On 09/02/2012 01:58 AM, poma wrote:
On 09/01/2012 07:37 PM, poma wrote:On 09/01/2012 05:48 PM, Antti Palosaari wrote:On 09/01/2012 06:35 PM, poma wrote:On 09/01/2012 03:54 PM, Antti Palosaari wrote:It is not allowed to call usb_clear_halt() after urbs are submitted. That causes oops sometimes. Move whole streaming_ctrl() logic to power_ctrl() in order to avoid wrong usb_clear_halt() use. Also, configuring streaming endpoint in streaming_ctrl() sounds like a little bit wrong as it is aimed for control stream gate. Reported-by: Hin-Tak Leung <htl10@xxxxxxxxxxxxxxxxxxxxx> Signed-off-by: Antti Palosaari <crope@xxxxxx> --- drivers/media/usb/dvb-usb-v2/rtl28xxu.c | 55 +++++++++++++++------------------ 1 file changed, 25 insertions(+), 30 deletions(-) diff --git a/drivers/media/usb/dvb-usb-v2/rtl28xxu.c b/drivers/media/usb/dvb-usb-v2/rtl28xxu.c index e29fca2..7d11c5d 100644 --- a/drivers/media/usb/dvb-usb-v2/rtl28xxu.c +++ b/drivers/media/usb/dvb-usb-v2/rtl28xxu.c @@ -825,37 +825,10 @@ err: return ret; } -static int rtl28xxu_streaming_ctrl(struct dvb_frontend *fe , int onoff) -{ - int ret; - u8 buf[2]; - struct dvb_usb_device *d = fe_to_d(fe); - - dev_dbg(&d->udev->dev, "%s: onoff=%d\n", __func__, onoff); - - if (onoff) { - buf[0] = 0x00; - buf[1] = 0x00; - usb_clear_halt(d->udev, usb_rcvbulkpipe(d->udev, 0x81)); - } else { - buf[0] = 0x10; /* stall EPA */ - buf[1] = 0x02; /* reset EPA */ - } - - ret = rtl28xx_wr_regs(d, USB_EPA_CTL, buf, 2); - if (ret) - goto err; - - return ret; -err: - dev_dbg(&d->udev->dev, "%s: failed=%d\n", __func__, ret); - return ret; -} - static int rtl2831u_power_ctrl(struct dvb_usb_device *d, int onoff) { int ret; - u8 gpio, sys0; + u8 gpio, sys0, epa_ctl[2]; dev_dbg(&d->udev->dev, "%s: onoff=%d\n", __func__, onoff); @@ -878,11 +851,15 @@ static int rtl2831u_power_ctrl(struct dvb_usb_device *d, int onoff) gpio |= 0x04; /* GPIO2 = 1, LED on */ sys0 = sys0 & 0x0f; sys0 |= 0xe0; + epa_ctl[0] = 0x00; /* clear stall */ + epa_ctl[1] = 0x00; /* clear reset */ } else { gpio &= (~0x01); /* GPIO0 = 0 */ gpio |= 0x10; /* GPIO4 = 1 */ gpio &= (~0x04); /* GPIO2 = 1, LED off */ sys0 = sys0 & (~0xc0); + epa_ctl[0] = 0x10; /* set stall */ + epa_ctl[1] = 0x02; /* set reset */ } dev_dbg(&d->udev->dev, "%s: WR SYS0=%02x GPIO_OUT_VAL=%02x\n", __func__, @@ -898,6 +875,14 @@ static int rtl2831u_power_ctrl(struct dvb_usb_device *d, int onoff) if (ret) goto err; + /* streaming EP: stall & reset */ + ret = rtl28xx_wr_regs(d, USB_EPA_CTL, epa_ctl, 2); + if (ret) + goto err; + + if (onoff) + usb_clear_halt(d->udev, usb_rcvbulkpipe(d->udev, 0x81)); + return ret; err: dev_dbg(&d->udev->dev, "%s: failed=%d\n", __func__, ret); @@ -972,6 +957,14 @@ static int rtl2832u_power_ctrl(struct dvb_usb_device *d, int onoff) goto err; + /* streaming EP: clear stall & reset */ + ret = rtl28xx_wr_regs(d, USB_EPA_CTL, "\x00\x00", 2); + if (ret) + goto err; + + ret = usb_clear_halt(d->udev, usb_rcvbulkpipe(d->udev, 0x81)); + if (ret) + goto err; } else { /* demod_ctl_1 */ ret = rtl28xx_rd_reg(d, SYS_DEMOD_CTL1, &val); @@ -1006,6 +999,10 @@ static int rtl2832u_power_ctrl(struct dvb_usb_device *d, int onoff) if (ret) goto err; + /* streaming EP: set stall & reset */ + ret = rtl28xx_wr_regs(d, USB_EPA_CTL, "\x10\x02", 2); + if (ret) + goto err; } return ret; @@ -1182,7 +1179,6 @@ static const struct dvb_usb_device_properties rtl2831u_props = { .tuner_attach = rtl2831u_tuner_attach, .init = rtl28xxu_init, .get_rc_config = rtl2831u_get_rc_config, - .streaming_ctrl = rtl28xxu_streaming_ctrl, .num_adapters = 1, .adapter = { @@ -1204,7 +1200,6 @@ static const struct dvb_usb_device_properties rtl2832u_props = { .tuner_attach = rtl2832u_tuner_attach, .init = rtl28xxu_init, .get_rc_config = rtl2832u_get_rc_config, - .streaming_ctrl = rtl28xxu_streaming_ctrl, .num_adapters = 1, .adapter = {OK, after patching with this one from http://goo.gl/5wtpT there is no OOPS, but this happened[1][2]: 1. mythtv-setup version: fixes/0.25 [v0.25.2-3-gf0e2ad8-dirty]: … E DVBChan(1:/dev/dvb/adapter0/frontend0): Getting Frontend uncorrected block count failed. eno: Operation not supported (95) 2012-09-01 17:08:20.577044 W DVBSM(/dev/dvb/adapter0/frontend0): Cannot count Uncorrected Blocks eno: Operation not supported (95) … 2. tzap/femon: … status 1f | signal 2f2f | snr 00f2 | ber 0000001e | unc 00000033 | FE_HAS_LOCK status 1f | signal 2f2f | snr 00f3 | ber 0000000b | unc 00000033 | FE_HAS_LOCK status 1f | signal 2f2f | snr 00f1 | ber 00000000 | unc 00000033 | FE_HAS_LOCK … … Problem retrieving frontend information: Operation not supported status SCVYL | signal 18% | snr 0% | ber 19 | unc 1 | FE_HAS_LOCK Problem retrieving frontend information: Operation not supported status SCVYL | signal 18% | snr 0% | ber 7 | unc 1 | FE_HAS_LOCK Problem retrieving frontend information: Operation not supported status SCVYL | signal 18% | snr 0% | ber 54 | unc 1 | FE_HAS_LOCK …It is correct as driver does not report uncorrected blocks at all. Those applications should be fixed. When I removed stub callback implementation I looked quite many frontend drivers and there is surely more than 10 other demod drivers reporting errors too. Unfortunately returned error codes varies from driver by driver. Correct error code for non-supported IOCTL is ENOTTY and DVB-frontend is changed to return it too, but you don't seems to have a such patch.Yeah, as you mentioned before this case; http://www.spinics.net/lists/linux-media/msg49869.html Thanks for explain that again!And also, patch in question has nothing to do with that error code.I'll go from 1st patch just to confirm ;)You are right! Actually none of this patches: - RFC-Fix-DVB-ioctls-failing-if-frontend-open-closed-too-fast.patch - 1-5-rtl28xxu-stream-did-not-start-after-stop-on-USB3.0.patch - 2-5-rtl28xxu-fix-rtl2832u-module-reload-fails-bug.patch - 3-5-rtl2832-implement-.get_frontend.patch - 4-5-rtl2832-implement-.read_snr.patch - 5-5-rtl2832-implement-.read_ber.patch - rtl28xxu-correct-usb_clear_halt-usage.patch cause errors from femon during tzap-ing and mythtv-setup during scan-ing. What is causing them is removed mentioned stub callback, actually: static int rtl2832_read_ucblocks(struct dvb_frontend *fe, u32 *ucblocks) { *ucblocks = 0; return 0; } … .read_ucblocks = rtl2832_read_ucblocks, in 'dvb-frontends/rtl2832.c' Yes, it is still part of media_build, although you send relative patches, right :) If 'rtl2832_read_ucblocks' stay in 'rtl2832.c' no problemo with femon and mythtv-setup.
It will not stay. And error code will change to ENOTTY.
I checked and compared current hg femon source - http://linuxtv.org/hg/dvb-apps/file/96025655e6e8/util/femon with current fedora - zero diff. I can understand non-engagement from mythtv devs, but femon's… All this is for the reader's sake ;) And excuse me for hurriedness.
Feel free to fix femon to show some vise values, eg. N/A when ENOTTY or EAGAIN is got.
Those are just visual problems. I am almost 100% nothing will break what ever error codes are returned. That is because there is so many existing demodulator (frontend) drivers returning many different error codes. Now that error code will be standardized to ENOTTY slowly. And EAGAIN will be returned to IOCTL when device is in state it cannot perform requested operation (like sleeping).
http://git.linuxtv.org/anttip/media_tree.git/commit/0467305ac2771ba301d6a59fae359edf20fe49d1 regards Antti -- http://palosaari.fi/ -- 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
