Gentoo Archives: gentoo-user

From: David W Noon <dwnoon@××××××××.com>
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] Boot partitions (WAS: migrating disks (from mounts to disklabels
Date: Wed, 24 Nov 2010 16:23:54
Message-Id: 20101124162336.0a0304a9@karnak.local
1 On Wed, 24 Nov 2010 01:10:02 +0100, Alan McKinnon wrote about Re:
2 [gentoo-user] Boot partitions (WAS: migrating disks (from mounts to
3 disklabels:
4
5 >Apparently, though unproven, at 01:14 on Wednesday 24 November 2010,
6 >David W Noon did opine thusly:
7 >
8 >> >Errm, not exactly. SCSI/SATAs are limited to 15 (inc. one extended
9 >> >partition) and old (legacy driven) IDEs are limited to some 63
10 >> >partitions if I recall correctly. If you use the new libata I think
11 >> >you only get 15 partitions for SATA/PATA.
12 >>
13 >> Well that's a software limitation. I am a little surprised that
14 >> the limit is so small, as Windows can support 24 drive letters
15 >> (C: through Z:) assigned to hard drive partitions. Of course,
16 >> accessing the CD-ROM would then be a bit sporty under Windows.
17 >
18 >Not quite, you are confused.
19
20 No, you should read again what I wrote.
21
22 >That's 24 'drives" of all kinds spread across all kinds:
23 >
24 >removeable media
25 >hard disks (all partitions)
26 >full disk (without partitions)
27 >network drives
28 >"$other_stuff" (a catchall for anything else MS might dream up).
29 >
30 >16 for SCSI is plenty in real life, and it's a hardware limitation not
31 >a software one so the driver can't be updated for this.
32
33 The disk hardware knows nothing about partition tables.
34
35 The partition table layout used on IA32 and AMD64 hardware is a
36 convention inherited from IBM's Industry Standard Architecture (ISA)
37 specification and the BIOS of the PC/AT (with revisions for the PS/2,
38 etc., plus extensions by Phoenix, AMI and Wang/Award). We still use
39 that under Linux, some 25+ years on.
40
41 Other operating systems can use radically different partition table
42 layouts. Any disk partitioned that way -- whatever it is -- would be
43 unreadable by Linux, Windows, OS/2 or any other OS that uses ISA
44 assumptions.
45
46 The SCSI/SATA/PATA hardware is totally agnostic about this, as it uses
47 absolute sector addresses, without any concept of partitioning, to
48 handle physical I/O requests. The sector address can be CHS (on older
49 disks) or LBA, but it *never* relates to a partition, only to a
50 complete disk.
51
52 >24 drive letters has *nothing* to do with partition number limits.
53 >They are not even vaguely related.
54
55 It seems you didn't get the joke. I probably should have put a "winky"
56 after the line ending "a bit sporty under Windows."
57 --
58 Regards,
59
60 Dave [RLU #314465]
61 *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
62 dwnoon@××××××××.com (David W Noon)
63 *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

Attachments

File name MIME type
signature.asc application/pgp-signature