- 21 Aug, 2019 22 commits
-
-
Maxime Ripard authored
The RC controllers have a bunch of generic properties that are needed in a device tree. Add a YAML schemas for those. Reviewed-by:
Rob Herring <robh@kernel.org> Signed-off-by:
Maxime Ripard <maxime.ripard@bootlin.com> Signed-off-by:
Sean Young <sean@mess.org> Signed-off-by:
Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
-
Tomas Bortoli authored
The function at issue does not always initialize each byte allocated for 'b' and can therefore leak uninitialized memory to a USB device in the call to usb_bulk_msg() Use kzalloc() instead of kmalloc() Signed-off-by:
Tomas Bortoli <tomasbortoli@gmail.com> Reported-by: syzbot+0522702e9d67142379f1@syzkaller.appspotmail.com Signed-off-by:
-
Nishka Dasgupta authored
Static structure en_templ, of type dvb_ca_en50221, is not used except to be copied into a local variable. Hence make it const to prevent unintended modification of the original fields. Issue found with Coccinelle. Signed-off-by:
Nishka Dasgupta <nishkadg.linux@gmail.com> Signed-off-by:
-
Nishka Dasgupta authored
The static structure ec168_props, of type dvb_usb_device_properties, is used only once, when it is assigned to field props of the constant structure ec168_driver_info. As this means that ec168_props will never be modified, make it const as well in order to prevent unintended modification of its fields. Issue found with Coccinelle. Signed-off-by:
-
Nishka Dasgupta authored
Static structure fdtv_ca, of type dvb_device, is only used when it is passed as the third argument to function dvb_register_device(). However, in the definition of dvb_register_device, the third parameter is declared as constant. Hence make fdtv_ca const as well. Issue found with Coccinelle. Signed-off-by:
-
Nishka Dasgupta authored
Static structure dst_config, of type dst_config, is not used except to be assigned as the value of field state of a variable having type dst_state *. In the definition of dst_state, field config is declared as const. Hence dst_config, when assigned to config, cannot be modified. Therefore, make dst_config const as well. Issue found with Coccinelle. Signed-off-by:
-
Jan Pieter van Woerkom authored
commit 5fa88151 ("[media] dvb-usb-cxusb: Geniatech T230 - resync TS FIFO after lock") does not solve the problem for all devices. This is the same issue reported on the T230C, which was moved to the dvbsky.c driver to work around this issue in commit 57422405 ("media: dvbsky: MyGica T230C support"). In addition, the cxusb driver causes an oops when unplugging the device while streaming; this problem does not happen any more. Signed-off-by:
Jan Pieter van Woerkom <jp@jpvw.nl> Tested-by:
James Hutchinson <jahutchinson99@googlemail.com> Signed-off-by:
-
Jan Pieter van Woerkom authored
Remove this device from the cxusb driver so it can be added to the dvbsky driver. Signed-off-by:
-
Wenwen Wang authored
In dvb_create_media_entity(), 'dvbdev->entity' is allocated through kzalloc(). Then, 'dvbdev->pads' is allocated through kcalloc(). However, if kcalloc() fails, the allocated 'dvbdev->entity' is not deallocated, leading to a memory leak bug. To fix this issue, free 'dvbdev->entity' before returning -ENOMEM. Signed-off-by:
Wenwen Wang <wenwen@cs.uga.edu> Signed-off-by:
-
Sean Young authored
This protocol responds to a real philips rc-mm remote; it does not respond to IR encoded with the encoder in ir-rcmm-decoder.c. Signed-off-by:
-
Wenwen Wang authored
In cx24117_load_firmware(), 'buf' is allocated through kmalloc() to hold the firmware. However, if i2c_transfer() fails, it is not deallocated, leading to a memory leak bug. Signed-off-by:
-
Wenwen Wang authored
In dib7000pc_detection(), 'tx' and 'rx' are allocated through kzalloc() respectively. However, if DiB7000PC is detected, they are not deallocated, leading to memory leaks. To fix this issue, create a label to free 'tx' and 'rx' before returning from the function. Signed-off-by:
-
A Sun authored
Fix and eliminate mceusb's IR length limit for IR signals transmitted to the MCE IR blaster ports. An IR signal TX exceeding 306 pulse/space samples presently causes -EINVAL return error. There's no such limitation nor error with the MCE device hardware. And valid IR signals exist with more than 400 pulse/space for the control of certain appliances (eg Panasonic ACXA75C00600 air conditioner). The scope of this patch is limited to the mceusb driver. There are still IR signal TX length and time constraints that related modules of rc core (eg LIRC) impose, further up the driver stack. Changes for mceusb_tx_ir(): Converts and sends LIRC IR pulse/space sequence to MCE device IR pulse/space format. Break long length LIRC sequence into multiple (unlimited number of) parts for sending to the MCE device. Reduce kernel stack IR buffer size: 128 (was 384) Increase MCE IR data packet size: 31 (was 5) Zero time LIRC pulse/space no longer copied to MCE IR data. Eliminate overwriting the source/input LIRC IR data in txbuf[]. Eliminate -EINVAL return; return number of IR samples sent (>0) or MCE write error code (<0). New mce_write() and mce_write_callback(): Implements synchronous blocking I/O, with timeout, for writing/sending data to the MCE device. An unlimited multipart IR signal sent to the MCE device faster than real time requires flow control absent with the original mce_request_packet() and mce_async_callback() asynchronous I/O implementation. Also absent is TX error feedback. mce_write() combines and replaces mce_request_packet() and mce_async_callback() with conversion to synchronous I/O. mce_write() returns bytes sent (>0) or MCE device write error (<0). Debug hex dump TX data before processing. Rename mce_async_out() -> mce_command_out(): The original name is misleading with underlying synchronous I/O implementation. Function renamed to mce_command_out(). Changes in mceusb_handle_command(): Add support for MCE device error case MCE_RSP_TX_TIMEOUT "IR TX timeout (TX buffer underrun)" Changes in mceusb_dev_printdata(): Changes support test and debug of multipart TX IR. Add buffer boundary information (offset and buffer size) to TX hex dump. Correct TX trace bug "Raw IR data, 0 pulse/space samples" Add trace for MCE_RSP_TX_TIMEOUT "IR TX timeout (TX buffer underrun)" Other changes: The driver's write to USB device architecture change (async to sync I/O) is significant so we bump DRIVER_VERSION to "1.95" (from "1.94"). Tests: $ cat -n irdata1 | head -3 1 carrier 36000 2 pulse 6350 3 space 6350 $ cat -n irdata1 | tail -3 76 pulse 6350 77 space 6350 78 pulse 6350 $ ir-ctl -s irdata1 [1549021.073612] mceusb 1-1.3:1.0: requesting 36000 HZ carrier [1549021.073635] mceusb 1-1.3:1.0: tx data[0]: 9f 06 01 45 (len=4 sz=4) [1549021.073649] mceusb 1-1.3:1.0: Request carrier of 35714 Hz (period 28us) [1549021.073848] mceusb 1-1.3:1.0: tx done status = 4 (wait = 100, expire = 100 (1000ms), urb->actual_length = 4, urb->status = 0) [1549021.074689] mceusb 1-1.3:1.0: rx data[0]: 9f 06 01 45 (len=4 sz=4) [1549021.074701] mceusb 1-1.3:1.0: Got carrier of 35714 Hz (period 28us) [1549021.102023] mceusb 1-1.3:1.0: tx data[0]: 9f 08 03 (len=3 sz=3) [1549021.102036] mceusb 1-1.3:1.0: Request transmit blaster mask of 0x03 [1549021.102219] mceusb 1-1.3:1.0: tx done status = 3 (wait = 100, expire = 100 (1000ms), urb->actual_length = 3, urb->status = 0) [1549021.131979] mceusb 1-1.3:1.0: tx data[0]: 9e ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f 9e ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f ff 7f 91 ff (len=81 sz=81) [1549021.131992] mceusb 1-1.3:1.0: Raw IR data, 30 pulse/space samples [1549021.133592] mceusb 1-1.3:1.0: tx done status = 81 (wait = 100, expire = 100 (1000ms), urb->actual_length = 81, urb->status = 0) Hex dumps limited to 64 bytes. 0xff is MCE maximum time pulse, 0x7f is MCE maximum time space. $ cat -n irdata2 | head -3 1 carrier 36000 2 pulse 50 3 space 50 $ cat -n irdata2 | tail -3 254 pulse 50 255 space 50 256 pulse 50 $ ir-ctl -s irdata2 [1549306.586998] mceusb 1-1.3:1.0: tx data[0]: 9f 08 03 (len=3 sz=3) [1549306.587015] mceusb 1-1.3:1.0: Request transmit blaster mask of 0x03 [1549306.587252] mceusb 1-1.3:1.0: tx done status = 3 (wait = 100, expire = 100 (1000ms), urb->actual_length = 3, urb->status = 0) [1549306.613275] mceusb 1-1.3:1.0: tx data[0]: 9e 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 9e 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 9e 81 (len=128 sz=128) [1549306.613291] mceusb 1-1.3:1.0: Raw IR data, 30 pulse/space samples [1549306.614837] mceusb 1-1.3:1.0: tx done status = 128 (wait = 100, expire = 100 (1000ms), urb->actual_length = 128, urb->status = 0) [1549306.614861] mceusb 1-1.3:1.0: tx data[0]: 9e 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 9e 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 01 81 9e 01 (len=128 sz=128) [1549306.614869] mceusb 1-1.3:1.0: Raw IR data, 30 pulse/space samples [1549306.620199] mceusb 1-1.3:1.0: tx done status = 128 (wait = 100, expire = 100 (1000ms), urb->actual_length = 128, urb->status = 0) [1549306.620212] mceusb 1-1.3:1.0: tx data[0]: 89 81 01 81 01 81 01 81 01 81 80 (len=11 sz=11) [1549306.620221] mceusb 1-1.3:1.0: Raw IR data, 9 pulse/space samples [1549306.633294] mceusb 1-1.3:1.0: tx done status = 11 (wait = 98, expire = 100 (1000ms), urb->actual_length = 11, urb->status = 0) Hex dumps limited to 64 bytes. 0x81 is MCE minimum time pulse, 0x01 is MCE minimum time space. TX IR part 3 sz=11 shows 20msec I/O blocking delay (100expire - 98wait = 2jiffies) Signed-off-by:A Sun <as1033x@comcast.net> Signed-off-by:
-
Christian Hewitt authored
This is a simple NEC remote control device shipped with the HardKernel ODROID range of SBC devices. Signed-off-by:
Christian Hewitt <christianshewitt@gmail.com> Signed-off-by:
-
Christian Hewitt authored
The WeTek Play 2 Android STB ships with an unusual remote where the main up/down/left/right/enter controls are surrounded with an outer ring of additional keys which are listed in clockwise order. Signed-off-by:
-
Christian Hewitt authored
The WeTek Hub Android STB ships with a simple NEC remote. Signed-off-by:
-
Christian Hewitt authored
The Tanix TX5 max Android STB ships with a simple NEC remote. Signed-off-by:
-
Christian Hewitt authored
The Tanix TX3 mini Android STB ships with a simple NEC remote. Signed-off-by:
-
Christian Hewitt authored
Khadas VIM and Edge SBC devices use the same NEC remote device. The remote includes a mouse button for Android use. This has been mapped to KEY_MUTE. Signed-off-by:
-
Christian Hewitt authored
The X96-Max Android STB ships with a simple NEC remote. It includes a TV section with preset buttons for controlling a TV. These are not configurable, but are noted to aid visual recognition of the device. Signed-off-by:
-
Marc Gonzalez authored
Use cmd_init() to fill a struct si2168_cmd command. Signed-off-by:
Marc Gonzalez <marc.w.gonzalez@free.fr> Signed-off-by:
-
Sean Young authored
KASAN: global-out-of-bounds Read in dvb_pll_attach Syzbot reported global-out-of-bounds Read in dvb_pll_attach, while accessing id[dvb_pll_devcount], because dvb_pll_devcount was 65, that is more than size of 'id' which is DVB_PLL_MAX(64). Rather than increasing dvb_pll_devcount every time, use ida so that numbers are allocated correctly. This does mean that no more than 64 devices can be attached at the same time, but this is more than sufficient. usb 1-1: dvb_usb_v2: will pass the complete MPEG2 transport stream to the software demuxer dvbdev: DVB: registering new adapter (774 Friio White ISDB-T USB2.0) usb 1-1: media controller created dvbdev: dvb_create_media_entity: media entity 'dvb-demux' registered. tc90522 0-0018: Toshiba TC90522 attached. usb 1-1: DVB: registering adapter 0 frontend 0 (Toshiba TC90522 ISDB-T module)... dvbdev: dvb_create_media_entity: media entity 'Toshiba TC90522 ISDB-T module' registered. ================================================================== BUG: KASAN: global-out-of-bounds in dvb_pll_attach+0x6c5/0x830 drivers/media/dvb-frontends/dvb-pll.c:798 Read of size 4 at addr ffffffff89c9e5e0 by task kworker/0:1/12 CPU: 0 PID: 12 Comm: kworker/0:1 Not tainted 5.2.0-rc6+ #13 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Workqueue: usb_hub_wq hub_event Call Trace: __dump_stack lib/dump_stack.c:77 [inline] dump_stack+0xca/0x13e lib/dump_stack.c:113 print_address_description+0x67/0x231 mm/kasan/report.c:188 __kasan_report.cold+0x1a/0x32 mm/kasan/report.c:317 kasan_report+0xe/0x20 mm/kasan/common.c:614 dvb_pll_attach+0x6c5/0x830 drivers/media/dvb-frontends/dvb-pll.c:798 dvb_pll_probe+0xfe/0x174 drivers/media/dvb-frontends/dvb-pll.c:877 i2c_device_probe+0x790/0xaa0 drivers/i2c/i2c-core-base.c:389 really_probe+0x281/0x660 drivers/base/dd.c:509 driver_probe_device+0x104/0x210 drivers/base/dd.c:670 __device_attach_driver+0x1c2/0x220 drivers/base/dd.c:777 bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454 __device_attach+0x217/0x360 drivers/base/dd.c:843 bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514 device_add+0xae6/0x16f0 drivers/base/core.c:2111 i2c_new_client_device+0x5b3/0xc40 drivers/i2c/i2c-core-base.c:778 i2c_new_device+0x19/0x50 drivers/i2c/i2c-core-base.c:821 dvb_module_probe+0xf9/0x220 drivers/media/dvb-core/dvbdev.c:985 friio_tuner_attach+0x125/0x1d0 drivers/media/usb/dvb-usb-v2/gl861.c:536 dvb_usbv2_adapter_frontend_init drivers/media/usb/dvb-usb-v2/dvb_usb_core.c:675 [inline] dvb_usbv2_adapter_init drivers/media/usb/dvb-usb-v2/dvb_usb_core.c:804 [inline] dvb_usbv2_init drivers/media/usb/dvb-usb-v2/dvb_usb_core.c:865 [inline] dvb_usbv2_probe.cold+0x24dc/0x255d drivers/media/usb/dvb-usb-v2/dvb_usb_core.c:980 usb_probe_interface+0x305/0x7a0 drivers/usb/core/driver.c:361 really_probe+0x281/0x660 drivers/base/dd.c:509 driver_probe_device+0x104/0x210 drivers/base/dd.c:670 __device_attach_driver+0x1c2/0x220 drivers/base/dd.c:777 bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454 __device_attach+0x217/0x360 drivers/base/dd.c:843 bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514 device_add+0xae6/0x16f0 drivers/base/core.c:2111 usb_set_configuration+0xdf6/0x1670 drivers/usb/core/message.c:2023 generic_probe+0x9d/0xd5 drivers/usb/core/generic.c:210 usb_probe_device+0x99/0x100 drivers/usb/core/driver.c:266 really_probe+0x281/0x660 drivers/base/dd.c:509 driver_probe_device+0x104/0x210 drivers/base/dd.c:670 __device_attach_driver+0x1c2/0x220 drivers/base/dd.c:777 bus_for_each_drv+0x15c/0x1e0 drivers/base/bus.c:454 __device_attach+0x217/0x360 drivers/base/dd.c:843 bus_probe_device+0x1e4/0x290 drivers/base/bus.c:514 device_add+0xae6/0x16f0 drivers/base/core.c:2111 usb_new_device.cold+0x8c1/0x1016 drivers/usb/core/hub.c:2534 hub_port_connect drivers/usb/core/hub.c:5089 [inline] hub_port_connect_change drivers/usb/core/hub.c:5204 [inline] port_event drivers/usb/core/hub.c:5350 [inline] hub_event+0x1ada/0x3590 drivers/usb/core/hub.c:5432 process_one_work+0x905/0x1570 kernel/workqueue.c:2269 process_scheduled_works kernel/workqueue.c:2331 [inline] worker_thread+0x7ab/0xe20 kernel/workqueue.c:2417 kthread+0x30b/0x410 kernel/kthread.c:255 ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352 The buggy address belongs to the variable: id+0x100/0x120 Memory state around the buggy address: ffffffff89c9e480: fa fa fa fa 00 00 fa fa fa fa fa fa 00 00 00 00 ffffffff89c9e500: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > ffffffff89c9e580: 00 00 00 00 00 00 00 00 00 00 00 00 fa fa fa fa ^ ffffffff89c9e600: 04 fa fa fa fa fa fa fa 04 fa fa fa fa fa fa fa ffffffff89c9e680: 04 fa fa fa fa fa fa fa 04 fa fa fa fa fa fa fa ================================================================== Reported-by: syzbot+8a8f48672560c8ca59dd@syzkaller.appspotmail.com Signed-off-by:
-
- 19 Aug, 2019 18 commits
-
-
Hans Verkuil authored
The existing documentation was incorrect and did not correspond to how actual codec drivers implemented this. Update the documentation to explicitly specify what is actually expected. Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by:
-
Tomasz Figa authored
Due to complexity of the video decoding process, the V4L2 drivers of stateful decoder hardware require specific sequences of V4L2 API calls to be followed. These include capability enumeration, initialization, decoding, seek, pause, dynamic resolution change, drain and end of stream. Specifics of the above have been discussed during Media Workshops at LinuxCon Europe 2012 in Barcelona and then later Embedded Linux Conference Europe 2014 in Düsseldorf. The de facto Codec API that originated at those events was later implemented by the drivers we already have merged in mainline, such as s5p-mfc or coda. The only thing missing was the real specification included as a part of Linux Media documentation. Fix it now and document the decoder part of the Codec API. Signed-off-by:
Tomasz Figa <tfiga@chromium.org> Signed-off-by:
-
Maxime Jourdan authored
Tag all the coded formats where the vicodec stateful decoder supports dynamic resolution switching and bytestream parsing. Signed-off-by:
Maxime Jourdan <mjourdan@baylibre.com> Reviewed-by:
Paul Kocialkowski <paul.kocialkowski@bootlin.com> Signed-off-by:
-
Maxime Jourdan authored
Tag all the coded formats where the mtk-vcodec decoder supports dynamic resolution switching. Signed-off-by:
Alexandre Courbot <acourbot@chromium.org> Signed-off-by:
-
Maxime Jourdan authored
Tag all the coded formats where the venus vdec supports dynamic resolution switching. Signed-off-by:
Stanimir Varbanov <stanimir.varbanov@linaro.org> Reviewed-by:
-
Maxime Jourdan authored
Add an enum_fmt format flag to specifically tag coded formats where dynamic resolution switching is supported by the device. This is useful for some codec drivers that can support dynamic resolution switching for one or more of their listed coded formats. It allows userspace to know whether it should extract the video parameters itself, or if it can rely on the device to send V4L2_EVENT_SOURCE_CHANGE when such changes are detected. Signed-off-by:
-
Hans Verkuil authored
Add an enum_fmt format flag to specifically tag coded formats where full bytestream parsing is supported by the device. Some stateful decoders are capable of fully parsing a bytestream, but others require that userspace pre-parses the bytestream into frames or fields (see the corresponding pixelformat descriptions for details). If this flag is set, then this pre-parsing step is not required (but still possible, of course). Signed-off-by:
-
Hertz Wong authored
Now that the generic bits have been added, we can activate H264 decoding on rk3288. Signed-off-by:
Hertz Wong <hertz.wong@rock-chips.com> Signed-off-by:
Boris Brezillon <boris.brezillon@collabora.com> Signed-off-by:
-
Hertz Wong authored
Add the G1 specific bits to support H264 decoding. Signed-off-by:
Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by:
-
Hertz Wong authored
Add helpers and patch hantro_{drv,v4l2}.c to prepare addition of H264 decoding support. Signed-off-by: -
Boris Brezillon authored
Some decoders use intra slice/frame references. The capture buffer pointed by these references might be new and thus have invalid timestamp which prevents the decoder logic from retrieving the vb2_buffer object based on the output buf timestamp. Copy all metadata (including the timestamp) before starting the decode operation. Suggested-by:
Jonas Karlman <jonas@kwiboo.se> Signed-off-by:
-
Ezequiel Garcia authored
The cedrus VPU is slice-based and expects V4L2_PIX_FMT_H264_SLICE buffers to contain H264 slices with no start code. Expose this to userspace with the newly added menu control. These two controls are specified as mandatory for applications, but we mark them as non-required on the driver side for backwards compatibility. Signed-off-by:
Ezequiel Garcia <ezequiel@collabora.com> Signed-off-by:
-
Ezequiel Garcia authored
In order to introduce other controls, the control initialization needs to support an initial struct v4l2_ctrl_control. While here, let's cleanup the control initialization, removing unneeded fields. Signed-off-by:
-
Boris Brezillon authored
Those lists can be extracted from the dpb, let's simplify userspace life and build that list kernel-side (generic helpers will be provided for drivers that need this list). Signed-off-by:
Nicolas Dufresne <nicolas.dufresne@collabora.com> Reviewed-by:
-
Ezequiel Garcia authored
Stateless decoders have different expectations about the start code that is prepended on H264 slices. Add a menu control to express the supported start code types (including no start code). Drivers are allowed to support only one start code type, but they can support both too. Note that this is independent of the H264 decoding mode, which specifies the granularity of the decoding operations. Either in frame-based or slice-based mode, this new control will allow to define the start code expected on H264 slices. Signed-off-by:
-
Boris Brezillon authored
Some stateless decoders don't support per-slice decoding granularity (or at least not in a way that would make them efficient or easy to use). Expose a menu to control the supported decoding modes. Drivers are allowed to support only one decoding but they can support both too. To fully specify the decoding operation, we need to introduce a start_byte_offset, to indicate where slices start. Signed-off-by:
-
Ezequiel Garcia authored
The V4L2_PIX_FMT_H264_SLICE_RAW name was originally suggested because the pixel format would represent H264 slices without any start code. However, as we will now introduce a start code menu control, give the pixel format a more meaningful name, while it's still early enough to do so. Signed-off-by:
-
Rasmus Villemoes authored
Our list_sort() utility has always supported a context argument that is passed through to the comparison routine. Now there's a use case for the similar thing for sort(). This implements sort_r by simply extending the existing sort function in the obvious way. To avoid code duplication, we want to implement sort() in terms of sort_r(). The naive way to do that is static int cmp_wrapper(const void *a, const void *b, const void *ctx) { int (*real_cmp)(const void*, const void*) = ctx; return real_cmp(a, b); } sort(..., cmp) { sort_r(..., cmp_wrapper, cmp) } but this would do two indirect calls for each comparison. Instead, do as is done for the default swap functions - that only adds a cost of a single easily predicted branch to each comparison call. Aside from introducing support for the context argument, this also serves as preparation for patches that will eliminate the indirect comparison calls in common cases. Requested-by:Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Tested-by:
-
