This is the bug/patch tracker for Rockbox. Click here for more information.
Quick links: Bugs · Patches · Rockbox frontpage
FS#2894 - cache refill on FF or REW, even if not necessary
Attached to Project:
Rockbox
Opened by Thomas Kobler (kobler) - Sunday, 08 January 2006, 17:18 GMT+2
Last edited by Daniel Stenberg (bagder) - Friday, 28 April 2006, 11:17 GMT+2
Opened by Thomas Kobler (kobler) - Sunday, 08 January 2006, 17:18 GMT+2
Last edited by Daniel Stenberg (bagder) - Friday, 28 April 2006, 11:17 GMT+2
|
DetailsRecorder V1 with 2 or 8MB, V2.5 and later(incl. 20060108)
When more than one file is in the cache (e.g. filesize of playing file is less than about 1,5/7,5MB) and skipping back or forth a few seconds, after releasing the button, data gets loaded from disk (red LED is blinking for several seconds) even when the needed data is still - or is already - in the cache. This doesn't happen with V2.4. Becaues the chance of more than one file beeing in the cache is much less on a 2MB RAM Recorder, it hasn't been noticed already, I guess. But with the new devices with bigger RAM sizes it will hurt more. I had a look at mpeg.c, but all I could find out, is, that the disk already spins up, when case MPEG_NEED_DATA: is *NOT YET REACHED*. And a little later (!filling && diffpos >= 0 && diffpos < audiobuflen) is false and also reloads cache. But than I got lost in ffwd_rew() and audio_ff_rewind().. |
This task depends upon
Closed by Daniel Stenberg (bagder)
Wednesday, 28 June 2006, 15:47 GMT+2
Reason for closing: Wont Fix
Additional comments about closing: This report is somewhat speculative and doesn\'t actually point out any error, just discussiong theories about what Rockbox perhaps could do better in a situation the reporter thinks he experienced.
Wednesday, 28 June 2006, 15:47 GMT+2
Reason for closing: Wont Fix
Additional comments about closing: This report is somewhat speculative and doesn\'t actually point out any error, just discussiong theories about what Rockbox perhaps could do better in a situation the reporter thinks he experienced.
Comment by Daniel Stenberg (bagder) -
Thursday, 30 March 2006, 10:24 GMT+2
I don't think this issue is of priority "urgent" and I don't think we MUST fix this for 3.0
Comment by Brandon Low (lostlogic) -
Thursday, 06 April 2006, 05:20 GMT+2
Are you sure the data is already in RAM? If you seek back a few seconds, it is possible (if somewhat unlikely) that the data has already been dropped from the buffer. In the same vein, if you seek forward a few seconds, that could trigger a rebuffer event which will drop old data from the buffer and buffer forward in the playlist / file.