Re: permission 600

Re: permission 600

[Tatsuro MATSUOKA]

> ----- Original Message -----
> 
> From: "Eliot Moss" 
> To: "cygwin
> Date: 2022/01/09 日 20:26
> Subject: Re: permission 600
> 
> 
> On 1/9/2022 6:11 AM, Tatsuro MATSUOKA wrote:
> >> ----- Original Message -----
> >>
> >> From: "Thomas Wolff"
> >> Date: 2022/01/09 日 17:51
> >> Subject: Re: permission 600
> >> Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
> >>>> ----- Original Message -----
> >>>>
> >>>> From: "Marco Atzeri"
> >>>> To: "cygwin> Date: 2022/01/09 日 14:39
> >>>> Subject: Re: permission 600
> >>>>
> >>>>
> >>>> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
> >>>>> $ echo aaa > test.txt
> >>>>> $ ls -l  test.txt
> >>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> >>>>> $ chmod 600 test.txt
> >>>>> $ ls -l test.txt
> >>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> >>>> it works for me
> >>>>
> >>>> $ ls -l test.txt
> >>>> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
> >>>>
> >>>> $ chmod 600 test.txt
> >>>>
> >>>> $ ls -l test.txt
> >>>> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
> >>>>
> >>>> I suspect that having user and group called same
> >>>> is the clue
> >>>>
> >>> Ah! Thanks!
> >>>
> >>> Tatsuro
> >> Did you verify it by using different names?
> >> It can hardly be an explanation by POSIX means. If so, it must be some
> >> weird consequence of Windows-specific stuff. Maybe a workaround could be
> >> found for cygwin?
> > 
> > I made another windows account and sign in PC with different user name.
> > But result for chmod 600 gave the same results.
> > I found the workaround for jupyter by readind the code of "paths.py" in jupyter.
> 
> You can find the underlying Window permissions using icacls.  Cygwin
> has to set up some funky work-arounds to model some of the POSIX
> permissions - the two models are substantially different.  Sometimes
> clearing back to some nominal Windows permissions and applying
> chmod again helps.  "setfacl -b" can help clear out permissions,
> etc.  Don't forget that permissions can be inherited from the
> containing directory as well.
> 
> Other people may be able to offer more specific guidance, especially
> if you post the output of icacls.
> 
> Best - Eliot Moss
> 
Because my C drive is not enough, I installed cygwin into USB drive and make home directory on it.
chmod issue are story files on the usb drive.
On files on C drive, chmod 600 works as expected.

My workround for jupyter changed :
export JUPYTER_ALLOW_INSECURE_WRITES=true
=>
export JUPYTER_RUNTIME_DIR=/cygdrive/c/Users/user/AppData/Local/Temp  
(/cygdrive/c/Users/user/AppData/Local/Temp "tmp" environmental variable value in my current environment.)

For another usb drive, chmod works as expected.
chmod does not work  => FAT32
chmod works => NTFS

I back up all files to the external HDD and  format FAT32 drive to NTFS drive and back them.

Thanks!

Tatsuro

Re: permission 600

[Thomas Wolff]

Am 10.01.2022 um 00:13 schrieb Tatsuro MATSUOKA:
>> ----- Original Message -----
>>
>> From: "Eliot Moss"
>> To: "cygwin
>> Date: 2022/01/09 日 20:26
>> Subject: Re: permission 600
>>
>>
>> On 1/9/2022 6:11 AM, Tatsuro MATSUOKA wrote:
>>>> ----- Original Message -----
>>>>
>>>> From: "Thomas Wolff"
>>>> Date: 2022/01/09 日 17:51
>>>> Subject: Re: permission 600
>>>> Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
>>>>>> ----- Original Message -----
>>>>>>
>>>>>> From: "Marco Atzeri"
>>>>>> To: "cygwin> Date: 2022/01/09 日 14:39
>>>>>> Subject: Re: permission 600
>>>>>>
>>>>>>
>>>>>> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
>>>>>>> $ echo aaa > test.txt
>>>>>>> $ ls -l  test.txt
>>>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
>>>>>>> $ chmod 600 test.txt
>>>>>>> $ ls -l test.txt
>>>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
>>>>>> it works for me
>>>>>>
>>>>>> $ ls -l test.txt
>>>>>> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
>>>>>>
>>>>>> $ chmod 600 test.txt
>>>>>>
>>>>>> $ ls -l test.txt
>>>>>> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
>>>>>>
>>>>>> I suspect that having user and group called same
>>>>>> is the clue
>>>>>>
>>>>> Ah! Thanks!
>>>>>
>>>>> Tatsuro
>>>> Did you verify it by using different names?
>>>> It can hardly be an explanation by POSIX means. If so, it must be some
>>>> weird consequence of Windows-specific stuff. Maybe a workaround could be
>>>> found for cygwin?
>>> I made another windows account and sign in PC with different user name.
>>> But result for chmod 600 gave the same results.
>>> I found the workaround for jupyter by readind the code of "paths.py" in jupyter.
>> You can find the underlying Window permissions using icacls.  Cygwin
>> has to set up some funky work-arounds to model some of the POSIX
>> permissions - the two models are substantially different.  Sometimes
>> clearing back to some nominal Windows permissions and applying
>> chmod again helps.  "setfacl -b" can help clear out permissions,
>> etc.  Don't forget that permissions can be inherited from the
>> containing directory as well.
>>
>> Other people may be able to offer more specific guidance, especially
>> if you post the output of icacls.
>>
>> Best - Eliot Moss
>>
> Because my C drive is not enough, I installed cygwin into USB drive and make home directory on it.
> chmod issue are story files on the usb drive.
> On files on C drive, chmod 600 works as expected.
>
> My workround for jupyter changed :
> export JUPYTER_ALLOW_INSECURE_WRITES=true
> =>
> export JUPYTER_RUNTIME_DIR=/cygdrive/c/Users/user/AppData/Local/Temp
> (/cygdrive/c/Users/user/AppData/Local/Temp "tmp" environmental variable value in my current environment.)
>
> For another usb drive, chmod works as expected.
> chmod does not work  => FAT32
> chmod works => NTFS
I was about to ask...
> I back up all files to the external HDD and  format FAT32 drive to NTFS drive and back them.
You can actually convert a USB drive to NTFS without copying files and 
reformatting,
using $SYSTEMROOT/System32/convert

Re: permission 600

[Tatsuro MATSUOKA]

> ----- Original Message ----- 
> From: "Tatsuro MATSUOKA" 
> To: "Thomas Wolff" ; "cygwin
> Date: 2022/01/10 月 20:10
> Subject: Re: permission 600
> 
> 
> > ----- Original Message -----
> > From: "Thomas Wolff" 
> > To: "cygwin@
> > Date: 2022/01/10 月 08:28
> > Subject: Re: permission 600
> > 
> > 
> > 
> > Am 10.01.2022 um 00:13 schrieb Tatsuro MATSUOKA:
> > >> ----- Original Message -----
> > >>
> > >> From: "Eliot Moss"
> > >> To: "cygwin
> > >> Date: 2022/01/09 日 20:26
> > >> Subject: Re: permission 600
> > >>
> > >>
> > >> On 1/9/2022 6:11 AM, Tatsuro MATSUOKA wrote:
> > >>>> ----- Original Message -----
> > >>>>
> > >>>> From: "Thomas Wolff"
> > >>>> Date: 2022/01/09 日 17:51
> > >>>> Subject: Re: permission 600
> > >>>> Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
> > >>>>>> ----- Original Message -----
> > >>>>>>
> > >>>>>> From: "Marco Atzeri"
> > >>>>>> To: "cygwin> Date: 2022/01/09 日 14:39
> > >>>>>> Subject: Re: permission 600
> > >>>>>>
> > >>>>>>
> > >>>>>> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
> > >>>>>>> $ echo aaa > test.txt
> > >>>>>>> $ ls -l  test.txt
> > >>>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> > >>>>>>> $ chmod 600 test.txt
> > >>>>>>> $ ls -l test.txt
> > >>>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> > >>>>>> it works for me
> > >>>>>>
> > >>>>>> $ ls -l test.txt
> > >>>>>> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
> > >>>>>>
> > >>>>>> $ chmod 600 test.txt
> > >>>>>>
> > >>>>>> $ ls -l test.txt
> > >>>>>> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
> > >>>>>>
> > >>>>>> I suspect that having user and group called same
> > >>>>>> is the clue
> > >>>>>>
> > >>>>> Ah! Thanks!
> > >>>>>
> > >>>>> Tatsuro
> > >>>> Did you verify it by using different names?
> > >>>> It can hardly be an explanation by POSIX means. If so, it must be some
> > >>>> weird consequence of Windows-specific stuff. Maybe a workaround could be
> > >>>> found for cygwin?
> > >>> I made another windows account and sign in PC with different user name.
> > >>> But result for chmod 600 gave the same results.
> > >>> I found the workaround for jupyter by readind the code of "paths.py" in jupyter.
> > >> You can find the underlying Window permissions using icacls.  Cygwin
> > >> has to set up some funky work-arounds to model some of the POSIX
> > >> permissions - the two models are substantially different.  Sometimes
> > >> clearing back to some nominal Windows permissions and applying
> > >> chmod again helps.  "setfacl -b" can help clear out permissions,
> > >> etc.  Don't forget that permissions can be inherited from the
> > >> containing directory as well.
> > >>
> > >> Other people may be able to offer more specific guidance, especially
> > >> if you post the output of icacls.
> > >>
> > >> Best - Eliot Moss
> > >>
> > > Because my C drive is not enough, I installed cygwin into USB drive and make home directory on it.
> > > chmod issue are story files on the usb drive.
> > > On files on C drive, chmod 600 works as expected.
> > >
> > > My workround for jupyter changed :
> > > export JUPYTER_ALLOW_INSECURE_WRITES=true
> > > =>
> > > export JUPYTER_RUNTIME_DIR=/cygdrive/c/Users/user/AppData/Local/Temp
> > > (/cygdrive/c/Users/user/AppData/Local/Temp "tmp" environmental variable value in my current environment.)
> > >
> > > For another usb drive, chmod works as expected.
> > > chmod does not work  => FAT32
> > > chmod works => NTFS
> > I was about to ask...
> > > I back up all files to the external HDD and  format FAT32 drive to NTFS drive and back them.
> > You can actually convert a USB drive to NTFS without copying files and 
> > reformatting,
> > using $SYSTEMROOT/System32/convert
> > 
> I will try after backup. 
> Thanks!
> 
> Tatsuro

I would like to clean up USB drive so that I format the drive and come back necesary file images were back.
So I did not use convert.
But thanks for your kindness.
After formatting USB drive in NTFS format, chmod 600 works as expected.
and 
$ python3.8 -m jupyter lab
now works without any patch work.

Thanks!

Tatsuro

Re: permission 600

[Tatsuro MATSUOKA]

> ----- Original Message -----
> From: "Thomas Wolff" 
> To: "cygwin@
> Date: 2022/01/10 月 08:28
> Subject: Re: permission 600
> 
> 
> 
> Am 10.01.2022 um 00:13 schrieb Tatsuro MATSUOKA:
> >> ----- Original Message -----
> >>
> >> From: "Eliot Moss"
> >> To: "cygwin
> >> Date: 2022/01/09 日 20:26
> >> Subject: Re: permission 600
> >>
> >>
> >> On 1/9/2022 6:11 AM, Tatsuro MATSUOKA wrote:
> >>>> ----- Original Message -----
> >>>>
> >>>> From: "Thomas Wolff"
> >>>> Date: 2022/01/09 日 17:51
> >>>> Subject: Re: permission 600
> >>>> Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
> >>>>>> ----- Original Message -----
> >>>>>>
> >>>>>> From: "Marco Atzeri"
> >>>>>> To: "cygwin> Date: 2022/01/09 日 14:39
> >>>>>> Subject: Re: permission 600
> >>>>>>
> >>>>>>
> >>>>>> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
> >>>>>>> $ echo aaa > test.txt
> >>>>>>> $ ls -l  test.txt
> >>>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> >>>>>>> $ chmod 600 test.txt
> >>>>>>> $ ls -l test.txt
> >>>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> >>>>>> it works for me
> >>>>>>
> >>>>>> $ ls -l test.txt
> >>>>>> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
> >>>>>>
> >>>>>> $ chmod 600 test.txt
> >>>>>>
> >>>>>> $ ls -l test.txt
> >>>>>> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
> >>>>>>
> >>>>>> I suspect that having user and group called same
> >>>>>> is the clue
> >>>>>>
> >>>>> Ah! Thanks!
> >>>>>
> >>>>> Tatsuro
> >>>> Did you verify it by using different names?
> >>>> It can hardly be an explanation by POSIX means. If so, it must be some
> >>>> weird consequence of Windows-specific stuff. Maybe a workaround could be
> >>>> found for cygwin?
> >>> I made another windows account and sign in PC with different user name.
> >>> But result for chmod 600 gave the same results.
> >>> I found the workaround for jupyter by readind the code of "paths.py" in jupyter.
> >> You can find the underlying Window permissions using icacls.  Cygwin
> >> has to set up some funky work-arounds to model some of the POSIX
> >> permissions - the two models are substantially different.  Sometimes
> >> clearing back to some nominal Windows permissions and applying
> >> chmod again helps.  "setfacl -b" can help clear out permissions,
> >> etc.  Don't forget that permissions can be inherited from the
> >> containing directory as well.
> >>
> >> Other people may be able to offer more specific guidance, especially
> >> if you post the output of icacls.
> >>
> >> Best - Eliot Moss
> >>
> > Because my C drive is not enough, I installed cygwin into USB drive and make home directory on it.
> > chmod issue are story files on the usb drive.
> > On files on C drive, chmod 600 works as expected.
> >
> > My workround for jupyter changed :
> > export JUPYTER_ALLOW_INSECURE_WRITES=true
> > =>
> > export JUPYTER_RUNTIME_DIR=/cygdrive/c/Users/user/AppData/Local/Temp
> > (/cygdrive/c/Users/user/AppData/Local/Temp "tmp" environmental variable value in my current environment.)
> >
> > For another usb drive, chmod works as expected.
> > chmod does not work  => FAT32
> > chmod works => NTFS
> I was about to ask...
> > I back up all files to the external HDD and  format FAT32 drive to NTFS drive and back them.
> You can actually convert a USB drive to NTFS without copying files and 
> reformatting,
> using $SYSTEMROOT/System32/convert
> 
I will try after backup. 
Thanks!

Tatsuro

Re: permission 600

[Eliot Moss]

On 1/9/2022 6:11 AM, Tatsuro MATSUOKA wrote:
>> ----- Original Message -----
>>
>> From: "Thomas Wolff"
>> Date: 2022/01/09 日 17:51
>> Subject: Re: permission 600
>> Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
>>>> ----- Original Message -----
>>>>
>>>> From: "Marco Atzeri"
>>>> To: "cygwin> Date: 2022/01/09 日 14:39
>>>> Subject: Re: permission 600
>>>>
>>>>
>>>> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
>>>>> $ echo aaa > test.txt
>>>>> $ ls -l  test.txt
>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
>>>>> $ chmod 600 test.txt
>>>>> $ ls -l test.txt
>>>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
>>>> it works for me
>>>>
>>>> $ ls -l test.txt
>>>> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
>>>>
>>>> $ chmod 600 test.txt
>>>>
>>>> $ ls -l test.txt
>>>> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
>>>>
>>>> I suspect that having user and group called same
>>>> is the clue
>>>>
>>> Ah! Thanks!
>>>
>>> Tatsuro
>> Did you verify it by using different names?
>> It can hardly be an explanation by POSIX means. If so, it must be some
>> weird consequence of Windows-specific stuff. Maybe a workaround could be
>> found for cygwin?
> 
> I made another windows account and sign in PC with different user name.
> But result for chmod 600 gave the same results.
> I found the workaround for jupyter by readind the code of "paths.py" in jupyter.

You can find the underlying Window permissions using icacls.  Cygwin
has to set up some funky work-arounds to model some of the POSIX
permissions - the two models are substantially different.  Sometimes
clearing back to some nominal Windows permissions and applying
chmod again helps.  "setfacl -b" can help clear out permissions,
etc.  Don't forget that permissions can be inherited from the
containing directory as well.

Other people may be able to offer more specific guidance, especially
if you post the output of icacls.

Best - Eliot Moss

Re: permission 600

[Tatsuro MATSUOKA]

> ----- Original Message -----
> 
> From: "Thomas Wolff" 
> Date: 2022/01/09 日 17:51
> Subject: Re: permission 600
> Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
> >> ----- Original Message -----
> >>
> >> From: "Marco Atzeri"
> >> To: "cygwin> Date: 2022/01/09 日 14:39
> >> Subject: Re: permission 600
> >>
> >>
> >> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
> >>> $ echo aaa > test.txt
> >>> $ ls -l  test.txt
> >>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> >>> $ chmod 600 test.txt
> >>> $ ls -l test.txt
> >>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> >> it works for me
> >>
> >> $ ls -l test.txt
> >> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
> >>
> >> $ chmod 600 test.txt
> >>
> >> $ ls -l test.txt
> >> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
> >>
> >> I suspect that having user and group called same
> >> is the clue
> >>
> > Ah! Thanks!
> >
> > Tatsuro
> Did you verify it by using different names?
> It can hardly be an explanation by POSIX means. If so, it must be some 
> weird consequence of Windows-specific stuff. Maybe a workaround could be 
> found for cygwin?

I made another windows account and sign in PC with different user name.
But result for chmod 600 gave the same results.
I found the workaround for jupyter by readind the code of "paths.py" in jupyter.

Tatsuro

Re: permission 600

[Thomas Wolff]

Am 09.01.2022 um 07:10 schrieb Tatsuro MATSUOKA:
>> ----- Original Message -----
>>
>> From: "Marco Atzeri"
>> To: "cygwin> Date: 2022/01/09 日 14:39
>> Subject: Re: permission 600
>>
>>
>> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
>>> $ echo aaa > test.txt
>>> $ ls -l  test.txt
>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
>>> $ chmod 600 test.txt
>>> $ ls -l test.txt
>>> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
>> it works for me
>>
>> $ ls -l test.txt
>> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
>>
>> $ chmod 600 test.txt
>>
>> $ ls -l test.txt
>> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
>>
>> I suspect that having user and group called same
>> is the clue
>>
> Ah! Thanks!
>
> Tatsuro
Did you verify it by using different names?
It can hardly be an explanation by POSIX means. If so, it must be some 
weird consequence of Windows-specific stuff. Maybe a workaround could be 
found for cygwin?
Thomas

Re: permission 600

[Tatsuro MATSUOKA]

> ----- Original Message -----
> 
> From: "Marco Atzeri" 
> To: "cygwin> Date: 2022/01/09 日 14:39
> Subject: Re: permission 600
> 
> 
> On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
> > $ echo aaa > test.txt
> > $ ls -l  test.txt
> > -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> > $ chmod 600 test.txt
> > $ ls -l test.txt
> > -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> 
> it works for me
> 
> $ ls -l test.txt
> -rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt
> 
> $ chmod 600 test.txt
> 
> $ ls -l test.txt
> -rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt
> 
> I suspect that having user and group called same
> is the clue
> 

Ah! Thanks!

Tatsuro

Re: permission 600

[Marco Atzeri]

On 09.01.2022 06:28, Tatsuro MATSUOKA wrote:
> $ echo aaa > test.txt
> $ ls -l  test.txt
> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
> $ chmod 600 test.txt
> $ ls -l test.txt
> -rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt

it works for me

$ ls -l test.txt
-rw-r--r-- 1 Marco Kein 4 Jan  9 06:35 test.txt

$ chmod 600 test.txt

$ ls -l test.txt
-rw------- 1 Marco Kein 4 Jan  9 06:35 test.txt

I suspect that having user and group called same
is the clue

> 
> I noticed this phenomena in launching the  jupyter lab.
> 
>    File "/usr/lib/python3.8/site-packages/jupyter_core/paths.py", line 909, in secure_write
>      raise RuntimeError("Permissions assignment failed for secure file: '{file}'."
> RuntimeError: Permissions assignment failed for secure file: '/home/user/.local/share/jupyter/runtime/jpserver-27580.json'. Got '0o644' instead of '0o0600'.
> 
> Workaround for the jupter lab error is
> export JUPYTER_ALLOW_INSECURE_WRITES=true
> 
> Tatsuro
> 

Regards
Marco

permission 600

[Tatsuro MATSUOKA]

$ echo aaa > test.txt
$ ls -l  test.txt
-rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt
$ chmod 600 test.txt
$ ls -l test.txt
-rw-r--r-- 1 user user 4 Jan  9 14:07 test.txt

I noticed this phenomena in launching the  jupyter lab.

  File "/usr/lib/python3.8/site-packages/jupyter_core/paths.py", line 909, in secure_write
    raise RuntimeError("Permissions assignment failed for secure file: '{file}'."
RuntimeError: Permissions assignment failed for secure file: '/home/user/.local/share/jupyter/runtime/jpserver-27580.json'. Got '0o644' instead of '0o0600'.

Workaround for the jupter lab error is
export JUPYTER_ALLOW_INSECURE_WRITES=true

Tatsuro