readme

将用户指定的符合iso-9660标准的文件的信息dump出来。

1999-11-14
==========

There seems to be a bug in drivers/block/ide-cd.c...
The user writes:

struct cdrom_read arg;
u_char buf[CD_FRAMESIZE_RAW];
int blk= 0;			/* block of cd to read raw */

arg.cdread_bufaddr= buf;
arg.cdread_buflen= CD_FRAMESIZE_RAW;

arg.cdread_lba= blk;
ioctl(,CDROMREADRAW,&arg);
write(STDOUT_FILENO,buf,CD_FRAMESIZE_RAW);


but ide_cdrom_dev_ioctl() of ide-cd.c takes the argument as struct cdrom_msf.
The function copy_from_user_ret() called by ide_cdrom_dev_ioctl() seems just
to return the lba value specified in arg.cdread_lba and not msf.


The program readraw.c only works with the patch ide-cd.c-v4.54.diff:
-------------------------------------------------------------------------------
2113a2114
> /*
2114a2116
> */
2126a2129
> /*
2131a2135,2139
> 
>               lba= *( (int*)&msf );
> */
> 
>               copy_from_user_ret(&lba, (void *)arg, sizeof (lba), -EFAULT);
-------------------------------------------------------------------------------

But, regarding unreadable blocks, dumping raw data is just as sensible as with
isodump.c which only reads /dev/cdrom by read().

This ioctl(CDROMREADRAW) can be used for reading data tracks not beeing the
first track of the cd. Perhaps I should use CDROMREADMODE1 for this, but
the problem with ide-cd.c will be the same, i.e the call of
ide_cdrom_dev_ioctl().


1999-11-17
==========
mail from Jens Axboe <[e-mail]>, Linux CD-ROM Maintainer: 
data CDROMREAD* ioctl takes arg buffer_address where a struct msf has
been copied to (in contrast to linux/cdrom.h), i.e

unsigned char buf[CD_FRAMESIZE_RAW];
struct cdrom_msf msf;
msf.cdmsf_min0= 0;	/* set up mfs */
memcpy( buf, &msf, sizeof(msf) );
ioctl( , CDROMREADRAW, buf );

Will not put such code to isodump.c for different reasons:
- isodump is based on read(2) and not on ioctl(2)
- even CDROMREADMODE1 does not apply the ECC code -- I hope that read(2)
  does (but I'm not shure)
- raw reading belongs rather to cdda than to data (see ECC)

Perhaps experiences with multisession data cds (not CD+) will change my mind...