Windows - psutil does not detect mounted folders without drive letters as disks

[MattGirolami]

Windows supports assigning NTFS disks a mounted folder/mount point value instead of (or in addition to) a drive letter. Psutil does not recognize drives that only have a mount point and not a drive letter.

In the example below, I've attached and formatted a hard drive, but instead of assigning it it's own drive letter, I've assigned it a mount point path of C:\mountedfolder\ . I can iterate the drive in powershell/wmi:

C:\>  Get-WmiObject -namespace "root/cimv2" -query "SELECT Name, Capacity, FreeSpace, deviceid FROM Win32_Volume WHERE DriveType = 3" | select deviceid,name,capacity | format-list


deviceid : \\?\Volume{343e9a8f-c1de-11e4-b98f-806e6f6e6963}\
name     : \\?\Volume{343e9a8f-c1de-11e4-b98f-806e6f6e6963}\
capacity : 523235328

deviceid : \\?\Volume{07ec83a6-0000-0000-0000-010000000000}\
name     : C:\mountedfolder\
capacity : 1070592000

deviceid : \\?\Volume{58dafacc-c28b-11e4-95bb-3863bbba0e0f}\
name     : C:\
capacity : 253377900544

deviceid : \\?\Volume{25083962-0000-0000-0000-c01d3b000000}\
name     : D:\
capacity : 2148532224`

If I attempt to iterate the disks in psutil, I do not see my disk at C:\mountedfolder\

C:\> py
Python 3.5.1 (v3.5.1:37a07cee5969, Dec  6 2015, 01:54:25) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import psutil
>>> psutil.disk_partitions()
[sdiskpart(device='C:\\', mountpoint='C:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='D:\\', mountpoint='D:\\', fstype='FAT32', opts='rw,fixed')]
>>>

If I assign a drive letter ( like E:) to the same disk at c:\mountedfolder\ I can list it in powershell

C:\>  Get-WmiObject -namespace "root/cimv2" -query "SELECT Name, Capacity, FreeSpace, deviceid FROM Win32_Volume WHERE DriveType = 3" | select deviceid,name,capacity | format-list


deviceid : \\?\Volume{343e9a8f-c1de-11e4-b98f-806e6f6e6963}\
name     : \\?\Volume{343e9a8f-c1de-11e4-b98f-806e6f6e6963}\
capacity : 523235328

deviceid : \\?\Volume{07ec83a6-0000-0000-0000-010000000000}\
name     : E:\
capacity : 1070592000

deviceid : \\?\Volume{58dafacc-c28b-11e4-95bb-3863bbba0e0f}\
name     : C:\
capacity : 253377900544

deviceid : \\?\Volume{25083962-0000-0000-0000-c01d3b000000}\
name     : D:\
capacity : 2148532224

Note the deviceid of E:\ matches the deviceid of C:\mountedfolder\ from the previous powershell block.

Now that the same disk is assigned E:, I can iterate it in psutil

C:\> py
Python 3.5.1 (v3.5.1:37a07cee5969, Dec  6 2015, 01:54:25) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import psutil
>>> psutil.disk_partitions()
[sdiskpart(device='C:\\', mountpoint='C:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='D:\\', mountpoint='D:\\', fstype='FAT32', opts='rw,fixed'), 
 sdiskpart(device='E:\\', mountpoint='E:\\', fstype='NTFS', opts='rw,fixed')]
>>> 

[giampaolo]

I think the issue may be here:

psutil/psutil/_psutil_windows.c

Line 2372 in ab9f29b

(type == DRIVE_NO_ROOT_DIR) ||

(skipping drives with no root dir).
If you can compile psutil you can try to remove that (type == DRIVE_NO_ROOT_DIR) || line and see if it makes any difference.
If you can't compile psutil, I should try myself: how do you mount a folder such as C:\mountedfolder?

[MattGirolami]

Having some issues getting it to compile myself, never hacked on python so trying to get all of the prereqs figured out 😄 .

To try it yourself on Windows, do Start -> Run -> mmc

Under File, select add/remove snapin, then add the Disk Management Snapin and connect to your local computer when prompted.

Right click on Disk Management in the left hand pane and select 'Create VHD' . This will allow you to create a virtual hard drive. Set a size and a file location, you will get a .vhd file. You could also attach a new hard drive, usb drive, etc. instead.

In the lower center pane you should have a new disk that is offline/not initialized. Right click on the disk and select initialize. Then, right click on box to the right that says X MB unallocated. Select new simple volume. When prompted to assign a drive letter, select the radio button for 'Mount in the following empty NTFS folder' and select any empty folder on a different drive. When prompted, format the volume as NTFS.

You can then browse to that file path in Windows explorer. It will have a hard drive icon in Windows explorer rather than the standard folder icon.

In the upper center pane of Disk Management you can right click on any volume to assign/unassign/change mounted folders and drive letters at any time.

[giampaolo]

I forgot about all argument. Can you try psutil.disk_partitions(all=True)? That will skip the DRIVE_NO_ROOT check.

[jomann09]

I can confirm that even with all=True it still does not show up. A good way to do it is mount a partition as a drive letter, say Z: and in the computer management window, under disk management, right click on the Z: drive > Change Drive Letter and Paths, then click on "Add..." which will bring up a menu to select a folder to mount the partition to. I did it with some extra space on one of my spare hard drives since it was the most reliable test.

[masci]

Any update on this?

[giampaolo]

@JHomann can you try removing this line and see what happens?

psutil/psutil/_psutil_windows.c

Line 2372 in ab9f29b

(type == DRIVE_NO_ROOT_DIR) ||

[giampaolo]

Actually nevermind, you already stated that all=True does not make any difference.

[jomann09]

So I know this isn't done "properly" (sorry I don't really program in C) because I don't know how we would want to do the loop, but essentially I have a way to show mount points - and I restricted it to only show if all=True and to only check if mount points can exist.

Let me know what you think, and whether or not we can do something to refactor this. I did compile it and test that it does output mount points properly.

jomann09@988f7d4

Also side note, it's not checking for rw/etc specifically on these mount points. Just using opts that were already set for the time being.

Test output:

C:\Users\jomann\Documents\psutil_jake>python
Python 2.7.14 (v2.7.14:84471935ed, Sep 16 2017, 20:19:30) [MSC v.1500 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import psutil
>>> psutil.disk_partitions()
[sdiskpart(device='C:\\', mountpoint='C:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='D:\\', mountpoint='D:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='E:\\', mountpoint='E:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='G:\\', mountpoint='G:\\', fstype='', opts='cdrom')]
>>> psutil.disk_partitions(all=True)
[sdiskpart(device='C:\\', mountpoint='C:\\xampp\\test2mount\\', fstype='NTFS', opts='rw'), 
 sdiskpart(device='C:\\', mountpoint='C:\\test\\', fstype='NTFS', opts='rw'), 
 sdiskpart(device='C:\\', mountpoint='C:\\testmount\\', fstype='NTFS', opts='rw'), 
 sdiskpart(device='C:\\', mountpoint='C:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='D:\\', mountpoint='D:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='E:\\', mountpoint='E:\\', fstype='NTFS', opts='rw,fixed'), 
 sdiskpart(device='G:\\', mountpoint='G:\\', fstype='', opts='cdrom'), 
 sdiskpart(device='I:\\', mountpoint='I:\\', fstype='NTFS', opts='rw,remote'), 
 sdiskpart(device='K:\\', mountpoint='K:\\', fstype='NTFS', opts='rw,remote'), 
 sdiskpart(device='S:\\', mountpoint='S:\\', fstype='NTFS', opts='rw,remote'), 
 sdiskpart(device='T:\\', mountpoint='T:\\', fstype='NTFS', opts='rw,remote'), 
 sdiskpart(device='Z:\\', mountpoint='Z:\\', fstype='NTFS', opts='rw,remote')]

[giampaolo]

Nice! This is certainly desirable.
About your patch, I would use a utility function to reuse this part:

+                        py_tuple = Py_BuildValue(
 +                            "(ssss)",
 +                            drive_letter,
 +                            mp_path,
 +                            fs_type,  // either FAT, FAT32, NTFS, HPFS, CDFS, UDF or NWFS
 +                            opts);
 +                        if (!py_tuple)
 +                            goto error;
 +                        if (PyList_Append(py_retlist, py_tuple))
 +                            goto error;
 +                        Py_DECREF(py_tuple);

...which is repeated 3 times (2 by "you", 1 by "me" later).
Also, I think it makes sense to show these mountpoints also if all=False. That would be in line with the Linux implementation:

>>> pp(psutil.disk_partitions())
[...
 sdiskpart(device='/dev/nvme0n1p4', mountpoint='/var/lib/docker/plugins', fstype='ext4', opts='rw,relatime,errors=remount-ro,data=ordered'),
 sdiskpart(device='/dev/nvme0n1p4', mountpoint='/var/lib/docker/aufs', fstype='ext4', opts='rw,relatime,errors=remount-ro,data=ordered')]
>>> 

A PR for this would be great.

[jomann09]

Awesome thanks for the info, I will re-work it a little bit with that in mind and will make a PR soon!

Thanks for the quick response.

[giampaolo]

Fixed in #1192.