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

Groups > alt.os.linux > #50703 > unrolled thread

Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you?

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

Contents

  Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Arlen Holder <arlenholder@nospam.net> - 2018-09-17 03:59 +0000
    Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Jasen Betts <jasen@xnet.co.nz> - 2018-09-17 07:01 +0000
      Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Arlen Holder <arlenholder@nospam.net> - 2018-09-18 03:25 +0000
    Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Dan Purgert <dan@djph.net> - 2018-09-17 10:47 +0000
      Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Arlen Holder <arlenholder@nospam.net> - 2018-09-18 03:27 +0000
        Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Dan Purgert <dan@djph.net> - 2018-09-18 10:00 +0000
          Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Arlen Holder <arlenholder@nospam.net> - 2018-09-19 04:22 +0000
    Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Paul <nospam@needed.invalid> - 2018-09-17 09:53 -0400
      Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Arlen Holder <arlenholder@nospam.net> - 2018-09-18 03:37 +0000
        Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Paul <nospam@needed.invalid> - 2018-09-18 00:33 -0400
          Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? "Arlen H. Holder" <arlenholder@nospam.net> - 2018-09-20 02:00 +0000
            Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Paul <nospam@needed.invalid> - 2018-09-20 05:55 -0400
              Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? "Arlen H. Holder" <arlenholder@nospam.net> - 2018-09-20 19:52 +0000
                Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? "Arlen H. Holder" <arlenholder@nospam.net> - 2018-09-20 19:58 +0000
          Re: Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you? Mike Easter <MikeE@ster.invalid> - 2018-09-20 11:31 -0700

#50703 — Do these USB transfer times on Ubuntu 18.04 to mobile devices make sense to you?

I just ran these six tests moments ago on Ubuntu 18.04 on an old desktop.

Do the widely disparate times make any sense to you?
(I took lots of screenshots - but they take too long to edit & post.)

1. On Windows, I copied a photo 400 times, named ant001.jpg to ant400.jpg
   Each photo = 4,287,961 bytes   Entire directory = 1.7 GB
--- at this point I dual booted to Ubuntu ---
2. I slid those 400 files over USB from that Windows filesystem to Android
   Time = 18 minutes, 21 seconds
7. I slid those 400 files over USB from that Windows filesystem to iOS
   Time = 1 minute, 42 seconds
3. I slid those 400 files over USB from Android to the Windows filesystem
   Time = 2 minutes, 20 seconds
6. I slid those 400 files over USB from iOS to the Windows filesystem
   Time = 1 minute, 55 seconds
5. I slid those 400 files over USB from iOS to Android
   Time = 11 minutes, 49 seconds
4. I slid those 400 files over USB from Android to iOS
   Time = 3 minutes, 5 seconds

Note: I ran the tests in the order of the numbers above so that the
screenshots would be easier to take without having to re-arrange windows.

Equipment:
Desktop = HP Pavilion P6230 Windows 10 (latest) dual booted to Ubuntu 18.04
iOS = 2017 9.7" iOS 11.2.6 iPad, 128GB storage 
Android = LG Stylo 3 Plus, 32GB 

Do the widely disparate times make any sense to you?

[toc] | [next] | [standalone]

#50707

On 2018-09-17, Arlen Holder <arlenholder@nospam.net> wrote:
> I just ran these six tests moments ago on Ubuntu 18.04 on an old desktop.
>
> Do the widely disparate times make any sense to you?
> (I took lots of screenshots - but they take too long to edit & post.)
>
> 1. On Windows, I copied a photo 400 times, named ant001.jpg to ant400.jpg
>    Each photo = 4,287,961 bytes   Entire directory = 1.7 GB
> --- at this point I dual booted to Ubuntu ---
> 2. I slid those 400 files over USB from that Windows filesystem to Android
>    Time = 18 minutes, 21 seconds
> 7. I slid those 400 files over USB from that Windows filesystem to iOS
>    Time = 1 minute, 42 seconds
> 3. I slid those 400 files over USB from Android to the Windows filesystem
>    Time = 2 minutes, 20 seconds
> 6. I slid those 400 files over USB from iOS to the Windows filesystem
>    Time = 1 minute, 55 seconds
> 5. I slid those 400 files over USB from iOS to Android
>    Time = 11 minutes, 49 seconds
> 4. I slid those 400 files over USB from Android to iOS
>    Time = 3 minutes, 5 seconds

They don't seem completely crazy. it could be that the android has a slower flash
writing speed or that the iOS did some short-cut specially because the 400
files had identical content but 18:21 vs 11:49 is puzzling.

Using distinct files and copying to/from a linux ramdisk might be informative.

-- 
     ت

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

#50733

On 17 Sep 2018 07:01:20 GMT, Jasen Betts wrote:

> They don't seem completely crazy. it could be that the android has a slower flash
> writing speed or that the iOS did some short-cut specially because the 400
> files had identical content but 18:21 vs 11:49 is puzzling.
> 
> Using distinct files and copying to/from a linux ramdisk might be informative.

Thanks for suggesting that the disparity of speeds isn't completely crazy.
It was disconcerting, for sure, the slowness of the slides *to* Android.

Perhaps the fact my phone only has 4GB of remaining storage mattered?
Dunno.

All other speeds were an order of magnitude faster.


NOTE: There is a trick to writing to iOS that most people don't know.

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

#50708

Arlen Holder wrote:
> [...]
>
> Do the widely disparate times make any sense to you?

Yes, the first go you had to read from (most likely) spinning rust into
cache (RAM).

After that, you were reading from (and potentially writing to) cache,
rather than the harddrive itself.


-- 
|_|O|_| Registered Linux user #585947
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: 05CA 9A50 3F2E 1335 4DC5  4AEE 8E11 DDF3 1279 A281

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

#50734

On 17 Sep 2018 10:47:06 GMT, Dan Purgert wrote:

> Yes, the first go you had to read from (most likely) spinning rust into
> cache (RAM).
> 
> After that, you were reading from (and potentially writing to) cache,
> rather than the harddrive itself.

There may be merit to that since I "re-used" already selected files.
In the next test, I'll switch the order around.
And I'll close all the already-open windows.

I had left them open to make the screenshots more consistent.
That is, I'd select the files from the Windows filesystem, and slide them
to, say, iOS, and then I'd have them still selected in iOS and slide them
to, say, Android.

Next time I'll close all the windows so that if selections are cached, that
might clear the cache.

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

#50745

Arlen Holder wrote:
> On 17 Sep 2018 10:47:06 GMT, Dan Purgert wrote:
>
>> Yes, the first go you had to read from (most likely) spinning rust into
>> cache (RAM).
>> 
>> After that, you were reading from (and potentially writing to) cache,
>> rather than the harddrive itself.
>
> There may be merit to that since I "re-used" already selected files.
> In the next test, I'll switch the order around.
> And I'll close all the already-open windows.

That won't change anything.  You'll need to flush the disk buffers /
cache.

Minimally, youll need this: 
  sync; echo 1 > /proc/sys/vm/drop_caches


-- 
|_|O|_| Registered Linux user #585947
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: 05CA 9A50 3F2E 1335 4DC5  4AEE 8E11 DDF3 1279 A281

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

#50763

On 18 Sep 2018 10:00:31 GMT, Dan Purgert wrote:

> That won't change anything.  You'll need to flush the disk buffers /
> cache.
> 
> Minimally, youll need this: 
>   sync; echo 1 > /proc/sys/vm/drop_caches

Thanks for that suggestion of how to purge the related cache.

I kind of gave up on editing the dozens of screenshots because Windows just
won't load them in the order I click them which makes editing a large
number of files in series a bitch.
  Why does a digital photo editor on Windows order a numbered series of incoming files capriciously? 
 <https://groups.google.com/forum/#!topic/rec.photo.digital/jlNvYuH8bmQ>

Here is an edited session of just the Android to iOS copy, where everyone
who 'says" it's obvious, is bullshitting us, because it's not obvious the
trick to write to any desired directory in the iOS visible file system.

Plug in any number of iOS & Android devices to your desktop:
 <http://www.bild.me/bild.php?file=9261136speedt001.jpg>

On Windows I created 400 copies of a test file using Irfanview:
 <http://www.bild.me/bild.php?file=6748333speedt002.jpg>

Each of the 400 files was 4,188KB (1.7GB in toto):
 <http://www.bild.me/bild.php?file=3783685speedt003.jpg>

The Android and iOS file systems easily mount on the desktop:
 <http://www.bild.me/bild.php?file=6176902speedt004.jpg>

The trick that nobody knows is how to WRITE to the iOS filesystem!
 <http://www.bild.me/bild.php?file=7155870speedt005.jpg>

Once you know the trick, you just create any folder where you want:
 <http://www.bild.me/bild.php?file=2335527speedt006.jpg>

In this case, I created a folder in the normally read-only DCIM folder:
 <http://www.bild.me/bild.php?file=8602708speedt007.jpg>

In one fell swoop, I used the desktop to copy 400 files from Android to
iOS:
 <http://www.bild.me/bild.php?file=9242923speedt008.jpg>

I could just as well have copied the other way around:
 <http://www.bild.me/bild.php?file=2204403speedt009.jpg>

Or, from either Android and iOS to or from the Windows filesystem:
 <http://www.bild.me/bild.php?file=3283618speedt010.jpg>

That's the sheer beauty and elegance of this brilliant solution:
 <http://www.bild.me/bild.php?file=6181611speedt011.jpg>

Moving the 400 files from Android to iOS took 3 minutes 5 seconds:
 <http://www.bild.me/bild.php?file=7049277speedt012.jpg>

If I did the math right, moving the files using Ubuntu 18.04 from the
Android 7 LG Stylo 3 Plus to the iOS 11.2.6 2017 iPad 9.7-inch went at the
rate of about 9.4 megabits per second or about a half gigabyte per minute.

I don't know what's "possible" but that's what I got.

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

#50714

Arlen Holder wrote:
> I just ran these six tests moments ago on Ubuntu 18.04 on an old desktop.
> 
> Do the widely disparate times make any sense to you?
> (I took lots of screenshots - but they take too long to edit & post.)
> 
> 1. On Windows, I copied a photo 400 times, named ant001.jpg to ant400.jpg
>    Each photo = 4,287,961 bytes   Entire directory = 1.7 GB
> --- at this point I dual booted to Ubuntu ---
> 2. I slid those 400 files over USB from that Windows filesystem to Android
>    Time = 18 minutes, 21 seconds
> 7. I slid those 400 files over USB from that Windows filesystem to iOS
>    Time = 1 minute, 42 seconds
> 3. I slid those 400 files over USB from Android to the Windows filesystem
>    Time = 2 minutes, 20 seconds
> 6. I slid those 400 files over USB from iOS to the Windows filesystem
>    Time = 1 minute, 55 seconds
> 5. I slid those 400 files over USB from iOS to Android
>    Time = 11 minutes, 49 seconds
> 4. I slid those 400 files over USB from Android to iOS
>    Time = 3 minutes, 5 seconds
> 
> Note: I ran the tests in the order of the numbers above so that the
> screenshots would be easier to take without having to re-arrange windows.
> 
> Equipment:
> Desktop = HP Pavilion P6230 Windows 10 (latest) dual booted to Ubuntu 18.04
> iOS = 2017 9.7" iOS 11.2.6 iPad, 128GB storage 
> Android = LG Stylo 3 Plus, 32GB 
> 
> Do the widely disparate times make any sense to you?

A review of the "LG Stylo 3 Plus" says it heats up quickly
and has short battery life. Would it be entering thermal
throttle state while you're testing ?

Your miserable times probably aren't all that uncommon.

https://forums.oneplus.com/threads/otg-usb-file-transfer-speed.503797/

    "I have never seen transfer speeds over 3Mbs
     even for USB 2.0 thats pretty slow ? "

https://vxlabs.com/2014/11/06/use-adb-to-bypass-dog-slow-mtp-transfer-of-files-from-android-to-linux/

    "This is beyond embarrassing."

   Paul

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

#50735

On 17 Sep 2018 06:53:34 GMT, Paul wrote:

> A review of the "LG Stylo 3 Plus" says it heats up quickly
> and has short battery life. Would it be entering thermal
> throttle state while you're testing ?

I bought a handful of these LG Stylo 3 Plus phones at Costco for $130 each
last Christmas, where I haven't seen any "heating" or "short battery" life.
<https://groups.google.com/d/msg/misc.phone.mobile.iphone/HDI8moW_4Pw/fFqp2LudAQAJ>

> 
> Your miserable times probably aren't all that uncommon.
> 
> https://forums.oneplus.com/threads/otg-usb-file-transfer-speed.503797/
> 
>     "I have never seen transfer speeds over 3Mbs
>      even for USB 2.0 thats pretty slow ? "

You always bring up good stuff Paul, where that forum showed flaky speeds,
depending on a host of factors.

In my case, the desktop is old, and even worse perhaps, I'm using the USB
hub inside the monitor where that's another level of abstraction.

Still, some speeds were an order of magnitude faster than others.
But, as noted, it could be the order of my testing procedure
(if cache is involved where I admit I never understood cache).

Likely I should reboot each time so as to have a valid start point.

> 
> https://vxlabs.com/2014/11/06/use-adb-to-bypass-dog-slow-mtp-transfer-of-files-from-android-to-linux/
> 
>     "This is beyond embarrassing."
> 

That was very useful Paul, as they got 1 megabit/second, and they said that
ADB is faster than MTP which is "dog slow". They did surmise it might be a
Linux-only problem, which I would disagree with since I tested to and from
iOS which was an order of magnitude faster using the same Linux desktop as
the OS host.

While they found ADB to be four times faster than MTP, any solution I come
up with has to be cross platform, where iOS is always going to be the
hardest one to overcome.

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

#50736

Arlen Holder wrote:
> On 17 Sep 2018 06:53:34 GMT, Paul wrote:
> 
>> A review of the "LG Stylo 3 Plus" says it heats up quickly
>> and has short battery life. Would it be entering thermal
>> throttle state while you're testing ?
> 
> I bought a handful of these LG Stylo 3 Plus phones at Costco for $130 each
> last Christmas, where I haven't seen any "heating" or "short battery" life.
> <https://groups.google.com/d/msg/misc.phone.mobile.iphone/HDI8moW_4Pw/fFqp2LudAQAJ>
> 
>> Your miserable times probably aren't all that uncommon.
>>
>> https://forums.oneplus.com/threads/otg-usb-file-transfer-speed.503797/
>>
>>     "I have never seen transfer speeds over 3Mbs
>>      even for USB 2.0 thats pretty slow ? "
> 
> You always bring up good stuff Paul, where that forum showed flaky speeds,
> depending on a host of factors.
> 
> In my case, the desktop is old, and even worse perhaps, I'm using the USB
> hub inside the monitor where that's another level of abstraction.
> 
> Still, some speeds were an order of magnitude faster than others.
> But, as noted, it could be the order of my testing procedure
> (if cache is involved where I admit I never understood cache).
> 
> Likely I should reboot each time so as to have a valid start point.
> 
>> https://vxlabs.com/2014/11/06/use-adb-to-bypass-dog-slow-mtp-transfer-of-files-from-android-to-linux/
>>
>>     "This is beyond embarrassing."
>>
> 
> That was very useful Paul, as they got 1 megabit/second, and they said that
> ADB is faster than MTP which is "dog slow". They did surmise it might be a
> Linux-only problem, which I would disagree with since I tested to and from
> iOS which was an order of magnitude faster using the same Linux desktop as
> the OS host.
> 
> While they found ADB to be four times faster than MTP, any solution I come
> up with has to be cross platform, where iOS is always going to be the
> hardest one to overcome.

Don't click on this PDF while reading it. Do control-A to select all,
if you plan on copying something. If you click on the surface of the
document, the document tries to open a web page.

This pretends to be a performance analysis.

https://www.ijert.org/phocadownload/V1I9/IJERTV1IS9375.pdf

   "3. Even though double buffering is enabled in Nexus, both the read
       and write speeds are lesser than Snowball.
       From the bus transactions, it is seen that there is a lot of data
       corruption due to CRC16 errors. This results in the re-transmission
       of data every time data corruption occurs.
   "

Now, that's an effect I would never have considered.

Do you release products that constantly emit transmission
errors ? Is there a high demand for that ? If the customers
only knew. I wonder if there is a performance counter for
that one.

If you read other parts of that paper, you'll see the author
has drunk the koolaid. Yes, our friend DRM. It's what users
always wanted.

    "Lastly, MTP is also more secure than MSC, as it allows
     digital rights management (DRM) protected content to be
     transferred to a host.

     Thus, MTP is more user-friendly than MSC.
    "

   Paul

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

#50811

On Tue, 18 Sep 2018 00:33:00 -0400, Paul wrote:

>      Thus, MTP is more user-friendly than MSC.

BTW, Paul, may I ask you a personal question?

If so, here's the question.

These are facts:
a. I proposed a system which NOBODY seems to understand
b. Everyone *thinks* is the typical USB method
c. But the typical USB method can't read/write to iOS

May I ask you Paul, since I know you're smarter than I am.

Did _anyone_ comprehend the method I proposed?
Or is everyone like a brick, unable to comprehend what I said?

If nobody understood it, then I'm a genius, because I understand it, but I
know I'm only average in intelligence, but I tower over everyone who
responded in that original thread simply becuase I understand it and they
don't (and it's not hard to understand).

I admit I'm frustrated.
I admit I don't have the social skills required to deal with morons.

But all I want to know is whether a _single person_ actually _understood_
the simple method I proposed.

Do you think _anyone_ actually understood it?
Or am I the only one?

Did you understand it Paul?

HINT: It's nowhere described in any news article and I've scoured the net
(where the tidbits are described but not the whole solution).

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

#50817

Arlen H. Holder wrote:
> On Tue, 18 Sep 2018 00:33:00 -0400, Paul wrote:
> 
>>      Thus, MTP is more user-friendly than MSC.
> 
> BTW, Paul, may I ask you a personal question?
> 
> If so, here's the question.
> 
> These are facts:
> a. I proposed a system which NOBODY seems to understand
> b. Everyone *thinks* is the typical USB method
> c. But the typical USB method can't read/write to iOS
> 
> May I ask you Paul, since I know you're smarter than I am.
> 
> Did _anyone_ comprehend the method I proposed?
> Or is everyone like a brick, unable to comprehend what I said?
> 
> If nobody understood it, then I'm a genius, because I understand it, but I
> know I'm only average in intelligence, but I tower over everyone who
> responded in that original thread simply becuase I understand it and they
> don't (and it's not hard to understand).
> 
> I admit I'm frustrated.
> I admit I don't have the social skills required to deal with morons.
> 
> But all I want to know is whether a _single person_ actually _understood_
> the simple method I proposed.
> 
> Do you think _anyone_ actually understood it?
> Or am I the only one?
> 
> Did you understand it Paul?
> 
> HINT: It's nowhere described in any news article and I've scoured the net
> (where the tidbits are described but not the whole solution).

It looks to me like MTPFS mounts done from a host which
supports such things. Of all the devices mentioned in these
various threads, probably only a Linux host has MTPFS. Some
of the platforms have "MTP transfer" where the file system
isn't mounted quite the same way.

In the case of IOS, I don't know if it supports MTP pull
over OTG. I can find references to both iTunes and MTP existing
at the same time. Or, of MTP failing to work properly when
it used to work properly. It suggests the Apple end could
have MTP, but I couldn't find a nice clean reference to it.

In any case, this is the closest model I have of what you're
doing. I certainly cannot reproduce this, because I don't have
any smartphones or tablets. I was gifted a laptop,
and that's the only reason I have a laptop to test with.

         host      MTPFS mount   periph
        Ubuntu <---------------- Android
transfer     \
   agent       \
                +--------------> IOS
                  MTPFS mount    periph

Like FTP, there's pushing and pulling, reading and writing.
And the user selects the best or most convenient of those
options, to do transfers. In this case, the MTPFS mounting
method, makes it possible for the Ubuntu box to work as
a transfer agent. Since not all possible combos exist in
this case, you have to use the combos that do exist.

But none of your posts to date, have shown technical details.
You didn't "show your skirt". Maybe in the above diagram,
showing two lines from /etc/mtab (the mount table) or using
some other status command, would have highlighted in a
minimal amount of space, what you were doing.

Linux has two tables. The /etc/fstab is a table for things
like permanent mounts, to be set up at startup. A "mount -a"
would attempt to mount everything listed in fstab. The /etc/mtab
file on the other hand, keeps a table of what is mounted
right now (the output resulting from successful mounts).

A number of mounts are necessary for system operation.
(Perhaps /proc is mounted on slash or something.) That's
the basic principle. By showing some sort of evidence of
the mount system (MTPFS) it would be more apparent what
you're doing.

I don't think I have any MTP devices in the house, so
I can't even mock up that part to show you what needs
showing. My camera is USB mass storage.

The first reference I could find, is to the notion of
an fstab entry. It hints, that in the year 2011, it's
still a fusermount, and that may be what the fuse means.

https://forums.linuxmint.com/viewtopic.php?t=74101

    mtpfs    /media/xoom    fuse    user,noauto,allow_other    0    0

And the method can get corrupted by other mounting
methods crashing or something.

https://stackoverflow.com/questions/24966676/transport-endpoint-is-not-connected

But there's not a strong signal there.

The current year is 2018, and all I can find is these
older references when the setup was still crude. Once mounts
move from fuse space (file system user space) to the kernel,
the automation improves dramatically.

https://ubuntuforums.org/showthread.php?t=2226702

    Paul

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

#50828

On Thu, 20 Sep 2018 05:55:18 -0400, Paul wrote:

> It looks to me like MTPFS mounts done from a host which
> supports such things.

Ah, thanks. Since I know you know more than I do about networking, it's
especially revealing that you said that.

In addition to understanding the Android file system, there's also the
ability to handle the iOS file system (both read and write):
 How to read/write access iOS file systems on Ubuntu/Windows over USB cable 
 <https://groups.google.com/forum/#!topic/misc.phone.mobile.iphone/IFC52JXBQ1c>

Note the *critical* feature pair, which is that there is *write* access to
the *entire* visible file system on iOS.
 <http://s1.bild.me/bilder/110417/9185854windows_to_ios.jpg>

I think you just explained to me what the problem is with people
comprehending the solution, which is that they don't realize that this is
more than just interfacing to Android.

It's a true cross platform solution.

> Of all the devices mentioned in these
> various threads, probably only a Linux host has MTPFS. Some
> of the platforms have "MTP transfer" where the file system
> isn't mounted quite the same way.

You bring up a good point about Linux having MTPFS, but that's the *easy*
part. That's also the *obvious* part.

What's not obvious to everyone is that Linux also has the ability to read
and write to the iOS file system.

And yet, I have only seen the trick stated in one place on the entire net,
in addition to myriad posts by me.

It's two things that are hard (until you learn the clever secret):
a. Viewing the entire visible file system on iOS, and, 
b. Writing to it!

Just google for the number of people who _want_ an iOS device to act like a
read/write USB stick, which, is trivial for me. 

But 99.999999% of people with iOS don't know how (IMHO) to do it.

> In the case of IOS, I don't know if it supports MTP pull
> over OTG. I can find references to both iTunes and MTP existing
> at the same time. Or, of MTP failing to work properly when
> it used to work properly. It suggests the Apple end could
> have MTP, but I couldn't find a nice clean reference to it.

Ubuntu 18.04 natively uses libimobiledevice drivers, which, AFAIK, speak
iOSFS (if that's even a term).

Gory details here:
 <https://groups.google.com/d/msg/misc.phone.mobile.iphone/IFC52JXBQ1c/Y6vQAYtSCgAJ>

> In any case, this is the closest model I have of what you're
> doing. I certainly cannot reproduce this, because I don't have
> any smartphones or tablets. I was gifted a laptop,
> and that's the only reason I have a laptop to test with.
> 
>          host      MTPFS mount   periph
>         Ubuntu <---------------- Android
> transfer     \
>    agent       \
>                 +--------------> IOS
>                   MTPFS mount    periph

I think that's *almost* the case.
I think there is no such thing as MTPFS on iOS.
It's iOSFS (or whatever it's called).


> Like FTP, there's pushing and pulling, reading and writing.

No no no no no no no no no.

You have to realize that *everything* on iOS is *different* than it is on
all other common consumer platforms!

FTP on iOS is severely limited in what it can "see".
USB on iOS is even more limited in what it can see.

Write is even worse.

My advice to you is to try to never make the pretty big mistake of thinking
what you know about FTP on all other platforms, applies equally to iOS. 

I wish it did.
But it doesn't (empirical observations of my own).

Neither FTP nor USB work on iOS like they do with all other platforms.

Why?
I don't know why.
I just know that they don't.

> And the user selects the best or most convenient of those
> options, to do transfers. In this case, the MTPFS mounting
> method, makes it possible for the Ubuntu box to work as
> a transfer agent. Since not all possible combos exist in
> this case, you have to use the combos that do exist.

At the top level, if we assume you meant MTPFS for Android and iOSFS for
Apple, and if we assume you meant NTFS for Windows, then that top-level
statement is true.

The inherent brilliance of this method is that it simultaneously allows
write/read access to the entire visible file system of all 4 platforms:
1. Linux (via ntfs)
2. Windows (via ntfs)
3. Android (via mtpfs)
4. iOS (via iOSfs)

The hard part is #4 (the rest are obvious).

> But none of your posts to date, have shown technical details.

See this thread for full & complete details, which you know, Paul, that I
always supply. I was just holding off on the "trick" to write to iOS
because of all those fools who said it was "obvious".

Every one of those fools was bullshitting us.
I hate bullshitters. I really do.

They waste everyone's time.
And they don't move the ball forward one inch.

Since I respond to them (they're bullies), the entire thread is a waste of
time for everyone. I didn't learn a thing. You didn't either.

All you need to know is in this thread:
 How to read/write access iOS file systems on Ubuntu/Windows over USB cable 
 <https://groups.google.com/forum/#!topic/misc.phone.mobile.iphone/IFC52JXBQ1c>

If adults can improve on *that* process, we'd move the ball forward.
Otherwise, everyone is wasting my time, and I'm wasting theirs.

> You didn't "show your skirt". Maybe in the above diagram,
> showing two lines from /etc/mtab (the mount table) or using
> some other status command, would have highlighted in a
> minimal amount of space, what you were doing.

The only reason I didn't lift my skirt was because the naysayers are so
Dunning Kruger canonical examples, that I wanted to prove they were full of
shit.

If they only googled my name, they'd have found _this_ thread which lifts
my skirt like you can't believe - since I'm one of the *rare* people here
who actually has purposefully helpful intent on improving our tribal
knowledge.
 <https://groups.google.com/forum/#!topic/misc.phone.mobile.iphone/IFC52JXBQ1c>

Again, if adults can improve on _that_ process - we'd get somewhere.
Just pissing on me helps nobody ... least of all you and me.

> Linux has two tables. The /etc/fstab is a table for things
> like permanent mounts, to be set up at startup. A "mount -a"
> would attempt to mount everything listed in fstab. The /etc/mtab
> file on the other hand, keeps a table of what is mounted
> right now (the output resulting from successful mounts).

The funny thing, Paul, is that I don't understand networking, and yet, all
of this 'fstab' stuff, happens completely automatically.

Hence, I (and therefore all of us) can "mount" iOS file systems, Windows
filesytems, Linux filesystems, and Android filesystems, without knowing
*anything* about the /etc/fstab file.

I think that's pretty neat, don't you?
It's part of the brilliance of this system!

> A number of mounts are necessary for system operation.
> (Perhaps /proc is mounted on slash or something.) That's
> the basic principle. By showing some sort of evidence of
> the mount system (MTPFS) it would be more apparent what
> you're doing.

See the aforementioned tutorial.
If something is missing, let me know.

I would *love* for someone to *improve* upon the process.
So far, nobody knows more than I do.

And, since I already said I don't even know what the fstab syntax is,
that's really sad, don't you think?

HINT: Everyone says the method is "obvious", but not a single person
understood it, while I understood that it works, but I can't explain how it
works (it's all "magic" to me).

If you can improve on this process - _that_ would be great!
 <https://groups.google.com/forum/#!topic/misc.phone.mobile.iphone/IFC52JXBQ1c>

> I don't think I have any MTP devices in the house, so
> I can't even mock up that part to show you what needs
> showing. My camera is USB mass storage.

I haven't dealt with the cameras, but I suspect, having dealt with cross
platform issues for decades, that the _only_ hard device to deal with will
always be iOS.

Any cross platform solution that doesn't include iOS, isn't a cross
platform solution. 

It should include the Mac also, by the way, but I don't have a Mac anymore.

> The first reference I could find, is to the notion of
> an fstab entry. It hints, that in the year 2011, it's
> still a fusermount, and that may be what the fuse means.
> 
> https://forums.linuxmint.com/viewtopic.php?t=74101
> 
>     mtpfs    /media/xoom    fuse    user,noauto,allow_other    0    0
> 
> And the method can get corrupted by other mounting
> methods crashing or something.
> 
> https://stackoverflow.com/questions/24966676/transport-endpoint-is-not-connected
> 
> But there's not a strong signal there.

I've been mounting iOS devices for years, where Ubuntu is just getting
better and better and better with each release (I think I started with
14.10, as I recall). 

This latest Ubuntu 18.04 if *fantastic* compared to 17.04 in terms of not
having to do _anything_ to be able to *read* the full visible iOS file
system.

NOTE: It's is NOT obvious how to _write_ to the full iOS fileystem!
(The note I pointed you to explains it - where I'm sure now all those
morons like Jasen and Diesel and nospam will claim they figured it out, but
all they'd have to do is read what I'm telling you - which - since they
haven't even done that yet - means they're all full of shit).

As an aside, the number of people full of shit on Usenet is something like
99% (personal experience), but luckily, the number of people full of shit
on the linux groups is vastly less than that.

> The current year is 2018, and all I can find is these
> older references when the setup was still crude. Once mounts
> move from fuse space (file system user space) to the kernel,
> the automation improves dramatically.
> 
> https://ubuntuforums.org/showthread.php?t=2226702
> 
>     Paul

As of Ubuntu 18.04, the setup is *fantastic*.
Take that from someone you know has never told a mistruth ever on Usenet.

I tell you where I get my data (personal experience), and I give all the
facts like I did in this tutorial (I've written perhaps thousands).

Thank you Paul for answering my heart-felt question.
I know you're sincere - and you probably am sceptical if I'm sincere.
But I am.

I _care_ to edify people.
It bothers me that people are fools (99% of them, IMHO).
They can't learn.

To be blunt, I don't have the social skills to deal with fools.

I know you're smart, and that you're caring, but it's indicative that my
assessment from your answer is that even you didn't understand the iOS
part. 

That (most likely) means nobody else did either.
Sigh.

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

#50830

On Thu, 20 Sep 2018 19:52:10 -0000 (UTC), Arlen H. Holder wrote:

> What's not obvious to everyone is that Linux also has the ability to read
> and write to the iOS file system.

BTW, in case it's not obvious ... Windows does NOT have this capability.

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

#50824

Paul wrote:
> This pretends to be a performance analysis.
> 
> https://www.ijert.org/phocadownload/V1I9/IJERTV1IS9375.pdf

Comparision of Data Transfer Protocols over USB  (2012)

(a discussion of MSC vs MTP)

> If you read other parts of that paper, you'll see the author
> has drunk the koolaid. Yes, our friend DRM. It's what users
> always wanted.

>     "Lastly, MTP is also more secure than MSC, as it allows
>      digital rights management (DRM) protected content to be
>      transferred to a host.
> 
>      Thus, MTP is more user-friendly than MSC. >     "

For those of us who are uninformed about this MSC vs MTP business and 
particularly how it affects USB transfers between these dissimilar 
devices, I found the wp article helpful for some basic understanding.

https://en.wikipedia.org/wiki/Media_Transfer_Protocol Media Transfer 
Protocol

     1 History
     2 Comparison with USB Mass Storage
         2.1 File oriented instead of block oriented protocol
         2.2 Performance
         2.3 Transparency to MTP unaware software
         2.4 Drivers know a fixed set of supported devices
         2.5 The spec knows a fixed set of defined file formats
     3 Direct modification and partial transfer features
         3.1 Android extensions
     4 MTP support
         4.1 Windows
         4.2 Unix-like systems
             4.2.1 Graphical
             4.2.2 Virtual file system
             4.2.3 macOS
         4.3 Other operating systems
         4.4 Other manufacturers
     5 Alternatives
     6 See also
     7 References


-- 
Mike Easter

[toc] | [prev] | [standalone]

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

csiph-web