Linux-Development-Sys Digest #668, Volume #6 Sat, 1 May 99 21:14:04 EDT
Contents:
Re: Q: Notification about termination of parent process. (David Wragg)
Help installing Red Hat 5.2 on Dell PowerEdge SP 5166 (Leslie Smith)
Re: stdio SMBd - name your price (Jeremy Allison)
Re: Possible?: glibc-2 system and (old?) Motif libraries (Remco van den Berg)
Re: CVS (Re: Bill Gates, self made man, NOT!) ("G. Sumner Hayes")
Re: Thing about Linux you do not know... (Mark Tranchant)
Re: stdio SMBd - name your price (Kyler Laird)
kernel service to contro process cpu time (Pablo Osinaga)
system() in glibc (Pablo Osinaga)
Re: Unix98 ptys and glibc-2.0 (Juergen Heinzl)
Re: how to access raw memory? (Kent Friis)
kernel crash in bttv driver (Joseph Dane)
Re: CVS (Re: Bill Gates, self made man, NOT!) (H. Peter Anvin)
Re: Q: Notification about termination of parent process. (Jan D.)
Re: system() in glibc (Juergen Heinzl)
Re: Possible?: glibc-2 system and (old?) Motif libraries (Juergen Heinzl)
kernel 2.2.6 - Oops then Kernel Panic ([EMAIL PROTECTED])
Re: stdio SMBd - name your price (Jeremy Allison)
----------------------------------------------------------------------------
From: David Wragg <[EMAIL PROTECTED]>
Subject: Re: Q: Notification about termination of parent process.
Date: 01 May 1999 13:26:14 +0000
Nils Henrik Lorentzen <[EMAIL PROTECTED]> writes:
> What I want is the
> subprocess to terminate, when the
> parent process dies (because of user pressing quit or ctrl-c or it
> segfaults
> or whatever). The way I currently do this is by in the
> subprocess regularly checking if getppid() == 1,
> but this is not a very elegant solution IMO.
> Does anyone know a better way to handle this ?
Yes, but only in 2.2. It adds a prctl system call which supports this,
with something like:
#include <syscall.h>
#include <linux/prctl.h>
_syscall2(int, prctl, unsigned long, opt, unsigned long, arg)
...
prctl(PR_SET_DEATHSIG, SIGKILL);
(You'll need 2.2 headers installed for that to compile).
On 2.0 you have to poll, but to take advantage of 2.2 you can try
prctl and only poll if it returns -1 with errno == ENOSYS
(Perhaps glibc2.1 adds a prctl wrapper function, making the syscall
part unnecessary).
Dave Wragg
------------------------------
From: [EMAIL PROTECTED] (Leslie Smith)
Subject: Help installing Red Hat 5.2 on Dell PowerEdge SP 5166
Date: 1 May 1999 13:35:52 +0100
Can help me out here, I have got a Dell PowerEdge server
SP 5166 Dual CPU.
I'm using Red Hat 5.2, when I am using the boot disk to boot
the machine I only get three dot's then the machine stops dead,
this also happens with Red Hat 5.1 too. I am able to load Red
Hat 5.0 OK.
Machine Spec:
CPU = 2 x 166
SCSI HD = 1 x 1GB
2 x 2GB
1 x 450GB
MEM = 128MB
Display = 1 x Vodoo 4MB
REgards
Leslie...UK:-)
------------------------------
Crossposted-To: comp.protocols.smb,comp.unix.solaris
From: [EMAIL PROTECTED] (Jeremy Allison)
Subject: Re: stdio SMBd - name your price
Date: Sat, 1 May 1999 18:30:51 GMT
[EMAIL PROTECTED] (Kyler Laird) writes:
>>>[I've asked for this before, usually with a few hundred
>>>The result would, of course, be open to the public.
>>>(I'll leave licensing to the author.)
>>Well if it's based on the Samba code the licensing
>>would be GPL of course.....
>I didn't realize a diff would necessarily
>fall under the GPL.
Yes if it's based on Samba (GPL'ed) code, read the
GPL itself for details.
Jeremy Allison,
Samba Team.
------------------------------
From: Remco van den Berg <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: Re: Possible?: glibc-2 system and (old?) Motif libraries
Date: Sat, 01 May 1999 20:39:26 +0200
Chris Rankin wrote:
>
> Remco van den Berg wrote:
> > When I try to compile motif applications like Nedit I get linker errors
> > of undefined references:
> >
> > libXm.so: undefined reference to '_xstat'
> > '_fxstat'
> > '_Xsetlocale'
>
> Hummm..
> I don't know if this will work, but off the top of my head I can't see
> why not:
>
> Add these commands to the VERY END of your link line:
>
> -lc /usr/lib/libc.a
>
> Hopefully, this will cause your compiler to use libc.so for as much as
> it can, and resort to the static archive for everything else. As I
> understand it, the _fxstat symbol is private to the shared library but
> must still be in the static archive somewhere.
>
> Cheers,
> Chris.
This works for _xstat and _fxstat, but it stil has problems finding
_Xsetlocale
-Remco
------------------------------
From: "G. Sumner Hayes" <[EMAIL PROTECTED]>
Subject: Re: CVS (Re: Bill Gates, self made man, NOT!)
Date: Sat, 01 May 1999 14:34:52 -0400
Arun Sharma wrote:
>
> "G. Sumner Hayes" <[EMAIL PROTECTED]> writes:
> > Bitkeeper is Larry Wall's attempt...
>
> You mean Larry McVoy ?
Ugh. Yes, of course, Larry McVoy. *sigh* Sometimes the fingers go
faster than the brain.
> Larry has a comparision on his website. Apart
> from that I don't know of anything negative about CVS. Some people
> dislike it's tendency to "automerge" things etc.
I have no strong personal opinion on the matter. Linus does,
apparently, and that's what matters.
--Sumner
------------------------------
From: Mark Tranchant <[EMAIL PROTECTED]>
Subject: Re: Thing about Linux you do not know...
Date: Fri, 23 Apr 1999 13:12:40 +0100
Reply-To: [EMAIL PROTECTED]
I think we should change the OS's name to LIN����X - much catchier.
Mark.
Gordon Scott wrote:
>
> Considering this spam has been posted about once a week for a couple of
> years now, I fid it very hard to believe that there's anyone who doesn't
> know about it. Including the views of those people who've used it.
>
> I'm prepared to consider applications that are announced in the appropriate
> places and in the appropriate ways. I, and I think many others, will
> mot use spammed products on principle.
>
> : YO���� CAN START ����SING WORKING, BREATHING LIN����XCAD IN
> : 24 HO����RS AFTER PAYING BY CREDIT CARD !!!
>
> Come on, spit it out. What are you trying to say here.
>
> --
> Gordon Scott Opinions expressed are my own.
> [EMAIL PROTECTED] (official) [EMAIL PROTECTED] (backdoor)
> [EMAIL PROTECTED] (home) http://www.apis.demon.co.uk
> Linux ............... Because I like to _get_ there today.
------------------------------
From: [EMAIL PROTECTED] (Kyler Laird)
Crossposted-To: comp.protocols.smb,comp.unix.solaris
Subject: Re: stdio SMBd - name your price
Date: 1 May 1999 21:05:18 GMT
[EMAIL PROTECTED] (Jeremy Allison) writes:
>>I didn't realize a diff would necessarily
>>fall under the GPL.
>Yes if it's based on Samba (GPL'ed) code, read the
>GPL itself for details.
Neat. We can take that to silly extremes.
This must be GPL'ed because it's based on Samba
code:
s/it's name/its name/g;
s/read it's old/read its old/g;
s/it's segment/its segment/g;
s/it's local name/its local name/g;
s/it's own/its own/g;
s/it's size/its size/g;
.
.
.
--kyler
------------------------------
From: Pablo Osinaga <[EMAIL PROTECTED]>
Subject: kernel service to contro process cpu time
Date: Sat, 01 May 1999 17:16:50 +0000
im having some troubles with processes taking over the whole cpu ussage.
is any kernel machanism to know when a process is taking more than X% of
the CPU.
My idea is not to limit the CPU time that the process can use, but
warning me in some way when the process is taking so much resources.
(something like a signal). ??
any solutions?
------------------------------
From: Pablo Osinaga <[EMAIL PROTECTED]>
Subject: system() in glibc
Date: Sat, 01 May 1999 17:25:15 +0000
there is a problem with system in new glibc versions.
when i do system() in a process, the new created process has problems
with signals,
in practice:
// dosomething.c
void handler()
{
printf("signal received\n");
system("dosomething &");
}
void main(void)
{
signal(SIGIO, handler);
pause();
}
if i do:
$ dosomething &
and then 'killall -SIGIO dosomething'
a new 'dosomething' process is created but when i do
killall -SIGIO dosomething
nothing is done, 'dosomething' doesn't even die. neither print nor
system any stuff....
i've fixed this recompiling with older versions of libc, and it worked
ok, but i need to compile with glibc due to some other restrictions...
what can i do? i've also tried the fork-exec method with the same
result...
any solutions from you?
------------------------------
From: [EMAIL PROTECTED] (Juergen Heinzl)
Subject: Re: Unix98 ptys and glibc-2.0
Date: Fri, 23 Apr 1999 22:08:06 GMT
In article <[EMAIL PROTECTED]>, "Stefan Monnier wrote:
>>> And in the same vein, is it harmful to turn on Unix98 pty support in
>>> the kernel if I don't have glibc-2.1 installed yet ?
>> You cannot use them and so the main result is going to be a larger
>> kernel; pain without a gain.
>
>But if I don't plan on rebooting within the next year (while I do plan on
>trying out glibc-2.1 during this time)...
You can use glibc-2.1 quite fine without Unix98 kernel support, no
problem. I've to know, I'd that configuration running until four
days ago ... no, no crash, just upgraded to 2.2.6 + Unix98 PTY's.
Say if you want to use them you need glibc-2.x but not so the other
way round.
Cheers,
Juergen
--
\ Real name : J�rgen Heinzl \ no flames /
\ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: Kent Friis <[EMAIL PROTECTED]>
Subject: Re: how to access raw memory?
Date: Thu, 29 Apr 1999 20:28:17 +0200
"Jacek Pop=B3awski" wrote:
> =
> heh... the problem is, that I WANT to write module, but now - i have no=
> idea HOW to write ANY module, I am experienced with programming in
> Assembler (protected mode) under DOS, but Linux is another system...
> yes... i know.. i should read Module-HOWTO or LDP ;-) but I want begin
> from writing some demos in 320x200x256,
Writing demos shouldn't be done in a module.
> i can't do it with svgalib
> or X, becouse it's slow, big, and need suid, it's sad,
Take a look at ggi http://www.ggi-project.org/
It's fast, you can write (almost) directly to the videoram,
and it can use both X (slows it down), SVGAlib (still needs
setuid) and the new fbcon(linux 2.2) and kgi (linux 2.3?)
kernel drivers, that doesn't have any of these problems.
> that Linux don't
> has normal way to programming gfx... or maybe i don't know something?
> I want to access video ram or Vesa-framebuffer and gfx-card ports,
> and write gfx demo, if it was so easy in DOS - why it is so hard in Lin=
ux?
Becaus you are still thinking DOS. actually it's much easier in
linux/GGI - you don't need to write a single line of assembler[1]
> i need to know how to:
> a) set 320x200x256 mode
> b) copy 64000bytes from memory to videoram
> c) back to textmode
> steps a) and c) should be easy (of course i can be wrong), but I have n=
o idea
> how to make step b)
> please tell me what should i read
All the docs are at the link above
> and - what are the reasons that i should
> not write that driver/demo...
you shouldn't write the driver because other people are already
doing this, and it looks good already.
But go ahead and write your demo - and let us see it, when it
works.
Kent
[1] You don't want to use assembler in userlevel programs - it
won't work on that quad UltraSparc or Alpha you are going to buy
in a few years anyway.
------------------------------
From: Joseph Dane <[EMAIL PROTECTED]>
Subject: kernel crash in bttv driver
Date: Sat, 01 May 1999 11:42:52 -1000
Hi
Just installed a Hauppage WinTV/PCI board in a fairly old HP Vectra
system (Pentium, unknown board type, some sort of SIS chipset, 16 MB)
running an unpatched 2.2.6. I get a kernel oops message as soon as
capturing begins. I'm using the "streamer" app from the xawtv
distribution.
The kernel messages never seem to make it to the log file. I don't know
a hell of a lot about linux kernel hacking, but it seems like something
is trying to sleep in an interrupt, and the kernel doesn't think it wise
to try writing any log stuff to the disk at that point. In any case, I
tried to write down what seemed important off the screen when the crash
happened:
bttv0: PLL: 28636363 => 35468950 .... Scheduling in interrupt
Unable to handle kernel NULL pointer dereference at virtual address
00000000
*pde = 00000000
Oops = 0002
EIP : c010ee26
And from my System.map:
c010ebdc T schedule
c010ee3c T __wake_up
So, is this a known problem and is there a patch? Does it look from the
info above like someone called schedule() from within an interrupt? Or
is the problem that the driver tried to dereference a NULL pointer? Or
both? Am I right in thinking that this sort of error will not be logged
to the disk, or is there something simply wrong in my configuration?
Thanks for any info and/or pointers.
Joe
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: Re: CVS (Re: Bill Gates, self made man, NOT!)
Date: 1 May 1999 21:53:34 GMT
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Followup to: <[EMAIL PROTECTED]>
By author: Arun Sharma <[EMAIL PROTECTED]>
In newsgroup: comp.os.linux.development.system
>
> "G. Sumner Hayes" <[EMAIL PROTECTED]> writes:
>
> >
> > Linus really hates CVS because it messes with his way of doing
> > business. Bitkeeper is Larry Wall's attempt to make something that
> > draws on his experience with Sun and SGI development tools and is
> > conducive to the Linux development model.
> >
>
> You mean Larry McVoy ? Larry has a comparision on his website. Apart
> from that I don't know of anything negative about CVS. Some people
> dislike it's tendency to "automerge" things etc.
>
There are a number of things about CVS that suck:
1. Doesn't handle moved files well.
2. Doesn't maintain the notion of a single commit that touches
multiple files.
3. Doesn't handle deleted files well.
4. Doesn't handle added directories well.
5. Handles deleted directories *really* poorly.
6. The CVS client/server model is not very clean; CVS itself expects
each user to touch the repository (yuck!) and the client/server is
a graft-on; for cleanliness it really should have been built around
a client/server model from the start.
-hpa
--
"The user's computer downloads the ActiveX code and simulates a 'Blue
Screen' crash, a generally benign event most users are familiar with
and that would not necessarily arouse suspicions."
-- Security exploit description on http://www.zks.net/p3/how.asp
------------------------------
From: Jan D. <[EMAIL PROTECTED]>
Subject: Re: Q: Notification about termination of parent process.
Date: Sat, 01 May 1999 22:03:11 GMT
In article <[EMAIL PROTECTED]>,
David Wragg <[EMAIL PROTECTED]> wrote:
>
> (Perhaps glibc2.1 adds a prctl wrapper function, making the syscall
> part unnecessary).
Yes.
#include <sys/prctl.h>
Jan D.
------------------------------
From: [EMAIL PROTECTED] (Juergen Heinzl)
Subject: Re: system() in glibc
Date: Sat, 01 May 1999 23:11:33 GMT
In article <[EMAIL PROTECTED]>, Pablo Osinaga wrote:
>there is a problem with system in new glibc versions.
>when i do system() in a process, the new created process has problems
>with signals,
>
>in practice:
>
>// dosomething.c
>
>void handler()
>{
> printf("signal received\n");
> system("dosomething &");
>}
Using system() in a signal handler is anything but a good
idea. If you call system(), then send a SIGIO to the new
process it is killed, quite fine.
Cheers,
Juergen
--
\ Real name : J�rgen Heinzl \ no flames /
\ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: [EMAIL PROTECTED] (Juergen Heinzl)
Crossposted-To: comp.os.linux.development.apps
Subject: Re: Possible?: glibc-2 system and (old?) Motif libraries
Date: Sat, 01 May 1999 23:11:36 GMT
In article <[EMAIL PROTECTED]>, Remco van den Berg wrote:
>Juergen Heinzl wrote:
>>
>> In article <7gdjoh$nfa$[EMAIL PROTECTED]>, Peter Samuelson wrote:
>> >
>> > [Remco van den Berg]
>> >> > Is it possible to solve this problem with a linker trick, or do I
>> >> > have to buy a new Motif release?
[...]
>> >Or, if you are lucky enough to be running Debian (I don't know the
>> >situation with Red Hat et al...), install the `altgcc', `libc5' and
>> >`libc5-altdev' packages, along with whatever `libxxx-altdev' packages
>> >you need for the apps in question. Then compile with `linux-libc1-gcc'
>> >(or something like that) instead of `gcc'.
>>
>> Yup, it is just that then you do need all that libc5 stuff 8( ... I've
>> no distribution but Motif came in both flavours; perhaps a second look
>> on the CDROM results in a nice surprise ...
[...]
>Nope, I'm unlucky. The CD-Rom is far too old. The INSTALL file on the
>CD-Rom is talking about RedHat version 2 and 3, ha ha... :-)
>
>I think I'll have to buy a new Motif release....
Depends due to see above. You can use libc5 and libc6 on the same system
no problem, just a little fiddling around if it is not set up that way.
I'd that for some weeks, the libc5 compiler in /usr and the libc6 in
/usr/local ... problems: 0
Mind too that you'll probably find "only" Motif 2.x for glibc, just in
case you do need and older version. Even so, some come with the XRT
toolkit as a bonus, so it might be worth it anyway, take a look at
Metrolinks site and there is still lesstif.
Cheers,
Juergen
--
\ Real name : J�rgen Heinzl \ no flames /
\ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: [EMAIL PROTECTED]
Subject: kernel 2.2.6 - Oops then Kernel Panic
Date: Sun, 02 May 1999 00:40:23 GMT
Hi all,
does anyone have an idea as to what might be causing the following
error? (is it the kernel or could it be hardware related?)
My machine locks up in a way that I can't reproduce reliably (although
often at the same when I make a remote file system access (mounted
through
autofs)) (for all I know this is just a coincidence) I keep getting "nfs
warning: mount version older than kernel". This warning pops up when I
first mount my home directory which is being served up by another linux
box running kernel 2.0.35. (I have other boxes with older kernels
accessing
the same server with no problems)
Here's what comes to the console as the machine crashes:
Scheduling in interrupt
Unable to handle kernel NULL pointer dereference at virtual address
00000000
current->tss.cr3 = 00101000, %cr3 = 00101000
*pde = 00000000
Oops: 0002 (but I've seen 0007 also)
CPU: 0
EIP: 0010: [<c01122ae>]
EFLAGS: 00010286
eax: 00000018 ebs: c01e4000 ecx: c01e4000 edx: cf7dc000
esi: 0009e200 edi: c0106000 ebp: c01e5fdc esp: c01e5fc0
ds: 0018 es: 0018 ss: 0018
Process swapper: (pid: 0, process nr: 0, stackpage=c01e5000)
Stack: 0009e200 c0106000 00000000 00000018 c0110018 00000000 c01f1000
00009000
c0107352 00000000 c01e6b9d 00000000 c0106000 00000000 c01cc7e0
c01001b1
Call Trace: [<c0106000>] [<c0110018>] [<c0107352>] [<c0106000>]
[<c01001b1>]
Code: c7 05 00 00 00 00 00 00 00 00 8d 65 e0 5b 5e 5f 89 ec 5d c3
Aiee, killing interrupt handler
Kernel panic: Attempted to kill the idle task!
In swapper task - not syncing
After this the system locks up (although the ethernet card is still
replying to pings)
I'm running stock RedHat 5.1 but upgraded the kernel to 2.2.6 on:
dual 450MHz PII
SuperMicro P6DBE motherboard
6.4 GB Western Digital Hard drive
256MB
I've upgraded all the system libs and utilities as per the
Documents/Changes
file in the linux-2.2.6 distribution.
Many thanks for any help
Here's a bit more info about the CPU's from /proc/cpuinfo:
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 5
model name : Pentium II (Deschutes)
stepping : 2
cpu MHz : 451.032162
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
sep_bug : no
f00f_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat pse36 mmx osfxsr
bogomips : 448.92
processor : 1
vendor_id : GenuineIntel
cpu family : 6
model : 5
model name : Pentium II (Deschutes)
stepping : 2
cpu MHz : 451.032162
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
sep_bug : no
f00f_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat pse36 mmx osfxsr
bogomips : 450.56
Mike
===== Posted via Deja.com, The People=Powered Information Exchange =====
====== http://www.deja.com/ Discussions * Ratings * Communities ======
------------------------------
Crossposted-To: comp.protocols.smb,comp.unix.solaris
From: [EMAIL PROTECTED] (Jeremy Allison)
Subject: Re: stdio SMBd - name your price
Date: Sun, 2 May 1999 00:59:22 GMT
[EMAIL PROTECTED] (Kyler Laird) writes:
>[EMAIL PROTECTED] (Jeremy Allison) writes:
>>>I didn't realize a diff would necessarily
>code:
> s/it's name/its name/g;
> s/read it's old/read its old/g;
> s/it's segment/its segment/g;
> s/it's local name/its local name/g;
> s/it's own/its own/g;
> s/it's size/its size/g;
> .
> .
> .
Err, no. It's not silly. It's the same reason you
cannot take Linux, change parts of it and call it
Kylix (and refuse to release source code).
I take a very dim view of people attempting
to fence in intellectual property held in the common
trust (like GPL'ed code is). But that's maybe because I've
seen people attempt this before with GPL code :-).
If you started with Samba code, the resulting program
must be GPL'ed. It's as simple as that. If you want
any more detail than that, go see a lawyer.
Regards,
Jeremy Allison,
Samba Team.
------------------------------
** 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
******************************
