Messages in this thread | | | From | Ondrej Zary <> | Subject | Re: [RFC PATCH 74/71] ncr5380: Enable PDMA for NCR53C400A | Date | Mon, 30 Nov 2015 14:40:04 +0100 |
| |
On Monday 30 November 2015, Finn Thain wrote: > > On Sun, 29 Nov 2015, Ondrej Zary wrote: > > > Add I/O register mapping for NCR53C400A and enable PDMA mode to > > improve performance and fix non-working IRQ. > > > > Is CONFIG_SCSI_GENERIC_NCR53C400 is still needed? Can the driver be fully > configured at runtime now?
Things depending on CONFIG_SCSI_GENERIC_NCR53C400:
#ifdef CONFIG_SCSI_GENERIC_NCR53C400 #define PSEUDO_DMA #endif
Defining PSEUDO_DMA should not break anything, just makes code bigger a bit. We can probably just define it always in g_NCR5380.c.
This looks weird. I don't have any card with a BIOS to test. #ifdef CONFIG_SCSI_GENERIC_NCR53C400 #define BIOSPARAM #define NCR5380_BIOSPARAM generic_NCR5380_biosparam #else #define NCR5380_BIOSPARAM NULL #endif
This looks very wrong and should be done at runtime: #ifdef CONFIG_SCSI_GENERIC_NCR53C400 #define NCR5380_region_size 16 #else #define NCR5380_region_size 8 #endif
> > Tested with HP C2502 (and user-space enabler). > > > > It would be nice if the enabler was documented. > Documentation/scsi/g_NCR5380.txt might be a good place for that. That file > could perhaps be updated (?) Certainly obsolete and misleading information > should be removed if you see any.
I'll integrate the enabler into the driver. It's like "wakeup sequence for the NCR53C400A and DTC3181E */" in g_NCR5380.c, only the magic numbers are different.
-- Ondrej Zary
| |