Re: [PATCH] Fix accesses at LBA28 boundary (old bug, but nasty)

[Mark Lord <kernel@teksavvy.com>]

On 07/04/10 01:11 AM, Tejun Heo wrote:
> Hello,
>
> On 04/07/2010 12:30 PM, Mark Lord wrote:
>> The cast was there already, so I left it there.
>> Just moved it to cover the entire compile-time expression as a unit,
>> rather than just the (u64)1  as the existing code did it.
>
> Oh, but there's functional difference between the two.  In the
> original place, the initial casting ripples through the whole
> expression making each step of the calculation u64.  It's meant to
> prevent int overflow during calculation (which isn't necessary in the
> current expression BTW) but if you cast the end result, it doesn't
> really mean anything.  If you want to kill the casting, kill it but
> moving the casting outside doesn't make much sense.  Can you repost
> without moving the cast?
..

No.  But I can/did repost with it completely gone.