sil_blacklist - are all those entries necessary?

sil_blacklist - are all those entries necessary?

[Witold Krecicki]

in sata_sil.c there is:
sil_blacklist [] = {
        { "ST320012AS",         SIL_QUIRK_MOD15WRITE },
        { "ST330013AS",         SIL_QUIRK_MOD15WRITE },
        { "ST340017AS",         SIL_QUIRK_MOD15WRITE },
        { "ST360015AS",         SIL_QUIRK_MOD15WRITE },
        { "ST380023AS",         SIL_QUIRK_MOD15WRITE },
        { "ST3120023AS",        SIL_QUIRK_MOD15WRITE },
        { "ST3160023AS",        SIL_QUIRK_MOD15WRITE },
        { "ST3120026AS",        SIL_QUIRK_MOD15WRITE },
        { "ST340014ASL",        SIL_QUIRK_MOD15WRITE },
        { "ST360014ASL",        SIL_QUIRK_MOD15WRITE },
        { "ST380011ASL",        SIL_QUIRK_MOD15WRITE },
        { "ST3120022ASL",       SIL_QUIRK_MOD15WRITE },
        { "ST3160021ASL",       SIL_QUIRK_MOD15WRITE },
        { "Maxtor 4D060H3",     SIL_QUIRK_UDMA5MAX },
        { }
};
I've got ST3120026AS and I've been using it with SIL3112 without this hack for 
a long time - without any negative effects. The same impression on 
ST3200822AS - is there any way to check if it is REALLY necessary? 15MB/s is 
not what I'd expect on SATA...
-- 
Witold Kręcicki (adasi) adasi [at] culm.net
GPG key: 7AE20871
http://www.culm.net

Re: sil_blacklist - are all those entries necessary?

[Jeff Garzik]

Witold Krecicki wrote:
> in sata_sil.c there is:
> sil_blacklist [] = {
>         { "ST320012AS",         SIL_QUIRK_MOD15WRITE },
>         { "ST330013AS",         SIL_QUIRK_MOD15WRITE },
>         { "ST340017AS",         SIL_QUIRK_MOD15WRITE },
>         { "ST360015AS",         SIL_QUIRK_MOD15WRITE },
>         { "ST380023AS",         SIL_QUIRK_MOD15WRITE },
>         { "ST3120023AS",        SIL_QUIRK_MOD15WRITE },
>         { "ST3160023AS",        SIL_QUIRK_MOD15WRITE },
>         { "ST3120026AS",        SIL_QUIRK_MOD15WRITE },
>         { "ST340014ASL",        SIL_QUIRK_MOD15WRITE },
>         { "ST360014ASL",        SIL_QUIRK_MOD15WRITE },
>         { "ST380011ASL",        SIL_QUIRK_MOD15WRITE },
>         { "ST3120022ASL",       SIL_QUIRK_MOD15WRITE },
>         { "ST3160021ASL",       SIL_QUIRK_MOD15WRITE },
>         { "Maxtor 4D060H3",     SIL_QUIRK_UDMA5MAX },
>         { }
> };
> I've got ST3120026AS and I've been using it with SIL3112 without this hack for 
> a long time - without any negative effects. The same impression on 
> ST3200822AS - is there any way to check if it is REALLY necessary? 15MB/s is 
> not what I'd expect on SATA...

It's necessary until we can prove otherwise.  Simply running well 
without your drive in the blacklist means nothing -- you just haven't 
hit the error condition yet.

	Jeff

Re: sil_blacklist - are all those entries necessary?

[Witold Krecicki]

Dnia wtorek 15 luty 2005 17:52, napisałeś:
> Witold Krecicki wrote:
> > in sata_sil.c there is:
> > sil_blacklist [] = {
> >         { "ST320012AS",         SIL_QUIRK_MOD15WRITE },
> >         { "ST330013AS",         SIL_QUIRK_MOD15WRITE },
> >         { "ST340017AS",         SIL_QUIRK_MOD15WRITE },
> >         { "ST360015AS",         SIL_QUIRK_MOD15WRITE },
> >         { "ST380023AS",         SIL_QUIRK_MOD15WRITE },
> >         { "ST3120023AS",        SIL_QUIRK_MOD15WRITE },
> >         { "ST3160023AS",        SIL_QUIRK_MOD15WRITE },
> >         { "ST3120026AS",        SIL_QUIRK_MOD15WRITE },
> >         { "ST340014ASL",        SIL_QUIRK_MOD15WRITE },
> >         { "ST360014ASL",        SIL_QUIRK_MOD15WRITE },
> >         { "ST380011ASL",        SIL_QUIRK_MOD15WRITE },
> >         { "ST3120022ASL",       SIL_QUIRK_MOD15WRITE },
> >         { "ST3160021ASL",       SIL_QUIRK_MOD15WRITE },
> >         { "Maxtor 4D060H3",     SIL_QUIRK_UDMA5MAX },
> >         { }
> > };
> > I've got ST3120026AS and I've been using it with SIL3112 without this
> > hack for a long time - without any negative effects. The same impression
> > on ST3200822AS - is there any way to check if it is REALLY necessary?
> > 15MB/s is not what I'd expect on SATA...
>
> It's necessary until we can prove otherwise.  Simply running well
> without your drive in the blacklist means nothing -- you just haven't
> hit the error condition yet.
So how can I proove it? Are there any tests? It's been running for over a 
year, almost 24/7 and nothing...
-- 
Witold Kręcicki (adasi) adasi [at] culm.net
GPG key: 7AE20871
http://www.culm.net

Re: sil_blacklist - are all those entries necessary?

[Witold Krecicki]

Dnia wtorek 15 luty 2005 21:29, napisałeś:
> Dnia wtorek 15 luty 2005 17:52, napisałeś:
> > Witold Krecicki wrote:
> > > in sata_sil.c there is:
> > > sil_blacklist [] = {
> > >         { "ST320012AS",         SIL_QUIRK_MOD15WRITE },
> > >         { "ST330013AS",         SIL_QUIRK_MOD15WRITE },
> > >         { "ST340017AS",         SIL_QUIRK_MOD15WRITE },
> > >         { "ST360015AS",         SIL_QUIRK_MOD15WRITE },
> > >         { "ST380023AS",         SIL_QUIRK_MOD15WRITE },
> > >         { "ST3120023AS",        SIL_QUIRK_MOD15WRITE },
> > >         { "ST3160023AS",        SIL_QUIRK_MOD15WRITE },
> > >         { "ST3120026AS",        SIL_QUIRK_MOD15WRITE },
> > >         { "ST340014ASL",        SIL_QUIRK_MOD15WRITE },
> > >         { "ST360014ASL",        SIL_QUIRK_MOD15WRITE },
> > >         { "ST380011ASL",        SIL_QUIRK_MOD15WRITE },
> > >         { "ST3120022ASL",       SIL_QUIRK_MOD15WRITE },
> > >         { "ST3160021ASL",       SIL_QUIRK_MOD15WRITE },
> > >         { "Maxtor 4D060H3",     SIL_QUIRK_UDMA5MAX },
> > >         { }
> > > };
> > > I've got ST3120026AS and I've been using it with SIL3112 without this
> > > hack for a long time - without any negative effects. The same
> > > impression on ST3200822AS - is there any way to check if it is REALLY
> > > necessary? 15MB/s is not what I'd expect on SATA...
> >
> > It's necessary until we can prove otherwise.  Simply running well
> > without your drive in the blacklist means nothing -- you just haven't
> > hit the error condition yet.
>
> So how can I proove it? Are there any tests? It's been running for over a
> year, almost 24/7 and nothing...
Still no response - so again:
is there ANY way to test if this hack is necessary for specific model of a 
disk?
-- 
Witold Kręcicki (adasi) adasi [at] culm.net
GPG key: 7AE20871
http://www.culm.net

Re: sil_blacklist - are all those entries necessary?

[Jeff Garzik]

Witold Krecicki wrote:
> Dnia wtorek 15 luty 2005 21:29, napisałeś:
> 
>>Dnia wtorek 15 luty 2005 17:52, napisałeś:
>>
>>>Witold Krecicki wrote:
>>>
>>>>in sata_sil.c there is:
>>>>sil_blacklist [] = {
>>>>        { "ST320012AS",         SIL_QUIRK_MOD15WRITE },
>>>>        { "ST330013AS",         SIL_QUIRK_MOD15WRITE },
>>>>        { "ST340017AS",         SIL_QUIRK_MOD15WRITE },
>>>>        { "ST360015AS",         SIL_QUIRK_MOD15WRITE },
>>>>        { "ST380023AS",         SIL_QUIRK_MOD15WRITE },
>>>>        { "ST3120023AS",        SIL_QUIRK_MOD15WRITE },
>>>>        { "ST3160023AS",        SIL_QUIRK_MOD15WRITE },
>>>>        { "ST3120026AS",        SIL_QUIRK_MOD15WRITE },
>>>>        { "ST340014ASL",        SIL_QUIRK_MOD15WRITE },
>>>>        { "ST360014ASL",        SIL_QUIRK_MOD15WRITE },
>>>>        { "ST380011ASL",        SIL_QUIRK_MOD15WRITE },
>>>>        { "ST3120022ASL",       SIL_QUIRK_MOD15WRITE },
>>>>        { "ST3160021ASL",       SIL_QUIRK_MOD15WRITE },
>>>>        { "Maxtor 4D060H3",     SIL_QUIRK_UDMA5MAX },
>>>>        { }
>>>>};
>>>>I've got ST3120026AS and I've been using it with SIL3112 without this
>>>>hack for a long time - without any negative effects. The same
>>>>impression on ST3200822AS - is there any way to check if it is REALLY
>>>>necessary? 15MB/s is not what I'd expect on SATA...
>>>
>>>It's necessary until we can prove otherwise.  Simply running well
>>>without your drive in the blacklist means nothing -- you just haven't
>>>hit the error condition yet.
>>
>>So how can I proove it? Are there any tests? It's been running for over a
>>year, almost 24/7 and nothing...
> 
> Still no response - so again:
> is there ANY way to test if this hack is necessary for specific model of a 
> disk?

You need a bus analyzer, and need to test different sizes of FIS's.  If 
all possible sizes (2048 combinations) work on your device, the 
blacklist entry is not needed.

	Jeff

Re: sil_blacklist - are all those entries necessary?

[Witold Krecicki]

Dnia czwartek 17 luty 2005 01:59, napisałeś:
> > is there ANY way to test if this hack is necessary for specific model of
> > a disk?
>
> You need a bus analyzer, and need to test different sizes of FIS's.  If
> all possible sizes (2048 combinations) work on your device, the
> blacklist entry is not needed.
is there any software bus analyzer? And if so, is there  
Testing-different-sizes-of-FIS's-for-dummies anywhere ?

-- 
Witold Kręcicki (adasi) adasi [at] culm.net
GPG key: 7AE20871
http://www.culm.net

Re: sil_blacklist - are all those entries necessary?

[Jeff Garzik]

Witold Krecicki wrote:
> Dnia czwartek 17 luty 2005 01:59, napisałeś:
> 
>>>is there ANY way to test if this hack is necessary for specific model of
>>>a disk?
>>
>>You need a bus analyzer, and need to test different sizes of FIS's.  If
>>all possible sizes (2048 combinations) work on your device, the
>>blacklist entry is not needed.
> 
> is there any software bus analyzer? And if so, is there  
> Testing-different-sizes-of-FIS's-for-dummies anywhere ?

You could modify a few drivers to receive an "unknown FIS", or receive 
all FIS's into a DMA-able area that the system could then read.  So it's 
doable.  I just don't have time to teach all of SATA and ATA to you :) 
See http://linux.yyz.us/sata/devel.html

	Jeff