Re: [ANNOUNCEMENT] TEST RELEASE: Cygwin 1.7.35-0.3

[Corinna Vinschen <corinna-cygwin at cygwin dot com>]

On Feb 20 12:48, Achim Gratz wrote:
> Corinna Vinschen writes:
> > And it's fast with other shells?  Bash is reading/writing a history
> > file, too, for instance...
> 
> Yes, that's the puzzling bit; although I don't have nearly the same
> number of history lines in .bash_history, but the difference still is at
> least an order of magnitude.
> 
> > I can't debug this.  Care to strace it to find out where the problem
> > might be?  Is that a 1.7.34+ problem?
> 
> I think so.  Since I switched very early to the LDAP integration code
> and have been running on snapshots most of the time since, I can't
> really say when exactly this was introduced.
> 
> > Hmm.  On second thought, this might still be an LDAP problem.  I have a
> > hunch, and it's related to
> >
> >   https://cygwin.com/cygwin-ug-net/ntsec.html#ntsec-mapping-nfs
> >   https://cygwin.com/cygwin-ug-net/ntsec.html#ntsec-mapping-samba
> >
> > But we need to discuss and debug this.  From me crummy little home
> > network I have a hard time to see issues like that.
> 
> Me too, I usually only see this when I'm on business travel, although I
> can use the VPN from home.

Would be cool if you could test that.  Assuming you're not accessing
NFS shares, it may be the Samba stuff.  The culprits are probably
the calls to convert_samba_sd in check_file_access and the cygpsid::get_id
method which explicitely checks for S-1-22 SIDs and tries to convert them
to a Windows account per RFC 2307.

If that's the problem, it might be prudent to switch this off by default
and to enable that only via a setting in /etc/nsswitch.conf.

> > One bug report at a time would be nice...  This could be related to the
> > changes in tty read/write related to
> > https://cygwin.com/ml/cygwin-developers/2014-11/msg00000.html
> > but I have only 24h per day for hacking and debugging, too.  Sans sleep,
> > sorry.
> 
> Understood.  The timing seems about right, I first noticed this
> beginning of December.
> 
> > Yep.  We should contemplate to allow sending a signal to cygserver to
> > invalidate its cache.  Somebody would have to code that.  Sigh.
> 
> Yes, yes. :-) In the meantime, restarting the service manually does
> work, unfortunately I can't add this to the script that opens the VPN
> tunnel.

Erm... if you can't add *this* to the script, you wouldn't be able to
add the kill -USR1 call to the script either, isn't it?


Corinna

P.S.: IRC?

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

Attachment: pgppl8Y5Dp1bz.pgp
Description: PGP signature