python / python/cpython

os.path.ismounts: further situations it doesn't detect correctly

Ouverte
#96,328 18 commentaires 3 réactions 0 personnes assignées Voir sur GitHub

Personne n'a encore pris cette issue.

docs type-bug
Langage dominant
Python
Étoiles
77.2k
Forks
36k
Métriques de merge des PR
Métriques de PR en attente

Description

Bug report

I had a look at os.path.ismount():
https://github.com/python/cpython/blob/43a6deadbb40bc93e0eaebb3c56c34f8a3502132/Lib/posixpath.py#L186-L216

And it may have some issues:

  1. The documentation already says, that it isn't capable of detecting bind-mounts of some filesystem on the very same filesystem. While it's good that it mentions this, the results of ismount() is IMO still simply wrong in these cases, cause it is a mountpoint.
  2. The same problem in the detection algorithm is not only limited to bind-mounts, but also to any other mount points, where the path’s st_dev equals that of its parent. For example, btrfs subvolumes. This is however not documented.

Looking a the code, it effectively does: lstat(), check whether it’s a symlink, lstat() the parent, check whether the device numbers are different (then it's a mountpoint), check whether the inodes are the same.

Now I thought about the latter, and what it is meant for.

The only case I could think of at first is /, where /.. is the same inode. [Is there any other?]

  1. This however already requires, that not only the inodes are the same, but also the device numbers. which is given because of the order of these checks, however, I'd suggest to comment that accordingly.
  2. If the above were the only case, wouldn't it be faster to simply check the pathname for being "/" or b"/". Would perhaps require some realpath() though, to also catch things like "/../.." and so on.
  3. However, I found that / is not the only case: Within a chroot, the chroot's / may (or may not) be a mountpoint - the problem is just, the current algorithm doesn't detect this either. Instead, it wrongly claims that a chroot’s "/" would be a mountpoint, even when it isn't.

Not sure what one could do about all that… /etc/mtab is dead... /proc/mounts may not be available on all POSIX systems and even on Linux it's nowadays rather superseded by /proc/self/mountinfo.

Your environment

  • CPython versions tested on: 3.10.6
  • Operating system and architecture: Debian sid

Cheers,
Chris.

Guide de contribution

Ouvrir le guide de contribution

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Piste de recherche

Commencez par os.path.ismount() dans Lib/posixpath.py, en particulier par l’implémentation liée, et examinez sa documentation. Reproduisez les cas signalés de bind-mount, de sous-volume btrfs et de chroot dans l’environnement Debian indiqué. Le travail est considéré comme terminé lorsqu’une approche de détection approuvée, un comportement corrigé et une documentation des limitations restantes de la plateforme sont disponibles.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Évaluation

Stack technique
python
Domaine
operating-systems
Type d'issue
Bug
Difficulté
5/5
Temps estimé
Plus d'une semaine
Activité
À l'abandon
Clarté
À clarifier
Accessibilité débutants
30/100

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.