Linux-Development-Sys Digest #178, Volume #7 Fri, 10 Sep 99 02:13:57 EDT
Contents:
Re: LILO and System.map ("Ross Crawford")
Re: make linux disk only one ("Ross Crawford")
HELP with MAC sharing ("..Luca T..")
Re: Any floating point support in kernel (Juergen Heinzl)
Re: Any floating point support in kernel (Tony Angelo)
Re: IDE for c++ dev? (Warren Young)
Re: Any floating point support in kernel (Tony Angelo)
Re: Linux standards compliance (Warren Young)
Re: survey linux project. (Chris Gregory)
Re: Disabling control-alt-delete from a program (Fred Christiansen)
Re: loadkeys/compose (Peter Samuelson)
Re: pci libraries (Mark McDougall)
Where can I find the handy SWEEP utility? (Joey McAlerney)
Re: Embedded X-server anyone ? ? (Mario Klebsch)
Re: select() and write descriptors ("Bigwoof!")
Re: 497.2 days ought to be enough for everybody (David Schwartz)
device to /dev/sd* binding, and other Q's (Fred Christiansen)
----------------------------------------------------------------------------
From: "Ross Crawford" <[EMAIL PROTECTED]>
Subject: Re: LILO and System.map
Date: Fri, 10 Sep 1999 09:04:32 +1000
Bram Bouwens <[EMAIL PROTECTED]> wrote in message
news:7r0l1q$hen$[EMAIL PROTECTED]...
> Allin Cottrell <[EMAIL PROTECTED]> writes:
>
> >and also note that current klogd will happily read
> >/boot/System.map-<kernel version>, e.g.
> >/boot/System.map-2.2.12
> >which allows you to keep more than one map in play
> >if you wish.
>
> That's quite a useful remark!
> And if I would have several variants of the same version number?
>
> Bram Bouwens
Dunno, but you could investigate using the EXTRAVERSION or whatever it that
RedHat use, to create, for example, version 2.2.12-1, 2.2.12-2, etc. I'm not
sure if this is supported in System.maps, though.
Regards,
ROSCO
------------------------------
From: "Ross Crawford" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup
Subject: Re: make linux disk only one
Date: Fri, 10 Sep 1999 09:08:01 +1000
B.Becking <[EMAIL PROTECTED]> wrote in message
news:7r7138$mi6$[EMAIL PROTECTED]...
>
> Manut's english isnt to good, true, but does that mean he should refrain
> from using english ?? Your native tongue is, de facto, the lingua franca
of
> the internet, and thats precisely the reason you cant go around telling
> people not to use it (or at least try to use it). If you cant be bothered,
> thats fine, just dont answer.
>
> Boudewijn
Get a grip. Leonard offered it as a suggestion - didn't "tell" anyone to do
anything. Surely it'd be worth a try? And if it turns out than Manut gets a
better response posting in native language, then it was a good suggestion,
wasn't it?
Regards,
ROSCO
------------------------------
From: "..Luca T.." <[EMAIL PROTECTED]>
Subject: HELP with MAC sharing
Date: Fri, 10 Sep 1999 01:54:40 +0200
Hi,
I setted up my linux box like a server with netatalk and my 3 Macs can see
the server in the menu.
The problems are that when i try to set the linux box like server for the
Mac from the Mac I receive a message like "..unnable to locate the server"
(but i can see it from the machine)
And another problem is that I can make pings on the Macs from linux and when
I fire netatalk it seems to work well (I receive the message of confirm for
the start) but after a minute I receive 2 messages like "unnable to register
@workstation....afpd not start)
I suppose that I have to make a folder owned by the netatalk users ( like I
did with SAMBA clients for windows) but i didn't understand where should I
register the clients (i saw the config file but i didn't undrestand it)
Do you know what's wrong??
Thanx.
LUCA.
------------------------------
From: [EMAIL PROTECTED] (Juergen Heinzl)
Subject: Re: Any floating point support in kernel
Date: Thu, 09 Sep 1999 21:27:23 GMT
In article <7r92b5$tbv$[EMAIL PROTECTED]>, [EMAIL PROTECTED] wrote:
>Hi,
>
> Is there any floating point support in Linux kernel such as
> to save/restore the states of FP registers and allow usage
Yes. As a process might use the FP unit the kernel has to save
all registers and the FPU's state.
> of sin()/cos()/...etc.?
No, since it just does not make sense. I presume you do not have
the math emulation in mind here though. Of course if necessary
within some driver you can fall back to using assembler functions
or link in static routines, but this can become a tricky one if
it runs on an architecture where a FP error results in a hardware
trap.
Cheers,
Juergen
--
\ Real name : J�rgen Heinzl \ no flames /
\ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: Tony Angelo <[EMAIL PROTECTED]>
Subject: Re: Any floating point support in kernel
Date: Thu, 09 Sep 1999 19:19:53 -0400
Unfortunately, this isn't true. The fp registers are undefined inside
the kernel. Those registers are saved only on context switches, and
moving
from userspace to kernelspace isn't a context switch. My sources:
http://www.progressive-comp.com/Lists/?l=linux-kernel&m=92508726800511&w=2
also,
http://www.progressive-comp.com/Lists/?l=linux-kernel&m=93311455928363&w=2
http://www.progressive-comp.com/Lists/?l=linux-kernel&m=92508126430281&w=2
Tony
Juergen Heinzl wrote:
>
> In article <7r92b5$tbv$[EMAIL PROTECTED]>, [EMAIL PROTECTED] wrote:
> >Hi,
> >
> > Is there any floating point support in Linux kernel such as
> > to save/restore the states of FP registers and allow usage
>
> Yes. As a process might use the FP unit the kernel has to save
> all registers and the FPU's state.
>
> > of sin()/cos()/...etc.?
> No, since it just does not make sense. I presume you do not have
> the math emulation in mind here though. Of course if necessary
> within some driver you can fall back to using assembler functions
> or link in static routines, but this can become a tricky one if
> it runs on an architecture where a FP error results in a hardware
> trap.
>
> Cheers,
> Juergen
>
> --
> \ Real name : J�rgen Heinzl \ no flames /
> \ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: Warren Young <[EMAIL PROTECTED]>
Subject: Re: IDE for c++ dev?
Date: Thu, 09 Sep 1999 17:34:49 -0600
Johan Kullstam wrote:
>
> [EMAIL PROTECTED] () writes:
>
> > So
> > be it, plenty of people have testified that they find emacs to be
> > well worth the time it took to learn to use it. What puzzles and
> > bothers me is why SO many emacs fans bristle and shout otherwise when
> > ANYONE states the obvious, that emacs is EXTREMELY difficult to use.
>
> you seem to be describing vi and its users.
Not at all. We vi users don't try to pretend that our editor is
intuitive. We tell newbies that it's hard to learn, but that it'll be
worth the effort of learning it. Emacs users tell newbies that their
editor is easier to learn and use than vi -- hogwash.
I'm not trying to say that emacs is worthless or broken or anything
else. I don't use it because it can't do anything I want that vim (my
vi of choice) can't -- since it's slower to start than vi and not as
ubiquitous, it loses my vote. No, I don't consider browsing the web
from within my editor to be a necessary feature. B-)
That brings up something else: many of emacs' features (mainly the ones
integrating virtually every aspect of Unix within the editor -- dired,
webbing, news and email reading, eliza... ;-> ) made sense for text
terminals, but aren't all that helpful when you have a windowing
system. You used to have to shut down (or background) the editor to
read email, but today you can just open up another xterm to launch pine.
You should also know that I do know emacs: I am reasonably productive
with it. It just isn't more valuable to me than vi.
--
= Warren Young: www.cyberport.com/~tangent | Yesterday it worked.
= ICBM Address: 36.8274040N, 108.0204086W, | Today it is not working.
= alt. 1714m | Windows is like that.
------------------------------
From: Tony Angelo <[EMAIL PROTECTED]>
Subject: Re: Any floating point support in kernel
Date: Thu, 09 Sep 1999 19:22:32 -0400
Ok, so I posted to the wrong article. Apologies.
Tony
Tony Angelo wrote:
>
> Unfortunately, this isn't true. The fp registers are undefined inside
> the kernel. Those registers are saved only on context switches, and
> moving
> from userspace to kernelspace isn't a context switch. My sources:
>
> http://www.progressive-comp.com/Lists/?l=linux-kernel&m=92508726800511&w=2
>
> also,
>
> http://www.progressive-comp.com/Lists/?l=linux-kernel&m=93311455928363&w=2
> http://www.progressive-comp.com/Lists/?l=linux-kernel&m=92508126430281&w=2
>
> Tony
>
> Juergen Heinzl wrote:
> >
> > In article <7r92b5$tbv$[EMAIL PROTECTED]>, [EMAIL PROTECTED] wrote:
> > >Hi,
> > >
> > > Is there any floating point support in Linux kernel such as
> > > to save/restore the states of FP registers and allow usage
> >
> > Yes. As a process might use the FP unit the kernel has to save
> > all registers and the FPU's state.
> >
> > > of sin()/cos()/...etc.?
> > No, since it just does not make sense. I presume you do not have
> > the math emulation in mind here though. Of course if necessary
> > within some driver you can fall back to using assembler functions
> > or link in static routines, but this can become a tricky one if
> > it runs on an architecture where a FP error results in a hardware
> > trap.
> >
> > Cheers,
> > Juergen
> >
> > --
> > \ Real name : J�rgen Heinzl \ no flames /
> > \ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: Warren Young <[EMAIL PROTECTED]>
Subject: Re: Linux standards compliance
Date: Thu, 09 Sep 1999 17:42:28 -0600
Phil Howard wrote:
>
> On Mon, 06 Sep 1999 21:43:48 GMT [EMAIL PROTECTED] wrote:
>
> | Easy. Microsoft acknowledges that one of the main causes for
> | NT4 instability are bad drivers (and that they intend to fix that
>
> I will second that. Somewhere around half the problems I have with
> NT4 are due to device drivers and related DLLs. One of them I have
> constantly had problems with was with the HP laser printer driver
> while doing page setups in Outlook. Certain actions always got BSoD.
> Uninstalling the drivers, changing printer, and installing the new
> drivers, and all worked fine in that area.
I don't doubt HP is partly to blame for your problem, but Microsoft is
at fault here as well: NT does way too many things in kernel space.
Samba/NFS (mostly), printing, graphics and windowing all exist in Linux
as user-land processes, but are in the kernel in Microsoft OSes. If my
Linux box's Epson printer driver crashes, I just get an aborted print
job.
--
= Warren Young: www.cyberport.com/~tangent | Yesterday it worked.
= ICBM Address: 36.8274040N, 108.0204086W, | Today it is not working.
= alt. 1714m | Windows is like that.
------------------------------
From: [EMAIL PROTECTED] (Chris Gregory)
Crossposted-To:
comp.os.linux.development.apps,linux.dev.gcc,linux.dev.kernel,linux.dev.x11
Subject: Re: survey linux project.
Reply-To: [EMAIL PROTECTED]
Date: Fri, 10 Sep 1999 00:59:21 GMT
On Thu, 09 Sep 1999 16:04:49 +1000, Karlo Szabo <[EMAIL PROTECTED]> wrote:
>Why not start a Linux DVD project?
Or, you could write a scanner driver (for my scanner).
>"Kim,Taesung" wrote:
>> Hello!
>> We( I and my friends) have plan to make soem application on linux.
>> First of all, we want to survey on going project on linux.
>> We want to know any kind of projects about linux.
>> Where can we find?
>> Thanks for regard.
Seriously, though, I don't know of any one central location for all ongoing
linux projects. I don't think there is one. If there is one, everyone's
been hiding it from me.
--
Chris Gregory <[EMAIL PROTECTED]>
------------------------------
From: Fred Christiansen <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: Disabling control-alt-delete from a program
Date: Thu, 09 Sep 1999 17:59:11 -0600
Steve Houseman wrote:
> Dont know how to disable ctl-alt-del but but two connected items
> are sysctl and /proc/sys/kernel/ctrl-alt-del
RedHat 6.0 has a program, ctrlaltdel, that will diddle /proc/sys/kernel/ctrl-alt-del
for you.
--
Fred Christiansen, a Canajan (Eh?) in Idaho ................................
[EMAIL PROTECTED] (specific) // [EMAIL PROTECTED] (general)
Folk inside HP's firewall can access: http://hpbs1668.boi.hp.com/~fredch/
and: http://hpbs1668.boi.hp.com/hazard/
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Crossposted-To: comp.os.linux.setup
Subject: Re: loadkeys/compose
Date: 9 Sep 1999 20:15:26 -0500
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[Bert Douglas]
> > 2) are you limited to one <compose> key in Linux?
[Dirk Foersterling <[EMAIL PROTECTED]>]
> You can do virtually anything with that one key. Why would you need a
> second one? Maybe you ask yourself after you read what it really is.
Hey, he had a valid point. Apparently in 'doze, you can configure the
keyboard so that you only need two keystrokes (a dead key and a regular
key) to get a composed char. On Linux you have to hit `compose' first.
Of course, the easy workaround is global-set-key, but then again that
only works if you never leave Emacs. <g>
--
Peter Samuelson
<sampo.creighton.edu!psamuels>
------------------------------
From: Mark McDougall <[EMAIL PROTECTED]>
Subject: Re: pci libraries
Date: Fri, 10 Sep 1999 11:47:50 +1000
Chris Naylor wrote:
> I have a problem that I'm sure has a very simple answer - I am developing a
> device driver for a pci interface card. As a result I am using functions in
> the pci.h header. My problem is that I don't know which libraries to
> include for these functions - in fact, I can't even find the file where
> these functions are implemented. I have worked on this for a few days and
> am getting nowhere - any help would be greatly appreciated!!
Being very new myself (about 1 week into my own pci device driver) to
the Linux programming world, I can't answer you directly other than to
guess some required #defines or headers are missing, which means the
functions aren't being prototyped correctly?!?
Your best bet is to go to the O'Reilly Books site and find "Linux Device
Drivers" by Alessandro Rubini. You'll find an archive of source from the
book available for download, which contains a PCI sample ready to run.
BTW I'd *strongly* suggest you buy this book - I read it from
cover-to-cover and I've basically got a functioning driver in 1 week
(PCI bus-mastering DMA)!
Are you developing the driver as a module? I'd suggest this is the way
to go, at least in the early stages of development. You can base your
driver on Rubini's source which are all modules. Sure beats the hell out
of writing Windows device drivers - and I'm speaking from experience!
Regards,
--
| Mark McDougall |
| Engineer |
| Virtual Logic Pty Ltd |
| http://www.vl.com.au |
------------------------------
From: Joey McAlerney <[EMAIL PROTECTED]>
Crossposted-To:
alt.os.linux,comp.os.linux.questions,redhat.rpm.general,linux.redhat.rpm
Subject: Where can I find the handy SWEEP utility?
Date: Thu, 09 Sep 1999 17:39:38 +0100
Hi everyone,
While working on Irix boxes over the summer, I got to take advantage of
the "sweep" utility. It simply allowed you to gain root access to a
machine by simply typing "sweep" (that is, of course, someone else with
root access specified that you could do so). I am looking for this
utility for Redhat, and couldn't find it so far. If it exists, could
you please point me to it?
Many thanks,
-Joe McAlerney
------------------------------
From: [EMAIL PROTECTED] (Mario Klebsch)
Crossposted-To: comp.os.linux.x,comp.windows.x,comp.sys.palmtops.pilot
Subject: Re: Embedded X-server anyone ? ?
Date: Thu, 9 Sep 1999 22:41:58 +0200
Nash Aragam <[EMAIL PROTECTED]> writes:
>Am lookin' for any/all info that might be available on the
>idea/concept/design/implementation of a SMALL_FOOTPRINT X-server for use
>in embedded OSs and embedded/handheld palmtops.
The original MIT X11R5 server ran on the Sun 3/50 with 68020 CPU /
25MHz?? and 4MB RAM / no swap. Is this footprint small enough?
73, Mario
--
Mario Klebsch [EMAIL PROTECTED]
------------------------------
Date: Fri, 10 Sep 1999 11:37:29 +0800
From: "Bigwoof!" <[EMAIL PROTECTED]>
Subject: Re: select() and write descriptors
[EMAIL PROTECTED] (Kaz Kylheku) <[EMAIL PROTECTED]> wrote:
: On Thu, 9 Sep 1999 11:00:46 +0800 , Bigwoof! <[EMAIL PROTECTED]> wrote:
:>Hi,
:>when i use the select call to scan a list of read fds and write fds, the
:>call always seems to take a very long time to process the write fds.
:>i.e. select(&nfds, &rfds, &wfds, 0, something)
: Umm, the first argument isn't a pointer, but an int. It must be one plus the
: highest numbered descriptor that you want to check in any of your sets.
:) i wrote it from memory. hence the reason for the messed up calls. :)
:>Why does it block so long on the write fds? does it take that long to
:>check if a socket is writable?
: You should only select sockets for writability if you have pending data.
You : see, sockets are write ready almost always! This means that select
will return : immediately and you will end up calling it over and over
again rapidly, wasting : CPU time.
okay. but why does this cause such a large delay? i figured that sockets
are always writable and thus checking them should not block so long.
for eg. i have some programs that tunnel through a transport layer
abstraction using select calls. if i check the write descriptors, end to
end pings take 200-500 ms. if i remove the write descriptors, it takes
1.4-1.8ms.
I can't figure out why there is such a large delay.
I don't need to check the sockets for writeability. I just left it in
there for future expansion purposes. (and because i didn't believe it
would affect performance so greatly)
Thanks for your help! :)
Rajesh
------------------------------
From: David Schwartz <[EMAIL PROTECTED]>
Subject: Re: 497.2 days ought to be enough for everybody
Date: Thu, 09 Sep 1999 12:46:41 -0700
Kaz Kylheku wrote:
> A mission-critical OS couldn't support such a function in its core API because
> the next reboot might not occur for months.
Had they thought about having jiffies wrap, they could have ensured
that the system is at least rebooted once every 497.2 days. It just
shows how the Linux people have more ideas than the NT people.
DS
------------------------------
From: Fred Christiansen <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.hardware
Subject: device to /dev/sd* binding, and other Q's
Date: Thu, 09 Sep 1999 18:35:28 -0600
Several questions, gentle readers, stemming from my experience
with Red Hat Linux 6.0:
1) For anyone familiar with HP-UX's ioscan tool .. Is there such a thing
for Linux? It would probably behave something like this:
for each device file in /dev/sd*
open device file
if it could be opened
send a test-unit-ready msg
if the unit is ready
print host.channel.target.lun CLAIMED device file
else
print host.channel.target.lun NO_HW device file
Someone who _really_ knows ioscan could probably describe it better.
2) For a given hardware configuration -- adapters, SCSI devices -- that
does not change from reboot to reboot, can I depend upon the binding
between a given host.channel.target.lun and its /dev/sd? file
remaining constant? E.g., if scsi2.0.10.3 is associated with /dev/sdno,
will that remain true from reboot to reboot if I don't change the
h/w config? My reason for asking is that a colleague tells me that
Win NT 4.0 does not guarantee the mapping between port.bus.target.lun
and \\.\PhysicalDriveX from reboot to reboot. I'm concerned this may
be a PC platform-ism, rather than an NT-ism, that could affect Linux
as well. A toolset I'm porting from a certain Unix to Linux and NT
makes the assumption that things remain constant.
3) I wanted to create but one partition on a large-ish (4GB disk), and
the docs for sfdisk suggested doing
sfdisk <<!
;
!
It worked, but the results surprised me. From reading the "sd" SCSI
driver code, I was assuming it would do the 63 sectors-per-track,
255 heads thing. Instead, it created a partition with 32 spt,
64 heads, and 4095 cylinders. (This was a fresh drive, previously
unpartitioned.) I can create an ext2 file system on it, mount it,
create files, and so on, so things look cool. When, then, does the
63/255 thing come into play? I'm not going to be booting from the
extra drive(s), so I figure I don't have to worry about extended
partitions, that one huge primary partition is OK. Yes?
--
Fred Christiansen, a Canajan (Eh?) in Idaho ................................
[EMAIL PROTECTED] (specific) // [EMAIL PROTECTED] (general)
Folk inside HP's firewall can access: http://hpbs1668.boi.hp.com/~fredch/
and: http://hpbs1668.boi.hp.com/hazard/
------------------------------
** 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
******************************
