• Status Closed
  • Percent Complete
  • Task Type Bugs
  • Category Codecs
  • Assigned To No-one
  • Operating System Another
  • Severity Low
  • Priority Very Low
  • Reported Version Daily build (which?)
  • Due in Version Undecided
  • Due Date Undecided
  • Votes
  • Private
Attached to Project: Rockbox
Opened by Lutzero - 2010-06-19
Last edited by funman - 2010-06-22

FS#11419 - flac problem clipv2

Sansa Clip v2

While playing tracks (flac) it skips to next song after around 30 seconds.

Closed by  funman
2010-06-22 20:04
Reason for closing:  Fixed

Happens on APE files too. Codec failure and then skips to the next track, or sometimes just gives static, presumably because the codec state is screwed up.

This also happens on my Sansa Clip+ (r26992-100620), every time between 43s and 47s, with ~1060kbps FLAC files. Lower bitrate FLAC files take a little longer before skipping to the next track (an 800kbps FLAC after 55s, 743kbps after 1:05). The same files always skip to next track at the same time.

I know this problem is introduced recently as I play FLAC files often, so I tried ROLO to revert to older daily builds. r26978 had the same issue, but r26946-100619 didn’t have this issue, here the files played normally.

So for Clip+, this has been introduced in a revision after r26946.

Also, in r26979 there’s mentioned that this might solve playback problems for lossless files on Clipv2. This had no effect on my Clip+, as r26992 still skips to next track.

Problem seems fixed now, i only test flac, but it seems to work with r26997 (todays build). No more skipping after around half a minute playing.
Thanks, i was getting pretty sick of the standard sansa firmware already ;)

Summarizing the posts above:
1) introcuded between r26947 – r26955
2) reproducible on clipv2 and clip+
3) duration until codec failure seems to depend on buffered data (after ~6000KB for each given example)
⇒ looks like r26953 and r26954 are hot candidates.

wow thats quick :p

Problem seems fixed now, i only test flac, but it seems to work with r26997 (todays build). No more skipping after around half a minute playing.
Thanks, i was getting pretty sick of the standard sansa firmware already ;) , assuming you typed at the same time i did and didn’t read my post.

Yes, we typed at the same time. And yes, I read your comment. Nevertheless I wonder why this should be fixed, there has been no change since r26992 (which you stated shows the failure) and now (r26997) that should have had any effect.

Do FLAC files now play completely to the end on your Clipv2 with r26997?

Because on my Clip+ this revision still skips to the next track, no change here.
The FLAC files I tested in my first post still skip at the same time.

If I’m reading right, r26997 also seems to change something for e200v2, not for Clips.

Lennert, can you build rockbox or are you using downloaded builds? If you can build it → can you check whether r26953/r26954 introduced the issue?

Sorry, I’m using downloaded builds…

Sorry, my bad, the problem still exists with r26997, r27003 still has the problem too.
Im using downloaded builds for your information.

OK, I built Rockbox from SVN and used ROLO to test a few revisions:

v26952: No problems, FLACs play to the end.
v26953: First FLAC I played made my player crash (WPS showed track name, time 0:00 and ?kbit, total freeze), I had to perform a hard reset.

              Also with MP3's this revision crashes (one file had a data abort at 3000AEE8, another file had a prefetch abort at E5D43014) and reboots back to my default rockbox after I press home.
              This seems to give even more problems on my Clip+...

v26954: Has the issue as above, FLAC files skip to next track

Just let me know if I can test something else.

⇒ looks like r26953 and r26954 are hot candidates.

Indeed they are the faulty commits.

Not sure if I should revert them yet, because I don’t see what could be wrong with these commits, unless:

- *_dcache_range() are buggy (I’ve looked at them several times but more eyes are welcome)
- STORAGE_WANTS_ALIGN is buggy (TheSeven is looking at this code atm so I would like to wait if he finds bugs)
- read/write buffer is accessed before sd_transfer_sectors() returns, which would be pretty bad

I would revert the code changes to have a working trunk again. You could then put your patch into an own flyspray entry, link this entry as a related task and try to find a working solution.

r27055 = fixed. No more flac skipping. Don’t have mpc files but thats another thread ;)


Available keyboard shortcuts


Task Details

Task Editing