From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from cdptpa-omtalb.mail.rr.com ([75.180.132.120]:42761 "EHLO cdptpa-omtalb.mail.rr.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750782AbZJ2ErN (ORCPT ); Thu, 29 Oct 2009 00:47:13 -0400 Message-ID: <4AE91E54.2030409@acm.org> Date: Wed, 28 Oct 2009 21:47:16 -0700 From: Bob Cunningham MIME-Version: 1.0 To: Devin Heitmueller CC: linux-media@vger.kernel.org Subject: Re: HVR-950Q problem under MythTV References: <4AE8F99E.5010701@acm.org> <829197380910282040t6fce747aoca318911e76aa23f@mail.gmail.com> In-Reply-To: <829197380910282040t6fce747aoca318911e76aa23f@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-media-owner@vger.kernel.org List-ID: On 10/28/2009 08:40 PM, Devin Heitmueller wrote: > On Wed, Oct 28, 2009 at 10:10 PM, Bob Cunningham wrote: >> I just completed a fresh install of MythTV 0.22 RC1 on my fully-updated >> Fedora 11 system. My tuner is an HVR-950Q, connected to cable. The tuner >> works fine under tvtime (SD) and xine (HD). >> >> All MythTV functions work, except LiveTV. The problem is that mythfrontend >> times out waiting for the HVR-950Q to tune to the first station. This >> appears to be due to the very long HVR-950Q firmware load time, since no >> errors are reported by the backend. >> >> Unfortunately, mythfrontend has a hard-wired 7 second timeout for most >> requests sent to the backend. It seems this timeout works fine under normal >> circumstances for every other tuner MythTV works with. >> >> The following is repeated in dmesg after every attempt: >> >> xc5000: waiting for firmware upload (dvb-fe-xc5000-1.6.114.fw)... >> usb 1-2: firmware: requesting dvb-fe-xc5000-1.6.114.fw >> xc5000: firmware read 12401 bytes. >> xc5000: firmware uploading... >> xc5000: firmware upload complete... >> >> It looks like the HVR-950Q driver reloads the firmware at every possible >> opportunity, independent of the hardware state, each time either the SD or >> HD device is opened, such as when changing from an SD channel on /dev/video0 >> to an HD channel on /dev/dvb/adapter0. Is this necessary? >> >> Is it possible to tell the driver to ease up on the firmware reloads? I >> don't mind if the first attempt fails, but the second attempt should succeed >> (without a reload). >> >> Alternatively, are faster firmware loads possible? >> >> Should I open a bug on this? > > Hello Bob, > > In order to avoid the firmware reloading condition, you need to add a > modprobe option called "no_poweroff=1" for the xc5000 driver to your > modprobe.conf file and then reboot your computer. I agree that this > is a very annoying workaround, but have not had a chance to try to > find another solution (the i2c master in the au0828 hardware is poorly > designed and this same problem occurs in Windows but the problem is > not as noticeable because the Windows application doesn't as > aggressively power down the tuner). For F11, I appended the line "options xc5000 no_poweroff=1" to /etc/modprobe.d/local.conf Rather than power down (shudder), I did the following: 1. Unplug HVR-950Q 2. rmmod xc5000 3. modprobe xc5000 no_poweroff=1 4. Plug in HVR-950Q > Also, in order for the video to be rendered properly, you need to make > sure your capture resolution for LiveTV mode and the various capture > modes is set to 720x480 (the default in MythTV is 480x480). Without > this change, the picture will appear to be vertically stretched. This > is actually a bug in MythTV not properly handling analog capture > products that do not have an onboard hardware scaler (I did work in > 0.22 to get the analog support working but have not had an opportunity > to fix this bug yet). Done. > If you still have trouble, feel free to reply to this message. > > Cheers, > > Devin All is well with the world: The tuner is tuning, MythTV is mythic, and I am a vidiot. Thanks!