Web lists-archives.com

Re: [ANNOUNCEMENT] Updated: Cygwin 2.8.1-1

On Jul  7 16:07, Adam Dinwoodie wrote:
> On Mon, Jul 03, 2017 at 03:31:22PM +0200, Corinna Vinschen wrote:
> > I uploaded a new Cygwin release 2.8.1-1
> This has introduced a regression that I'm seeing when running `ls` on
> some network shares.  I can reproduce the behaviour with an install of
> only the base Cygwin packages, and the behaviour disappears if I
> downgrade back to v2.8.0-1.
> (Apologies for the obfuscation in the below report; I'm not clear on
> what I'm authorised to disclose about my work network, and so erring
> on the side of caution.)
> Specifically, if I run `ls -l` or `ls --color=always` over certain
> directories on one of my company's Windows network shares, I sometimes
> see errors stating:
>     ls: cannot access '//path/to/file/in/listed/share': Bad address
> The file that is listed in the error message appears as below in the
> `ls -l` output:
>     -?????????? ? ?                  ?                    ?
> ?  <filename>
> When this happens, the file is also coloured by `ls` as if it were not
> executable; with v2.8.0-1 the file is correctly marked as executable.
> Alternatively, in some circumstances when `ls`ing that directory, I
> see no output whatsoever.  This seems to happen in particular when
> accessing the directory via a two-hop symlink, i.e. something created
> like this:
>     $ ln -s //path/to/share symlink1
>     $ ln -s symlink1 symlink2
>     $ ls -l symlink1/
>     <list of files>
>     $ ls -l symlink2/
>     $
> The behaviour doesn't seem to be entirely consistent, and I haven't
> been able to characterise when this behaviour occurs and when it
> doesn't, even on the same directory.
> Given the behaviour seems to reliably not occur when running a bare
> `ls`, I'm guessing the problem is relating to how Cygwin is parsing
> the file permissions.

I have not been able to reproduce this.  The only relevant changes in
2.8.1 in terms of `ls' were related to reparse points (native symlinks,
directory junctions, etc) and a change of st_blocks computation related
to CompactOS compression.

For a start, can you please check if you can reproduce the problem
with the devolper snapshot from 20170606?

Also, can you reproduce this with 2.8.1 under strace perhaps? If so,
I'd be interested to see the strace output.  If you fear to compromise
company data, we can arrange to do this in private email.


Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

Attachment: signature.asc
Description: PGP signature