Linux-Development-Sys Digest #654, Volume #7 Mon, 6 Mar 00 03:13:19 EST
Contents:
Re: What's GNU/Linux? (Peter Samuelson)
How to determine the Maximum nymber of system call per seconds?
([EMAIL PROTECTED])
Re: 'file' command source (Peter Samuelson)
mkfs.ext2? (Weiguang Shi)
Re: unmmaped area? (Takeyasu Wakabayashi)
Re: GCC and EGCS, SuSE say they can co-exist, I don't see how (Peter Samuelson)
Re: GCC and EGCS, SuSE say they can co-exist, I don't see how (Peter Samuelson)
Re: mkfs.ext2? (Peter Samuelson)
Re: mkfs.ext2? (David T. Blake)
Re: Binary compatibility: what kind of crack are they smoking? (Nix)
Re: Binary compatibility: what kind of crack are they smoking? (Nix)
Re: Binary compatibility: what kind of crack are they smoking? (Nix)
Re: Binary compatibility: what kind of crack are they smoking? (Nix)
Re: Binary compatibility: what kind of crack are they smoking? (Nix)
Re: glibc development language (Nix)
Re: What's GNU/Linux? (Alexander Viro)
Re: Binary compatibility: what kind of crack are they smoking? (Alexander Viro)
Solaris NFS incompatibility ("Scott L. Burson")
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: What's GNU/Linux?
Date: 5 Mar 2000 22:00:05 -0600
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[Christopher Browne <[EMAIL PROTECTED]>]
> If the FSF had decided to create a Linux distribution, and called it
> GNU/Linux, this would have been well and fine.
[...]
> I think he'd accomplish more by "fighting" other battles.
Slashdot doesn't usually yield a lot of signal, but occasionally
someone says something quotable:
"I call it GNU/Linux. Except the GNU/ is silent."
Ben Reiter <[EMAIL PROTECTED]>
Peter
------------------------------
From: [EMAIL PROTECTED]
Subject: How to determine the Maximum nymber of system call per seconds?
Date: Mon, 06 Mar 2000 04:24:14 GMT
How to determine the Maximum nymber of system call per seconds?
Is there any formular I can use to get this value??
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Subject: Re: 'file' command source
Date: 5 Mar 2000 22:33:44 -0600
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[Mark Barlow <[EMAIL PROTECTED]>]
> Don't suppose anyone knows which file contains the source for the
> 'file' command which uses a magic file to guess filetypes??
Debian mirrors never fail.[tm]
http://ftp.uk.debian.org/debian/dists/frozen/main/source/utils/file_3.28.orig.tar.gz
ftp://ftp.uk.debian.org/debian/dists/frozen/main/source/utils/file_3.28.orig.tar.gz
Peter
------------------------------
From: Weiguang Shi <[EMAIL PROTECTED]>
Subject: mkfs.ext2?
Date: Sun, 5 Mar 2000 21:36:54 -0700
Hi, all:
Could you please tell me where to get the source code of mkfs.ext2? It
seems it is not in the util-linux-2.10 package.
Thanks a lot.
Weiguang
------------------------------
From: Takeyasu Wakabayashi <[EMAIL PROTECTED]>
Subject: Re: unmmaped area?
Date: 06 Mar 2000 13:53:20 +0900
Fabrice Peix <[EMAIL PROTECTED]> writes:
> I thing you are wrong , paging system is made bye processor and all the
> memory
> is manage by MMU and so paging...
Then why do you have `struct page' structure and `mem_map' list
made thereof in the Linux kernel?
--
Takeyasu Wakabayashi
Faculty of Economics, Toyama University
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Subject: Re: GCC and EGCS, SuSE say they can co-exist, I don't see how
Date: 5 Mar 2000 22:58:38 -0600
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[Markus Kossmann <[EMAIL PROTECTED]>]
> > To compile the kernel, edit the toplevel kernel Makefile and add that
> > switch to the line
> > CC =$(CROSS_COMPILE)gcc -D__KERNEL__ -I$(HPATH)
[Martin von Loewis <[EMAIL PROTECTED]>]
> Alternatively, after installing 2.7.2.3, you can rename the front-end
> to, say, kgcc. Then you install egcs, and 'gcc' will be egcs; 'kgcc'
> will continue to be 2.7.2.3. If you set CROSS_COMPILE to 'k', you
> don't even need to edit the Makefile - but you'll also need 'kas' and
> 'kld'.
...And possibly kobjcopy, kobjdump, etc.
Or use Linux 2.3 (or wait for 2.4), where $(CC) no longer needs any
flags. (I wrote that patch for exactly this reason.) Then you can
just say something like
make bzImage CC=gcc272 # works on Debian
make bzImage CC='gcc -V2.7.2.3' # should work anywhere
Peter
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Subject: Re: GCC and EGCS, SuSE say they can co-exist, I don't see how
Date: 5 Mar 2000 23:02:42 -0600
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[Lee Reynolds <[EMAIL PROTECTED]>]
> Why use different compilers? I downloaded and compiled gcc 2.95.2
> and I use it for everything. I've never had any problems and I
> compile all kinds of things, including the 2.3.x kernels.
Do you use it for 3c59x.o on 2.2.14? Does your network card behave
strangely?
egcs 1.1.2 is generally considered "safer" for 2.2 kernels. 3c59x is
one known instance of breakage with 2.95; there could be other unknown
bugs. Yes, they are fixed as soon as found, but if you want a stable
2.2.x system you may wish to use an older compiler for it, just in
case.
Peter
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Subject: Re: mkfs.ext2?
Date: 5 Mar 2000 23:12:48 -0600
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[Weiguang Shi <[EMAIL PROTECTED]>]
> Could you please tell me where to get the source code of mkfs.ext2? It
> seems it is not in the util-linux-2.10 package.
$ dpkg -S mkfs.ext2
e2fsprogs: /sbin/mkfs.ext2
e2fsprogs: /usr/share/man/man8/mkfs.ext2.8.gz
Therefore:
http://http.us.debian.org/debian/dists/frozen/main/source/base/e2fsprogs_1.18.orig.tar.gz
ftp://ftp.us.debian.org/debian/dists/frozen/main/source/base/e2fsprogs_1.18.orig.tar.gz
Peter
------------------------------
From: [EMAIL PROTECTED] (David T. Blake)
Subject: Re: mkfs.ext2?
Date: 6 Mar 2000 05:23:35 GMT
Reply-To: [EMAIL PROTECTED]
Weiguang Shi <[EMAIL PROTECTED]> wrote:
> Hi, all:
> Could you please tell me where to get the source code of
> mkfs.ext2? It seems it is not in the util-linux-2.10 package.
> Thanks a lot.
e2fsprogs
And on my box it is called /sbin/mke2fs
--
Dave Blake
[EMAIL PROTECTED]
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 05 Mar 2000 23:17:19 +0000
[EMAIL PROTECTED] (Mario Klebsch) writes:
> Err, WinNT is UNIX,
WTF?
WinNT is POSIX-compliant if and only if you take the most restricted and
impoverished subset of POSIX you can find; it most definitely is not a
UNIX. Internally it is more like a sort of sick distorted warmed-over
VMS.
--
`> KNOWLEDGE AND SKILLS
You must have some, but I don't see any evidence of it.'
--- Craig Hardie flames a luser recruitment consultant
advertising `Microsoft based solutions' on uk.comp.os.linux
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 05 Mar 2000 23:35:10 +0000
[EMAIL PROTECTED] (Mario Klebsch) writes:
> The ABI should be a core pice of the OS, since it is the basic
> requirement for binary distribution. This should not be tied to a
> compiler.
Er, the compiler generates code to an ABI; the ABI's requirements are
set by the requirements of the compiler.
It can't be tied to anything *other* than the compiler!
(FWIW the C++ ABI should be settling down fairly soon. One hopes a last
change for gcc-3.0 is all that will be required...)
--
`> KNOWLEDGE AND SKILLS
You must have some, but I don't see any evidence of it.'
--- Craig Hardie flames a luser recruitment consultant
advertising `Microsoft based solutions' on uk.comp.os.linux
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 05 Mar 2000 23:37:47 +0000
[EMAIL PROTECTED] (Mario Klebsch) writes:
> So, please tell me, how can I recompile my applix office suite, the
> Adabas-D database or acroread? I tried my best, but I was not able to
> find the source code.
Serves you right for closed-source software, then.
(Well, you're crossposting this to comp.os.linux.development.system,
what kind of response do you *expect*? The only way of trolling more
effectively would be to add gnu.misc.discuss to the crosspost.
Furrfu.)
--
`> KNOWLEDGE AND SKILLS
You must have some, but I don't see any evidence of it.'
--- Craig Hardie flames a luser recruitment consultant
advertising `Microsoft based solutions' on uk.comp.os.linux
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 05 Mar 2000 23:41:54 +0000
[EMAIL PROTECTED] (Mario Klebsch) writes:
> IMHO this should have changed years ago. Linux should change from
> being just a kernel to bein an entire operating system. Put the big
> players in the linux distribution market probably have no interest is
> this, because they would loose their position. :-(
Good god, not just a troll, a conspiracy theorist too.
--
`> KNOWLEDGE AND SKILLS
You must have some, but I don't see any evidence of it.'
--- Craig Hardie flames a luser recruitment consultant
advertising `Microsoft based solutions' on uk.comp.os.linux
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 05 Mar 2000 23:58:30 +0000
[EMAIL PROTECTED] (Mario Klebsch) writes:
> I do not see the argument about C++. The only C++ compiler widely used
> on Linux systems is gcc, and it shlould be sufficient to get stable
> calling and name mangling conventions.
Oh, my god, you really don't get it, do you.
Issues:- name mangling (as you said)
- class layout
- RTTI slot location and layout
- virtual base class handling
- exception throwing, especially from shared libraries,
across C code; the C code may be compiled by non-gcc
compilers and the shared library loader may not be GNU
(gcc runs in proprietary vendor environments too, and
should not stop working there)
- class emission semantics (i.e. where to emit the vtables &c
for classes)
- intercalling with Java/C/ObjC (yes, this may require ABI
changes; although I doubt it I don't know much about this
area)
- and doubtless more stuff I've forgotten.
This is *not* a small problem, and your brushing it off shows how little
you really know about this area. I was wondering if you had some cause
to be so condescendingly arrogant; maybe you knew something the rest of
us did not. It unfortunately appears that this is not the case :(
> The remaining unstability of the ABI is the contents of the
> library.
The arbitrarily large set of libraries, mostly under active and
continuous development.
> An ABI defines a required minimum, and it really should be
> possible to agree upon this.
Yes. POSIX/Unix98, except that there isn't yet support for POSIX
shm/ipc/msg in place. This is coming.
Also, there is no support for Ritchie's Streams I/O system, and it
doesn't look like there will ever be (apparent consensus; lovely and
elegant, shame about the performance hit).
> However, I see that the use of inline
> functions and templates does put library implementation details into
> user code. :-( I have no answer to this problem.
Templates will be done with `export'; God alone knows how that'll be
implemented. Mark Mitchell or another of the true gcc C++ gurus may
know. I haven't a clue.
> So, things are going to get really worse, if the first binary only
> programs for kde appear, aren't they?
They cannot. That would violate the GPL, unless I am much mistaken;
they'd be linking against GPLed libraries, while not themselves being
GPLed.
Likewise binary-only GNOME programs.
--
`> KNOWLEDGE AND SKILLS
You must have some, but I don't see any evidence of it.'
--- Craig Hardie flames a luser recruitment consultant
advertising `Microsoft based solutions' on uk.comp.os.linux
------------------------------
From: Nix <$}xinix{[email protected]>
Subject: Re: glibc development language
Date: 06 Mar 2000 00:16:55 +0000
[EMAIL PROTECTED] (Kaz Kylheku) writes:
> Secondly, the GNU libc does not require the GNU compiler. This library supports
> systems other than Linux: installations of these systems may not have the GNU
> compiler suite, just some bundled compiler.
He's talking about compilation. It's a lost cause trying to *compile*
glibc with any compiler other than gcc, although you can use it with
lesser compilers.
> Also note that the library headers have the __P () wrapper around parameter
> list declarations.
This is in the headers, though, which have to be inter-compiler portable.
> And here you are talking about going to C++!
Crazy, I admit. The ABI is still changing, the optimizer isn't yet as
good as for C (although whole-function trees should help fix *that*),
it's only been standardized for a year, and most free software
developers --- certainly most working on low-level projects like glibc
or the kernel --- prefer C, sometimes vehemently (eg Dave Miller).
> >The "C++ slower than C" argument is quite old - dating back to the early
> >nineties.
... because it's true. Not *inherently*, but with current compilers.
> >Most programmers use C++ or Java nowadays.
>
> You can't use Java for serious system development, so you can forget about that
> one.
On top of that, arguments from popularity are inherently flawed
(although `if you switch to using language {foo} for project {bar} all
project {bar}'s developers will leave' is not a flawed argument IMNSHO).
--
`> KNOWLEDGE AND SKILLS
You must have some, but I don't see any evidence of it.'
--- Craig Hardie flames a luser recruitment consultant
advertising `Microsoft based solutions' on uk.comp.os.linux
------------------------------
From: [EMAIL PROTECTED] (Alexander Viro)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: What's GNU/Linux?
Date: 6 Mar 2000 02:25:45 -0500
In article <[EMAIL PROTECTED]>,
Edward Rosten <[EMAIL PROTECTED]> wrote:
>I agree with RMS on this one. Most of what we know as 'Linux' is
>actually GNU stuff.
<sarcasm>
Like X, for one, right? Or Perl. Or TeX. Or MTA-of-your-choice.
Or BIND. Or trn. Or <arbitrary-network-related-stuff>. Or vi. Or...
</sarcasm>
>The kernel is only one bit of the OS.
Sure. And you know what? About the only piece of GNU code we really depend
on is gcc, with GNU make as distant second. Basic utilities? No Thanks.
Look at them someday. As in "read through the code". And then tell whether
you needed a barf-bag. Bloated, overdesigned, trying to outsmart the kernel/
users/everything, full of ifdefs... <bletch>
>Calling
>GNU/Linux Linux kind of takes away the recognition of the FSF.
Umhm. You know, in their place I would be ashamed of such code, but tastes
differ...
>Many
>people think that Linus did the whole lot (more or less), and whilts I
>think that what Linus has done is fantastic, it is worth recognising
>that much of it had already been done - the FSF was just looking for a
>kernel.
FSF would beg to differ... They were (are) making their own kernel.
--
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid. Get yourself a better computer" - Dilbert.
------------------------------
From: [EMAIL PROTECTED] (Alexander Viro)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 6 Mar 2000 02:35:04 -0500
In article <[EMAIL PROTECTED]>,
Nix <$}xinix{[email protected]> wrote:
>Also, there is no support for Ritchie's Streams I/O system, and it
>doesn't look like there will ever be (apparent consensus; lovely and
>elegant, shame about the performance hit).
Erm... AFAICS it's rather "_Ritchie's_ variant of STREAMS is lovely and
elegant, but nobody cares about it; the only thing really wanted is the
abortion USG made out of it and _here_ performance hit is nasty (along
with many other things)". IOW, Missed'em'V STREAMS are seriously heavier
(and less elegant), but everybody who wants STREAMS asks for them, not
for original thing. And yes, it's a fscking shame.
--
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid. Get yourself a better computer" - Dilbert.
------------------------------
From: "Scott L. Burson" <[EMAIL PROTECTED]>
Subject: Solaris NFS incompatibility
Date: Sun, 05 Mar 2000 23:59:03 -0800
Greetings all,
It has become clear to me that there is a subtle incompatibility between
the Linux 2.2 NFS client and the Solaris NFS server (at least, the one
in Solaris 2.5.1). What I have observed is occasional object file
corruption during large C builds, when the build is being done over NFS
to the Solaris server. I first noticed this with 2.2.5 (RedHat
6.0/SPARC) and have now seen it also with 2.2.13 (SuSE 6.3/Intel). I
have not, repeat not, seen it with 2.0.30 (RedHat 4.2/SPARC) or 2.0.35
(RedHat 5.2/SPARC). It doesn't happen when building on a local
filesystem, of course, nor have I observed it with a 2.2.13 client and a
different server running Linux 2.0.35.
In more detail: one of the first things I do when setting up a new OS is
to build Kerberos (currently 1.0.6). This is a pretty large build, with
some 800 C source files; the resulting build tree contains some 130MB of
object files, libraries, and executables. What I have observed with the
2.2 kernels is that I will get roughly halfway through the build before
`ld' or `ar' chokes (SEGV, typically) on one of the object files that
has been written out. Once this happens, attempts to restart the build
(by just running `make' again without changing anything) always crash in
the same way, suggesting that the problem has to do with the NFS write
rather than the subsequent read (i.e. the object file is actually
corrupted and is not just being read incorrectly).
I have seen it happen twice in the same place in the build (in fact, it
did that to me the first two times I tried it, and I thought I had
encountered some bizarre bug in EGCS). But I just tried it three times
with this 2.2.13 machine I'm installing, and it crashed in different
places each time. Interestingly, the crash points do not seem to be
uniformly distributed through the build; so far they are all between 1/2
and 3/4 of the way through. I have no idea what that could mean.
While this problem is not a total show-stopper for me, it's fairly
serious: it means I can't trust anything written from this 2.2.13
machine to the Solaris server. So I'm fairly motivated to help track it
down. But I don't really know anything about the guts of NFS, so the
best I can offer is to perform various experiments under direction of
one of the Linux implementors. Or maybe this is a known problem, fixed
in a more recent kernel?
-- Scott
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.development.system) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Development-System Digest
******************************
