Am 05.01.2014 13:56, schrieb Mauro Carvalho Chehab:
> Em Sun, 05 Jan 2014 11:47:00 +0100
> Frank Schäfer <fschaefer.oss@xxxxxxxxxxxxxx> escreveu:
>
>> Am 04.01.2014 11:55, schrieb Mauro Carvalho Chehab:
>>> Now that all analog-specific code are at em28xx-video, convert
>>> it into an em28xx extension and load it as a separate module.
>>>
>>> Signed-off-by: Mauro Carvalho Chehab <m.chehab@xxxxxxxxxxx>
>>> ---
>>> drivers/media/usb/em28xx/Kconfig | 6 ++-
>>> drivers/media/usb/em28xx/Makefile | 5 ++-
>>> drivers/media/usb/em28xx/em28xx-camera.c | 1 +
>>> drivers/media/usb/em28xx/em28xx-cards.c | 45 ++++---------------
>>> drivers/media/usb/em28xx/em28xx-core.c | 12 ++++++
>>> drivers/media/usb/em28xx/em28xx-v4l.h | 20 +++++++++
>>> drivers/media/usb/em28xx/em28xx-vbi.c | 1 +
>>> drivers/media/usb/em28xx/em28xx-video.c | 74 +++++++++++++++++++++++---------
>>> drivers/media/usb/em28xx/em28xx.h | 23 ++--------
>>> 9 files changed, 107 insertions(+), 80 deletions(-)
>>> create mode 100644 drivers/media/usb/em28xx/em28xx-v4l.h
>>>
>>> diff --git a/drivers/media/usb/em28xx/Kconfig b/drivers/media/usb/em28xx/Kconfig
>>> index d6ba514d31eb..838fc9dbb747 100644
>>> --- a/drivers/media/usb/em28xx/Kconfig
>>> +++ b/drivers/media/usb/em28xx/Kconfig
>>> @@ -1,8 +1,12 @@
>>> config VIDEO_EM28XX
>>> - tristate "Empia EM28xx USB video capture support"
>>> + tristate "Empia EM28xx USB devices support"
>>> depends on VIDEO_DEV && I2C
>>> select VIDEO_TUNER
>>> select VIDEO_TVEEPROM
>>> +
>>> +config VIDEO_EM28XX_V4L2
>>> + tristate "Empia EM28xx analog TV, video capture and/or webcam support"
>>> + depends on VIDEO_EM28XX && I2C
>> VIDEO_EM28XX already depends on I2C.
>>
>>> select VIDEOBUF2_VMALLOC
>>> select VIDEO_SAA711X if MEDIA_SUBDRV_AUTOSELECT
>>> select VIDEO_TVP5150 if MEDIA_SUBDRV_AUTOSELECT
>>> diff --git a/drivers/media/usb/em28xx/Makefile b/drivers/media/usb/em28xx/Makefile
>>> index ad6d48557940..3f850d5063d0 100644
>>> --- a/drivers/media/usb/em28xx/Makefile
>>> +++ b/drivers/media/usb/em28xx/Makefile
>>> @@ -1,10 +1,11 @@
>>> -em28xx-y += em28xx-video.o em28xx-i2c.o em28xx-cards.o
>>> -em28xx-y += em28xx-core.o em28xx-vbi.o em28xx-camera.o
>>> +em28xx-y += em28xx-core.o em28xx-i2c.o em28xx-cards.o em28xx-camera.o
>>>
>>> +em28xx-v4l-objs := em28xx-video.o em28xx-vbi.o
>>> em28xx-alsa-objs := em28xx-audio.o
>>> em28xx-rc-objs := em28xx-input.o
>>>
>>> obj-$(CONFIG_VIDEO_EM28XX) += em28xx.o
>>> +obj-$(CONFIG_VIDEO_EM28XX_V4L2) += em28xx-v4l.o
>>> obj-$(CONFIG_VIDEO_EM28XX_ALSA) += em28xx-alsa.o
>>> obj-$(CONFIG_VIDEO_EM28XX_DVB) += em28xx-dvb.o
>>> obj-$(CONFIG_VIDEO_EM28XX_RC) += em28xx-rc.o
>>> diff --git a/drivers/media/usb/em28xx/em28xx-camera.c b/drivers/media/usb/em28xx/em28xx-camera.c
>>> index d666741797d4..c29f5c4e7b40 100644
>>> --- a/drivers/media/usb/em28xx/em28xx-camera.c
>>> +++ b/drivers/media/usb/em28xx/em28xx-camera.c
>>> @@ -454,3 +454,4 @@ int em28xx_init_camera(struct em28xx *dev)
>>>
>>> return ret;
>>> }
>>> +EXPORT_SYMBOL_GPL(em28xx_init_camera);
>> em28xx-camera should also be part of the em28xx-v4l module.
> I tried that. Moving em28xx-camera to em28xx-v4l would cause a recursive
> dependency, due to the camera probing logic, that should be called by
> em28xx-cards.c.
I see...
> Moving that probing part to em28xx-v4l is too complex, due to the code
> that detects the board.
>
> One solution would be to break em28xx-camera into two parts: the detection
> code, to be merged with the core, and the sensor part, to be merged with
> em28xx-v4l, but that sounds ugly.
>
> Other solution would be to use something like the dvb_attach() macro,
> to avoid the recursive dependency, but that would be a dirty solution.
Another solution would be to move the sensor probing to em28xx-v4l, too.
Then we need to rely on dev->board.is_webcam to load the v4l module.
This would of course also require to split the board hint stuff.
Even more ugly I think...
> As the code there is not big, the amount of overhead of having everything
> at em28xx-core is not high. So, I opted to keep the simplest path here,
> avoiding the risk of breaking something with a complex changeset.
I agree.
>
>>> diff --git a/drivers/media/usb/em28xx/em28xx-cards.c b/drivers/media/usb/em28xx/em28xx-cards.c
>>> index 175cd776e0a1..938daaabd8e0 100644
>>> --- a/drivers/media/usb/em28xx/em28xx-cards.c
>>> +++ b/drivers/media/usb/em28xx/em28xx-cards.c
>>> @@ -2159,6 +2159,8 @@ struct em28xx_board em28xx_boards[] = {
>>> .ir_codes = RC_MAP_PINNACLE_PCTV_HD,
>>> },
>>> };
>>> +EXPORT_SYMBOL_GPL(em28xx_boards);
>>> +
>>> const unsigned int em28xx_bcount = ARRAY_SIZE(em28xx_boards);
>>>
>>> /* table of devices that work with this driver */
>>> @@ -2827,10 +2829,6 @@ static void em28xx_card_setup(struct em28xx *dev)
>>> "tuner", dev->tuner_addr, NULL);
>>> }
>>> }
>>> -
>>> - em28xx_tuner_setup(dev);
>>> -
>>> - em28xx_init_camera(dev);
>>> }
>> Here you are fixing half of the em28xx_card_setup() oopses which you've
>> introduced with patch 3/22.
>> This needs to be done together with patch 5/22 before patch 3/22.
> Ok.
>
>>> @@ -2848,11 +2846,12 @@ static void request_module_async(struct work_struct *work)
>>> em28xx_init_extension(dev);
>>>
>>> #if defined(CONFIG_MODULES) && defined(MODULE)
>>> + if (dev->has_video)
>>> + request_module("em28xx-v4l");
>>> if (dev->has_audio_class)
>>> request_module("snd-usb-audio");
>>> else if (dev->has_alsa_audio)
>>> request_module("em28xx-alsa");
>>> -
>>> if (dev->board.has_dvb)
>>> request_module("em28xx-dvb");
>>> if (dev->board.buttons ||
>>> @@ -2881,18 +2880,12 @@ void em28xx_release_resources(struct em28xx *dev)
>>> {
>>> /*FIXME: I2C IR should be disconnected */
>>>
>>> - em28xx_release_analog_resources(dev);
>>> -
>>> if (dev->def_i2c_bus)
>>> em28xx_i2c_unregister(dev, 1);
>>> em28xx_i2c_unregister(dev, 0);
>>> if (dev->clk)
>>> v4l2_clk_unregister_fixed(dev->clk);
>>>
>>> - v4l2_ctrl_handler_free(&dev->ctrl_handler);
>>> -
>>> - v4l2_device_unregister(&dev->v4l2_dev);
>>> -
>>> usb_put_dev(dev->udev);
>>>
>>> /* Mark device as unused */
>>> @@ -3043,7 +3036,7 @@ static int em28xx_init_dev(struct em28xx *dev, struct usb_device *udev,
>>> if (retval < 0) {
>>> em28xx_errdev("%s: em28xx_i2c_register bus 0 - error [%d]!\n",
>>> __func__, retval);
>>> - goto unregister_dev;
>>> + return retval;
>>> }
>>>
>>> /* register i2c bus 1 */
>>> @@ -3057,30 +3050,14 @@ static int em28xx_init_dev(struct em28xx *dev, struct usb_device *udev,
>>> if (retval < 0) {
>>> em28xx_errdev("%s: em28xx_i2c_register bus 1 - error [%d]!\n",
>>> __func__, retval);
>>> - goto unregister_dev;
>>> + return retval;
>>> }
>>> }
>> Hmm... if registering of bus 1 fails, we need to unregister bus 0.
>> But that's an old bug which we can fix later...
> Good point. Yes, this should be on a separate patch.
>
>>> /* Do board specific init and eeprom reading */
>>> em28xx_card_setup(dev);
>>>
>>> - retval = em28xx_register_analog_devices(dev);
>>> - if (retval < 0) {
>>> - goto fail;
>>> - }
>>> -
>>> return 0;
>>> -
>>> -fail:
>>> - if (dev->def_i2c_bus)
>>> - em28xx_i2c_unregister(dev, 1);
>>> - em28xx_i2c_unregister(dev, 0);
>>> - v4l2_ctrl_handler_free(&dev->ctrl_handler);
>>> -
>>> -unregister_dev:
>>> - v4l2_device_unregister(&dev->v4l2_dev);
>>> -
>>> - return retval;
>>> }
>>>
>>> /* high bandwidth multiplier, as encoded in highspeed endpoint descriptors */
>>> @@ -3283,6 +3260,7 @@ static int em28xx_usb_probe(struct usb_interface *interface,
>>> dev->alt = -1;
>>> dev->is_audio_only = has_audio && !(has_video || has_dvb);
>>> dev->has_alsa_audio = has_audio;
>>> + dev->has_video = has_video;
>>> dev->audio_ifnum = ifnum;
>>>
>>> /* Checks if audio is provided by some interface */
>>> @@ -3322,10 +3300,9 @@ static int em28xx_usb_probe(struct usb_interface *interface,
>>>
>>> /* allocate device struct */
>>> mutex_init(&dev->lock);
>>> - mutex_lock(&dev->lock);
>>> retval = em28xx_init_dev(dev, udev, interface, nr);
>>> if (retval) {
>>> - goto unlock_and_free;
>>> + goto err_free;
>>> }
>>>
>>> if (usb_xfer_mode < 0) {
>>> @@ -3368,7 +3345,7 @@ static int em28xx_usb_probe(struct usb_interface *interface,
>>> if (retval) {
>>> printk(DRIVER_NAME
>>> ": Failed to pre-allocate USB transfer buffers for DVB.\n");
>>> - goto unlock_and_free;
>>> + goto err_free;
>>> }
>>> }
>>>
>>> @@ -3377,13 +3354,9 @@ static int em28xx_usb_probe(struct usb_interface *interface,
>>> /* Should be the last thing to do, to avoid newer udev's to
>>> open the device before fully initializing it
>>> */
>>> - mutex_unlock(&dev->lock);
>>>
>>> return 0;
>>>
>>> -unlock_and_free:
>>> - mutex_unlock(&dev->lock);
>>> -
>>> err_free:
>>> kfree(dev->alt_max_pkt_size_isoc);
>>> kfree(dev);
>>> diff --git a/drivers/media/usb/em28xx/em28xx-core.c b/drivers/media/usb/em28xx/em28xx-core.c
>>> index 3012912d2997..1113d4e107d8 100644
>>> --- a/drivers/media/usb/em28xx/em28xx-core.c
>>> +++ b/drivers/media/usb/em28xx/em28xx-core.c
>>> @@ -33,6 +33,18 @@
>>>
>>> #include "em28xx.h"
>>>
>>> +#define DRIVER_AUTHOR "Ludovico Cavedon <cavedon@xxxxxxxx>, " \
>>> + "Markus Rechberger <mrechberger@xxxxxxxxx>, " \
>>> + "Mauro Carvalho Chehab <mchehab@xxxxxxxxxxxxx>, " \
>>> + "Sascha Sommer <saschasommer@xxxxxxxxxx>"
>>> +
>>> +#define DRIVER_DESC "Empia em28xx based USB core driver"
>>> +
>>> +MODULE_AUTHOR(DRIVER_AUTHOR);
>>> +MODULE_DESCRIPTION(DRIVER_DESC);
>>> +MODULE_LICENSE("GPL");
>>> +MODULE_VERSION(EM28XX_VERSION);
>>> +
>>> /* #define ENABLE_DEBUG_ISOC_FRAMES */
>>>
>>> static unsigned int core_debug;
>>> diff --git a/drivers/media/usb/em28xx/em28xx-v4l.h b/drivers/media/usb/em28xx/em28xx-v4l.h
>>> new file mode 100644
>>> index 000000000000..bce438691e0e
>>> --- /dev/null
>>> +++ b/drivers/media/usb/em28xx/em28xx-v4l.h
>>> @@ -0,0 +1,20 @@
>>> +/*
>>> + em28xx-video.c - driver for Empia EM2800/EM2820/2840 USB
>>> + video capture devices
>> The information about supported chips is outdated everywhere in the driver,
>> but if you introduce a new header file it should really be up to date.
>> Just talk about EM27xx/EM28xx.
> We need to do a cleanup on all those headers: they don't follow Kernel
> CodingStyle and the old headers use my previous email.
>
> For the sake of keeping a coherency, I deliberately opted to just use the
> same way as the other headers.
>
> My plan is to write a patch series fixing this and other CodingStyle
> issues on em28xx after merging this changeset.
Ok, fine.
>
>>> +
>>> + Copyright (C) 2013-2014 Mauro Carvalho Chehab <m.chehab@xxxxxxxxxxx>
>>> +
>>> + This program is free software; you can redistribute it and/or modify
>>> + it under the terms of the GNU General Public License as published by
>>> + the Free Software Foundation version 2 of the License.
>>> +
>>> + This program is distributed in the hope that it will be useful,
>>> + but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>> + GNU General Public License for more details.
>>> + */
>>> +
>>> +
>>> +int em28xx_start_analog_streaming(struct vb2_queue *vq, unsigned int count);
>>> +int em28xx_stop_vbi_streaming(struct vb2_queue *vq);
>>> +extern struct vb2_ops em28xx_vbi_qops;
>>> diff --git a/drivers/media/usb/em28xx/em28xx-vbi.c b/drivers/media/usb/em28xx/em28xx-vbi.c
>>> index 39f39c527c13..db3d655600df 100644
>>> --- a/drivers/media/usb/em28xx/em28xx-vbi.c
>>> +++ b/drivers/media/usb/em28xx/em28xx-vbi.c
>>> @@ -27,6 +27,7 @@
>>> #include <linux/init.h>
>>>
>>> #include "em28xx.h"
>>> +#include "em28xx-v4l.h"
>>>
>>> static unsigned int vbibufs = 5;
>>> module_param(vbibufs, int, 0644);
>>> diff --git a/drivers/media/usb/em28xx/em28xx-video.c b/drivers/media/usb/em28xx/em28xx-video.c
>>> index 85284626dbd6..d615bff8ac09 100644
>>> --- a/drivers/media/usb/em28xx/em28xx-video.c
>>> +++ b/drivers/media/usb/em28xx/em28xx-video.c
>>> @@ -38,6 +38,7 @@
>>> #include <linux/slab.h>
>>>
>>> #include "em28xx.h"
>>> +#include "em28xx-v4l.h"
>>> #include <media/v4l2-common.h>
>>> #include <media/v4l2-ioctl.h>
>>> #include <media/v4l2-event.h>
>>> @@ -141,11 +142,13 @@ static struct em28xx_fmt format[] = {
>>> },
>>> };
>>>
>>> +static int em28xx_vbi_supported(struct em28xx *dev);
>>> +
>> Or move em28xx_vbi_supported() before em28xx_set_outfmt() and
>> em28xx_resolution_set() again.
>> If you had not changed the functions order in patch 1/22, this wouldn't
>> be necessary.
> True. I'll put an extra cleanup patch to remove this reference on the
> CodingStyle cleanup patchset I'm planning to do.
>
>>> /*
>>> * em28xx_wake_i2c()
>>> * configure i2c attached devices
>>> */
>>> -void em28xx_wake_i2c(struct em28xx *dev)
>>> +static void em28xx_wake_i2c(struct em28xx *dev)
>>> {
>>> v4l2_device_call_all(&dev->v4l2_dev, 0, core, reset, 0);
>>> v4l2_device_call_all(&dev->v4l2_dev, 0, video, s_routing,
>>> @@ -153,7 +156,7 @@ void em28xx_wake_i2c(struct em28xx *dev)
>>> v4l2_device_call_all(&dev->v4l2_dev, 0, video, s_stream, 0);
>>> }
>>>
>>> -int em28xx_colorlevels_set_default(struct em28xx *dev)
>>> +static int em28xx_colorlevels_set_default(struct em28xx *dev)
>>> {
>>> em28xx_write_reg(dev, EM28XX_R20_YGAIN, CONTRAST_DEFAULT);
>>> em28xx_write_reg(dev, EM28XX_R21_YOFFSET, BRIGHTNESS_DEFAULT);
>>> @@ -171,7 +174,7 @@ int em28xx_colorlevels_set_default(struct em28xx *dev)
>>> return em28xx_write_reg(dev, EM28XX_R1A_BOFFSET, 0x00);
>>> }
>>>
>>> -int em28xx_set_outfmt(struct em28xx *dev)
>>> +static int em28xx_set_outfmt(struct em28xx *dev)
>>> {
>>> int ret;
>>> u8 fmt, vinctrl;
>>> @@ -278,7 +281,7 @@ static int em28xx_scaler_set(struct em28xx *dev, u16 h, u16 v)
>>> }
>>>
>>> /* FIXME: this only function read values from dev */
>>> -int em28xx_resolution_set(struct em28xx *dev)
>>> +static int em28xx_resolution_set(struct em28xx *dev)
>>> {
>>> int width, height;
>>> width = norm_maxw(dev);
>>> @@ -310,7 +313,7 @@ int em28xx_resolution_set(struct em28xx *dev)
>>> return em28xx_scaler_set(dev, dev->hscale, dev->vscale);
>>> }
>>>
>>> -int em28xx_vbi_supported(struct em28xx *dev)
>>> +static int em28xx_vbi_supported(struct em28xx *dev)
>>> {
>>> /* Modprobe option to manually disable */
>>> if (disable_vbi == 1)
>>> @@ -330,7 +333,7 @@ int em28xx_vbi_supported(struct em28xx *dev)
>>> }
>>>
>>> /* Set USB alternate setting for analog video */
>>> -int em28xx_set_alternate(struct em28xx *dev)
>>> +static int em28xx_set_alternate(struct em28xx *dev)
>>> {
>>> int errCode;
>>> int i;
>>> @@ -1020,7 +1023,7 @@ static struct vb2_ops em28xx_video_qops = {
>>> .wait_finish = vb2_ops_wait_finish,
>>> };
>>>
>>> -int em28xx_vb2_setup(struct em28xx *dev)
>>> +static int em28xx_vb2_setup(struct em28xx *dev)
>>> {
>>> int rc;
>>> struct vb2_queue *q;
>>> @@ -1088,7 +1091,7 @@ static void video_mux(struct em28xx *dev, int index)
>>> em28xx_audio_analog_set(dev);
>>> }
>>>
>>> -void em28xx_ctrl_notify(struct v4l2_ctrl *ctrl, void *priv)
>>> +static void em28xx_ctrl_notify(struct v4l2_ctrl *ctrl, void *priv)
>>> {
>>> struct em28xx *dev = priv;
>>>
>>> @@ -1625,7 +1628,7 @@ static int vidioc_g_register(struct file *file, void *priv,
>>> reg->val = ret;
>>> } else {
>>> __le16 val = 0;
>>> - ret = em28xx_read_reg_req_len(dev, USB_REQ_GET_STATUS,
>>> + ret = dev->em28xx_read_reg_req_len(dev, USB_REQ_GET_STATUS,
>>> reg->reg, (char *)&val, 2);
>> Urgh... is it really necessary to start using these pointers again ?
>> Ok... keep it as is, I'll send a patch to clean this up later.
> This is one of the few places where we weren't using those pointers:
>
> $ git grep em28xx_read_reg_req_len drivers/media/usb/em28xx/
> drivers/media/usb/em28xx/em28xx-cards.c: dev->em28xx_read_reg_req_len = em28xx_read_reg_req_len;
> drivers/media/usb/em28xx/em28xx-core.c:int em28xx_read_reg_req_len(struct em28xx *dev, u8 req, u16 reg,
> drivers/media/usb/em28xx/em28xx-core.c: ret = em28xx_read_reg_req_len(dev, req, reg, &val, 1);
> drivers/media/usb/em28xx/em28xx-core.c: ret = dev->em28xx_read_reg_req_len(dev, 0, EM28XX_R40_AC97LSB,
> drivers/media/usb/em28xx/em28xx-i2c.c: ret = dev->em28xx_read_reg_req_len(dev, 0x00, 4-len, buf2, len);
> drivers/media/usb/em28xx/em28xx-i2c.c: ret = dev->em28xx_read_reg_req_len(dev, 2, addr, buf, len);
> drivers/media/usb/em28xx/em28xx-i2c.c: ret = dev->em28xx_read_reg_req_len(dev, 0x06, addr, buf, len);
> drivers/media/usb/em28xx/em28xx-input.c: rc = dev->em28xx_read_reg_req_len(dev, 0, EM28XX_R45_IR,
> drivers/media/usb/em28xx/em28xx-input.c: rc = dev->em28xx_read_reg_req_len(dev, 0, EM2874_R51_IR,
> drivers/media/usb/em28xx/em28xx-video.c: ret = dev->em28xx_read_reg_req_len(dev, USB_REQ_GET_STATUS,
> drivers/media/usb/em28xx/em28xx.h: int (*em28xx_read_reg_req_len) (struct em28xx *dev, u8 req, u16 reg,
> drivers/media/usb/em28xx/em28xx.h:int em28xx_read_reg_req_len(struct em28xx *dev, u8 req, u16 reg,
It seems like your "grep" works better than mine... ;)
> For the sake of coherency, we should either use one or the other way.
Agreed.
>
>>> if (ret < 0)
>>> return ret;
>>> @@ -1872,11 +1875,11 @@ static int em28xx_v4l2_open(struct file *filp)
>>> }
>>>
>>> /*
>>> - * em28xx_realease_resources()
>>> + * em28xx_v4l2_fini()
>>> * unregisters the v4l2,i2c and usb devices
>>> * called when the device gets disconected or at module unload
>>> */
>>> -void em28xx_release_analog_resources(struct em28xx *dev)
>>> +static int em28xx_v4l2_fini(struct em28xx *dev)
>>> {
>>>
>>> /*FIXME: I2C IR should be disconnected */
>>> @@ -1906,6 +1909,8 @@ void em28xx_release_analog_resources(struct em28xx *dev)
>>> video_device_release(dev->vdev);
>>> dev->vdev = NULL;
>>> }
>>> +
>>> + return 0;
>>> }
>>>
>>> /*
>>> @@ -1927,12 +1932,12 @@ static int em28xx_v4l2_close(struct file *filp)
>>> if (dev->users == 1) {
>>> /* the device is already disconnect,
>>> free the remaining resources */
>>> +
>>> if (dev->disconnected) {
>>> - em28xx_release_resources(dev);
>> Who releases the resources now ?
> This is tricky.
It is. And this part of the code really looks fishy.
> The hole idea is that em28xx-v4l release the resources allocated for V4L
> only, and not all resources, the same way as the other em28xx modules
> do.
>
> If you take a look at the original em28xx_release_resources(), what it does
> is to call this function, and then to de-allocate and unregister v4l2:
>
> v4l2_ctrl_handler_free(&dev->ctrl_handler);
> v4l2_device_unregister(&dev->v4l2_dev);
>
> after that, it deallocates the rest of allocated data.
>
> Now, em28xx_release_resources() is only called when em28xx is removed
> from memory or when the device is removed.
>
> The code there first calls em28xx_close_extension(dev). So, at device
> removal, em28xx-v4l will release the V4L2 specific resources, and
> em28xx_release_resources() will drop the common dev struct.
Up to now, em28xx_release_resources() was called after the last close.
If you remove this call, the resources are never released.
Or am I missing another place where this function is called ?
> I suspect that it might still be a bug there, but, if so, this bug is
> also present on em28xx-dvb, em28xx-alsa and em28xx-rc: what happens if
> the device is removed while the file descriptors is still opened?
At least it doesn't produce any oopses with the old code.
I'm not sure if I checked after the core/v4l split...
Did you observe an problems when you keep the em28xx_release_resources()
call in em28xx_v4l_fini() ?
I agree that the whole thing smells like bugs and close() probably isn't
the correct place for this.
But never calling this function is definitely wrong.
> Maybe the driver core already prevent such things, but I'm not sure.
Hmm... we currently close the extensions unconditionally in
em28xx_usb_disconnect()...
That seems to avoid trouble.
> If there's a bug out there, the proper fix seems to use kref for
> struct em28xx, increasing refcount for every em28xx extension and/or
> file open, decreasing it at either extension fini call, or at file close.
>
> This way, em28xx_release_resources() would be called only after refcount
> reaching zero, e. g. after being sure that nobody is using it.
>
> My plan is to take a look on it after having this changeset merged,
> as such change, if needed, will be complex and would require lots of
> testing. Also, it is independent on those changes.
Yes, let's fix things step by step and not all at the same time.
>
>>> + v4l2_ctrl_handler_free(&dev->ctrl_handler);
>>> + v4l2_device_unregister(&dev->v4l2_dev);
>>> kfree(dev->alt_max_pkt_size_isoc);
>>> - mutex_unlock(&dev->lock);
>>> - kfree(dev);
>>> - return 0;
>>> + goto exit;
>>> }
>>>
>>> /* Save some power by putting tuner to sleep */
>>> @@ -1951,6 +1956,7 @@ static int em28xx_v4l2_close(struct file *filp)
>>> }
>>> }
>>>
>>> +exit:
>>> dev->users--;
>> Nice bugfix.
>>
>>> mutex_unlock(&dev->lock);
>>> return 0;
>>> @@ -2047,8 +2053,6 @@ static struct video_device em28xx_radio_template = {
>>>
>>> /******************************** usb interface ******************************/
>>>
>>> -
>>> -
>>> static struct video_device *em28xx_vdev_init(struct em28xx *dev,
>>> const struct video_device *template,
>>> const char *type_name)
>>> @@ -2122,7 +2126,7 @@ static void em28xx_setup_xc3028(struct em28xx *dev, struct xc2028_ctrl *ctl)
>>> }
>>> }
>>>
>>> -void em28xx_tuner_setup(struct em28xx *dev)
>>> +static void em28xx_tuner_setup(struct em28xx *dev)
>>> {
>>> struct tuner_setup tun_setup;
>>> struct v4l2_frequency f;
>>> @@ -2181,14 +2185,14 @@ void em28xx_tuner_setup(struct em28xx *dev)
>>> v4l2_device_call_all(&dev->v4l2_dev, 0, tuner, s_frequency, &f);
>>> }
>>>
>>> -int em28xx_register_analog_devices(struct em28xx *dev)
>>> +static int em28xx_v4l2_init(struct em28xx *dev)
>>> {
>>> u8 val;
>>> int ret;
>>> unsigned int maxw;
>>> struct v4l2_ctrl_handler *hdl = &dev->ctrl_handler;
>>>
>>> - if (!dev->is_audio_only) {
>>> + if (!dev->has_video) {
>>> /* This device does not support the v4l2 extension */
>>> return 0;
>>> }
>>> @@ -2196,6 +2200,8 @@ int em28xx_register_analog_devices(struct em28xx *dev)
>>> printk(KERN_INFO "%s: v4l2 driver version %s\n",
>>> dev->name, EM28XX_VERSION);
>>>
>>> + mutex_lock(&dev->lock);
>>> +
>>> ret = v4l2_device_register(&dev->udev->dev, &dev->v4l2_dev);
>>> if (ret < 0) {
>>> em28xx_errdev("Call to v4l2_device_register() failed!\n");
>>> @@ -2212,6 +2218,10 @@ int em28xx_register_analog_devices(struct em28xx *dev)
>>> dev->vinctl = EM28XX_VINCTRL_INTERLACED |
>>> EM28XX_VINCTRL_CCIR656_ENABLE;
>>>
>>> + /* Initialize tuner and camera */
>>> + em28xx_tuner_setup(dev);
>>> + em28xx_init_camera(dev);
>>> +
>>> /* Configure audio */
>>> ret = em28xx_audio_setup(dev);
>>> if (ret < 0) {
>>> @@ -2422,6 +2432,28 @@ int em28xx_register_analog_devices(struct em28xx *dev)
>>>
>>> /* initialize videobuf2 stuff */
>>> em28xx_vb2_setup(dev);
>>> +
>>> err:
>>> - return 0;
>>> + mutex_unlock(&dev->lock);
>>> + return ret;
>>> +}
>>> +
>>> +static struct em28xx_ops v4l2_ops = {
>>> + .id = EM28XX_V4L2,
>>> + .name = "Em28xx v4l2 Extension",
>>> + .init = em28xx_v4l2_init,
>>> + .fini = em28xx_v4l2_fini,
>>> +};
>>> +
>>> +static int __init em28xx_video_register(void)
>>> +{
>>> + return em28xx_register_extension(&v4l2_ops);
>>> }
>>> +
>>> +static void __exit em28xx_video_unregister(void)
>>> +{
>>> + em28xx_unregister_extension(&v4l2_ops);
>>> +}
>>> +
>>> +module_init(em28xx_video_register);
>>> +module_exit(em28xx_video_unregister);
>>> diff --git a/drivers/media/usb/em28xx/em28xx.h b/drivers/media/usb/em28xx/em28xx.h
>>> index 7ae05ebc13c1..9d6f43e4681f 100644
>>> --- a/drivers/media/usb/em28xx/em28xx.h
>>> +++ b/drivers/media/usb/em28xx/em28xx.h
>>> @@ -26,7 +26,7 @@
>>> #ifndef _EM28XX_H
>>> #define _EM28XX_H
>>>
>>> -#define EM28XX_VERSION "0.2.0"
>>> +#define EM28XX_VERSION "0.2.1"
>>>
>>> #include <linux/workqueue.h>
>>> #include <linux/i2c.h>
>>> @@ -474,6 +474,7 @@ struct em28xx_eeprom {
>>> #define EM28XX_AUDIO 0x10
>>> #define EM28XX_DVB 0x20
>>> #define EM28XX_RC 0x30
>>> +#define EM28XX_V4L2 0x40
>>>
>>> /* em28xx resource types (used for res_get/res_lock etc */
>>> #define EM28XX_RESOURCE_VIDEO 0x01
>>> @@ -527,6 +528,7 @@ struct em28xx {
>>>
>>> unsigned int is_em25xx:1; /* em25xx/em276x/7x/8x family bridge */
>>> unsigned char disconnected:1; /* device has been diconnected */
>>> + unsigned int has_video:1;
>>> unsigned int has_audio_class:1;
>>> unsigned int has_alsa_audio:1;
>>> unsigned int is_audio_only:1;
>>> @@ -723,14 +725,9 @@ int em28xx_write_ac97(struct em28xx *dev, u8 reg, u16 val);
>>> int em28xx_audio_analog_set(struct em28xx *dev);
>>> int em28xx_audio_setup(struct em28xx *dev);
>>>
>>> -int em28xx_colorlevels_set_default(struct em28xx *dev);
>>> const struct em28xx_led *em28xx_find_led(struct em28xx *dev,
>>> enum em28xx_led_role role);
>>> int em28xx_capture_start(struct em28xx *dev, int start);
>>> -int em28xx_vbi_supported(struct em28xx *dev);
>>> -int em28xx_set_outfmt(struct em28xx *dev);
>>> -int em28xx_resolution_set(struct em28xx *dev);
>>> -int em28xx_set_alternate(struct em28xx *dev);
>>> int em28xx_alloc_urbs(struct em28xx *dev, enum em28xx_mode mode, int xfer_bulk,
>>> int num_bufs, int max_pkt_size, int packet_multiplier);
>>> int em28xx_init_usb_xfer(struct em28xx *dev, enum em28xx_mode mode,
>>> @@ -742,31 +739,17 @@ void em28xx_uninit_usb_xfer(struct em28xx *dev, enum em28xx_mode mode);
>>> void em28xx_stop_urbs(struct em28xx *dev);
>>> int em28xx_set_mode(struct em28xx *dev, enum em28xx_mode set_mode);
>>> int em28xx_gpio_set(struct em28xx *dev, struct em28xx_reg_seq *gpio);
>>> -void em28xx_wake_i2c(struct em28xx *dev);
>>> int em28xx_register_extension(struct em28xx_ops *dev);
>>> void em28xx_unregister_extension(struct em28xx_ops *dev);
>>> void em28xx_init_extension(struct em28xx *dev);
>>> void em28xx_close_extension(struct em28xx *dev);
>>>
>>> -/* Provided by em28xx-video.c */
>>> -void em28xx_tuner_setup(struct em28xx *dev);
>>> -int em28xx_vb2_setup(struct em28xx *dev);
>>> -int em28xx_register_analog_devices(struct em28xx *dev);
>>> -void em28xx_release_analog_resources(struct em28xx *dev);
>>> -void em28xx_ctrl_notify(struct v4l2_ctrl *ctrl, void *priv);
>>> -int em28xx_start_analog_streaming(struct vb2_queue *vq, unsigned int count);
>>> -int em28xx_stop_vbi_streaming(struct vb2_queue *vq);
>>> -extern const struct v4l2_ctrl_ops em28xx_ctrl_ops;
>>> -
>>> /* Provided by em28xx-cards.c */
>>> extern struct em28xx_board em28xx_boards[];
>>> extern struct usb_device_id em28xx_id_table[];
>>> int em28xx_tuner_callback(void *ptr, int component, int command, int arg);
>>> void em28xx_release_resources(struct em28xx *dev);
>>>
>>> -/* Provided by em28xx-vbi.c */
>>> -extern struct vb2_ops em28xx_vbi_qops;
>>> -
>>> /* Provided by em28xx-camera.c */
>>> int em28xx_detect_sensor(struct em28xx *dev);
>>> int em28xx_init_camera(struct em28xx *dev);
>> Nice clean-up ! :)
> Thanks!
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
