Currently the default bMaxBurst is 0. Set default bMaxBurst to 15 (i.e.
16 bursts) to Data IN and OUT endpoints to improve performance. It
should be fine for a controller that supports less than 16 bursts. It
should be able to negotiate properly with the host at packet level for
the end of burst.
If the controller can't handle a burst of 16, and high performance isn't
important, the user can use BOT protocol from mass_storage gadget driver
instead.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/9d6265db4d138e542f281988362bc4392b034d43.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This commit addresses an issue related to below kernel panic where
panic_on_warn is enabled. It is caused by the unnecessary use of WARN_ON
in functionsfs_bind, which easily leads to the following scenarios.
1.adb_write in adbd 2. UDC write via configfs
================= =====================
->usb_ffs_open_thread() ->UDC write
->open_functionfs() ->configfs_write_iter()
->adb_open() ->gadget_dev_desc_UDC_store()
->adb_write() ->usb_gadget_register_driver_owner
->driver_register()
->StartMonitor() ->bus_add_driver()
->adb_read() ->gadget_bind_driver()
<times-out without BIND event> ->configfs_composite_bind()
->usb_add_function()
->open_functionfs() ->ffs_func_bind()
->adb_open() ->functionfs_bind()
<ffs->state !=FFS_ACTIVE>
The adb_open, adb_read, and adb_write operations are invoked from the
daemon, but trying to bind the function is a process that is invoked by
UDC write through configfs, which opens up the possibility of a race
condition between the two paths. In this race scenario, the kernel panic
occurs due to the WARN_ON from functionfs_bind when panic_on_warn is
enabled. This commit fixes the kernel panic by removing the unnecessary
WARN_ON.
Kernel panic - not syncing: kernel: panic_on_warn set ...
[ 14.542395] Call trace:
[ 14.542464] ffs_func_bind+0x1c8/0x14a8
[ 14.542468] usb_add_function+0xcc/0x1f0
[ 14.542473] configfs_composite_bind+0x468/0x588
[ 14.542478] gadget_bind_driver+0x108/0x27c
[ 14.542483] really_probe+0x190/0x374
[ 14.542488] __driver_probe_device+0xa0/0x12c
[ 14.542492] driver_probe_device+0x3c/0x220
[ 14.542498] __driver_attach+0x11c/0x1fc
[ 14.542502] bus_for_each_dev+0x104/0x160
[ 14.542506] driver_attach+0x24/0x34
[ 14.542510] bus_add_driver+0x154/0x270
[ 14.542514] driver_register+0x68/0x104
[ 14.542518] usb_gadget_register_driver_owner+0x48/0xf4
[ 14.542523] gadget_dev_desc_UDC_store+0xf8/0x144
[ 14.542526] configfs_write_iter+0xf0/0x138
Fixes: ddf8abd259 ("USB: f_fs: the FunctionFS driver")
Cc: stable <stable@kernel.org>
Signed-off-by: Akash M <akash.m5@samsung.com>
Link: https://lore.kernel.org/r/20241219125221.1679-1-akash.m5@samsung.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Considering that in some extreme cases, when performing the
unbinding operation, gserial_disconnect has cleared gser->ioport,
which triggers gadget reconfiguration, and then calls gs_read_complete,
resulting in access to a null pointer. Therefore, ep is disabled before
gserial_disconnect sets port to null to prevent this from happening.
Call trace:
gs_read_complete+0x58/0x240
usb_gadget_giveback_request+0x40/0x160
dwc3_remove_requests+0x170/0x484
dwc3_ep0_out_start+0xb0/0x1d4
__dwc3_gadget_start+0x25c/0x720
kretprobe_trampoline.cfi_jt+0x0/0x8
kretprobe_trampoline.cfi_jt+0x0/0x8
udc_bind_to_driver+0x1d8/0x300
usb_gadget_probe_driver+0xa8/0x1dc
gadget_dev_desc_UDC_store+0x13c/0x188
configfs_write_iter+0x160/0x1f4
vfs_write+0x2d0/0x40c
ksys_write+0x7c/0xf0
__arm64_sys_write+0x20/0x30
invoke_syscall+0x60/0x150
el0_svc_common+0x8c/0xf8
do_el0_svc+0x28/0xa0
el0_svc+0x24/0x84
Fixes: c1dca562be ("usb gadget: split out serial core")
Cc: stable <stable@kernel.org>
Suggested-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Lianqin Hu <hulianqin@vivo.com>
Link: https://lore.kernel.org/r/TYUPR06MB621733B5AC690DBDF80A0DCCD2042@TYUPR06MB6217.apcprd06.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently afunc_bind sets std_ac_if_desc.bNumEndpoints to 1 if
controls (mute/volume) are enabled. During next afunc_bind call,
bNumEndpoints would be unchanged and incorrectly set to 1 even
if the controls aren't enabled.
Fix this by resetting the value of bNumEndpoints to 0 on every
afunc_bind call.
Fixes: eaf6cbe099 ("usb: gadget: f_uac2: add volume and mute support")
Cc: stable <stable@kernel.org>
Signed-off-by: Prashanth K <quic_prashk@quicinc.com>
Link: https://lore.kernel.org/r/20241211115915.159864-1-quic_prashk@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Considering that in some extreme cases,
when u_serial driver is accessed by multiple threads,
Thread A is executing the open operation and calling the gs_open,
Thread B is executing the disconnect operation and calling the
gserial_disconnect function,The port->port_usb pointer will be set to NULL.
E.g.
Thread A Thread B
gs_open() gadget_unbind_driver()
gs_start_io() composite_disconnect()
gs_start_rx() gserial_disconnect()
... ...
spin_unlock(&port->port_lock)
status = usb_ep_queue() spin_lock(&port->port_lock)
spin_lock(&port->port_lock) port->port_usb = NULL
gs_free_requests(port->port_usb->in) spin_unlock(&port->port_lock)
Crash
This causes thread A to access a null pointer (port->port_usb is null)
when calling the gs_free_requests function, causing a crash.
If port_usb is NULL, the release request will be skipped as it
will be done by gserial_disconnect.
So add a null pointer check to gs_start_io before attempting
to access the value of the pointer port->port_usb.
Call trace:
gs_start_io+0x164/0x25c
gs_open+0x108/0x13c
tty_open+0x314/0x638
chrdev_open+0x1b8/0x258
do_dentry_open+0x2c4/0x700
vfs_open+0x2c/0x3c
path_openat+0xa64/0xc60
do_filp_open+0xb8/0x164
do_sys_openat2+0x84/0xf0
__arm64_sys_openat+0x70/0x9c
invoke_syscall+0x58/0x114
el0_svc_common+0x80/0xe0
do_el0_svc+0x1c/0x28
el0_svc+0x38/0x68
Fixes: c1dca562be ("usb gadget: split out serial core")
Cc: stable@vger.kernel.org
Suggested-by: Prashanth K <quic_prashk@quicinc.com>
Signed-off-by: Lianqin Hu <hulianqin@vivo.com>
Acked-by: Prashanth K <quic_prashk@quicinc.com>
Link: https://lore.kernel.org/r/TYUPR06MB62178DC3473F9E1A537DCD02D2362@TYUPR06MB6217.apcprd06.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The UMP Function Block info m1.0 field (represented by is_midi1 sysfs
entry) is an enumeration from 0 to 2, while the midi2 gadget driver
incorrectly copies it to the corresponding snd_ump_block_info.flags
bits as-is. This made the wrong bit flags set when m1.0 = 2.
This patch corrects the wrong interpretation of is_midi1 bits.
Fixes: 29ee7a4ddd ("usb: gadget: midi2: Add configfs support")
Cc: stable@vger.kernel.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20241127070213.8232-1-tiwai@suse.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the streamoff call was triggered by some previous disconnect
or userspace application shutdown the uvc_function_setup_continue
should not be called and the state should not be overwritten.
For this situation the set_alt(0) was never called and the streaming ep
has no USB_GADGET_DELAYED_STATUS pending.
Since the state then was already updated before we also omit the state
update.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Link: https://lore.kernel.org/r/20240403-uvc_request_length_by_interval-v7-9-e224bb1035f0@pengutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Compressed formats generate content depending amount of data that is set
in the vb2 buffer by the payload_size. When streaming those formats it
is better to scatter that smaller data over all requests. This patch is
doing that by introducing the calculated req_payload_size which is
updated by each frame. It the uses this amount of data to fill the
isoc requests instead of the video->req_size.
For uncompressed formats it will not make a difference since the payload
size will be equal to the imagesize. Therefore the code will have no
effecta as req_payload_size will be equal to req_size.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Link: https://lore.kernel.org/r/20240403-uvc_request_length_by_interval-v7-6-e224bb1035f0@pengutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch is removing the initial imprecise and limited calculation of
requests needed to be used from the queue_setup callback. It instead
introduces the uvc_video_prep_requests function which is called
immediately before the request allocation.
With the information of the usb frame interval length it is possible to
calculate the number of requests needed during one frame duration.
Based on the calculated number of requests and the imagesize we
calculate the actual size per request. This calculation has the benefit
that the frame data is equally distributed over all allocated requests.
When the req_size is not in the range for the actually configured
max_req_size configured for the overall bandwidth we fallback
to use the max_req_size instead.
Since this calculations are only important for isoc transfers we just
use max_request_size for bulk and skip it.
As video->req_size will be recalculated on every video_enable resetting
it to 0 is not necessary anymore.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Link: https://lore.kernel.org/r/20240403-uvc_request_length_by_interval-v7-5-e224bb1035f0@pengutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We install an kthread with pfifo priority that is iterating over all
prepared requests and keeps the isoc queue busy. This way it will be
scheduled with the same priority as the interrupt handler.
As the kthread is triggered with video_enable it will immediately
queue some zero length requests into the hw if there is no buffer data
available. It also watches the level of needed zero length requests in
the hardware not to fall under the UVCG_REQ_MAX_ZERO_COUNT threshold.
This way we can drop the function uvc_video_ep_queue_initial_requests
entirely.
By using the kthread to do the actual request handling the interrupt
handler will not be running into the time consuming and eventually
locking work of actually enqueueing the requests back into its own
pipeline. This work can now even be scheduled on another cpu.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Link: https://lore.kernel.org/r/20240403-uvc_request_length_by_interval-v7-3-e224bb1035f0@pengutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The complete handler will at least be called after 16 requests have
completed, but will still handle all finisher requests. Since we have
to maintain a costant filling in the isoc queue we ensure this by
adding zero length requests.
By counting the amount enqueued requests we can ensure that the queue is
never underrun and only need to get active if the queue is running
critical. This patch is setting 32 as the critical level, which
is twice the request amount that is needed to create interrupts.
To properly solve the amount of zero length requests that needs to
be held in the hardware after one interrupt needs to be measured
and depends on the runtime of the first enqueue run after the interrupt
triggered. For now we just use twice the amount of requests between an
interrupt.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Link: https://lore.kernel.org/r/20240403-uvc_request_length_by_interval-v7-2-e224bb1035f0@pengutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add support for frame-based frame format, which can be used to support
multiple formats like H264 or H265, in addition to MJPEG and YUV frames.
The frame-based format is set to H264 by default, but it can be updated
to other formats by modifying the GUID through the guid configfs
attribute. Different structures are used for all three formats, as
H264 has a different structure compared to MJPEG and uncompressed
formats. These structures will be passed to the frame make function
based on the active format, using a common frame structure with
additional parameters needed only for frame-based formats. These
parameters are handled at runtime in the UVC driver.
Signed-off-by: Akash Kumar <quic_akakum@quicinc.com>
Link: https://lore.kernel.org/r/20240927152138.31416-1-quic_akakum@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The configfs store callback should return the number of bytes consumed
not the total number of bytes we actually stored. These could differ if
for example the passed in string had a newline we did not store.
If the returned value does not match the number of bytes written the
writer might assume a failure or keep trying to write the remaining bytes.
For example the following command will hang trying to write the final
newline over and over again (tested on bash 2.05b):
echo foo > function_name
Fixes: 993a44fa85 ("usb: gadget: f_uac2: allow changing interface name via configfs")
Cc: stable <stable@kernel.org>
Signed-off-by: Kevin Groeneveld <kgroeneveld@lenbrook.com>
Link: https://lore.kernel.org/r/20241006232637.4267-1-kgroeneveld@lenbrook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
asm/unaligned.h is always an include of asm-generic/unaligned.h;
might as well move that thing to linux/unaligned.h and include
that - there's nothing arch-specific in that header.
auto-generated by the following:
for i in `git grep -l -w asm/unaligned.h`; do
sed -i -e "s/asm\/unaligned.h/linux\/unaligned.h/" $i
done
for i in `git grep -l -w asm-generic/unaligned.h`; do
sed -i -e "s/asm-generic\/unaligned.h/linux\/unaligned.h/" $i
done
git mv include/asm-generic/unaligned.h include/linux/unaligned.h
git mv tools/include/asm-generic/unaligned.h tools/include/linux/unaligned.h
sed -i -e "/unaligned.h/d" include/asm-generic/Kbuild
sed -i -e "s/__ASM_GENERIC/__LINUX/" include/linux/unaligned.h tools/include/linux/unaligned.h
no_llseek had been defined to NULL two years ago, in commit 868941b144
("fs: remove no_llseek")
To quote that commit,
At -rc1 we'll need do a mechanical removal of no_llseek -
git grep -l -w no_llseek | grep -v porting.rst | while read i; do
sed -i '/\<no_llseek\>/d' $i
done
would do it.
Unfortunately, that hadn't been done. Linus, could you do that now, so
that we could finally put that thing to rest? All instances are of the
form
.llseek = no_llseek,
so it's obviously safe.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
