Re: [PATCH] size-in-bytes

[Martin Dalecki <dalecki@evision-ventures.com>]

Andries.Brouwer@cwi.nl wrote:

>    From dalecki@evision-ventures.com Sat Feb 16 22:09:26 2002
>
>    Andries.Brouwer@cwi.nl wrote:
>
>    > /*
>    >+ * blk_size_in_bytes contains the size of all block-devices in bytes
>    >+ * (blk_size has too low a resolution, since we really need the size
>    >+ * in 512 byte sectors, and fails on devices > 2 TB)
>    >+ *
>    >+ * blk_size_in_bytes[MAJOR][MINOR]
>    >+ *
>    >+ * if (!blk_size_in_bytes[MAJOR]) then no minor size checking is done.
>    >+ */
>    >+loff_t * blk_size_in_bytes[MAX_BLKDEV];
>    >+
>    >+/*
>
>    Please pin it up the block device structure not to just another 
>     arbitrary global array.
>
>You miss the point of the patch, perhaps forgot to read the introduction :-)
>
>The point of the patch is that all applied occurrences of blk_size[][]
>have disappeared. All places in the code where a size is needed, that size
>comes from blkdev_size_in_bytes(), and the size comes in bytes, and the size
>is a loff_t.
>
>You want to remove these arrays - fine, so do I.
>(And of course I removed them a few times, but that is another story.)
>But that is an independent action. Whatever one planned to do with
>blk_size[][] now applies to blk_size_in_bytes[][].
>So, this patch is a step ahead but does not solve all the worlds problems
>at once.
>

Well I see that you have killed all places where it get's used - that's 
fine!
But why don't just kill (fix) the initializations as well. You have already:

g->part[minor(dev)].nr_sects;

and friends right there at your hands ;-).