Re: [PATCH #upstream-fixes UPDATED] sata_nv: fix generic, nf2/3 detection regression

[Jeff Garzik <jeff@garzik.org>]

Tejun Heo wrote:
> All three flavors of sata_nv's are different in how their hardreset
> behaves.
> 
> * generic: Hardreset is not reliable.  Link often doesn't come online
>   after hardreset.
> 
> * nf2/3: A little bit better - link comes online with longer debounce
>   timing.  However, nf2/3 can't reliable wait for the first D2H
>   Register FIS, so it can't wait for device readiness or classify the
>   device after hardreset.  Follow-up SRST required.
> 
> * ck804: Hardreset finally works.
> 
> The core layer change to prefer hardreset and follow up changes
> exposed the above issues and caused various detection regressions for
> all three flavors.  This patch, hopefully, fixes all the known issues
> and should make sata_nv error handling more reliable.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>
> ---
> Forgot to update comment for longer debounce timing required for
> nf2/3.  Commit message and comment updated.
> 
>  drivers/ata/sata_nv.c |   53 +++++++++++++++++++++++---------------------------
>  1 file changed, 25 insertions(+), 28 deletions(-)

applied