Rockbox

This is the bug/patch tracker for Rockbox. Click here for more information.

Quick links: Bugs · Patches · Rockbox frontpage

Tasklist

FS#10114 - Sansa AMS : use IRAM for codecs on >2MB targets

Attached to Project: Rockbox
Opened by Rafaël Carré (funman) - Thursday, 09 April 2009, 15:44 GMT+2
Last edited by Thomas Martitz (kugel.) - Friday, 10 April 2009, 19:05 GMT+2
Task Type Patches
Category Codecs
Status Closed
Assigned To No-one
Player Type All players
Severity Low
Priority Normal
Reported Version Version 3.2
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Private No

Details

Stable mp3 decoding on Fuze

some codecs (mad_iram.h for example) use .icode section and not ICODE_ATTR

This caused a problem because we defined incorrectly IRAMSIZE to be >0 in plugin.lds for these targets
   0002-Sansa-AMS-make-models-with-more-than-2MB-of-... (2.7 KiB) 
 b/apps/plugins/plugin.lds            |    4 ++--
 b/firmware/export/config.h           |    1 +
 b/firmware/target/arm/as3525/app.lds |    8 +++-----
 3 files changed, 6 insertions(+), 7 deletions(-)

This task depends upon

Closed by  Thomas Martitz (kugel.)
Friday, 10 April 2009, 19:05 GMT+2
Reason for closing:  Accepted
Additional comments about closing:  Committed r20678. Thank you very much again :) Great job, really!
Comment by Thomas Martitz (kugel.) - Thursday, 09 April 2009, 15:57 GMT+2
Weird. I'm fairly sure I tried this very changes too (exactly these!), and I couldn't get it to work :S
Comment by Rafaël Carré (funman) - Thursday, 09 April 2009, 17:11 GMT+2
strange :/

are you sure you kept first range of IRAM (0x0 - something) for rockbox (because the vectors are there)?
Comment by Dustin Skoracki (sko) - Thursday, 09 April 2009, 18:34 GMT+2
Now mp3 works really fine on my e250v2 (even my 320k/bit files)!
Comment by Jack Halpin (FlynDice) - Thursday, 09 April 2009, 19:21 GMT+2
Let me demonstrate my lack of practical experience here by asking if it makes a difference if you've used 0x50000 instead of 0x4FFFF as the end of IRAM. The memory map at the end of the 3525 datasheet lists the "embedded T-ram" from 0x0000_0000 to 0x0004_FFFF which is a size of 0x50000. But if you're using the endpoint as a reference to count back from shouldn't it be 0x4FFFF?
Comment by Thomas Martitz (kugel.) - Thursday, 09 April 2009, 22:24 GMT+2
Well, guess how the patch is working for me,

Not at all. I get the same symptoms which I got with my work: Every codec just skips every track, except MP3 which throws a undefined instruction :(

Is it my hardware? That would explain why it didn't work for me when I implemented it :( I cannot really explain it.
Comment by Thomas Martitz (kugel.) - Friday, 10 April 2009, 18:16 GMT+2
Alright, seems to be a dependecy problem. I now built from a fresh checkout and it works.

If I knew that before we already had mp3 3 weeks ago :(

Loading...