On Wed, Sep 24, 2003 at 01:58:20PM -0700, Jason Thorpe wrote:
> 
> On Wednesday, September 24, 2003, at 01:55  PM, David Laight wrote:
> 
> >>Hm..  Less than a page size.  But 2 sectors worth.
> >
> >possibly 1 fragment?
> 
> Possibly.  The point is that it's not a *page*.

Sorry - my sloppy use of language :/
FWIW
Disk with corruption written to it:
#        size    offset     fstype [fsize bsize cpg/sgs]
 a: 490234752         0     4.2BSD   2048 16384 28624  # (Cyl.      0 - 486343)
Disk from which bad data came:
 h:   4200600  30622374     4.2BSD   1024  8192 46376  # (Cyl.   6028*-   6854*)

It was bytes 0x1000-13ff ie 0x400 or 1024 bytes of block 71716, block size
8192 (here "block" is pgsql term).
This is i386, so a page is 0x1000 or 4096 bytes.

So either 2 disk blocks, or 1 read fragment. Not 1 written fragment :/

Cheers,

Patrick