Re: Working on ATA: Less internal state, more asking

On 3 August 2012 23:51, Jonas Wielicki <j.wielicki <at> sotecware.net> wrote:
> Hi all,
>
> I'm (still *sigh*) hunting my SSD issues on iriver (as posted previously
> on the list[1][2]). I think it basically reduces to a “have you tried
> turning it off and on again” fix, because somehow the SSD seems to get
> stuck sometimes, a powercycle can mitigiate that in most of the cases.
>
> BUT, in the course I implemented some ATA parts which might come in
> handy with devices which are behaving weirdly, like going to sleep
> automatically, namely, I removed the internal rockbox state about the
> drive's power management state and added requests to get the current
> state instead (using CMD_CHECK_POWER_MODE).
>
> I realize that I'll probably have to keep some of my rockbox kernel
> patched to work more or less with that SSD (I really should try to
> return it, maybe it's just broken). But some of these changes may give
> rockbox advantages when dealing with buggy devices. Here is a summarized
> list of changes:
>
> * check the current power mode of the device before using it
> * explicitly wake it up, if needed
> * two levels of error handling:
>   1. do a hard reset
>   2. powercycle the IDE bus
>
> I wonder what, if any, of this is appreciated by rockbox so that I can
> wrap it up as one or more nice commits for review at gerrit.
>
> best regards,