Bug with inaccessible folders

[resistancelion]

When you're trying to open locked folder it results in recursion bug:

[DDoSolitary]

What do you mean by "locked"?

[resistancelion]

What do you mean by "locked"?

A folder to which ssh user do not have access to read/write, i've locked X11 folder in openSSH config, so it cannot be opened via remote control (and it works so in FileZilla)

[DDoSolitary]

i've locked X11 folder in openSSH config

Could you please share the exact config line?

[the-Arioch]

i tried to enter into /dev or /dev/char and yasfw seems to really spawn a lot of warnings, which perhaps can significantly slow down console window!

when i did normal run - i waited for almost a minute to enter /dev then i killed yafsw

when i run with redirection echo @#$@#$@#$| yasfw -r -c --server 23213213213 --user root --mount-point W: >yasfw_log.txt - the Explorer worked fast, but the log quickly grew to 1.3MB

i also did grep < yasfw_log.txt name: | sort >> yasfw_sorted.txt - and there seems to be many repeated tries, especially over /dev/zero like 20-30 times. yasfw_sorted.zip

not an infinite loop, but truly many iterations dunno why.

That being said, i saw no "recursion" on windows side.

BTW, symlinks are implemented in modern NTFS so should be in WinNT core too, so i guess in theory Dokan or yafsw could implement them too.
Albeit it would require careful file path parsing and modification, or yasfw would run into security nightmare like those elegant in theory HURD translators did.

[DDoSolitary]

@the-Arioch I think it's better to open a separate issue for this.

The repeated tries are performed by Windows so there's no way to prevent that. Probably I can do some rate-limiting for the logs.

so i guess in theory Dokan or yafsw could implement them too.

It's possible in theory, but unfortunately not supported by Dokan: dokan-dev/dokany#343

[the-Arioch]

Just tried an ad hoc repro on a stalled ticket.

I doubt they are performed by Windows alone as there is clearly different number of attempts at different files.
Why so many attempts at /dev/zero and much less of others?

Also, how does Windows even know there is /dev/zero ?
I wonder if it would make sense to make a mode where yasfw would not report to Windows files that it can not process?

[DDoSolitary]

@the-Arioch PLEASE OPEN A SEPARATE ISSUE FOR THIS

The OP is mainly talking about the recursion issue

[the-Arioch]

i would if I HAD an issue with it. Thus could make tests, confirm fixes, etc. ...but i gonna use yafsw in a very chrooted mode so won't be able to test and confirm. I just made an non-biased extra input on a stalled ticket.

[DDoSolitary]

Well, the ticket is stalled because I don't even know how the X11 folder is "locked"...

[the-Arioch]

That is what i tried to guess, but the very minimised Centos 6.0 box does not even have that folder. From my memory, X11 there is probably a symlink. And i suspect /dev is full of symlinks, so *something* similar manifests itself with so many /dev/zero knocks. But i failed to repro it as well.