Subject: Re: Quotas
To: Bill Sommerfeld <email@example.com>
From: Bill Studenmund <skippy@macro.Stanford.EDU>
Date: 04/01/1999 12:06:06
On Wed, 31 Mar 1999, Bill Sommerfeld wrote:
> >From a quick look at this, this limit looks like it can be fixed
> without changing the kernel at all.
> Quotas are maintained by the kernel and passed to/from userland in
> terms of disk blocks, not bytes.
> However, edquota and friends convert from blocks to kilobytes by way
> of bytes. A few judicious (off_t) casts of the arguments to the btodb
> and dbtob macro invocations in quota, edquota and repquota should do
> the trick, at least until we get to around 2 terabytes. (and FFS as a
> whole will run into similar limits at that point).
Actually ffs is fine as long as we don't want more than 2^31 frags on a
disk. With 1k frags, we get the 2 TB number above. But if you grow frag
size, each doubling doubles the max size. So 8k frags == 16 TB, etc.
Though I think eventually FFS will need help.