Re: [PATCH 4/4] MTD: drivers return max_bitflips, mtd returns -EUCLEAN

[Shmulik Ladkani <shmulik.ladkani@gmail.com>]

On Mon, 02 Apr 2012 09:40:08 -0700 Mike Dunn <mikedunn@newsguy.com> wrote:
> Hi Shmulik,
> 
> On 03/30/2012 11:37 PM, Shmulik Ladkani wrote:
> > Please re-consider having the 'euclean_threshold' comparison within the
> > NAND infrastructure (nand_base.c and clones), instead of within the
> > generic 'mtd_read()' wrapper, as discussed in [1].
> 
> 
> This isn't possible given the intended functionality of the patches.  The plan
> is to allow the user to adjust the bitflip_threshold through sysfs, and do it
> uniquely for each partition.  Since the driver itself receives as its argument
> the mtd_info for the master, not the partition, it can't compare the bitflip
> count with the appropriate threshold to determine whether or not to return
> -EUCLEAN.  

Ok.

Now I understand why you submitted [l2-mtd ac5b1dd mtd: fix partition
wrapper functions]...
You rely on ac5b1dd in order for the comparison to be executed per
partition.
Reasnoable, however delicate. The reslationship between the
'euclean_threeshold' feature and this patch is implicit and not obvious.
It raises a concern that one might later change the behavior of
'part_read' to use the mtd_read wrapper for some reason, without
understanding the affects.

Anyway, looking on it in a broader perspective, can you please state
use-case examples where one might set different thresholds for different
partitions of same device?
I mean, do we really need that granularity?

Regards,
Shmulik