vdr April 2006

vdr@linuxtv.org

121 participants
140 discussions

sysinfo plugin 0.1.0 compile problem
by Simon Baxter 30 Apr '06

30 Apr '06

Hi. I'm getting the following error on the new plugin version plugin against vdr 1.3.46: Plugin sysinfo: make[1]: Entering directory `/home/vdruser/vdr-1.3.46/PLUGINS/src/sysinfo-0.1.0' make[1]: Leaving directory `/home/vdruser/vdr-1.3.46/PLUGINS/src/sysinfo-0.1.0' make[1]: Entering directory `/home/vdruser/vdr-1.3.46/PLUGINS/src/sysinfo-0.1.0' g++ -fPIC -g -O2 -Wall -Woverloaded-virtual -c -D_GNU_SOURCE -DPLUGIN_NAME_I18N='"sysinfo"' -I../../../include -I/lib/modules/2.6.16.9/build/include sysinfo.c sysinfoosd.h:43: error: extra qualification 'cSysInfoOsd::' on member 'PleaseWait' sysinfoosd.h:44: error: extra qualification âcSysInfoOsd::' on member 'ShowData' make[1]: *** [sysinfo.o] Error 1 make[1]: Leaving directory `/home/vdruser/vdr-1.3.46/PLUGINS/src/sysinfo-0.1.0' Any ideas?

2 1

pvrinput and calculation of time to record
by Daniel Dornacher 29 Apr '06

29 Apr '06

Hi, my vdr is doing very well. pvr 150 and vdr-xine output with xvmc acceleration. But i have one "big" problem. vdr calculates the time which is left wrong. I have 4500 kbits for the video stream. (and x for audio) but i doubt that it's the right value in hours. I think it is computing the videostream rate with something like near 6000 kbit. Have i overlooked something? Regards

4 5

Linking plugins against shared libs
by Laz 29 Apr '06

29 Apr '06

This isn't exactly a vdr question but hopefully someone can give me an answer here! I'm in the process of trying to add hardware MPEG decoding to the softdevice plugin. I have just built a shared library which does the decoding. I can link a test application against this and call functions from the library with no problems. However, when I link the softdevice plugin against my library, it links with no errors but I get an "undefined symbol" error when I try to run vdr and it trys to load the plugin! This is the first time I've had the need to build a shared library in this way so I'm a bit confused as to what's wrong here! My library is being linked with "-Wl,-soname,libcle266decode.so.0.1 -o libcle266decode.so.0.1", and I've added the relevant -I, -l, and -L to the plugin makefile so that it finds it. "ldd libvdr-softdevice.so.1.3.47" shows that it is finding my library and "strings libcle266decode.so.0.1" shows the symbol for the function which is being called. This must be something really basic that I'm missing here... :-s Cheers, Laz

2 2

VDR 1.4.0 - the final countdown
by Klaus Schmidinger 29 Apr '06

29 Apr '06

There have been three last minute fixes since version 1.3.49, which I'm attaching as a 'diff' to this message: - Fixed handling the "Power" key in case a recording is going on and no plugin is active (thanks to Udo Richter; bug reported by Dominique Simon). - Fixed a memory leak in handling external EPG data (thanks to Tobias Grimm). - Fixed a memory leak in closing the video file during replay (thanks to Tobias Grimm). Maybe some of you can find the time to apply this small patch and verify it. There have also been patches to i18n.c for updates on the Polish and Italian texts, but that's uncritical. If nothing really serious goes wrong, I plan to release VDR 1.4.0 tomorrow afternoon. Klaus

1 0

missing piece at end of cutted recording
by Joerg Riechardt 29 Apr '06

29 Apr '06

* I record a film * I set 2 marks (time a and time b) and cut * in the cutted film there are 2 marks as well (time c and time d) * I compare the length of the piece between the 2 marks in the original with the cutted film: d-c is shorter than b-a, the differenc is b-(last i-frame before b) * if the cutted film is very short, like 10 seconds, you see that after d there is a little more film, but not enough that you could shift mark d further to the end this is with the last vdr versions and vdr-xine. do others have the same effect? is this known? is there a fix? Jörg

2 2

[PATCH] cUnbufferedFile - another approac to readahead
by Artur Skawina 29 Apr '06

29 Apr '06

i suspected the readahead was being too aggressive, hence this -- a more autotuning readahead + general cleanups. I think the only bugfix in here is the one i just sent separately, but this code is from vdr-1.3.45 era, i might have forgotten something. (.46+ kept segfaulting here, found the cause only today -- a missing dependency in the streamdev plugin) Posting this just in case somebody has trouble w/ the fadvise code and wants to try something else. Commenting out "#define USE_FADVISE_RA" should disable the manual readahead completely, but still flush written data out, preventing an ongoing recording from eating up all free ram etc. artur --- vdr-1.3.49.org/cutter.c 2006-02-12 10:07:23.000000000 +0000 +++ vdr-1.3.49/cutter.c 2006-04-29 08:54:19.000000000 +0000 @@ -66,7 +66,8 @@ void cCuttingThread::Action(void) toFile = toFileName->Open(); if (!fromFile || !toFile) return; - fromFile->SetReadAhead(MEGABYTE(20)); + fromFile->CuttingSrc(); + toFile->CuttingDst(); int Index = Mark->position; Mark = fromMarks.Next(Mark); int FileSize = 0; @@ -91,7 +92,7 @@ void cCuttingThread::Action(void) if (fromIndex->Get(Index++, &FileNumber, &FileOffset, &PictureType, &Length)) { if (FileNumber != CurrentFileNumber) { fromFile = fromFileName->SetOffset(FileNumber, FileOffset); - fromFile->SetReadAhead(MEGABYTE(20)); + fromFile->CuttingSrc(); CurrentFileNumber = FileNumber; } if (fromFile) { @@ -124,6 +125,7 @@ void cCuttingThread::Action(void) error = "toFile 1"; break; } + toFile->CuttingDst(); FileSize = 0; } LastIFrame = 0; @@ -164,6 +166,7 @@ void cCuttingThread::Action(void) error = "toFile 2"; break; } + toFile->CuttingDst(); FileSize = 0; } } --- vdr-1.3.49.org/tools.c 2006-04-21 15:12:47.000000000 +0000 +++ vdr-1.3.49/tools.c 2006-04-29 08:54:19.000000000 +0000 @@ -1060,9 +1060,14 @@ bool cSafeFile::Close(void) // --- cUnbufferedFile ------------------------------------------------------- -#define USE_FADVISE +#define USE_FADVISE_RA +#define USE_FADVISE_DB -#define WRITE_BUFFER KILOBYTE(800) +//#define dfsyslog dsyslog // turn on fadvise related logging +#define dfsyslog(a...) do {} while (0) + +#define FADVGRAN KILOBYTE(4) // AKA fadvise-chunk-size; PAGE_SIZE or getpagesize(2) would also work. +#define READCHUNK MEGABYTE(8) cUnbufferedFile::cUnbufferedFile(void) { @@ -1079,13 +1084,16 @@ int cUnbufferedFile::Open(const char *Fi Close(); fd = open(FileName, Flags, Mode); curpos = 0; -#ifdef USE_FADVISE - begin = lastpos = ahead = 0; - cachedstart = 0; - cachedend = 0; - readahead = KILOBYTE(128); +#if defined(USE_FADVISE_RA) || defined(USE_FADVISE_DB) + lastpos = 0; + lastjump = 0; + cachedstart = cachedend = 0; + //readahead = READCHUNK; + readahead = KILOBYTE(768); + writebuffer = KILOBYTE(800); written = 0; totwritten = 0; + cutting = 0; if (fd >= 0) posix_fadvise(fd, 0, 0, POSIX_FADV_RANDOM); // we could use POSIX_FADV_SEQUENTIAL, but we do our own readahead, disabling the kernel one. #endif @@ -1094,7 +1102,7 @@ int cUnbufferedFile::Open(const char *Fi int cUnbufferedFile::Close(void) { -#ifdef USE_FADVISE +#if defined(USE_FADVISE_DB) if (fd >= 0) { if (totwritten) // if we wrote anything make sure the data has hit the disk before fdatasync(fd); // calling fadvise, as this is our last chance to un-cache it. @@ -1111,18 +1119,35 @@ int cUnbufferedFile::Close(void) // We try to handle the common cases such as PLAY->FF->PLAY, small // jumps, moving editing marks etc. -#define FADVGRAN KILOBYTE(4) // AKA fadvise-chunk-size; PAGE_SIZE or getpagesize(2) would also work. -#define READCHUNK MEGABYTE(8) +void cUnbufferedFile::CuttingSrc(void) +{ + readahead = MEGABYTE(16); + cutting = 1; +} -void cUnbufferedFile::SetReadAhead(size_t ra) +void cUnbufferedFile::CuttingDst(void) { - readahead = ra; + writebuffer = MEGABYTE(16); + cutting = 2; } int cUnbufferedFile::FadviseDrop(off_t Offset, off_t Len) { - // rounding up the window to make sure that not PAGE_SIZE-aligned data gets freed. - return posix_fadvise(fd, Offset - (FADVGRAN - 1), Len + (FADVGRAN - 1) * 2, POSIX_FADV_DONTNEED); + // Round up the window to make sure that not PAGE_SIZE-aligned data gets freed. + // Note that that also means calling this with Len==0 isn't special (unlike fadvise). + dfsyslog("DROP: fd:%3d %9zd .. %9zd SIZE: %6zd", fd, Offset, Offset+Len, Len); + off_t prewin = min(Offset ,(off_t)FADVGRAN - 1); // must not wrap below 0. + return posix_fadvise(fd, Offset - prewin, Len + prewin + (FADVGRAN - 1), POSIX_FADV_DONTNEED); +} + +// Trigger background readahead on the specified range and add +// it to the "cached" area so that we can drop the data later. +int cUnbufferedFile::FadviseRead(off_t Offset, off_t Len) +{ + dfsyslog("WANT: fd:%3d %9zd .. %9zd SIZE: %6zd", fd, Offset, Offset+Len, Len); + cachedstart = min(cachedstart, Offset); + cachedend = max(cachedend, Offset+Len); + return posix_fadvise(fd, Offset, Len, POSIX_FADV_WILLNEED); } off_t cUnbufferedFile::Seek(off_t Offset, int Whence) @@ -1136,54 +1161,78 @@ off_t cUnbufferedFile::Seek(off_t Offset ssize_t cUnbufferedFile::Read(void *Data, size_t Size) { if (fd >= 0) { -#ifdef USE_FADVISE +#ifdef USE_FADVISE_RA off_t jumped = curpos-lastpos; // nonzero means we're not at the last offset - if ((cachedstart < cachedend) && (curpos < cachedstart || curpos > cachedend)) { + + dfsyslog("READ: fd:%3d %9zd .. %9zd SIZE: %6zd jump: %9zd ra: %7zd", fd, curpos, curpos+Size, Size, jumped, readahead); + + if (curpos < cachedstart || curpos > cachedend) { // current position is outside the cached window -- invalidate it. - FadviseDrop(cachedstart, cachedend-cachedstart); + if (cachedstart != cachedend) + FadviseDrop(cachedstart, cachedend-cachedstart); cachedstart = curpos; cachedend = curpos; } - cachedstart = min(cachedstart, curpos); #endif ssize_t bytesRead = safe_read(fd, Data, Size); -#ifdef USE_FADVISE +#ifdef USE_FADVISE_RA if (bytesRead > 0) { + cachedstart = min(cachedstart, curpos); curpos += bytesRead; - cachedend = max(cachedend, curpos); - // Read ahead: - // no jump? (allow small forward jump still inside readahead window). - if (jumped >= 0 && jumped <= (off_t)readahead) { - // Trigger the readahead IO, but only if we've used at least - // 1/2 of the previously requested area. This avoids calling - // fadvise() after every read() call. - if (ahead - curpos < (off_t)(readahead / 2)) { - posix_fadvise(fd, curpos, readahead, POSIX_FADV_WILLNEED); - ahead = curpos + readahead; - cachedend = max(cachedend, ahead); - } - if (readahead < Size * 32) { // automagically tune readahead size. - readahead = Size * 32; + // no jump or small forward jump still inside readahead window. + if (jumped >= 0 && curpos <= cachedend) { + // Trigger the readahead IO, but only if we've used at least some of the previously + // requested area. This avoids calling fadvise() after every read() call. + size_t cachedsize = cachedend - curpos; + size_t ra = cachedsize + Size*1 + (size_t)jumped*1; + if (cutting) + ra += KILOBYTE(128); + ra = min(readahead, ra); + // Start I/O if we A) used some of the data or B) can read sufficiently large new chunk. + // (A) is important when starting w/ a small readahead. + if (cachedsize < (ra-ra/4) || cachedsize+KILOBYTE(128) <= ra) + FadviseRead(curpos, ra); + } + else if (jumped >= 0) { // either large forward jump, or FF (jumps by ~4xSize) + FadviseRead(curpos, ((size_t)jumped < Size*8)?(jumped+Size)*2:Size*2); + } + else /*if (jumped < 0)*/ { // backward jump: + // We don't want any readahead, otherwise e.g. fast-rewind gets in trouble. +#if 0 + // But we'll do some read behind for sequential short backward jumps. + size_t rbsize = -jumped * 16; + if (lastjump < 0 && (size_t)-jumped < Size * 8 && (size_t)((curpos-Size) - cachedstart ) < rbsize) { + // current position has moved back enough, grow tail window. + off_t start = max((off_t)0, (off_t)((curpos - Size) - rbsize)); + FadviseRead(start, rbsize); } +#endif + // We could reduce readahead window here. But this would lead to already + // prefetched data being thrown out by the code below; not exactly ideal + // if this jump was only caused by play mode transition etc. + //readahead = Size * 8; } - else - ahead = curpos; // jumped -> we really don't want any readahead, otherwise e.g. fast-rewind gets in trouble. + cachedend = max(cachedend, curpos); } if (cachedstart < cachedend) { - if (curpos - cachedstart > READCHUNK * 2) { + off_t maxtail = cutting ? KILOBYTE(32) : READCHUNK; + off_t maxhead = max(readahead, (size_t)READCHUNK); + if (jumped >= 0 && curpos - cachedstart >= maxtail * 2) { // current position has moved forward enough, shrink tail window. - FadviseDrop(cachedstart, curpos - READCHUNK - cachedstart); - cachedstart = curpos - READCHUNK; + FadviseDrop(cachedstart, (curpos - maxtail) - cachedstart); + cachedstart = curpos - maxtail; } - else if (cachedend > ahead && cachedend - curpos > READCHUNK * 2) { + else if (jumped < 0 && lastjump < 0 && cachedend - curpos >= maxhead * 2) { // current position has moved back enough, shrink head window. - FadviseDrop(curpos + READCHUNK, cachedend - curpos + READCHUNK); - cachedend = curpos + READCHUNK; + // (a large readahead value may prevent this) + FadviseDrop(curpos + maxhead, cachedend - (curpos + maxhead)); + cachedend = curpos + maxhead; } } lastpos = curpos; + lastjump = jumped; #endif return bytesRead; } @@ -1194,26 +1243,27 @@ ssize_t cUnbufferedFile::Write(const voi { if (fd >=0) { ssize_t bytesWritten = safe_write(fd, Data, Size); -#ifdef USE_FADVISE + //dsyslog("WRIT: fd:%3d %9zd .. %9zd SIZE: %6zd", fd, curpos, curpos+Size, Size); +#ifdef USE_FADVISE_DB if (bytesWritten > 0) { - begin = min(begin, curpos); + cachedstart = min(cachedstart, curpos); curpos += bytesWritten; written += bytesWritten; - lastpos = max(lastpos, curpos); - if (written > WRITE_BUFFER) { - if (lastpos > begin) { + cachedend = max(cachedend, curpos); + if (written > writebuffer) { + if (cachedend > cachedstart) { // Now do three things: - // 1) Start writeback of begin..lastpos range + // 1) Start writeback of cachedstart..cachedend range // 2) Drop the already written range (by the previous fadvise call) // 3) Handle nonpagealigned data. - // This is why we double the WRITE_BUFFER; the first time around the + // This is why we double the writebuffer; the first time around the // last (partial) page might be skipped, writeback will start only after // second call; the third call will still include this page and finally // drop it from cache. - off_t headdrop = min(begin, WRITE_BUFFER * 2L); - posix_fadvise(fd, begin - headdrop, lastpos - begin + headdrop, POSIX_FADV_DONTNEED); + off_t headdrop = min(cachedstart, (off_t)writebuffer * 2); + posix_fadvise(fd, cachedstart - headdrop, cachedend - cachedstart + headdrop, POSIX_FADV_DONTNEED); } - begin = lastpos = curpos; + cachedstart = cachedend = curpos; totwritten += written; written = 0; // The above fadvise() works when writing slowly (recording), but could @@ -1223,7 +1273,7 @@ ssize_t cUnbufferedFile::Write(const voi // So we do another round of flushing, just like above, but at larger // intervals -- this should catch any pages that couldn't be released // earlier. - if (totwritten > MEGABYTE(32)) { + if (totwritten > MEGABYTE(32) + writebuffer ) { // It seems in some setups, fadvise() does not trigger any I/O and // a fdatasync() call would be required do all the work (reiserfs with some // kind of write gathering enabled), but the syncs cause (io) load.. --- vdr-1.3.49.org/tools.h 2006-04-16 10:40:45.000000000 +0000 +++ vdr-1.3.49/tools.h 2006-04-29 08:54:19.000000000 +0000 @@ -245,19 +245,22 @@ private: off_t curpos; off_t cachedstart; off_t cachedend; - off_t begin; off_t lastpos; - off_t ahead; + off_t lastjump; size_t readahead; size_t written; size_t totwritten; + int cutting; + size_t writebuffer; int FadviseDrop(off_t Offset, off_t Len); + int FadviseRead(off_t Offset, off_t Len); public: cUnbufferedFile(void); ~cUnbufferedFile(); int Open(const char *FileName, int Flags, mode_t Mode = DEFFILEMODE); int Close(void); - void SetReadAhead(size_t ra); + void CuttingSrc(void); + void CuttingDst(void); off_t Seek(off_t Offset, int Whence); ssize_t Read(void *Data, size_t Size); ssize_t Write(const void *Data, size_t Size);

1 0

New short summary view for vdradmin-am-3.4.4
by Joachim Banzhaf (compuserve) 28 Apr '06

28 Apr '06

Hi list, I implemented an extension for vdradmin-am-3.4.4 and posted it on http://www.vdr-portal.de/board/thread.php?postid=458137#post458137 It just occured to me, that not everyone likes reading german portals (I prefer international ML's too :-)). So here it comes again: actually I have two patches: the first (vdradmin-am-3.4.4-summary-next.patch) just implemented a next-button on the summary view to get me warm a bit with perl. It displays all the shows that immediatelly follow the current shows. It was quite easy to do, so I was encouraged to start the second patch which actually includes the first one and adds a second summary view with just two lines per show (one for the title an one for the subtitle). The patches are absolutely usable but there are still some rough edges in it. Maybe someone here can help out with them: There is no translation yet, beause I don't know how to do it (the words - at, after, next, ... - are not the main problem;-)) The first line of the summary list should include heading infos like title, show, station, ... and not be just empty. Just a matter of time, i guess. I didn't manage to display the percentage for current shows in a graphical manner. I thought it would be quite obvious to use something like this: <table height="8px" width="70px" border="1" cellspacing="0" cellpadding="0"> <tr> <td bgcolor=#111111 width="19%"> </td> <td bgcolor=#eeeeee> </td> </tr> </table> and it works standalone (see test.html.gz or test.png) but not in vdradmin-am. I suspect css is in my way, but my knowledge on this is not enough to solve it. The code for the new summary is just a copy and paste of the old view with only a few changes. Someone with more perl knowledge than me may want to merge the code. Even better (in my opinion) would be a persistent toggle in the header of the old summary view that switches to the new view and back. Last, Andreas Mair - the current author/maintainer of the vdradmin-am fork - didn't like the big icons that I reused too much. Maybe someone can provide us with a set of new small icons like in the autotimer view? For those of you not that familiar with using patches: untar vdradmin-am-3.4.4.tar.bz2 change in the new vdradmin-am-3.4.4 directory. Patch with patch -p0 /path/to/the/patchfile start the modified ./vdradmind.pl Thats it so far, keep your ideas, patches, icons etc. coming... Joachim Banzhaf

1 1

[ANNOUNCE] DVB subtitles 0.3.11
by Pekka Virtanen 28 Apr '06

28 Apr '06

There is a maintenance release of DVB subtitles plugin available at: http://virtanen.org/vdr/subtitles/ 2006-04-28: Version 0.3.11 -Compability with VDR versions >= 1.3.48 (Thanks to Olli Janatuinen) -Patch for VDR 1.3.48 (Thanks to Rolf Ahrenberg) -- pekka

1 0

internet exchange of vdr files for people who are not able to receive tv-programs
by b2069x 28 Apr '06

28 Apr '06

Hi! I did some internet research about finding a kind of a exchange server e.g. P2P with BitTorrent or whatever where it is possible to download vdr files. Unfortunaltely I didn't find something. Of cause - people who are dealing with VDR have their own DVB card(s) to record the things they want to watch. But what about people that are limited to e.g. one card and want to record 2 programs .. or people they have no TV access at all for what ever reason. E.g. in my former flat I ran a server with 2 DVB-S cards but I had to move and now for me its not possible to run a satelite system. Hmmmm. Does somebody know an exchange or download service/server for that ....? German and English speaking recordings would be fine. Thanks for all hints! Ben

2 1

Help! No lock on primary device when timer starts
by Andreas Mair 27 Apr '06

27 Apr '06

Hi, I'm experiencing problems with my VDR box. I've used different VDR releases (1.3.37 to 1.3.48), different firmwares (f62623 to fd2623) and different DVB drivers (this monday's hg and original kernel 2.6.14.2). The box features 2 FF cards (1.3 and 2.2). So what happens? When a timer starts while watching LiveTV (ZDF or BR) the live picture gets disturbed and audio consists only of some peaks. Even if I'm watching ZDF and the timer is recording ZDF. The recording is fine. It's impossible to continue viewing that channel. The picture is only fine when switching to a limited number of channels: Pro7, Sat1, Kabel1 (and what else belongs to this group). Switching to Das Erste, ZDF, BR, WDR, NDR... or the RTL group shows trash. The log is full of lines like: frontend 0 timed out while tuning to channel 2, tp 111953 regained lock... It doesn't help if the timer stops. But restarting VDR then (no DVB drivers are reloaded!) LiveTV is fine again. I don't know what else I can do/try :( Any help/hints appreciated! Regards, Andreas -- http://andreas.vdr-developer.org --- VDRAdmin-AM

4 6

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

vdr April 2006