Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > alt.os.linux > #32945 > unrolled thread

UEFI question

Back to article view | Back to alt.os.linux

Contents

  UEFI question crankypuss <invalid@invalid.invalid> - 2015-12-30 05:11 -0700
    Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2015-12-30 13:35 +0100
      Re: UEFI question crankypuss <invalid@invalid.invalid> - 2015-12-30 06:03 -0700
        Re: UEFI question Bit Twister <BitTwister@mouse-potato.com> - 2015-12-30 13:48 +0000
          Re: UEFI question crankypuss <invalid@invalid.invalid> - 2015-12-30 14:10 -0700
            Re: UEFI question "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2015-12-30 16:26 -0500
              Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2015-12-31 10:21 +0100
                Re: UEFI question "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2015-12-31 08:50 -0500
                  Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2015-12-31 16:19 +0100
                  Re: UEFI question crankypuss <invalid@invalid.invalid> - 2015-12-31 11:51 -0700
                    Re: UEFI question Bit Twister <BitTwister@mouse-potato.com> - 2015-12-31 19:34 +0000
                      Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2015-12-31 21:52 +0100
                        Re: UEFI question "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2016-01-01 18:58 -0500
                          Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2016-01-02 02:19 +0100
                            Re: UEFI question blind Pete <0123peter@gmail.com> - 2016-01-03 16:30 +1100
                              Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2016-01-03 11:12 +0100
                                Re: UEFI question Eef Hartman <E.J.M.Hartman@gmail.com> - 2016-01-03 12:44 +0000
                                  Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2016-01-03 15:16 +0100
                                Re: UEFI question blind Pete <0123peter@gmail.com> - 2016-01-03 23:12 +1100
                                  Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2016-01-03 15:22 +0100
                          Re: UEFI question crankypuss <invalid@invalid.invalid> - 2016-01-02 01:53 -0700
                            Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2016-01-02 12:08 +0100
                              Re: UEFI question crankypuss <invalid@invalid.invalid> - 2016-01-02 10:47 -0700
              Re: UEFI question crankypuss <invalid@invalid.invalid> - 2015-12-31 02:37 -0700
            Re: UEFI question Bit Twister <BitTwister@mouse-potato.com> - 2015-12-30 22:18 +0000
        Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2015-12-31 10:16 +0100
          Re: UEFI question crankypuss <invalid@invalid.invalid> - 2015-12-31 02:46 -0700
        Re: UEFI question blind Pete <0123peter@gmail.com> - 2016-01-01 21:20 +1100
          Re: UEFI question crankypuss <invalid@invalid.invalid> - 2016-01-01 05:08 -0700
            Re: UEFI question Pascal Hambourg <boite-a-spam@plouf.fr.eu.org> - 2016-01-01 21:51 +0100
              Re: UEFI question crankypuss <invalid@invalid.invalid> - 2016-01-01 15:17 -0700
                Re: UEFI question Bit Twister <BitTwister@mouse-potato.com> - 2016-01-01 23:06 +0000
                  Re: UEFI question crankypuss <invalid@invalid.invalid> - 2016-01-02 02:16 -0700
                    Re: UEFI question Bit Twister <BitTwister@mouse-potato.com> - 2016-01-02 10:04 +0000
                      Re: UEFI question crankypuss <invalid@invalid.invalid> - 2016-01-02 03:39 -0700
                    Re: UEFI question blind Pete <0123peter@gmail.com> - 2016-01-03 15:20 +1100
            Re: UEFI question blind Pete <0123peter@gmail.com> - 2016-01-03 15:14 +1100

Page 2 of 2 — ← Prev page 1 [2]

#33185

David W. Hodgins wrote:

> On Thu, 31 Dec 2015 15:52:07 -0500, Pascal Hambourg
> <boite-a-spam@plouf.fr.eu.org> wrote:
> 
>> Bit Twister a écrit :
>>> On Thu, 31 Dec 2015 11:51:44 -0700, crankypuss wrote:
>>>> David W. Hodgins wrote:
>>>>> Oops. Using gpt does require uefi. :-)
>>>>
>>>> Now I'm really confused.  I remember BitTwister saying that he had
>>>> moved to GPT some time ago and never looked back, and you're
>>>> implying that in order to do that he had to upgrade all his systems
>>>> to EFI firmware?
>>>
>>> Hodgins implied no such thing.
>>
>> Then how do you understand his "Using gpt does require uefi" ?
> 
> I was referring to a gpt partition table, not a hybrid table that has
> both the gpt entries and the mbr code/table.

I'm not up-to-snuff on this stuff, but it seems like there is always a 
MBR, even with a pure GPT disklabel.

And by the way, what kind of gonzo-bozos came up with new meanings for 
the word "label", probably some of them programmer types don't speek gud 
inglish or sumpin. <g>

> With a hybrid table, the drive can be used to boot on either bios or
> uefi firmware. With a non-hybrid gpt table, uefi is required.
> 
> Note that when using bios firmware, the 2TB limit still applies.

I keep hearing all this stuff about how important the 2T limit is, but 
I'm not sure what to do with the 256G this laptop has.  I guess a lot of 
people have bunches of videos and music, not sure how else you could 
ever need a 2T drive unless you're running a server, and even then it 
seems like you'd want to run some kind of RAID using many less costly 
drives so the ones that go bad can be swapped out without RAIDing the 
cookie jar.

It is kind of comforting that larger faster storage devices are getting 
cheaper and cheaper.  In the extrapolated case "free inexhaustible" 
storage is going to flip some algorithms upside down.

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33194

crankypuss a écrit :
> 
> I'm not up-to-snuff on this stuff, but it seems like there is always a 
> MBR, even with a pure GPT disklabel.

Yes, it is called "protective MBR". Its purpose is to protect the GPT
partition table from being overwritten by programs which do not
understand the GPT format.

[toc] | [prev] | [next] | [standalone]

#33235

Pascal Hambourg wrote:

> crankypuss a écrit :
>> 
>> I'm not up-to-snuff on this stuff, but it seems like there is always
>> a MBR, even with a pure GPT disklabel.
> 
> Yes, it is called "protective MBR". Its purpose is to protect the GPT
> partition table from being overwritten by programs which do not
> understand the GPT format.

That's one of the earlier things I learned about GPT, but I'm still not 
entirely clear on it and don't want to distract myself by writing an 
editor to observe it.  One thing at a time. <g>

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33035

David W. Hodgins wrote:

> On Wed, 30 Dec 2015 16:10:38 -0500, crankypuss
> <invalid@invalid.invalid> wrote:
> 
>> partition where its second-level executable resides.  So the question
>> remains as to whether EFI Firmware itself recognizes filesystems, or
>> whether it views a partition as being identified as the ESP by its
>> flags alone, doesn't it?
> 
> Uefi recognizes and reads the filesystem, though it must be a vfat
> filesystem. The partition type also has to be type EF, so uefi
> can determine which partition to read.
> 
> Note that uefi does not require gpt. A uefi system can boot from an
> mbr style partition table. Using gpt though, does require gpt as
> bios firmware can't handle gpt partition tables.

Thanks Dave.

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#32997

On Wed, 30 Dec 2015 14:10:38 -0700, crankypuss wrote:
> Bit Twister wrote:
>
>> I have a Lenovo, HP, ASUS and an older system at my neighbors house
>> all using legacy grub booting from MBR on GPT formatted ext4 drives.
>
> What model Lenovo? 

DMI: LENOVO H215/Tilapia CRB, BIOS D2KT21AUS 03/12/2010

[toc] | [prev] | [next] | [standalone]

#33031

crankypuss a écrit :
> 
> I recognize that there are broken firmware implementations, that's how 
> you learn which makers to avoid. <g>
> 
> From the theoretical correctness of those statements, I tend to conclude 
> that a legacy BIOS system will remain unaffected by the presence/absence 
> of an ESP, it's just some other partition not to be concerned with.  

Indeed. A non-broken legacy BIOS implementation should not be concerned
about partition tables at all.

> However, the ESP only appears in GPT tables (doesnt' it?),

No. The ESP is present in either GPT or MS-DOS style partition tables in
order to boot in EFI native mode.

> so there's a 
> question of whether (all non-broken) legacy BIOS systems will boot from 
> a GPT-formatted device (and I've heard anecdotal comments that lead me 
> to believe they will, but I haven't yet validated that myself). 

As I wrote above, a non-broken legacy BIOS implementation should not be
concerned about partition tables at all. However I have seen some BIOS
implementations which expect that a partition in the MBR table has the
boot flag set, which is against the GPT specification.

> It also leads me to conclude that (non-broken) UEFI firmware recognizes 
> the ESP, but it's unclear whether it navigates the filesystem within the 
> ESP or simply recognizes it by some set of flags

Both. The ESP has a specific type identifier and filesystem (FAT).

[toc] | [prev] | [next] | [standalone]

#33036

Pascal Hambourg wrote:

> crankypuss a écrit :

[snip]
> 
>> However, the ESP only appears in GPT tables (doesnt' it?),
> 
> No. The ESP is present in either GPT or MS-DOS style partition tables
> in order to boot in EFI native mode.
> 
>> so there's a
>> question of whether (all non-broken) legacy BIOS systems will boot
>> from a GPT-formatted device (and I've heard anecdotal comments that
>> lead me to believe they will, but I haven't yet validated that
>> myself).
> 
> As I wrote above, a non-broken legacy BIOS implementation should not
> be concerned about partition tables at all. However I have seen some
> BIOS implementations which expect that a partition in the MBR table
> has the boot flag set, which is against the GPT specification.
> 
>> It also leads me to conclude that (non-broken) UEFI firmware
>> recognizes the ESP, but it's unclear whether it navigates the
>> filesystem within the ESP or simply recognizes it by some set of
>> flags
> 
> Both. The ESP has a specific type identifier and filesystem (FAT).

Thanks Pascal.

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33095

crankypuss wrote:

> Pascal Hambourg wrote:
> 
>> Hello,
>> 
>> crankypuss a écrit :
>>> from,
>>>    http://adam8157.info/blog/2015/04/hybrid-bootable-usb-stick-for-uefi-and-legacy-bios
>>> 
>>> "Legacy BIOS needs bootloaders to be installed into MBR
>> 
>> Yes.
>> 
>>> and the reserved space after it.
>> 
>> No. Anywhere else is fine too.
>> 
>>> UEFI only needs bootloaders in ESP(EFI System Partition).
>> 
>> Yes.
>> 
>>> There is no resource competition between legacy BIOS and UEFI
>>> requirements."
>> 
>> Yes, in theory. But practice may be different due to BIOS and UEFI
>> implementations.

Agreed, and you have to expect to trip over twice as many firmware 
bugs - that might interact in interesting ways.  

> Thanks Pascal.
> 
> I recognize that there are broken firmware implementations, that's how
> you learn which makers to avoid. <g>
> 
> From the theoretical correctness of those statements, I tend to conclude
> that a legacy BIOS system will remain unaffected by the presence/absence
> of an ESP, it's just some other partition not to be concerned with.
> However, the ESP only appears in GPT tables (doesnt' it?), so there's a
> question of whether (all non-broken) legacy BIOS systems will boot from
> a GPT-formatted device (and I've heard anecdotal comments that lead me
> to believe they will, but I haven't yet validated that myself).

Legacy BIOS firmware or UEFI firmware in legacy mode?  

It is possible to prepare a hybrid drive that looks like an MSDOS 
partitioned drive to a legacy BIOS, but is usable as a GPT drive.  
Such things are generally regarded as unreliable abominations.  

> It also leads me to conclude that (non-broken) UEFI firmware recognizes
> the ESP, but it's unclear whether it navigates the filesystem within the
> ESP or simply recognizes it by some set of flags and runs a first-stage
> loader from the recognized partition?

UEFI is required to recognize a fork of VFAT, but may recognize more.  

-- 
blind Pete
Sig goes here...  

[toc] | [prev] | [next] | [standalone]

#33097

blind Pete wrote:

> crankypuss wrote:

>> From the theoretical correctness of those statements, I tend to
>> conclude that a legacy BIOS system will remain unaffected by the
>> presence/absence of an ESP, it's just some other partition not to be
>> concerned with. However, the ESP only appears in GPT tables (doesnt'
>> it?), so there's a question of whether (all non-broken) legacy BIOS
>> systems will boot from a GPT-formatted device (and I've heard
>> anecdotal comments that lead me to believe they will, but I haven't
>> yet validated that myself).
> 
> Legacy BIOS firmware or UEFI firmware in legacy mode?
> 
> It is possible to prepare a hybrid drive that looks like an MSDOS
> partitioned drive to a legacy BIOS, but is usable as a GPT drive.
> Such things are generally regarded as unreliable abominations.

Unreliable abominations aside, I'm going to take a step back here and 
describe what I'm trying to do, as simply as I can.

I have multiple linux root partitions on a single USB stick.  Some of 
these partitions contain only one kernel, others contain several, but 
they are all bootable linux installs, and they are all 32-bit builds.

Each such USB stick is used for backup/rescue/recovery.  I need to boot 
any of these on 3 different machines which have different firmware:

Acer AOA150: 32-bit processor, legacy BIOS.

Dell XPS13: 64-bit processor, UEFI Firmware, but will boot in legacy 
mode if the boot partition is physical (not logical) and its boot flag 
is set.

ASUS T150: 64-bit processor with 32-bit UEFI Firmware, will not even 
consider booting a non-UEFI device.

I've been successfully handling the Acer/Dell cases with MSDOS partition 
table having one physical partition (boot flag set, grub2 installed) 
followed by an extended partition containing the other distros, one per 
partition.

I'm trying to understand how to add the ASUS case to the mix.  As far as 
I can tell the ASUS pushes me to using a GPT partition table, but beyond 
that the details get a little fuzzy.  I expect that by the time I get 
enough other things done that I can go back to the following two web 
pages, I'll be able to figure it out, cross fingers.  Any additional 
clarification would be helpful.

http://adam8157.info/blog/2015/04/hybrid-bootable-usb-stick-for-uefi-and-legacy-bios

https://www.debian-administration.org/users/dkg/weblog/112

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33131

crankypuss a écrit :
> 
> I have multiple linux root partitions on a single USB stick.  Some of 
> these partitions contain only one kernel, others contain several, but 
> they are all bootable linux installs, and they are all 32-bit builds.
> 
> Each such USB stick is used for backup/rescue/recovery.  I need to boot 
> any of these on 3 different machines which have different firmware:
> 
> Acer AOA150: 32-bit processor, legacy BIOS.
> 
> Dell XPS13: 64-bit processor, UEFI Firmware, but will boot in legacy 
> mode if the boot partition is physical (not logical) and its boot flag 
> is set.

"Physical" ? Do you mean "primary" ?

> ASUS T150: 64-bit processor with 32-bit UEFI Firmware, will not even 
> consider booting a non-UEFI device.
> 
> I've been successfully handling the Acer/Dell cases with MSDOS partition 
> table having one physical partition (boot flag set, grub2 installed) 
> followed by an extended partition containing the other distros, one per 
> partition.
> 
> I'm trying to understand how to add the ASUS case to the mix.  As far as 
> I can tell the ASUS pushes me to using a GPT partition table, but beyond 
> that the details get a little fuzzy.

If the UEFI firmware complies to the UEFI specification, it should
accept to boot from an MSDOS partition table. The drive must contain a
small EFI system partition of type 0xEF formated as FAT (I don't know if
it must be primary or can be logical) where you can install GRUB EFI or
another EFI compatible boot loader as the default loader
EFI/boot/bootx64.efi for a 64-bit firmware or EFI/boot/bootia32.efi for
a 32-bit firmware. Beware, there are 64-bit machines with 32-bit UEFI
firmware.

If the firmware cannot boot from an MSDOS partition table, you'll have
to convert the table to GPT with tools such as gdisk, and reinstall the
legacy and EFI bootloaders. GRUB PC (BIOS) likes to have a small "BIOS
boot" partition on a GPT disk, but it is not mandatory.

[toc] | [prev] | [next] | [standalone]

#33143

Pascal Hambourg wrote:

> crankypuss a écrit :
>> 
>> I have multiple linux root partitions on a single USB stick.  Some of
>> these partitions contain only one kernel, others contain several, but
>> they are all bootable linux installs, and they are all 32-bit builds.
>> 
>> Each such USB stick is used for backup/rescue/recovery.  I need to
>> boot any of these on 3 different machines which have different
>> firmware:
>> 
>> Acer AOA150: 32-bit processor, legacy BIOS.
>> 
>> Dell XPS13: 64-bit processor, UEFI Firmware, but will boot in legacy
>> mode if the boot partition is physical (not logical) and its boot
>> flag is set.
> 
> "Physical" ? Do you mean "primary" ?

Yes, sorry, my terminology is not always quite perfect, I still 
occasionally call a symlink a "synonym", and so forth; old terms die 
hard, my bad.

>> ASUS T150: 64-bit processor with 32-bit UEFI Firmware, will not even
>> consider booting a non-UEFI device.
>> 
>> I've been successfully handling the Acer/Dell cases with MSDOS
>> partition table having one physical partition (boot flag set, grub2
>> installed) followed by an extended partition containing the other
>> distros, one per partition.
>> 
>> I'm trying to understand how to add the ASUS case to the mix.  As far
>> as I can tell the ASUS pushes me to using a GPT partition table, but
>> beyond that the details get a little fuzzy.
> 
> If the UEFI firmware complies to the UEFI specification, it should
> accept to boot from an MSDOS partition table. The drive must contain a
> small EFI system partition of type 0xEF formated as FAT (I don't know
> if it must be primary or can be logical)

It also needs a couple flags set, and I don't know about the 
primary/logical requirement either.  I've been busy enough with other 
things and researching stuff I don't understand that I just got back to 
writing a little code for a change this morning.

I could just try a bunch of stuff, but testing it for bootability on 3 
different systems makes pure trial-and-error a bit too time-consuming.

> where you can install GRUB
> EFI or another EFI compatible boot loader as the default loader
> EFI/boot/bootx64.efi for a 64-bit firmware or EFI/boot/bootia32.efi
> for a 32-bit firmware. Beware, there are 64-bit machines with 32-bit
> UEFI firmware.

Yes, my ASUS T100 is a 64-bit machine with 32-bit UEFI firmware.  Since 
I'm maintaining binary compatability on 3 differing systems, and one of 
them is 32-bit only, all my installs are 32-bit.

> If the firmware cannot boot from an MSDOS partition table, you'll have
> to convert the table to GPT with tools such as gdisk, and reinstall
> the legacy and EFI bootloaders. GRUB PC (BIOS) likes to have a small
> "BIOS boot" partition on a GPT disk, but it is not mandatory.

The test stick I'm working with has been converted by gdisk (as I 
recall) but mostly I'm working on the code to collect information about 
the various partitions available on various block devices.

Thanks P.

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33151

On Fri, 01 Jan 2016 15:17:36 -0700, crankypuss wrote:
>
> I could just try a bunch of stuff, but testing it for bootability on 3 
> different systems makes pure trial-and-error a bit too time-consuming.

Oh woe is you. If only there was some virtual box thingy where you
could install/boot several different Operating Systems/releases for
research/regression/backup/restore/upgrade testing and whatnot.

I know on my bloated KDE Desktop Environment, I click my vbox desktop
shortcut icon which flips to virtual desktop 6 launches VirtualBox and
about 1.5 seconds later up pops a menu showing
    $ ls -1 | wc -l
     26
different virtual machines I have installed. 

I double click any one of them and about 8 seconds later I am at the
exact same place I was when I saved the virtual machine state.

If I feel whatever I am about to do may be lethal for that machine, I
can spend ~5 minutes creating a clone of the virtual machine or if really
pressed for time, create a snapshot which can be rolled back to previous state.

Oh, wait, I remember now, you indicated you could not see any value in
that when you could just create another install on your hardware and
boot it.

Let's look at that, 26 virtual machines, about 40 gig drive assigned
to each, comes out to around 1,040G of disk space. 
Actual space needed for all that works out to 
   $ df -h /vmguest
   Filesystem      Size  Used Avail Use% Mounted on
   /dev/sda9       357G  278G   76G  79% /vmguest

Hmmmm, not bad, not bad at all. Well, to each their own, I guess.

If it were me, I would install a 64 bit OS on the UEFI system, then
VirtualBox where I could create UEFI/legacy 64 or 32 bit installs
which would allow me to test ~95% of everything I might want to do.

Only thing missing would be the actual hardware of a/the real machine.

[toc] | [prev] | [next] | [standalone]

#33186

Bit Twister wrote:

> On Fri, 01 Jan 2016 15:17:36 -0700, crankypuss wrote:
>>
>> I could just try a bunch of stuff, but testing it for bootability on
>> 3 different systems makes pure trial-and-error a bit too
>> time-consuming.
> 
> Oh woe is you.

Life is tough. <g>

> If only there was some virtual box thingy where you
> could install/boot several different Operating Systems/releases for
> research/regression/backup/restore/upgrade testing and whatnot.

I tried virtualbox about 3 years ago and I thought it was crap, but a 
lot can change in that length of time, and this XPS13 has more oomph 
than the Acer AOD257 that I was using then.  It might run decently on 
this system, but until I have bootable backups that boot on all my 
systems, I'm not going to install a bunch of stuff like virtualbox, 
android studio, etc etc.  Later on maybe I'll try it just to see what 
the latest version is like, those are hours useless for anything else, 
right?  Nah, maybe I'll write some code instead, that's at least 
potentially useful, I have more coding projects than I can shake a stick 
at.  Or most times, even remember.

> I know on my bloated KDE Desktop Environment,

I tried the KDE DE in Mageia-5.  Run it on something with a slow 
processor and not much storage sometime.  It takes longer to come up 
than linux takes to boot.  But then so do most of the DE's if you run 
them on slow hardware.  Why run them on slow hardware?  To find out 
which busted places need fixed, is why.  Granted KDE is pretty, but more 
useful than something you don't have to wait for while it does fancy 
fades?  I kinda think not, really.

> I click my vbox desktop
> shortcut icon which flips to virtual desktop 6 launches VirtualBox and
> about 1.5 seconds later

A second and a half for it to start up?  I'll bet 90% of that time is 
wasted by GUI libraries.

> up pops a menu showing
>     $ ls -1 | wc -l
>      26
> different virtual machines I have installed.
> 
> I double click any one of them and about 8 seconds later I am at the
> exact same place I was when I saved the virtual machine state.

You're a patient man.  I recently fibbed and said that I wasn't a speed 
addict, but I believed that at the time.  Even on my slowest system 
there are few things that take 8 seconds.  Whatever, we are all 
different folks in different situations, most people would probably 
consider my setup hopelessly crude just because it doesn't run videos or 
tell me if I've flipped it upside-down.

> If I feel whatever I am about to do may be lethal for that machine, I
> can spend ~5 minutes creating a clone of the virtual machine or if
> really pressed for time, create a snapshot which can be rolled back to
> previous state.

Yeah, it takes something like 5 minutes to rsync jessie onto a new USB 
stick on this XPS13, it may be a bit funky in ways but to me it seems 
very fast.  About as fast as response seemed when I was sharing a 
mainframe with 600 other people back in the '80s.  But our perspectives 
change over time, it seems.

Thing is, I live in a forest, there are 150-year-old trees here, which 
means the odds of a forest fire are creeping up year by year.  I have 
all my work in my pocket, and I can buy some used laptop for $100 or 
less and keep on truckin' while the insurance company tries to remember 
that we've paid their premiums.

I have code that I've been working on since 2002... for that matter I 
have some code for a project I sold the rights to in... sometime in the 
'90s, you know how time is for old people, right?  Getting old to the 
point where you want to avoid starting over because of a forest fire or 
even a simple brainfart.

> Oh, wait, I remember now, you indicated you could not see any value in
> that when you could just create another install on your hardware and
> boot it.
> 
> Let's look at that, 26 virtual machines, about 40 gig drive assigned
> to each, comes out to around 1,040G of disk space.
> Actual space needed for all that works out to
>    $ df -h /vmguest
>    Filesystem      Size  Used Avail Use% Mounted on
>    /dev/sda9       357G  278G   76G  79% /vmguest
> 
> Hmmmm, not bad, not bad at all. Well, to each their own, I guess.
> 
> If it were me, I would install a 64 bit OS on the UEFI system, then
> VirtualBox where I could create UEFI/legacy 64 or 32 bit installs
> which would allow me to test ~95% of everything I might want to do.
> 
> Only thing missing would be the actual hardware of a/the real machine.

How much space does your root partition take?  Just curious.

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33189

On Sat, 02 Jan 2016 02:16:57 -0700, crankypuss wrote:

Yeah KDE startup is damn poor, but not a problem since machines are up 24/7.
I just turn off the monitor when not using it.

> How much space does your root partition take?  Just curious.

My install starts out at
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda3        41G  5.4G   33G  15% /

and after install of
# grep Elapsed: wb_*
wb_official_5_day_one_install.rpt:Elapsed: 0 00:01:59 for 20 packages
wb_official_5_install_addons.rpt:Elapsed:  0 00:08:12 for 138 packages
wb_official_5_install_changes.rpt:Elapsed: 0 00:10:32 for 18 packages
wb_official_5_install_toys.rpt:Elapsed:    0 00:24:57 for 56 packages
wb_official_5_new_install.rpt:Elapsed:     0 00:02:57 for 38 packages

$ df -h /
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda3        41G   16G   23G  42% /

[toc] | [prev] | [next] | [standalone]

#33193

Bit Twister wrote:

> On Sat, 02 Jan 2016 02:16:57 -0700, crankypuss wrote:
> 
> Yeah KDE startup is damn poor, but not a problem since machines are up
> 24/7. I just turn off the monitor when not using it.

From about 3:30pm to 8am (this time of year) we're running on batteries, 
I just turn off the laptops when not using them; besides, when they're 
set up it takes valuable table space.  Our total electric bill for 2015 
was about $10 because there were a few days when I ran the generator.  I 
get to spend the saved electric-bill-money on new toys. <g>

Quite often I hibernate my systems instead of shutting them down, 
because there's enough setup to make it worthwhile.  Too few apps save 
their state when exiting, imo; that's one of the things iceweasel does 
better than it did last year.

>> How much space does your root partition take?  Just curious.
> 
> My install starts out at
> Filesystem      Size  Used Avail Use% Mounted on
> /dev/sda3        41G  5.4G   33G  15% /
> 
> and after install of
> # grep Elapsed: wb_*
> wb_official_5_day_one_install.rpt:Elapsed: 0 00:01:59 for 20 packages
> wb_official_5_install_addons.rpt:Elapsed:  0 00:08:12 for 138 packages
> wb_official_5_install_changes.rpt:Elapsed: 0 00:10:32 for 18 packages
> wb_official_5_install_toys.rpt:Elapsed:    0 00:24:57 for 56 packages
> wb_official_5_new_install.rpt:Elapsed:     0 00:02:57 for 38 packages
> 
> $ df -h /
> Filesystem      Size  Used Avail Use% Mounted on
> /dev/sda3        41G   16G   23G  42% /


> df -h /
Filesystem      Size  Used Avail Use% Mounted on
/dev/sda9       9.9G  5.8G  3.6G  62% /

That's with a bunch of stuff installed that I don't use... games, office 
stuff, etc. 

-- 
http://totally-portable-software.blogspot.com
  [Sun Nov 22: "Total Portability is not binary"]

[toc] | [prev] | [next] | [standalone]

#33322

crankypuss wrote:

[snip]
> Thing is, I live in a forest, there are 150-year-old trees here, which
> means the odds of a forest fire are creeping up year by year.  I have
> all my work in my pocket, and I can buy some used laptop for $100 or
> less and keep on truckin' while the insurance company tries to remember
> that we've paid their premiums.
[snip]

Off site backup.  Encrypt it and copy it to the cloud.  

Don't lose the keys.  

-- 
blind Pete
Sig goes here...  

[toc] | [prev] | [next] | [standalone]

#33321

crankypuss wrote:

> blind Pete wrote:
> 
>> crankypuss wrote:
> 
>>> From the theoretical correctness of those statements, I tend to
>>> conclude that a legacy BIOS system will remain unaffected by the
>>> presence/absence of an ESP, it's just some other partition not to be
>>> concerned with. However, the ESP only appears in GPT tables (doesnt'
>>> it?), so there's a question of whether (all non-broken) legacy BIOS
>>> systems will boot from a GPT-formatted device (and I've heard
>>> anecdotal comments that lead me to believe they will, but I haven't
>>> yet validated that myself).
>> 
>> Legacy BIOS firmware or UEFI firmware in legacy mode?
>> 
>> It is possible to prepare a hybrid drive that looks like an MSDOS
>> partitioned drive to a legacy BIOS, but is usable as a GPT drive.
>> Such things are generally regarded as unreliable abominations.
> 
> Unreliable abominations aside, I'm going to take a step back here and
> describe what I'm trying to do, as simply as I can.
> 
> I have multiple linux root partitions on a single USB stick.  Some of
> these partitions contain only one kernel, others contain several, but
> they are all bootable linux installs, and they are all 32-bit builds.
> 
> Each such USB stick is used for backup/rescue/recovery.  I need to boot
> any of these on 3 different machines which have different firmware:
> 
> Acer AOA150: 32-bit processor, legacy BIOS.
> 
> Dell XPS13: 64-bit processor, UEFI Firmware, but will boot in legacy
> mode if the boot partition is physical (not logical) and its boot flag
> is set.
> 
> ASUS T150: 64-bit processor with 32-bit UEFI Firmware, will not even
> consider booting a non-UEFI device.

To the best of my knowledge the ASUS is broken, but I'm not 
authoritative on that.  

> I've been successfully handling the Acer/Dell cases with MSDOS partition
> table having one physical partition (boot flag set, grub2 installed)
> followed by an extended partition containing the other distros, one per
> partition.
> 
> I'm trying to understand how to add the ASUS case to the mix.  As far as
> I can tell the ASUS pushes me to using a GPT partition table, but beyond
> that the details get a little fuzzy.  I expect that by the time I get
> enough other things done that I can go back to the following two web
> pages, I'll be able to figure it out, cross fingers.  Any additional
> clarification would be helpful.
> 
> http://adam8157.info/blog/2015/04/hybrid-bootable-usb-stick-for-uefi-and-legacy-bios
> 
> https://www.debian-administration.org/users/dkg/weblog/112

Options are many; Fix the ASUS, buy a second USB stick, get devious.  
Remember that devious is dangerous.  If one OS hardware combination 
sees what looks like a genuine old style drive and changes it it 
might be no longer valid GPT drive.  Then the fun starts when 
something helpfully fixes the broken layout.  

Do you really need to have everything on one USB stick?  

-- 
blind Pete
Sig goes here...  

[toc] | [prev] | [standalone]

Page 2 of 2 — ← Prev page 1 [2]

Back to top | Article view | alt.os.linux

csiph-web