg_resume'
Cc: Ivan T. Ivanov <iivanov@mm-sol.com>
Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Josh Cartwright <joshc@codeaurora.org>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Fix possible NULL pointer dereference introduced in
commit 219580e (usb: f_fs: check quirk to pad epout
buf size when not aligned to maxpacketsize)
In cases we do wait with:
wait_event_interruptible(epfile->wait, (ep = epfile->ep));
for endpoint to be enabled, functionfs_bind() has not been called yet
and epfile->ffs->gadget is still NULL and the automatic variable 'gadget'
has been initialized with NULL at the point of its definition.
Later on it is used as a parameter to:
usb_ep_align_maybe(gadget, ep->ep, len)
which in turn dereferences it.
This patch fixes it by moving the actual assignment to the local 'gadget'
variable after the potential waiting has completed.
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
We need to use gadget_is_otg to check if the gadget is really
otg support at runtime, other composite gadget drivers have already
followed this method.
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Around DWC USB3 2.30a release another bit has been added to the
Device-Specific Event (DEVT) Event Information (EvtInfo) bitfield.
Because of that, what used to be 8 bits long, has become 9 bits long.
Per dwc3 2.30a+ spec in the Device-Specific Event (DEVT), the field of
Event Information Bits(EvtInfo) uses [24:16] bits, and it has 9 bits
not 8 bits. And the following reserved field uses [31:25] bits not
[31:24] bits, and it has 7 bits.
So in dwc3_event_devt, the bit mask should be:
event_info [24:16] 9 bits
reserved31_25 [31:25] 7 bits
This patch makes sure that newer core releases will work fine with
Linux and that we will decode the event information properly on new
core releases.
[ balbi@ti.com : improve commit log a bit ]
Cc: <stable@vger.kernel.org>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Debugfs function return an ERR_PTR if they compiled out. We don't need
to test for that here because if the debugfs file are compiled out then
it is ok to pass an ERR_PTR to debugfs_create_file() since it will just
be a no-op stub.
Debugfs return NULLs on error, but we don't need to test for that either
because debugfs_create_file() will accept NULL pointers.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
A number of Kconfig entries default to (uppercase) "N". It was clearly
intended to use "default n". But since (lowercase) "n" is the default
anyway, these lines might as well be removed.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
The SoCs earlier than sama5d3, they have the same number endpoints
and DMA channels. In driver code, they use the same definition
USBA_NR_ENDPOINTS for both endpoints and dma channels. However,
in sama5d3, it has different number for endpoints and DMA channels.
So, define a new macro USBA_NR_DMAs for DMA channels. And the
USBA_NR_ENDPOINS is not used anymore, remove it at the same time.
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Bo Shen <voice.shen@atmel.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
In sama5d3 SoC, there are 16 endpoints, which is different with
earlier SoCs (only have 7 endpoints). The USBA_NR_ENDPOINTS macro
is not suitable for sama5d3. So, get the endpoints number through
the udc->num_ep, which get from platform data for non-dt kernel,
or parse from dt node.
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Bo Shen <voice.shen@atmel.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch fix spelling typo in Documentation/DocBook.
It is because .html and .xml files are generated by make htmldocs,
I have to fix a typo within the source files.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
USB_ARCH_HAS_EHCI, USB_ARCH_HAS_OHCI, and USB_ARCH_HAS_XHCI were made
obsolete in v3.11. They have not been used ever since. Setting them has
no effect. They can safely be removed.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
schedule_delayed_work() takes the delay in jiffies, not msecs. Fix the
caller sites in musb. This bug caused regressions with the cleanups
that went in for 3.14 (8ed1fb790e: "usb: musb: finish suspend/reset
work independently from musb_hub_control()").
Signed-off-by: Daniel Mack <daniel@zonque.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
musb_port_reset() is called from musb_hub_control() which in turn holds
a spinlock, so musb_port_reset() is not allowed to call msleep().
With the asynchronous work helpers in place, this is fortunately easy to
fix by rescheduling the reset deassertion function to after the time
when the wait period is finished.
Note, however, that the MUSB_POWER_RESUME bit is only set on AM33xx
processors under rare conditions such as when to another driver
reporting an error during suspend. Hence, this didn't hit me yet in
normal operation.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This interface is to be handled by the qmi_wwan driver.
CC: Hans-Christoph Schemmel <hans-christoph.schemmel@gemalto.com>
CC: Christian Schmiedl <christian.schmiedl@gemalto.com>
CC: Nicolaus Colberg <nicolaus.colberg@gemalto.com>
CC: David McCullough <david.mccullough@accelecon.com>
Signed-off-by: Aleksander Morgado <aleksander@aleksander.es>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
High-speed USB connections revert back to full-speed signalling when
the device goes into suspend. This takes several milliseconds, and
during that time it's not possible to tell reliably whether the device
has been disconnected.
On some platforms, the Wake-On-Disconnect circuitry gets confused
during this intermediate state. It generates a false wakeup signal,
which can prevent the controller from going to sleep.
To avoid this problem, this patch adds a 5-ms delay to the
ehci_bus_suspend() routine if any ports have to switch over to
full-speed signalling. (Actually, the delay was already present for
devices using a particular kind of PHY power management; the patch
merely causes the delay to be used more widely.)
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reviewed-by: Peter Chen <Peter.Chen@freescale.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds asynchronous I/O support for FunctionFS endpoint files.
It adds ffs_epfile_aio_write() and ffs_epfile_aio_read() functions responsible
for preparing AIO operations.
It also modifies ffs_epfile_io() function, adding aio handling code. Instead
of extending list of parameters of this function, there is new struct
ffs_io_data which contains all information needed to perform I/O operation.
Pointer to this struct replaces "buf" and "len" parameters of ffs_epfile_io()
function. Allocated buffer is freed immediately only after sync operation,
because in async IO it's freed in complete funcion. For each async operation
an USB request is allocated, because it allows to have more than one request
queued on single endpoint.
According to changes in ffs_epfile_io() function, functions ffs_epfile_write()
and ffs_epfile_read() are updated to use new API.
For asynchronous I/O operations there is new request complete function named
ffs_epfile_async_io_complete(), which completes AIO operation, and frees
used memory.
Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch adds poll function for file representing ep0.
Ability of read from or write to ep0 file is related with actual state of ffs:
- When desctiptors or strings are not written yet, POLLOUT flag is set.
- If there is any event to read, POLLIN flag is set.
- If setup request was read, POLLIN and POLLOUT flag is set, to allow
send response (by performing I/O operation consistent with setup request
direction) or set stall (by performing I/O operation opposite setup
request direction).
Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch fixes __ffs_ep0_queue_wait() function, which now returns number of
bytes transferred in USB request or error code in case of failure. This is
needed by ffs_ep0_read() function, when read data is copied to userspace.
It also cleans up code by removing usused variable ep0req_status.
Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The FFS_SETUP_STATUS macro could be trivialy replaced with an static
inline function but more importantly its name was tad confusing.
The name suggested it was a simple accessor macro but it actually
did change the state of the ffs_data structure perfomring
a FFS_SETUP_CANCELLED -> FFS_NO_SETUP transition. The name of the
function -- ffs_setup_state_clear_cancelled -- should better
describe what the function actually does.
Signed-off-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Since “cancelled” is spelled with two “l”s, rename FFS_SETUP_CANCELED
to FFS_SETUP_CANCELLED.
Signed-off-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Consistently prefix function name with underscore if the function has to
be called with ffs_lock taken.
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
ffs_alloc_dev and ffs_free_dev are used only in f_fs.c,
so make them static.
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Remove trailing whitespace
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
ffs_dev->ffs_release_dev_callback should be accessed only if ffs_dev
is not NULL.
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Replace kzalloc by devm_kzalloc and remove the kfree() calls.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
devm_ioremap_resource does sanity checks on the given resource. No need to
duplicate this in the driver.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
When s3c_hsotg_ep_sethalt() function is called for ep0 it should be stalled
in the same way that it is in s3c_hsotg_process_control() function, because
SET_HALT for ep0 is delayed response for setup request. Endpoint 0, if
halted, it doesn't need CLEAR_HALT because it clears "stalled" state
automatically when next setup request is received.
For this reason this patch moves code setting ep0 to "stalled" state to new
function named s3c_hsotg_stall_ep0() which is called in
s3c_hsotg_process_control() function as an immediate response for setup
request, and in s3c_hsotg_ep_sethalt() function as a delayed response for
setup request.
Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
When read data from g_printer, we see a Segmentation fault. eg:
Unable to handle kernel paging request at virtual address bf048000 pgd
= cf038000 [bf048000] *pgd=8e8cf811, *pte=00000000, *ppte=00000000
Internal error: Oops: 7 [#1] PREEMPT ARM Modules linked in: bluetooth
rfcomm g_printer
CPU: 0 Not tainted (3.4.43-WR5.0.1.9_standard #1)
PC is at __copy_to_user_std+0x310/0x3a8 LR is at 0x4c808010
pc : [<c036e990>] lr : [<4c808010>] psr: 20000013
sp : cf883ea8 ip : 80801018 fp : cf883f24
r10: bf04706c r9 : 18a21205 r8 : 21953888
r7 : 201588aa r6 : 5109aa16 r5 : 0705aaa2 r4 : 5140aa8a
r3 : 0000004c r2 : 00000fdc r1 : bf048000 r0 : bef5fc3c
Flags: nzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user
Control: 10c5387d Table: 8f038019 DAC: 00000015 Process
g_printer_test. (pid: 661, stack limit = 0xcf8822e8)
Stack: (0xcf883ea8 to 0xcf884000)
3ea0: bf047068 00001fff bef5ecb9 cf882000 00001fff bef5ecb9
3ec0: 00001fff 00000000 cf2e8724 bf044d3c 80000013 80000013 00000001
bf04706c
3ee0: cf883f24 cf883ef0 c012e5ac c0324388 c007c8ac c0046298 00008180
cf29b900
3f00: 00002000 bef5ecb8 cf883f68 00000003 cf882000 cf29b900 cf883f54
cf883f28
3f20: c012ea08 bf044b0c c000eb88 00000000 cf883f7c 00000000 00000000
00002000
3f40: bef5ecb8 00000003 cf883fa4 cf883f58 c012eae8 c012e960 00000001
bef60cb8
3f60: 000000a8 c000eb88 00000000 00000000 cf883fa4 00000000 c014329c
00000000
3f80: 000000d4 41af63f0 00000003 c000eb88 cf882000 00000000 00000000
cf883fa8
3fa0: c000e920 c012eaa4 00000000 000000d4 00000003 bef5ecb8 00002000
bef5ecb8
3fc0: 00000000 000000d4 41af63f0 00000003 b6f534c0 00000000 419f9000
00000000
3fe0: 00000000 bef5ecac 000086d9 41a986bc 60000010 00000003 0109608a
0088828a
Code: f5d1f07c e8b100f0 e1a03c2e e2522020 (e8b15300) ---[ end trace
97e2618e250e3377 ]--- Segmentation fault
The root cause is the dev->rx_buffers list has been broken.
When we call printer_read(), the following call tree is triggered:
printer_read()
|
+---setup_rx_reqs(req)
| |
| +---usb_ep_queue(req)
| | |
| | +---...
| | |
| | +---rx_complete(req).
| |
| +---add the req to dev->rx_reqs_active
|
+---while(!list_empty(&dev->rx_buffers)))
The route happens when we don't use DMA or fail to start DMA in USB
driver. We can see: in the case, in rx_complete() it will add the req
to dev->rx_buffers. meanwhile we see that we will also add the req to
dev->rx_reqs_active after usb_ep_queue() return, so this adding will
break the dev->rx_buffers out.
After, when we call list_empty() to check dev->rx_buffers in while(),
due to can't check correctly dev->rx_buffers, so the Segmentation fault
occurs when copy_to_user() is called.
Signed-off-by: wenlin.kang <wenlin.kang@windriver.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The problem occurs in follow path.
printer_read()
|
+---setup_rx_reqs()
|
+---usb_ep_queue()
|
+---...
|
+---rx_complete()
Although it is clear from code, we can't get it normally.
only when we enable some spin_lock debug config option, we can find it.
eg:
BUG: spinlock lockup on CPU#0, g_printer_test_/584
lock: bf05e158, .magic: dead4ead, .owner: g_printer_test_/584, .owner_cpu: 0
[<c0016e1c>] (unwind_backtrace+0x0/0x104) from [<c067aef8>] (dump_stack+0x20/0x24)
[<c067aef8>] (dump_stack+0x20/0x24) from [<c0680bec>] (spin_dump+0x8c/0x94)
[<c0680bec>] (spin_dump+0x8c/0x94) from [<c039071c>] (do_raw_spin_lock+0x128/0x154)
[<c039071c>] (do_raw_spin_lock+0x128/0x154) from [<c0685618>] (_raw_spin_lock_irqsave+0x64/0x70)
[<c0685618>] (_raw_spin_lock_irqsave+0x64/0x70) from [<bf05b4e8>] (rx_complete+0x54/0x10c [g_printer])
[<bf05b4e8>] (rx_complete+0x54/0x10c [g_printer]) from [<c0480478>] (musb_g_giveback+0x78/0x88)
[<c0480478>] (musb_g_giveback+0x78/0x88) from [<c048060c>] (rxstate+0xa0/0x10c)
[<c048060c>] (rxstate+0xa0/0x10c) from [<c0480d50>] (musb_ep_restart+0x44/0x70)
[<c0480d50>] (musb_ep_restart+0x44/0x70) from [<c0480fe4>] (musb_gadget_queue+0xe8/0xf8)
[<c0480fe4>] (musb_gadget_queue+0xe8/0xf8) from [<bf05b2b0>] (setup_rx_reqs+0xa4/0x178 [g_printer])
[<bf05b2b0>] (setup_rx_reqs+0xa4/0x178 [g_printer]) from [<bf05bb58>] (printer_read+0x9c/0x3f4 [g_printer])
[<bf05bb58>] (printer_read+0x9c/0x3f4 [g_printer]) from [<c01387f0>] (vfs_read+0xb4/0x144)
[<c01387f0>] (vfs_read+0xb4/0x144) from [<c01388d0>] (sys_read+0x50/0x124)
[<c01388d0>] (sys_read+0x50/0x124) from [<c000e900>] (ret_fast_syscall+0x0/0x3c)
The root cause is that we use the same lock two time in a path, so to avoid
the deadlock, we need to unlock in setup_rx_reqs(), and only unlock.
Signed-off-by: wenlin.kang <wenlin.kang@windriver.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Fixes the following compilation warning:
drivers/usb/gadget/s3c-hsudc.c: In function ‘s3c_hsudc_probe’:
drivers/usb/gadget/s3c-hsudc.c:1347:1: warning: label ‘err_add_device’
defined but not used [-Wunused-label]
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Felipe Balbi <balbi@ti.com>
