Re: [Libcdio-devel] Read Sub-channel changes

[Rocky Bernstein]

>
>  - cd-info should report Sense Codes of unexpectedly failed SCSI commands.


Fair enough. Suggest some code to detect errors. You are probably the
person in the best position to know where and how to address.

Thanks.

On Tue, May 31, 2016 at 3:03 PM, Thomas Schmitt <address@hidden> wrote:

> Hi,
>
> the command READ SUB-CHANNEL seems to be quite a lottery.
>
> 2 out of 5 inspected CDs show glitches with MCN and ISRC.
> The result is about the same on three DVD/BD drives. (I do not
> have any real CD drive any more.)
>
> All 5 CDs show undamaged CD-TEXT. It is a riddle why the MCN and
> ISRC data are more prone to read problems than the CD-TEXT stuff.
>
> Well, there were no valgrind error messages. The failures were caused
> by unwilling drive replies, not by wrong processing in libcdio.
>
> --------------------------------------------------------------------
> Proposals:
> --------------------------------------------------------------------
>
> - cdio_get_mcn() should finally lead to mmc_get_mcn().
>
>   Currently it gets redirected to ioctl(CDROM_GET_MCN) in Linux driver.
>   Is mmc supported on all platforms ?
>   Can ((CdIo_t *) p_cdio)->op.get_mcn be skipped in favor of calling
>   mmc_get_mcn() directly ?
>   Or is it necessary to hop from each driver onto mmc_get_mcn() ?
>   (How to get the (CdIo_t *) parameter, then ?)
>
> - cd-info should report Sense Codes of unexpectedly failed SCSI commands.
>
>   I provoked some "illegal request" by wrong track number in order to
>   verify that really no Sense Code gets emitted by failed READ SUB-CHANNEL
>   with correct track numbers.
>   But there was no error indication to see with cd-info.
>   I had to put a printf() into run_mmc_cmd_linux() in order to see it.
>
>   The printf() shows 5,24,00 with wrong track number. No Sense Code is
>   replied with correct track numbers even if the returned TCVAL bit is 0.
>
> - Leon should have a look at
>
> http://git.savannah.gnu.org/gitweb/?p=libcdio.git;a=commitdiff;h=0f9fe770268bc74e571c999e67b74d4968d0e835
>   and judge whether
>     +  if (i_cdtext > CDTEXT_LEN_BINARY_MAX + 2)
>   and following lines are really correct. It seems inconsistent and it
>   has the "2" with a different sign than in the replaced code snippet.
>
> --------------------------------------------------------------------
>
> Have a nice day :)
>
> Thomas
>
>
>