Bill Studenmund | 1 Apr 2005 03:40
Picon

Re: upgrading 2.99.14

On Wed, Mar 30, 2005 at 11:56:52AM +0200, Johnny Billquist wrote:
> On Tue, 29 Mar 2005, Rick Kelly wrote:
> 
> > Martti Kuparinen said:
> >
> > >cvs update -dPA			# 3.99.x
> >
> > So 3.99.x is -current? How the heck did it get there from 2.99.xx?
> 
> That is the new order of the day.

Yes, this is the first release where we are doing it "right." 2.99.xx came 
a while after 2.0 was branched..

> Appearantly 3.0 have already been cut, and are now groomed for release.
> Though I haven't seen or heard anything about it yet. :-)

As others have mentioned, it's still getting started.

Take care,

Bill
John Nemeth | 1 Apr 2005 04:52
Picon
Favicon

OpenSSH and getpw*

     I did a sweep of the entire source tree looking for things that
use getpw*.  Right now, I'm looking at OpenSSH.  It is so convoluted
that it will be difficult to determine if its use of getpw* will
conflict with its use of PAM.  Does somebody else want to look at it?
Should I just change every use of getpw* in the source of sshd to
getpw*_r?  Should we just ignore OpenSSH now that the sweep of all the
PAM modules is done?

NetBSD source update | 1 Apr 2005 05:29
Picon

daily CVS update output


Updating src tree:
P src/distrib/utils/libhack/getpwent.c
P src/distrib/utils/libhack/syslog.c
P src/doc/CHANGES
P src/doc/CHANGES.prev
P src/gnu/lib/libstdc++-v3/arch/alpha/c++config.h
P src/gnu/lib/libstdc++-v3/arch/alpha/config.h
P src/gnu/lib/libstdc++-v3/arch/arm/c++config.h
P src/gnu/lib/libstdc++-v3/arch/arm/config.h
P src/gnu/lib/libstdc++-v3/arch/armeb/c++config.h
P src/gnu/lib/libstdc++-v3/arch/armeb/config.h
P src/gnu/lib/libstdc++-v3/arch/hppa/c++config.h
P src/gnu/lib/libstdc++-v3/arch/hppa/config.h
P src/gnu/lib/libstdc++-v3/arch/m68000/c++config.h
P src/gnu/lib/libstdc++-v3/arch/m68000/config.h
P src/gnu/lib/libstdc++-v3/arch/m68k/c++config.h
P src/gnu/lib/libstdc++-v3/arch/m68k/config.h
P src/gnu/lib/libstdc++-v3/arch/mipseb/c++config.h
P src/gnu/lib/libstdc++-v3/arch/mipseb/config.h
P src/gnu/lib/libstdc++-v3/arch/mipsel/c++config.h
P src/gnu/lib/libstdc++-v3/arch/mipsel/config.h
P src/gnu/lib/libstdc++-v3/arch/ns32k/c++config.h
P src/gnu/lib/libstdc++-v3/arch/ns32k/config.h
P src/gnu/lib/libstdc++-v3/arch/powerpc/c++config.h
P src/gnu/lib/libstdc++-v3/arch/powerpc/config.h
P src/gnu/lib/libstdc++-v3/arch/sh3eb/c++config.h
P src/gnu/lib/libstdc++-v3/arch/sh3eb/config.h
P src/gnu/lib/libstdc++-v3/arch/sh3el/c++config.h
P src/gnu/lib/libstdc++-v3/arch/sh3el/config.h
(Continue reading)

John Nemeth | 1 Apr 2005 09:36
Picon
Favicon

getpwent_r()

     libedit uses getpwent() to enumerate all users starting with a
particular string.  I think this means that we are going to need a
getpwent_r() function.  Solaris 8 has one that looks like this:

     struct passwd *getpwent_r(struct passwd *pwd, char  *buffer,
     int buflen);

     For enumeration in multithreaded applications, the  position
     within  the enumeration is a process-wide property shared by
     all threads. The setpwent() function may be used in  a  mul-
     tithreaded  application  but resets the enumeration position
     for all threads.  If multiple threads  interleave  calls  to
     getpwent_r(), the threads will enumerate disjoint subsets of
     the password database.

libedit also uses setpwent() and endpwent() which presents another
problem.  I don't see any simple solutions to the enumeration position
problem, so we'll probably just have to document that it does this.

John Nemeth | 1 Apr 2005 09:49
Picon
Favicon

Re: getpwent_r()

On Aug 21,  6:12pm, John Nemeth wrote:
}
}      libedit uses getpwent() to enumerate all users starting with a
} particular string.  I think this means that we are going to need a
} getpwent_r() function.  Solaris 8 has one that looks like this:
} 
}      struct passwd *getpwent_r(struct passwd *pwd, char  *buffer,
}      int buflen);

     On the other hand, FreeBSD has one that looks like:

     int
     getpwent_r(struct passwd *pwd, char *buffer, size_t bufsize,
	 struct passwd **result);

}-- End of excerpt from John Nemeth

Zbigniew Baniewski | 1 Apr 2005 13:14
Picon

Resource needs

I'm comparing NetBSD with Debian Linux running both on the (almost) same,
poorly equipped machines (TX, K6-2/300, 64 MB RAM). NetBSD V2.0, with kernel
recompiled to suit the equipment, and Debian Woody with kernel 2.2.26.

It seems, that NetBSD needs more memory (about twice as much?) - just
because, f.e. while on the Debian system I can use "full" Mozilla (with
very little swapping), ran in IceWM - on the NetBSD, using Firefox (which
is less "heavy" than Mozilla, and in fvwm2 environment) I notice regular
swapping. When trying to make use out of OpenOffice, I can still almost
normally work on the Debian system (although it's not especially
comfortable), but under NetBSD it's impossible - heavy swapping while in
editing mode of OOWriter (not even touching the keyboard).

And so I would to obtain some info: how can be the resource-needs of NetBSD
compared to that of Linux (and/or FreeBSD)?
--

-- 
				pozdrawiam / regards

						Zbigniew Baniewski

Piotr Meyer | 1 Apr 2005 13:14
Picon
Gravatar

Re: Resource needs

On Fri, Apr 01, 2005 at 01:14:13PM +0200, Zbigniew Baniewski wrote:

[...]
> comfortable), but under NetBSD it's impossible - heavy swapping while in
> editing mode of OOWriter (not even touching the keyboard).

You may try some VM-tuning, like described on:
http://www.selonen.org/arto/netbsd/vm_tune.html

--

-- 
Piotr 'aniou' Meyer

Zbigniew Baniewski | 1 Apr 2005 13:42
Picon

Re: Resource needs

On Fri, Apr 01, 2005 at 01:14:23PM +0200, Piotr Meyer wrote:

> You may try some VM-tuning, like described on:
> http://www.selonen.org/arto/netbsd/vm_tune.html

Thanks, I'll take a look at it - but I'm not going to say, that's something
wrong with NetBSD; just looking for some more info.

Of course, when running f.e. FreeDOS, I'll have the system which will take
much less resources.
--

-- 
				pozdrawiam / regards

						Zbigniew Baniewski

Martijn van Buul | 1 Apr 2005 13:45
Picon

Re: Resource needs

It occurred to me that Zbigniew Baniewski wrote in gmane.os.netbsd.current:

> Of course, when running f.e. FreeDOS, I'll have the system which will take
> much less resources.

Trolling again, aren't we?

Chris Wareham | 1 Apr 2005 15:04
Picon
Picon

Re: Resource needs

Zbigniew Baniewski wrote:
> I'm comparing NetBSD with Debian Linux running both on the (almost) same,
> poorly equipped machines (TX, K6-2/300, 64 MB RAM). NetBSD V2.0, with kernel
> recompiled to suit the equipment, and Debian Woody with kernel 2.2.26.
> 

Linux used to be praised for its performance on modest hardware,
especially when compared to MS Windows. However, after the 2.2.x series,
Linux development has been focused on support for 'server' hardware.
This has resulted in a considerable increase in resource usage to
support the features deemed necessary for an 'enterprise' class
operating system. More subjectively, I'd argue that the chaotic nature
of Linux kernel development (a large number of personal developer trees
and vendor versions) has had a negative impact on the size and quality
of the kernel.

As a result, I'd suggest that comparing Linux 2.2.x with its relatively
modest feature set to NetBSD 2.0 is a little unfair.

> It seems, that NetBSD needs more memory (about twice as much?) - just
> because, f.e. while on the Debian system I can use "full" Mozilla (with
> very little swapping), ran in IceWM - on the NetBSD, using Firefox (which
> is less "heavy" than Mozilla, and in fvwm2 environment) I notice regular
> swapping. When trying to make use out of OpenOffice, I can still almost
> normally work on the Debian system (although it's not especially
> comfortable), but under NetBSD it's impossible - heavy swapping while in
> editing mode of OOWriter (not even touching the keyboard).
> 
> And so I would to obtain some info: how can be the resource-needs of NetBSD
> compared to that of Linux (and/or FreeBSD)?
(Continue reading)


Gmane