lame - segmentation fault

[Gabor FUNK]

Hi,

I experience segmentation faults with several mp3 files.
They're archived from internet radio stations per hour,
and 99.9% of those files is fine by lame - however,
several files causes segmentation fault, when I try to
"lame" (lame --mp3input infile.mp3 outfile.mp3) them.

- Tried 3.97 and 3.98 beta 6, same result
- Tested on more computers, same result
- Windows media player plays "bad" file without problem

Here are some samples, strace logs, etc.
http://www.huweb.hu/v/lame/

Any ideas?

Thanks, Gabor

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Register now and save $200. Hurry, offer ends at 11:59 p.m.,
Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Lame-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/lame-dev

[Robert Hegemann]

Hello Gabor,

thanks for sharing the problem files with us.

Am 02.04.2008, 00:40 Uhr, schrieb Gabor FUNK <[hidden email]>:

> Hi,
>
> I experience segmentation faults with several mp3 files.
> They're archived from internet radio stations per hour,
> and 99.9% of those files is fine by lame - however,
> several files causes segmentation fault, when I try to
> "lame" (lame --mp3input infile.mp3 outfile.mp3) them.
>
> - Tried 3.97 and 3.98 beta 6, same result
> - Tested on more computers, same result
> - Windows media player plays "bad" file without problem
>
> Here are some samples, strace logs, etc.
> http://www.huweb.hu/v/lame/
>
> Any ideas?

There is something wrong with the mp3 files. I tried the smaller
problem mp3, the mpg123 decoder LAME uses can not decode it
and crashes. To work around the problem, use MAD decoder or
foobar for decoding. LAME encoder part has no problem.

Frame#   127/9222   192 kbps   MS    bitstream problem: resyncing...
Blocktype == 0 and window-switching == 1 not allowed.
big_values too large! 290
Frame#   128/9222    96 kbps

> Thanks, Gabor


Ciao Robert

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Register now and save $200. Hurry, offer ends at 11:59 p.m.,
Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Lame-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/lame-dev

[Gabor FUNK]

> There is something wrong with the mp3 files. I tried the smaller
> problem mp3, the mpg123 decoder LAME uses can not decode it
> and crashes. To work around the problem, use MAD decoder or
> foobar for decoding. LAME encoder part has no problem.
>
> Frame#   127/9222   192 kbps   MS    bitstream problem: resyncing...
> Blocktype == 0 and window-switching == 1 not allowed.
> big_values too large! 290
> Frame#   128/9222    96 kbps

Thanks for the quick reply.
The mp3 files are from internet radio stations, recorded by streamripper.
Could the corruption caused by
a) the internet radio station
b) streamreipper

And/or: should I go and report this to mpg123 developers?
My opinion is that "segmentation fault" is not the proper
way for error handling no matter what... :-]

G.

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Register now and save $200. Hurry, offer ends at 11:59 p.m.,
Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Lame-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/lame-dev

[Robert Hegemann]

Am 05.04.2008, 14:48 Uhr, schrieb Gabor FUNK <[hidden email]>:

>> There is something wrong with the mp3 files. I tried the smaller
>> problem mp3, the mpg123 decoder LAME uses can not decode it
>> and crashes. To work around the problem, use MAD decoder or
>> foobar for decoding. LAME encoder part has no problem.
>>
>> Frame#   127/9222   192 kbps   MS    bitstream problem: resyncing...
>> Blocktype == 0 and window-switching == 1 not allowed.
>> big_values too large! 290
>> Frame#   128/9222    96 kbps
>
> Thanks for the quick reply.
> The mp3 files are from internet radio stations, recorded by streamripper.
> Could the corruption caused by
> a) the internet radio station
> b) streamreipper
>
> And/or: should I go and report this to mpg123 developers?

Well, I don't know how current official mpg123 handles these
files. We forked libmpg123 several years ago and our decoder
isn't that good in handling stream errors. There was a plan
spending some time on the decoder for 3.99 release.

> My opinion is that "segmentation fault" is not the proper
> way for error handling no matter what... :-]

That is true.

> G.


Ciao Robert

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Register now and save $200. Hurry, offer ends at 11:59 p.m.,
Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Lame-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/lame-dev

[Gabor FUNK]

> Well, I don't know how current official mpg123 handles these
> files. We forked libmpg123 several years ago and our decoder
> isn't that good in handling stream errors. There was a plan
> spending some time on the decoder for 3.99 release.

Just for the record/googlers:

I tested mpg123 (0.67) and it seems to handle these "bad" files
as well, so now I use:

    mpg123 -s inputfile.mp3 | lame - outputfile.mp3

Maybe an entry could make it to your todo list to "refork"/update
libmpg123 eventually... :-)

Cheers, G.

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Lame-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/lame-dev