The UAC2 function driver currently responds to all packets at all times
with wMaxPacketSize packets. That results in way too fast audio
playback as the function driver (which is in fact supposed to define
the audio stream pace) delivers as fast as it can.
Fix this by sizing each packet correctly with the following steps:
a) Set the packet's size by dividing the nominal data rate by the
playback endpoint's interval.
b) If there is a residual value from the calculation in a), add
it to a accumulator to keep track of it across packets.
c) If the accumulator has gathered at least the number of bytes
that are needed for one sample frame, increase the packet size.
This way, the packet size calculation will get rid of any kind of
imprecision that would otherwise occur with a simple division over
time.
Some of the variables that are needed while processing each packet
are pre-computed for performance reasons.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
With packet sizes other than 512, payloads in the packets may wrap
around the ALSA dma buffer partially, which leads to memory corruption
and audible clicks and pops in the audio stream at the moment, because
there is no boundary check before the memcpy().
In preparation to an implementation for smaller and dynamically sized
packets, we have to address such cases, and copy the payload in two
steps conditionally.
The 'src' and 'dst' approach doesn't work here anymore, as different
behavior is necessary in playback and capture cases. Thus, this patch
open-codes the routine now.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Add a simple container_of() wrapper to get a struct f_uac2_opts from a
struct struct audio_dev. Use it in two places where it is currently
open-coded.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
In afunc_bind() and afunc_set_alt(), &uac2->pdev.dev are used multiple
times. Adding a short-hand for them makes lines shorter so we can
remove some line wraps.
No functional change.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Restructure some code to make it easier to read.
While at it, return -ENOMEM instead of -EINVAL if
usb_ep_alloc_request() fails, and omit the logging in such cases
(the mm core will complain loud enough).
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Flatten the start_transfer function by reversing the if condition and
returning early out of the function if everything went fine. It makes
the function look less complicated, at least to me, and easier to
understand.
Signed-off-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
commit f226708(usb: gadget: composite: dequeue cdev->req before free it in
composite_dev_cleanup) fixed a bug: free the usb request(i.e. cdev->req) but
does not dequeue it beforehand. This fix is not proper enough because it
dequeues the request after free its data buffer, considering the hardware can
access the buffer's memory anytime before the request's complettion rountine
runs, and usb_ep_dequeue always call the complettion rountine before it returns,
so the best way is to dequeue the request before free its buffer.
Suggested-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Li Jun <b47624@freescale.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
If NO_DMA=y:
drivers/built-in.o: In function `usbhsg_dma_map_ctrl':
mod_gadget.c:(.text+0x53b226): undefined reference to `usb_gadget_map_request'
mod_gadget.c:(.text+0x53b242): undefined reference to `usb_gadget_unmap_request'
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Using $(PWD) breaks builds when make was invoked from outside
of the kernel tree.
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch introduces the use of the functions usb_endpoint_type and
usb_endpoint_num.
The Coccinelle semantic patch that makes these changes is as follows:
@@ struct usb_endpoint_descriptor *epd; @@
- (epd->bEndpointAddress & \(USB_ENDPOINT_NUMBER_MASK\|0x0f\))
+ usb_endpoint_num(epd)
@@ struct usb_endpoint_descriptor *epd; @@
- (epd->bmAttributes & \(USB_ENDPOINT_XFERTYPE_MASK\|3\))
+ usb_endpoint_type(epd)
Signed-off-by: Himangi Saraogi <himangi774@gmail.com>
Acked-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Interrupt endpoints behave quite similar to the bulk endpoints with the
difference that the endpoints expect data sending/reception request at
particular intervals till the whole data has not been transmitted.
The interrupt EP support is added to gadget zero. A new alternate setting (=2)
has been added. It has 2 interrupt endpoints. The default parameters are set as:
bInterval: 1 ms for FS and 8 uFrames (implying 1 ms) for HS/SS
wMaxPacketSize: 64 bytes for FS and 1024 bytes for HS/SS
However, the same can be overridden through the module parameter interface.
The code is tested for HS and SS on a platform having DWC3 controller.
Signed-off-by: Amit Virdi <amit.virdi@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Remove the ttyGS PREFIX macro from u_serial.c and replace all occurences with
the hardcoded ttyGS string.
This macro was mostly used in a few debug/warning messages and a lot of
hardcoded ttyGS existed beneath. It may have been used for renaming the
tty, but if done so most debug messages would have ignored this.
Due to the fact the usage of this PREFIX in all debug calls would have
resulted in a hard to read/grep code it is removed completely.
Signed-off-by: Richard Leitner <richard.leitner@skidata.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Replace the VDBG and DBG macro with the kernels "proper" debug macros
(dev_vdbg and dev_dbg) in f_acm.c, f_obex.c & f_serial.c
Signed-off-by: Richard Leitner <richard.leitner@skidata.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
As the driver is in mainline we can remove the version numbers.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
If uvc->control_buf is successfuly allocated but uvc->control_req
is not, uvc->control_buf is not freed in the error recovery path.
With this patch applied uvc->control_buf is freed unconditionally;
if it happens to be NULL kfree on it is safe anyway.
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Activate the videobuf2 DMABUF support. As vb2-vmalloc supports the
importer role only, exporting buffers isn't supported yet. When the
exporter role will be implemented in vb2-vmalloc the UVC gadget driver
will automatically gain support for it.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The disconnect of the USB Device is a common pattern for
an UVC Camera. In many cases this will give us an meaningless
information for all buffers that couldn't be enqueued.
That patch changes this to KERN_DEBUG.
Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Add support for using f_uac1 function as a component of a gadget
composed with configfs.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
When configfs support is added the values in question will have to be
used in two different places. Substitute them with defined constants
to avoid duplicating magic numbers.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Use the new usb_gstring_attach interface.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
There are no users of the old interface left, so it can be removed.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Use the new interface so that the old one can be removed.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Converting uac1 to the new function interface requires converting
the USB uac1's function code and its users.
This patch converts the f_uac1.c to the new function interface.
The file is now compiled into a separate usb_f_uac1.ko module.
The old function interface is provided by means of a preprocessor
conditional directives. After all users are converted, the old interface
can be removed.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Integrating configfs requires converting f_uac1 to new function interface,
which in turn requires converting it to the new function interface,
which involves separate compilation of f_uac1.c into usb_f_uac1.ko.
u_uac1.c contains some module parameters. After this patch is applied
they are still a part of the resulting g_audio.ko, but can be guarded
with a compatiblity flag which will be removed when no users of the old
function interface of f_uac1 are left.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
uac1 function is missing strings. Add them.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Add support for using f_uac2 function as a component of a gadget
composed with configfs.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
When configfs is integrated the same values will have to be used
as defaults. Use symbolic names in order not to duplicate magic numbers.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Use the new usb_gstring_attach interface.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
There are no users of the old interface left, so it can be removed.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Use the new interface so that the old one can be removed.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Converting uac2 to the new function interface requires converting
the USB uac2's function code and its users.
This patch converts the f_uac2.c to the new function interface.
The file is now compiled into a separate usb_f_uac2.ko module.
The old function interface is provided by means of a preprocessor
conditional directives. After all users are converted, the old interface
can be removed.
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Eliminate static struct *agdev_g from f_uac2.c.
It is used for freeing its memory, but the same address can be found
by calling container_of in afunc_unbind(). This implies eliminating
uac2_unbind_config(). The audio_config_driver in audio.c does not have
its unbind method any more. It has been used only when uac2 is used,
so uac2 itself can handle unbinding in afunc_unbind().
Tested-by: Sebastian Reimers <sebastian.reimers@googlemail.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
There is a error message within devm_ioremap_resource
already, so remove the dev_err call to avoid redundant
error message.
Acked-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Pull networking updates from David Miller:
"Highlights:
1) Steady transitioning of the BPF instructure to a generic spot so
all kernel subsystems can make use of it, from Alexei Starovoitov.
2) SFC driver supports busy polling, from Alexandre Rames.
3) Take advantage of hash table in UDP multicast delivery, from David
Held.
4) Lighten locking, in particular by getting rid of the LRU lists, in
inet frag handling. From Florian Westphal.
5) Add support for various RFC6458 control messages in SCTP, from
Geir Ola Vaagland.
6) Allow to filter bridge forwarding database dumps by device, from
Jamal Hadi Salim.
7) virtio-net also now supports busy polling, from Jason Wang.
8) Some low level optimization tweaks in pktgen from Jesper Dangaard
Brouer.
9) Add support for ipv6 address generation modes, so that userland
can have some input into the process. From Jiri Pirko.
10) Consolidate common TCP connection request code in ipv4 and ipv6,
from Octavian Purdila.
11) New ARP packet logger in netfilter, from Pablo Neira Ayuso.
12) Generic resizable RCU hash table, with intial users in netlink and
nftables. From Thomas Graf.
13) Maintain a name assignment type so that userspace can see where a
network device name came from (enumerated by kernel, assigned
explicitly by userspace, etc.) From Tom Gundersen.
14) Automatic flow label generation on transmit in ipv6, from Tom
Herbert.
15) New packet timestamping facilities from Willem de Bruijn, meant to
assist in measuring latencies going into/out-of the packet
scheduler, latency from TCP data transmission to ACK, etc"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1536 commits)
cxgb4 : Disable recursive mailbox commands when enabling vi
net: reduce USB network driver config options.
tg3: Modify tg3_tso_bug() to handle multiple TX rings
amd-xgbe: Perform phy connect/disconnect at dev open/stop
amd-xgbe: Use dma_set_mask_and_coherent to set DMA mask
net: sun4i-emac: fix memory leak on bad packet
sctp: fix possible seqlock seadlock in sctp_packet_transmit()
Revert "net: phy: Set the driver when registering an MDIO bus device"
cxgb4vf: Turn off SGE RX/TX Callback Timers and interrupts in PCI shutdown routine
team: Simplify return path of team_newlink
bridge: Update outdated comment on promiscuous mode
net-timestamp: ACK timestamp for bytestreams
net-timestamp: TCP timestamping
net-timestamp: SCHED timestamp on entering packet scheduler
net-timestamp: add key to disambiguate concurrent datagrams
net-timestamp: move timestamp flags out of sk_flags
net-timestamp: extend SCM_TIMESTAMPING ancillary data struct
cxgb4i : Move stray CPL definitions to cxgb4 driver
tcp: reduce spurious retransmits due to transient SACK reneging
qlcnic: Initialize dcbnl_ops before register_netdev
...
Surprisingly enough, while a big set of patches, the majority is
composed of cleanups (using devm_*, fixing sparse errors, moving
code around, adding const, etc).
The highlights are addition of new support for PLX USB338x devices,
and support for USB 2.0-only configurations of the DWC3 IP core.
Signed-of-by: Felipe Balbi <balbi@ti.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=aJY4
-----END PGP SIGNATURE-----
Merge tag 'usb-for-v3.17' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-next
Felipe writes:
usb: patches for v3.17 merge window
Surprisingly enough, while a big set of patches, the majority is
composed of cleanups (using devm_*, fixing sparse errors, moving
code around, adding const, etc).
The highlights are addition of new support for PLX USB338x devices,
and support for USB 2.0-only configurations of the DWC3 IP core.
Signed-of-by: Felipe Balbi <balbi@ti.com>
f->os_desc_table[0].if_id is zero by default. If the actual id happens
to be different then no Feature Descriptors will be returned to the host
for this interface, so assign if_id as soon as it is known.
Cc: <stable@vger.kernel.org> # v3.16
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The pcm playback and capture sample size format was fixed
SNDRV_PCM_FMTBIT_S16_LE.
This patch respects also 16, 24 and 32 bit p_ssize and c_ssize values.
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sebastian Reimers <sebastian.reimers@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
No functional changes, just fixing one easy to spot
sparse error.
While fixing that sparse error, I had to add two
includes to a header to avoid a build error.
Signed-off-by: Felipe Balbi <balbi@ti.com>
The drivers/usb/gadget directory contains many files.
Files which are related can be distributed into separate directories.
This patch moves the USB functions implementations into a separate directory.
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The drivers/usb/gadget directory contains many files.
Files which are related can be distributed into separate directories.
This patch moves the UDC drivers into a separate directory.
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The drivers/usb/gadget directory contains many files.
Files which are related can be distributed into separate directories.
This patch moves the legacy gadgets (i.e. those not using configfs)
into a separate directory.
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch try to dequeue the cdev->req to guarantee the request is not queued
before free it.
Signed-off-by: Li Jun <b47624@freescale.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This fixes a problem with dropped packets over 16k CDC-NCM
when the connection is being heavily used.
The issue was that the extracted frames cloned from the
received frame were consuming more memory than necessary
resulting in the truesize being ~32KB instead of ~2KB, this
meant there was a high chance of reaching the sk_rcvbuf
limit.
Signed-off-by: Jim Baxter <jim_baxter@mentor.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This adds multi-frame support to the NCM NTB's for
the gadget driver. This allows multiple network
packets to be put inside a single USB NTB with a
maximum size of 16kB.
It has a time out of 300ms to ensure that smaller
number of packets still maintain a normal latency.
Also the .fp_index and .next_fp_index have been
changed to .ndp_index and .next_ndp_index to
match the latest CDC-NCM specification and
help with maintenance.
Results transmitting from gadget to host.
Before the change:
TCP_STREAM Throughput (10^6bits/sec): 22.72
UDP_STREAM Throughput (10^6bits/sec): 25.94
Latency:
netperf -H 192.168.1.101 -v2 -l 50 -t TCP_RR -- -r 16384,16384
Trans. RoundTrip Throughput
Rate Latency 10^6bits/s
per sec usec/Tran Outbound
100.83 9918.116 13.215
After the change:
TCP_STREAM Throughput (10^6bits/sec): 124.26
UDP_STREAM Throughput (10^6bits/sec): 227.48
Latency:
netperf -H 192.168.1.101 -v2 -l 50 -t TCP_RR -- -r 16384,16384
Trans. RoundTrip Throughput
Rate Latency 10^6bits/s
per sec usec/Tran Outbound
156.80 6377.730 20.552
Signed-off-by: Jim Baxter <jim_baxter@mentor.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The NDP was ignoring the wNextNdpIndex in the NDP which
means that NTBs containing multiple NDPs would have missed
frames.
Signed-off-by: Jim Baxter <jim_baxter@mentor.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Even if the skb is cloned, we still need a ZLP or USB will stall.
Signed-off-by: Nathan Sullivan <nathan.sullivan@ni.com>
Acked-by: Brad Mouring <brad.mouring@ni.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Daniel Mack