Re: ATAPI Problems

Matt Kucenski wrote:
> I have had this problem with my DVD drive since 5.3RC1.  I 
> just installed 6-LATEST hoping that the problem may have 
> been corrected, but no such luck.
> 
> A normal boot of the GENERIC kernel gives these messages:
> 
> acd0: FAILURE - MODE_SENSE_BIG timed out
> acd0: CDROM <NEC DVD+RW ND-1100A/10GE> at ata1-master UDMA33
 >
> Afterwards, I am unable to mount a CD.  I receive several 
> more of the same "timed out" messages, followed 
> by "Input/output error" message when running 'mount /cdrom'.

Same here, BETA7 did not exhibit this behaviour:

ad0: 78167MB <Maxtor 6Y080L0/YAR41BW0> [158816/16/63] at ata0-master UDMA133
acd0: FAILURE - MODE_SENSE_BIG timed out
acd0: CDRW <PHILIPS DVDRW416N/1.C0> at ata1-slave UDMA33

> If I run 'atacontrol mode 1 BIOSPIO BIOSPIO' (DVD drive is 
> only drive on second ata channel), I receive one
> 
> acd0: FAILURE - TEST_UNIT_READY timed out
> 
> message and then the drive will mount normally.

Not tried, I'll do..

> Also, if I boot verbosely, my dmesg shows this:
>  (...)
> No errors and the drive will mount normally.

So you say that booting -v you don't see the error? I'll try this too!

> I have tried several combinations of slave/master on 
> different ATA channels with the same results.  I have also 
> tried booting with/without ACPI and with/without DMA 
> (hw.ata.atapi_dma="0/1" in loader.conf) and combinations of 
> the two.

Yes, that's the first thing I thought of, but it did no difference for 
me too.
A bug with the same simptoms has existed ever since in 4-STABLE: the 
ATA/ATAPI code is compeletely new in 5.x so I think it's not related, 
but I quote the PRs for reference, who knows:

    http://www.freebsd.org/cgi/query-pr.cgi?pr=i386/66368
    http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/36610
    http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/56889

(the latter has been closed, but without being fixed on 4-STABLE, and 
the same thing happened to 36610. I hope this one is not beeing 'forgot' 
for some months and the closed because 'This bug is old, I think it's 
been fixed')

I've already declared in the past my availability to help in debugging, 
but nobody cared (and the bug was never fixed): nevertheless, here I am 
again.

Angelo Turetta.