[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Parted 1.5.5-pre4

From: Andrew Clausen
Subject: Re: Parted 1.5.5-pre4
Date: Mon, 3 Dec 2001 08:53:32 +1100
User-agent: Mutt/1.3.17i

On Sun, Dec 02, 2001 at 03:43:05PM -0600, address@hidden wrote:
> > So, what happens when the sector size != 512?
> Take for example 1K sectors.  Then the PMBR is 1 sector, the GPT headers are
> each 1 sector, and the partition table entries (normally 32 sectors) are
> instead 16 sectors (they're defined to be minimum of 16K, where the headers
> and pmbr are defined to be 1 sector each).

Nit-pick: the later isn't a semantic change then, although the
addressing changes ;)

> > Must partitions be aligned to the sector size?
> Yes.  All partitions are described as start/end pairs on logical blocks (the
> device's idea of a sector).  So far we've only used disks with 512-byte
> sectors, but that isn't strictly required.  One *could* use 2048 byte
> sectors.  I thought that dev->sector_size wasn't hard-coded to 512, but
> asked the device what its sector size is and uses that?  I tried to replace
> all references to 512 (except in the header struct definition) to be
> dev->sector_size for this reason.

dev->sector_size isn't hard-coded to 512, but the parted IO functions
ignore it.

Should this be changed?

The thing is, file systems use these functions heavily, etc.

I think it's best that the user of the ped_device_* does the calculation
(dev->sector_size / 512, etc.)

Perhaps byte addressing is better?  64 bits is enough?


reply via email to

[Prev in Thread] Current Thread [Next in Thread]