"Grant" <g_r_a_n_t_@dodo.com.au> wrote in message
news:umipj11h1vfn75mqmaimhpm36hr0lgcqtn@4ax.com...

> Yes, dual booting is safe, another option is to use Live CDs to
> try GNU/Linux -- unlike the inferior OS from Redmond, unix _can_
> run from read-only media. Windows is inherently unsafe and not
> secure.

Windows can also be run from a live CD, according to an ad for a $500 tool I
was just reading. Unfortunately, the important stuff like MS Office and the
security patches are unlikely to *FIT* on a live CD/

Charlie Gibbs wrote:
> Can any of you who use dual-booted systems provide any views?
> I don't know whether anything would let me come any closer than
> swappable hard drives, but I'd be interested in what any of you
> have to say.

At home, I currently have two dual-boot systems I've set up with Linux
on one partition and 'doze XP on another. One's a laptop and the
other's a desktop, FWIW. They're both connected to my network. I
rarely ever boot 'doze on either one, and when I do it's only to use one
or two specific programs I haven't been able to get working under wine.
ASAP I switch back to Linux.

To date I've had no problems whatsoever as a result of having them set
up to dual boot. Well...that's not exactly true! On one box, after
repartitioning the drive and installing Linux on the new partitions, my
first attempt to boot 'doze failed with some stupid M$ error. To fix
that, I had no choice but to boot from an installation disc and
reinstall 'doze. Not a problem, since I had nothing to lose on the
drive at that point. After reinstalling 'doze and rebooting, I found
that it had indeed clobbered my MBR. So I reinstalled Linux. From that
point forward it's been smooth sailing.

Given the infamous security flaws inherent in M$ products, and the M$
mindset of snooping, I certainly do understand your concerns, but, for
me, they're not enough to keep me from dual-booting. I should mention,
however, that I've had almost 20 years experience in system
administration and programming UNIX/Linux systems (anyone remember Tandy
Xenix?!), so I'm highly aware of security and keep close track of what's
happening on my networks.

On Fri, 30 Sep 2005 21:28:30 +0000, Jane Doe wrote:

> that it had indeed clobbered my MBR. So I reinstalled Linux. From that
> point forward it's been smooth sailing.

Why didn't you just boot the installation CD, mount your Linux /
partition, chroot into it and execute /sbin/lilo?

20 years experience!

Dave Uhring wrote:
>>that it had indeed clobbered my MBR. So I reinstalled Linux. From that
>>point forward it's been smooth sailing.

> Why didn't you just boot the installation CD, mount your Linux /
> partition, chroot into it and execute /sbin/lilo?

It's been some time ago and I really don't recall WHY that wouldn't
work, but it wouldn't.

On Fri, 30 Sep 2005 05:38:04 +0200, Charlie Gibbs <cgibbs@kltpzyxm.invalid> wrote:

> Hanlon's Razor states: "Never ascribe to malice that which can
> adequately be explained by stupidity." But this little voice
> in the back of my mind keeps saying, "But Microsoft isn't stupid!"

The back of your mind is right. In a way, at least. But MS is a large
organization, and sometimes the intelligence of an organization is
the intersection of the consitituents' intelligence. Sometimes it's
the union, but most of the time there are complicated patterns.

The management of MS are thought to be extremely focused. They were
always obsessed with keeping their staff lean. Everything always had
to be done in a hurry. MS tended to employ young hungry professionals
with sometimes little ballast. The result is that the cleverness
and intelligence went where the focus and the priorities were. Anything
not on the top of the agenda was done with the left hand if at all.
Hence the stupidity.

You might have heard about "snooping on users", but you should check
out what is known about that snooping. I do remember having seen something
about it, but I only remember it kinda made sense. (But questionable
morally, nevertheless.) Any systematic or random snooping in non-MS
partitions in random geeks' systems, what would MS win by that?

Technically it is easy to do. To report home without some users noticing
is not so easy, but it is possible if they invest enough patience. Wait
until the user visits a site controlled by MS, using IE. Keep the data
encrypted. Use the https protocol. Then nobody can check what the contents
of the communication is.

But the most I can imagine they would want to achieve in this way
would be statistics on the number of people having what other software
on their computers. Pure market research. If they were to gather
names or IP addresses of people having specific items in their Linux
partitions, what would they be using it for? How does such arcane
activities fit with their policy of lean staff? For the statistics,
the can get all the statistics they need using marked research
bureaus, and it does not cost too much. To sum up, I would not
waste much of my energy worrying about that. I might get killed by
an asteroid hitting me in my head any moment, but I don't waste
my energy on that either.

As for the dual boot. I've had it for many years, with few problems.
The worst part was finding enough stuff about it on the net, and
figuring out how to boot Windows from Lilo/Grub. I decided to keep the
windows partition as /dev/hda1, i.e., the first partition on the primary
IDE master drive. I never let fdisk.exe touch the partition table.
I do keep copies of the MBR around and copy it back if MS clobber it
for me. But I almost never needed to install Windows again. I almost
never use it. I have never installed XP. I have 2k and can't afford to
upgrade. I have more than one disk, and today's BIOS allow me to
configure booting from any disk. Then Windows can have the MBR
when I resintall, and I boot off the second disk's MBR all other
times. But I have never seen Windows clobber my non-Windows
partitions.

-Enrique

"Enrique Perez-Terron" <enrio@online.no> wrote in message
newsp.sxxyiucanxtvbs@apeiron.home.lan...

> The management of MS are thought to be extremely focused. They were
> always obsessed with keeping their staff lean. Everything always had
> to be done in a hurry. MS tended to employ young hungry professionals
> with sometimes little ballast. The result is that the cleverness
> and intelligence went where the focus and the priorities were. Anything
> not on the top of the agenda was done with the left hand if at all.
> Hence the stupidity.

Done with the left hand making a hard physical motion, yes. I'm pretty harsh
on that company.

> But the most I can imagine they would want to achieve in this way
> would be statistics on the number of people having what other software
> on their computers. Pure market research. If they were to gather
> names or IP addresses of people having specific items in their Linux
> partitions, what would they be using it for? How does such arcane
> activities fit with their policy of lean staff? For the statistics,
> the can get all the statistics they need using marked research
> bureaus, and it does not cost too much. To sum up, I would not
> waste much of my energy worrying about that. I might get killed by
> an asteroid hitting me in my head any moment, but I don't waste
> my energy on that either.

"Go through this script when the user tests the computer: download this
widget, have it report back to the mother ship lots of information, then
store it".

This is what automated vendor test tools are for, such as the "Genuine
Windows Validation" tool used to permit system updates these days. There are
very legitimate uses for such tools: there's nothing for debugging a Linux
box like having the user run a tool ro run and report these.:

/etc/cron.daily/rpm
cat /var/log/rpmpkgs # for RPM based systems
cat /etc/issue.net
cat /etc/passwd
cat /etc/group
cat /etc/issue.net
cat /etc/fstab
cat /etc/mtab
cat /proc/cpuinfo
cat /proc/meminfo # to report system information

And put that data in a stored directory for reference by a competent
sysadmin. But that information tells you a *lot* about the user's system and
their use of that machine, and could be misuesed for system cracking. So you
don't keep it in a public directory. If you're a minpulative weasel, such as
an angry ex-employee or a monotlithic and abusive monopoly, you could do
quite a lot with such a database of information. If you're a careless
administrator, you could encourage such behavior by making the database
available to less ethical or well-intentioned people by leaving the database
accessible to them on backup tapes or network shares.

It happens, especially in large workgroups. Heck, I've tried to encourage
company presidents and their paperwork khandles to rotate their passwords
and not use guessable ones. You can guess how far that's gotten.

> As for the dual boot. I've had it for many years, with few problems.
> The worst part was finding enough stuff about it on the net, and
> figuring out how to boot Windows from Lilo/Grub. I decided to keep the
> windows partition as /dev/hda1, i.e., the first partition on the primary
> IDE master drive. I never let fdisk.exe touch the partition table.
> I do keep copies of the MBR around and copy it back if MS clobber it
> for me. But I almost never needed to install Windows again. I almost
> never use it. I have never installed XP. I have 2k and can't afford to
> upgrade. I have more than one disk, and today's BIOS allow me to
> configure booting from any disk. Then Windows can have the MBR
> when I resintall, and I boot off the second disk's MBR all other
> times. But I have never seen Windows clobber my non-Windows
> partitions.

I've had it happen, when doing deep systems work such as building a
dual-boot network installer. But that's what test boxes and full system
backups are *for*.

On Sat, 01 Oct 2005 02:33:56 +0200, Enrique Perez-Terron wrote:

> times. But I have never seen Windows clobber my non-Windows
> partitions.

The partition ID assigned by Linux fdisk for an extended partition is 5.
Windows XP recognizes that ID type and asks if the user wishes to format
the logical partitions. Windows lusers, being habituated to click OK in
any Windows query box, then destroy their own Linux installations.

The solution is to change the extended partition type to 0x85. Windows
does not recognize that ID type and lusers are not tempted to overwrite
their Linux partitions.

On Sat, 01 Oct 2005 16:05:14 +0200, Dave Uhring <daveuhring@yahoo.com> wrote:

> On Sat, 01 Oct 2005 02:33:56 +0200, Enrique Perez-Terron wrote:
>
>> times. But I have never seen Windows clobber my non-Windows
>> partitions.
>
> The partition ID assigned by Linux fdisk for an extended partition is 5.
> Windows XP recognizes that ID type and asks if the user wishes to format
> the logical partitions.

Good work, Billy Boy.

But... Are you saying that they offer to format all logical partitions in
one fell swoop? Making it a primary partition? Doesn't Windows recognize
more than four partitions any more? Did they ever? What if one of the
logical partitions is a fat or ntfs partition... Was that never possible?

-Enrique

On Sat, 01 Oct 2005 19:24:39 +0200, Enrique Perez-Terron wrote:

> On Sat, 01 Oct 2005 16:05:14 +0200, Dave Uhring <daveuhring@yahoo.com> wrote:

>> The partition ID assigned by Linux fdisk for an extended partition is 5.
>> Windows XP recognizes that ID type and asks if the user wishes to format
>> the logical partitions.
>
> Good work, Billy Boy.

In the perverted Windows way of looking at the HDD it makes perfect
sense. The user is not expected to properly partition the drive himself
and Windows is just being helpful.

> But... Are you saying that they offer to format all logical partitions in
> one fell swoop? Making it a primary partition? Doesn't Windows recognize
> more than four partitions any more? Did they ever? What if one of the
> logical partitions is a fat or ntfs partition... Was that never possible?

Windows, when 2 or more partitions are generated, allocates the additional
partitions as logical partitions, the "D:", "E:", etc. "drives". I do not
know how many of those logical partitions Windows would have formatted
when it popped up that query box. I clicked "Cancel" and rebooted Linux
to change the extended partition type to 0x85.

If the user has a DOS or NTFS logical partition I suppose that keeping the
extended partition ID as 5 is necessary. Windows does not "see" the
logical partitions if the extended partition has type ID of 0x85.

Since a primary partition table requires 64 bytes and the extended
partition table has 512 bytes available, I assume that Windows could
recognize as many as 32 logical partitions in addition to the 3 primary
partitions.

On Sat, 01 Oct 2005 20:05:19 +0200, Dave Uhring <daveuhring@yahoo.com> wrote:

> On Sat, 01 Oct 2005 19:24:39 +0200, Enrique Perez-Terron wrote:
>
>> On Sat, 01 Oct 2005 16:05:14 +0200, Dave Uhring <daveuhring@yahoo.com> wrote:
>
>>> The partition ID assigned by Linux fdisk for an extended partition is 5.
>>> Windows XP recognizes that ID type and asks if the user wishes to format
>>> the logical partitions.
>>
>> Good work, Billy Boy.
>
> In the perverted Windows way of looking at the HDD it makes perfect
> sense. The user is not expected to properly partition the drive himself
> and Windows is just being helpful.
>
>> But... Are you saying that they offer to format all logical partitions in
>> one fell swoop? Making it a primary partition? Doesn't Windows recognize
>> more than four partitions any more? Did they ever? What if one of the
>> logical partitions is a fat or ntfs partition... Was that never possible?
>
> Windows, when 2 or more partitions are generated, allocates the additional
> partitions as logical partitions, the "D:", "E:", etc. "drives". I do not
> know how many of those logical partitions Windows would have formatted
> when it popped up that query box. I clicked "Cancel" and rebooted Linux
> to change the extended partition type to 0x85.
>
> If the user has a DOS or NTFS logical partition I suppose that keeping the
> extended partition ID as 5 is necessary. Windows does not "see" the
> logical partitions if the extended partition has type ID of 0x85.
>
> Since a primary partition table requires 64 bytes and the extended
> partition table has 512 bytes available, I assume that Windows could
> recognize as many as 32 logical partitions in addition to the 3 primary
> partitions.

The logical partitioning works differently: they are chained. Each partition
has a a partition table in its first sector, with two entries. One describes
the current logical partition, one describes the "rest".

I am not 100% sure of the details, but filling in the gaps of my knowledge
with a little guesswork, this is how I imagine it:

Master partition table:
parition X (1-4) is type "Extended", start sector X1, extends N
sectors.

Partition table in sector X1:
partition 5 (in slot 1) is type whaterver, starts in sector X1,
extends N1 sectors (with N1 < N)
Rest (in slot 2) is type "extended", starts in sector X2,
(usually X2 == X1 + N1), extends over N1.rest sectors
(whith N1.rest = N - N1, normally)

Partition table in sector X2:
partition 6 (in slot 1) is type whaterver, starts in sector X2,
extends N2 sectors (with N2 <= N1.rest)
Rest (in slot 2) is type "extended", starts in sector X3,
(usually X3 == X2 + N2, extends over N2.rest sectors

etc.

The last partition table in the chain does not have any slot of type
"Extended". I imagine that the layout of sector zero of each logical
partition is the same as the primary boot sector, so the partition
table is located near the end of the sector and has room for four slots.

This picture allows for a recursive function implementation, where at
each step the size of the extended partition replaces the size of the
disk, and each partition table is treated much the same way. But then
implementations are probably not all like that, most likely they even
conflict in a couple of ways.

-Enrique