On Wed, 14 Mar 2007 05:06:11 -0400, Jeff Garzik wrote:
> > - tmp = readl(mmio);
> > - if (tmp & PDC_ERR_MASK) {
> > - qc->err_mask |= AC_ERR_DEV;
> > - pdc_reset_port(ap);
>
> AFAICS pdc_reset_port() will no longer be called, after this change.
>
> Since you are freezing the port with ata_port_freeze() before
> pdc_error_handler() is called, the one remaining call to
> pdc_reset_port() never occurs:
>
> if (!(ap->pflags & ATA_PFLAG_FROZEN))
> pdc_reset_port(ap);
>
> And Promise hardware definitely needs to be "kicked" by pdc_reset_port()
> after most errors.
There's also a conditional pdc_reset_port() in pdc_post_internal_cmd().
I'll reinvestigate the reset logic.
> Also, PCI error are controller-wide, so ideally you should presume that
> all transactions on all ports are bad, and reset everything. See
> mv_pci_error() in sata_mv.c in libata-dev.git#mv-eh for mostly-generic
> code that initiates error handling on all ports, rather than just one.
Agreed.
Thanks,
/Mikael
-
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
