On Tue, 3 Feb 2015, Nicholas Mc Guire wrote:
> This is only an API consolidation to make things more readable.
Please remove the above; the g_NCR5380.c change invalidates it.
> Instances of var * HZ / 1000 are replaced by msecs_to_jiffies(var).
> In addition some timing constants that assumed HZ 100 were adjusted
> to HZ independent settings based on review of the original drivers in
> 1.0.31 and 2.2.16.
>
> Signed-off-by: Nicholas Mc Guire <hofrat@xxxxxxxxx>
> ---
>
> v2: the original patch was not taking care of all the dependencies
> as reported by Finn Thain <fthain@xxxxxxxxxxxxxxxxxxx> - this
> version now uses the suggested config to check the patch.
> v3: g_NCR5380.c changes fixed up based on feedback from Michael Schmitz
> <schmitzmic@xxxxxxxxx> as these settings were from around 1.0.31
> kernel (or earlier) where HZ was 100 only - thus the unit here is
> actually 10s of microseconds. This was "verified" by checking the
> setting changes against 2.2.16 indicating that it was forgotten in
> 1998/99.
>
> Converting milliseconds to jiffies by "val * HZ / 1000" is technically
> ok but msecs_to_jiffies(val) is the cleaner solution and handles all
> corner cases correctly. This is a minor API cleanup only.
>
> This patch was only compile tested with i386_defconfig + CONFIG_ISA=y
> as well as all dependent drivers enabled as modules:
> CONFIG_SCSI_LOWLEVEL=y, CONFIG_SCSI_GENERIC_NCR5380=m
> CONFIG_SCSI_DMX3191D=m, CONFIG_SCSI_DTC3280=m
> CONFIG_SCSI_GENERIC_NCR5380=m, CONFIG_SCSI_GENERIC_NCR5380_MMIO=m
> CONFIG_SCSI_PAS16=m, CONFIG_SCSI_T128=m
>
> Patch is against 3.19.0-rc7 (localversion-next = -next-20150203)
>
> drivers/scsi/NCR5380.c | 10 +++++-----
> drivers/scsi/g_NCR5380.c | 6 +++---
> 2 files changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
> index 8981701..d15411b 100644
> --- a/drivers/scsi/NCR5380.c
> +++ b/drivers/scsi/NCR5380.c
> @@ -474,11 +474,11 @@ static void NCR5380_print_phase(struct Scsi_Host *instance)
> */
> #ifndef USLEEP_SLEEP
> /* 20 ms (reasonable hard disk speed) */
> -#define USLEEP_SLEEP (20*HZ/1000)
> +#define USLEEP_SLEEP msecs_to_jiffies(20)
> #endif
> /* 300 RPM (floppy speed) */
> #ifndef USLEEP_POLL
> -#define USLEEP_POLL (200*HZ/1000)
> +#define USLEEP_POLL msecs_to_jiffies(200)
> #endif
> #ifndef USLEEP_WAITLONG
> /* RvC: (reasonable time to wait on select error) */
> @@ -576,7 +576,7 @@ static int __init __maybe_unused NCR5380_probe_irq(struct Scsi_Host *instance,
> if ((mask & possible) && (request_irq(i, &probe_intr, 0, "NCR-probe", NULL) == 0))
> trying_irqs |= mask;
>
> - timeout = jiffies + (250 * HZ / 1000);
> + timeout = jiffies + msecs_to_jiffies(250);
> probe_irq = NO_IRQ;
>
> /*
> @@ -634,7 +634,7 @@ static void prepare_info(struct Scsi_Host *instance)
> "sg_tablesize %d, this_id %d, "
> "flags { %s%s%s}, "
> #if defined(USLEEP_POLL) && defined(USLEEP_WAITLONG)
> - "USLEEP_POLL %d, USLEEP_WAITLONG %d, "
> + "USLEEP_POLL %lu, USLEEP_WAITLONG %lu, "
> #endif
> "options { %s} ",
> instance->hostt->name, instance->io_port, instance->n_io_port,
> @@ -1346,7 +1346,7 @@ static int NCR5380_select(struct Scsi_Host *instance, struct scsi_cmnd *cmd)
> * selection.
> */
>
> - timeout = jiffies + (250 * HZ / 1000);
> + timeout = jiffies + msecs_to_jiffies(250);
atari_NCR5380.c has exactly the same assignment. I'd like to keep
atari_NCR5380.c and NCR5380.c in agreement because further divergence will
make it harder it to re-unify them. Please make the same change there.
>
> /*
> * XXX very interesting - we're seeing a bounce where the BSY we
> diff --git a/drivers/scsi/g_NCR5380.c b/drivers/scsi/g_NCR5380.c
> index f35792f..a11b152 100644
> --- a/drivers/scsi/g_NCR5380.c
> +++ b/drivers/scsi/g_NCR5380.c
> @@ -57,9 +57,9 @@
> */
>
> /* settings for DTC3181E card with only Mustek scanner attached */
> -#define USLEEP_POLL 1
> -#define USLEEP_SLEEP 20
> -#define USLEEP_WAITLONG 500
> +#define USLEEP_POLL msecs_to_jiffies(10)
> +#define USLEEP_SLEEP msecs_to_jiffies(200)
> +#define USLEEP_WAITLONG msecs_to_jiffies(5000)
>
> #define AUTOPROBE_IRQ
>
Looks OK otherwise.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
