[Dsp-tremor-devel] Debugging update

[Simon Pickering]

For those who might be listening...

I've added the code to SVN, slightly more up to date than the tarballs  
I posted originally.

The end of the output from the dsp task looks like so:

[   99.523437] _make_words: Entered
[   99.523437] _make_decode_table: s->dec_table >> 8=4441
[   99.523437] _make_decode_table: work >> 8=4376
[   99.523437] _make_decode_table: s->dec_leafw==1
[   99.523437] _make_decode_table: End return 0
[   99.523437] vorbis_book_unpack: Entered
[   99.523437] vorbis_book_unpack: check alignment, byte 1 = 17218
[   99.523437] vorbis_book_unpack: check alignment, byte 2 = 86
[   99.523437] vorbis_book_unpack: Codeword ordering Case 0
[   99.523437] vorbis_book_unpack: Mapping Case 0
[   99.523437] _make_decode_table: Entered
[   99.523437] _make_words: Entered
[   99.523437] _make_words: Overpopulated tree, return -1
[   99.523437] _make_decode_table: _make_words() 2 True
[   99.523437] vorbis_book_unpack: Error in  
_make_decode_table(s,lengthlist,quantvals,opb,mapt
[   99.523437] ype)
[   99.523437] _vorbis_unpack_books: Error in  
vorbis_book_unpack(opb,ci->book_param+i)
[   99.523437] _fetch_headers: ret=vorbis_dsp_headerin(vi,&op) is TRUE
[   99.523437] _ov_open1: About to leave, ret=-133
[   99.523437] ov_open_callbacks: Entered
[   99.523437] Input does not appear to be an Ogg bitstream.

So there's an error occurring in _make_words() (in codebook.c iirc). I  
need to sit down and look at the code to work out what this code does,  
and what's going wrong. Any help appreciated of course.

I also note that at the end of codebook.c the  
vorbis_book_decodevs_add(), vorbis_book_decodev_add(),  
vorbis_book_decodev_set() and vorbis_book_decodevv_add() functions  
have an extra input argument (ogg_int32_t *v) as compared to the  
x86/standard lowmem-no-byte code.

This is presumably as the decode_map() function used to be an ASM  
function (called decode_mapa()) which may not have wanted to allocate  
memory. I don't know, but it would be worth looking to see if I've  
missed anything when I did the decode_mapa()/decode_map() swap.

There are probably some other things I should have mentioned, but I  
can't remember them off the top of my head. I'll email again as I  
remember/make more changes.

Cheers,


Simon