[9fans] Octopus installation on Plan9
Dear List, I am in the process of installing an Octopus PC on plan9. Following the INSTALL.Octopus instructions I have - downloaded the o.zip to /usr/octopus - unzipped the file there - On a plan I then run rc ./usr/octopus/lib/Install.Plan9.rc which starts asking for network setup and a PC password After accepting the password change the script prints: sh: /usr/octopus/lib/setup.sh: file does not exist How am I supposed to proceed from here? In earlier attempts I managed to get further where the script shows me an rc script to start octopus but then the files /dis/o/pcrc seem to miss the executable flag set which results in broken startups scripts. (nameservice and more). Where is my mistake? Thanks, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen) pgptMj3qGANMe.pgp Description: PGP signature
[9fans] P9P libthread on Debian ARM vs. makecontext et al.
Hi, I recently got myself a lovely Nokia N810 internet tablet. That runs Debian ARM and I thought I'd have a stab at compiling P9P for it. Unfortunately, the makecontext/swapcontext calls are not supported on that platform by Debian. I noticed that libthread/Linux.c has implementations of those two functions for ARM. So I thought I try compiling libthread without pthread support. This time, it's the mcontext_t (aka sigcontext) struct that is causing an issue. The Debian ARM sigcontext struct looks like this: struct sigcontext { unsigned long trap_no; unsigned long error_code; unsigned long oldmask; unsigned long arm_r0; ... unsigned long arm_r10; unsigned long arm_fp; unsigned long arm_ip; unsigned long arm_sp; unsigned long arm_lr; unsigned long arm_pc; unsigned long arm_cpsr; unsigned long fault_address; }; The code in Linux.c that implements makecontext() for ARM expects this struct to contain a gregs array. I'm guessing that gregs is the same as the enumerated arm_* ones above. So, would I be correct in assuming that gregs[13] is arm_sp and gregs[14] is arm_lr? Thanks, Robby
Re: [9fans] P9P libthread on Debian ARM vs. makecontext et al.
The code in Linux.c that implements makecontext() for ARM expects this struct to contain a gregs array. I'm guessing that gregs is the same as the enumerated arm_* ones above. So, would I be correct in assuming that gregs[13] is arm_sp and gregs[14] is arm_lr? Hi, I would say yes, as SP is r13 and LR is r14 on arm. To run Russ' libtask on arm7 I used the Linux-arm-asm.s and the arm-specific makecontext implementation from Linux.c, as well as the following structures. M. -- enum { NREG = 16, }; typedef struct Mcontext { u32int gregs[NREG];/* general registers */ } Mcontext; typedef struct ucontext { Mcontextuc_mcontext; sigset_t uc_sigmask; struct { void*ss_sp; u16int ss_size; } uc_stack; } ucontext_t;
Re: [9fans] lisp
cmucl is directly executable but it has only enough intelligence to load a big lisp.core, which contains all the smarts. If these core files aren't generated all *that* often, one could write a tool which would turn a core file into a byte array, link a new lisp executable, and exec() that. I realize that's not an answer to the question as posed, but... Dave Eckhardt
[9fans] A question on remotemail
Dear List, I have been fiddling with the mail setup on my CPU server. This machine has to deliver all mail to my ISPs smtp so I set the rewrite script to rewrite.gateway. The ISPs smtp rejected my mail though because it expected a From: line different to my [EMAIL PROTECTED] or the default $fd set in remotemail. I added a 'From: ckeen [EMAIL PROTECTED]' line to /mail/box/ckeen/headers but this does not seem to have any effect. I thus added a names.remote file with 'ckeen [EMAIL PROTECTED]' and changed remotemail with this diff: 14c15,19 exec /bin/upas/smtp -h $fd $addr $sender $* --- remotename=`{/bin/upas/aliasmail $sender} if ( test -n $remotename ) { sender = $remotename } exec /bin/upas/smtp -a -h $fd $addr $sender $* This does the work for me. Questions: - shouldn't the headers file work too? - how is this done right? Thanks for your opinions, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen) pgp5burawDsdC.pgp Description: PGP signature
Re: [9fans] lisp
On Tue, Jul 8, 2008 at 7:08 AM, Dave Eckhardt [EMAIL PROTECTED][EMAIL PROTECTED] wrote: cmucl is directly executable but it has only enough intelligence to load a big lisp.core, which contains all the smarts. If these core files aren't generated all *that* often, one could write a tool which would turn a core file into a byte array, link a new lisp executable, and exec() that. I realize that's not an answer to the question as posed, but... Dave Eckhardt I'm not sure if any action is being taken just yet on which Lisp to go for but, if I were porting a lisp I'd be looking at either CMUCL or SBCL, and probably lean more towards SBCL, as they've done a bit of work to make bootstrapping a little nicer. I like Dave Eckhardt's idea. I don't think core files get changed that often, however sometimes when one produces a standalone product from one of the above SBCL or CMUCL, you do want to make your own core file and distribute it. That's still a pretty static process if my memory serves me. Dave
Re: [9fans] Octopus installation on Plan9
Has the setup.sh execute permissions? Is /usr/octopus unpacked correctly under inferno, so that it might be something like /usr/inferno/usr/octopus on the Plan 9 system? I think all the problem is that the files are being extracted without execute permissions. This did not happen when we used tar instead of zip. Anyway, you can either give the execute permissions to /dis/... yourself or wait a little until I fix the install script to make sure this does not happen again. On Tue, Jul 8, 2008 at 1:22 PM, Christian Kellermann [EMAIL PROTECTED] wrote: Dear List, I am in the process of installing an Octopus PC on plan9. Following the INSTALL.Octopus instructions I have - downloaded the o.zip to /usr/octopus - unzipped the file there - On a plan I then run rc ./usr/octopus/lib/Install.Plan9.rc which starts asking for network setup and a PC password After accepting the password change the script prints: sh: /usr/octopus/lib/setup.sh: file does not exist How am I supposed to proceed from here? In earlier attempts I managed to get further where the script shows me an rc script to start octopus but then the files /dis/o/pcrc seem to miss the executable flag set which results in broken startups scripts. (nameservice and more). Where is my mistake? Thanks, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen)
Re: [9fans] Octopus installation on Plan9
* Francisco J Ballesteros [EMAIL PROTECTED] [080708 17:30]: Has the setup.sh execute permissions? Is /usr/octopus unpacked correctly under inferno, so that it might be something like /usr/inferno/usr/octopus on the Plan 9 system? I think all the problem is that the files are being extracted without execute permissions. This did not happen when we used tar instead of zip. Anyway, you can either give the execute permissions to /dis/... yourself or wait a little until I fix the install script to make sure this does not happen again. Thanks for your help. Just to make sure we are both talking about the same thing: I create a /usr/octopus on plan9, unzip the o.zip there which already contains inferno and a octopus inferno user in ./usr/octopus. I need to chmod +x the files in /dis/* also somewhere else? Kind regards, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen) pgpNtVZCGkRVB.pgp Description: PGP signature
[9fans] why not Lvx for Plan 9?
I downloaded Russ's 9vx and vx32 source tree Ford's web page, and build those on my Debian stable machine. I also read the paper of vx32 last week. Now, everything goes fine, and got a fun to play with games/mahjongg on that virtual machine. Thank you very much Russ! However, I have somewhat confused mind Why it's not Linux vx for Plan9? I have no intention to make it bad, however, if we believe Plan 9 is better than Linux, then Linux vx for Plan9 should be more neccessary, shouldn't ? Kenji
Re: [9fans] A question on remotemail
This machine has to deliver all mail to my ISPs smtp so I set the rewrite script to rewrite.gateway. The ISPs smtp rejected my mail though because it expected a From: line different to my [EMAIL PROTECTED] or the default $fd set in remotemail. I added a is the problem that you require that the smtp helo domain be different from the envelope sender (the $address in the smtp mail from $address line?) is this why you can't set $fd to something that won't be rejected and still have your From: lines properly rewritten? 'From: ckeen [EMAIL PROTECTED]' line to /mail/box/ckeen/headers but this does not seem to have any effect. I thus added a names.remote file with 'ckeen [EMAIL PROTECTED]' according to the man page, the standard place for this information would be a file listed in /mail/lib/fromfiles, and alias mail should be given the -f switch. (maybe there's a difference for gatways that i'm missing?) - shouldn't the headers file work too? the envelope sender isn't the same as the rfc822 sender so no, i don't think that should work. - how is this done right? - erik
Re: [9fans] Octopus installation on Plan9
probably just /dis/* and /dis/*/* It would be safe to add usr/octopus/lib/*.{sh, rc} I'll try to reproduce what's happening to you here and fix the install. On Tue, Jul 8, 2008 at 5:35 PM, Christian Kellermann [EMAIL PROTECTED] wrote: * Francisco J Ballesteros [EMAIL PROTECTED] [080708 17:30]: Has the setup.sh execute permissions? Is /usr/octopus unpacked correctly under inferno, so that it might be something like /usr/inferno/usr/octopus on the Plan 9 system? I think all the problem is that the files are being extracted without execute permissions. This did not happen when we used tar instead of zip. Anyway, you can either give the execute permissions to /dis/... yourself or wait a little until I fix the install script to make sure this does not happen again. Thanks for your help. Just to make sure we are both talking about the same thing: I create a /usr/octopus on plan9, unzip the o.zip there which already contains inferno and a octopus inferno user in ./usr/octopus. I need to chmod +x the files in /dis/* also somewhere else? Kind regards, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen)
Re: [9fans] P9P libthread on Debian ARM vs. makecontext et al.
On 7/8/08, Michael Teichgräber [EMAIL PROTECTED] wrote: The code in Linux.c that implements makecontext() for ARM expects this struct to contain a gregs array. I'm guessing that gregs is the same as the enumerated arm_* ones above. So, would I be correct in assuming that gregs[13] is arm_sp and gregs[14] is arm_lr? Hi, I would say yes, as SP is r13 and LR is r14 on arm. To run Russ' libtask on arm7 I used the Linux-arm-asm.s and the arm-specific makecontext implementation from Linux.c, as well as the following structures. M. -- enum { NREG = 16, }; typedef struct Mcontext { u32int gregs[NREG];/* general registers */ } Mcontext; typedef struct ucontext { Mcontextuc_mcontext; sigset_t uc_sigmask; struct { void*ss_sp; u16int ss_size; } uc_stack; } ucontext_t; Thanks, that's gotten me quite a bit further. I can compile the whole lot now and run quite a few of the commands, including rc, under the ARM qemulation. Hurrah. Next I need to figure out why I get segfaults for some. And then try it on real HW. Robby
Re: [9fans] A question on remotemail
* erik quanstrom [EMAIL PROTECTED] [080708 17:40]: This machine has to deliver all mail to my ISPs smtp so I set the rewrite script to rewrite.gateway. The ISPs smtp rejected my mail though because it expected a From: line different to my [EMAIL PROTECTED] or the default $fd set in remotemail. I added a is the problem that you require that the smtp helo domain be different from the envelope sender (the $address in the smtp mail from $address line?) is this why you can't set $fd to something that won't be rejected and still have your From: lines properly rewritten? As far as I understood $fd just sets the domain, I need the user and the domain name to be different from my local settings though. 'From: ckeen [EMAIL PROTECTED]' line to /mail/box/ckeen/headers but this does not seem to have any effect. I thus added a names.remote file with 'ckeen [EMAIL PROTECTED]' according to the man page, the standard place for this information would be a file listed in /mail/lib/fromfiles, and alias mail should be given the -f switch. (maybe there's a difference for gatways that i'm missing?) I read the page as if the -f switch is for domain names only. Am I wrong? - shouldn't the headers file work too? the envelope sender isn't the same as the rfc822 sender so no, i don't think that should work. right, thanks. Kind regards, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen) pgp2gY6KolHyf.pgp Description: PGP signature
Re: [9fans] why not Lvx for Plan 9?
I believe the reasoning is as such: Linux has more drivers than Plan 9, therefore Plan 9 should run on linux. Use Linux as your driver repository... this is an approach used by some microkernel systems like L4. Dave On Tue, Jul 8, 2008 at 8:28 AM, [EMAIL PROTECTED] wrote: I downloaded Russ's 9vx and vx32 source tree Ford's web page, and build those on my Debian stable machine. I also read the paper of vx32 last week. Now, everything goes fine, and got a fun to play with games/mahjongg on that virtual machine. Thank you very much Russ! However, I have somewhat confused mind Why it's not Linux vx for Plan9? I have no intention to make it bad, however, if we believe Plan 9 is better than Linux, then Linux vx for Plan9 should be more neccessary, shouldn't ? Kenji
Re: [9fans] A question on remotemail
* erik quanstrom [EMAIL PROTECTED] [080708 17:40]: This machine has to deliver all mail to my ISPs smtp so I set the rewrite script to rewrite.gateway. The ISPs smtp rejected my mail though because it expected a From: line different to my [EMAIL PROTECTED] or the default $fd set in remotemail. I added a is the problem that you require that the smtp helo domain be different from the envelope sender (the $address in the smtp mail from $address line?) is this why you can't set $fd to something that won't be rejected and still have your From: lines properly rewritten? As far as I understood $fd just sets the domain, I need the user and the domain name to be different from my local settings though. smtp goes to the trouble of transforming lines like From: erik quanstrom quanstro to From: erik quanstrom [EMAIL PROTECTED] provided fd=example.com - erik
Re: [9fans] A question on remotemail
* Christian Kellermann [EMAIL PROTECTED] [080708 18:23]: * erik quanstrom [EMAIL PROTECTED] [080708 18:10]: * erik quanstrom [EMAIL PROTECTED] [080708 17:40]: Instead of having From: [EMAIL PROTECTED] I now get: [EMAIL PROTECTED] And I would need From: [EMAIL PROTECTED] -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen) pgpQigQ5AD3rX.pgp Description: PGP signature
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 9:04 AM, erik quanstrom [EMAIL PROTECTED] wrote: if that's the argument, wouldn't it make sense to get rid of plan 9? Think of 9vx and lguest and friends as software tools. Software tools did a lot to popularize the ideas of Unix, and made it easier for people to consider using the real thing. ron
Re: [9fans] why not Lvx for Plan 9?
Think of 9vx and lguest and friends as software tools. Software tools did a lot to popularize the ideas of Unix, and made it easier for people to consider using the real thing. yes, but that was when the underlying system was System 370, VMS, PRIME, GCOS, ... which didn't do all that much for you. now the underlying system has all the fun. (unless you need to program it.)
Re: [9fans] A question on remotemail
There are two different From addresses associated with a particular mail message. There is the RFC822 From: line and then there is the MAIL FROM return address that is sent as part of the SMTP conversation. (This is the address recorded in the From line at the top of the Plan 9 mbox-format messages. See thread last month.) The two are not always the same. For example, this message says it is From: [EMAIL PROTECTED], but the SMTP return address is something like [EMAIL PROTECTED], so that if a bounce comes back, it goes to the mailing list software, which uses the last bit to determine which address on the list is bouncing. I would be a little surprised if your ISP is rejecting based on the From: line rather than the SMTP return address, but maybe spammers have driven them to that. The /mail/box/$user/headers file sets RFC822 headers, including the From: header. The upas/smtp arguments set the SMTP return address. As for how to do it right, editing remotemail sounds fine. The nice thing about upas is you can understand the shell scripts and edit them, instead of having to shoehorn everything into some preordained config file. Russ
Re: [9fans] A question on remotemail
* Christian Kellermann [EMAIL PROTECTED] [080708 18:23]: * erik quanstrom [EMAIL PROTECTED] [080708 18:10]: * erik quanstrom [EMAIL PROTECTED] [080708 17:40]: Instead of having From: [EMAIL PROTECTED] I now get: [EMAIL PROTECTED] And I would need From: [EMAIL PROTECTED] i think now your headers file will fix that problem. but i'm getting a bit confused about your setup. ;-) - erik
Re: [9fans] why not Lvx for Plan 9?
However, I have somewhat confused mind Why it's not Linux vx for Plan9? I have no intention to make it bad, however, if we believe Plan 9 is better than Linux, then Linux vx for Plan9 should be more neccessary, shouldn't ? i'm not sure if this was the authors' intent, but making plan9 easily accessible to a larger group (mac, linux and someday windows users) will make plan9 more popular. vxlinux under plan9 will be useful, but only to us 9fans.
Re: [9fans] 9vx dns funny
here's the symptom ndb/dnsquery 9hal.ath.cx 9hal.ath.cx ip 9.0.0.0 Fixed. Instead of the tacky solution, one can check the return value from v4parseip. on a related note, would it be worth while to put effort into supporting ptr queries, ip6 c using the host's lookup facilities or should that effort be directed into adding bits to enable ndb/dns to run? If someone sends me code to answer other DNS queries or to convert the current devip to IPv6, I would gladly take it. I intend 9vx to be a very thin layer atop the host's resources, not a second machine that you have to administer. If you want a second machine, there is always VMware. That said, if some people do want to switch over to Plan 9-provided versions of some host resources, that doesn't bother me, as long as they aren't required. I'm happy to make it possible to run ndb/dns or fossil instead of using the host's DNS and file system, but it's important that they're not required. i guess this gets to a more philosophical question on how 9vx networking relates to the host. personally, i feel it would be more useful to be able to use plan 9's native network stack. but i'm biased. i want to send aoe/cec/il packets. I'm happy to accomodate these kinds of things, but I don't want to make it required. Running the native TCP/IP stack requires being able to send and receive raw ethernet frames on an otherwise unused ethernet device, something most machines don't have and also something that usually requires being root. This would hamper portability quite a bit. It would be trivial to write a devether, like devip, that handles raw ethernet frames. You'd still have to be root on most systems, but if you stay away from sending TCP and UDP packets, you can coexist with the host OS on a single ethernet device. That would be make sending aoe and cec very easy, and if you did a bit more work and the host OS turned a blind eye, you might even get away with IL. I'd rather see that than drag in the entire network stack from the Plan 9 kernel. The relevant low-level Linux ethernet device code is in p9p. In fact, you could probably just add a new ether0 protocol directory to devip instead of writing a new devether wrapper driver. Russ
Re: [9fans] why not Lvx for Plan 9?
However, I have somewhat confused mind Why it's not Linux vx for Plan9? I have no intention to make it bad, however, if we believe Plan 9 is better than Linux, then Linux vx for Plan9 should be more neccessary, shouldn't ? I wrote 9vx for people like me, who would prefer to use Plan 9 but, for various reasons, must use FreeBSD or Linux or Mac OS X. Linux vx for Plan 9 would help people who prefer to use Linux but must use Plan 9. I am sure such people exist, but I feel no sympathy toward them. ;-) Russ
Re: [9fans] why not Lvx for Plan 9?
Well, there's also people like me: I prefer and am able to use Plan 9 the bulk of the time, but have a few particular tasks I need Linux for. It'd be nice to be able to stick the Linux box in a little jail. I'm very glad 9vx exists: I now have Plan 9 on my OS X laptop. I'd like Lvx to stick on my Plan 9 cpu servers. As it stands, I'll be converting one of my older cpu servers into a linux box in a few weeks. That makes me feel unclean. Anthony ---BeginMessage--- However, I have somewhat confused mind Why it's not Linux vx for Plan9? I have no intention to make it bad, however, if we believe Plan 9 is better than Linux, then Linux vx for Plan9 should be more neccessary, shouldn't ? I wrote 9vx for people like me, who would prefer to use Plan 9 but, for various reasons, must use FreeBSD or Linux or Mac OS X. Linux vx for Plan 9 would help people who prefer to use Linux but must use Plan 9. I am sure such people exist, but I feel no sympathy toward them. ;-) Russ ---End Message---
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 11:04 AM, erik quanstrom [EMAIL PROTECTED] wrote: I believe the reasoning is as such: Linux has more drivers than Plan 9, therefore Plan 9 should run on linux. in this model, all plan 9 does is add an extra layer of goo on top of linux. it's not like you can avoid admining linux by hiding on a vm running on linux. That's not entirely true depending on the virtualization layer used. I'm not experienced yet with vx32, but for example, lguest/kvm/xen can be setup to pass-through device access to network, disk, audio, whatever. The logical partition running Plan 9 can be essentially pinned to a processor (or processors) and on that processor it rules the roost. Linux just deals with device access. I don't really think this undermines Plan 9 in any way unless you are keen on optimizing device performance -- in which case you do indeed most likely want native. But my point is, the Linux I/O layer is essentially non-administered. It doesn't need user accounts, an IP address, or even much of a file system (just enough to boot Plan 9 like in THX). -ericvh
Re: [9fans] why not Lvx for Plan 9?
in this model, all plan 9 does is add an extra layer of goo on top of linux. it's not like you can avoid admining linux by hiding on a vm running on linux. That's not entirely true depending on the virtualization layer used. I'm not experienced yet with vx32, but for example, lguest/kvm/xen can be setup to pass-through device access to network, disk, audio, whatever. The logical partition running Plan 9 can be essentially pinned to a processor (or processors) and on that processor it rules the roost. Linux just deals with device access. you didn't explain the part where you avoid admining the linux host. - erik
Re: [9fans] A question on remotemail
* Russ Cox [EMAIL PROTECTED] [080708 18:56]: As for how to do it right, editing remotemail sounds fine. The nice thing about upas is you can understand the shell scripts and edit them, instead of having to shoehorn everything into some preordained config file. Thanks for clarifying that Russ. It all makes sense to me now. I will leave my remotemail script as it is and be happy :) Thanks to erik for making me rething the behaviour of the scripts. Regards, Christian -- You may use my gpg key for replies: pub 1024D/47F79788 2005/02/02 Christian Kellermann (C-Keen) pgplwzfZHzXru.pgp Description: PGP signature
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 9:04 AM, erik quanstrom [EMAIL PROTECTED] wrote: I believe the reasoning is as such: Linux has more drivers than Plan 9, therefore Plan 9 should run on linux. if that's the argument, wouldn't it make sense to get rid of plan 9? I'm just saying I would never consider running linux on plan 9. I can't think of a single reason I'd ever want to do that, because, linux is so much easier to get installed on real hardware than plan 9. in this model, all plan 9 does is add an extra layer of goo on top of linux. it's not like you can avoid admining linux by hiding on a vm running on linux. If the goal was to avoid admining linux then one shouldn't run linux. That's not much of an argument. May as well run 9vx on FreeBSD :-) Same argument holds. i don't mean to use a broad brush. there are good reasons for running plan 9 in a vm on linux -- like you want to use a linux hosting company. but linux didn't get where it is by using windows as a device driver. Nope Linux got where it is by apache. - erik
Re: [9fans] lisp
On Tue, 08 Jul 2008 08:07:59 PDT David Leimbach [EMAIL PROTECTED] wrote: On Tue, Jul 8, 2008 at 7:08 AM, Dave Eckhardt [EMAIL PROTECTED][EMAIL PROTECTED] wrote: cmucl is directly executable but it has only enough intelligence to load a big lisp.core, which contains all the smarts. If these core files aren't generated all *that* often, one could write a tool which would turn a core file into a byte array, link a new lisp executable, and exec() that. I realize that's not an answer to the question as posed, but... Dave Eckhardt I'm not sure if any action is being taken just yet on which Lisp to go for but, if I were porting a lisp I'd be looking at either CMUCL or SBCL, and probably lean more towards SBCL, as they've done a bit of work to make bootstrapping a little nicer. I had asked my friend who ported CMUCL to Windows about this. He too first looked at SBCL but found it hard to port and the bootstrapping process was a lot longer. He was then able to create the initial CMUCL port by porting just the runtime -- he added just enough of a compatibility layer to allow use of a FreeBSD core image. Compiled Lisp object would think it is running on FreeBSD. I like Dave Eckhardt's idea. I don't think core files get changed that often, In the olden days Lisps used to create a custom executable by storing the heap in a data segment. Something to look into.
Re: [9fans] why not Lvx for Plan 9?
there's a certain level of administration required, sure, but i think eric's point was that the level of administration required just to get a good VM environment up is pretty minimal. if your VM has its own access to disk and network, you needn't have linux users or full network information. especially if your linux has been gutted so you don't have helpful demons running around in the background, this can be a fairly light - and front-loaded - burden. non-zero, to be sure, but not what you'd expect from a normal production linux box. anthony are you taking security updates into account. they're the #1 biggest timewaster for me and they're not front loaded. - erik
Re: [9fans] why not Lvx for Plan 9?
i believe i am, but perhaps your experience is different from mine. certainly there's less stuff to worry about patches for if you've got less stuff on the box. again, the idea is not to take ubuntu (or whatever) and stick a VM on top, but rather to strip the linux down to just what's needed to run the VM, like THNX. far fewer security updates, although i suppose one could argue it'll be more manual, ad hoc, and time consuming per update. i've not done enough of that to really say. i'm certainly not trying to convince you to run linux. if you don't have a pre-existing compelling reason to run it, don't. all i (and i believe eric) am saying is that if you *do* have such a reason, a tailored VM environment can reduce the burden quite a lot. anthony
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 3:15 PM, William Josephson [EMAIL PROTECTED] wrote: On Tue, Jul 08, 2008 at 03:03:45PM -0500, Eric Van Hensbergen wrote: I setup every machine on my network to tftpboot (BIOS), and they all tftpboot a kernel+ramdisk which has everything necessary to startup lguest/kvm plan9 and passthrough I/O requests to the disk/network. Admin done. I've found setting up diskless boot with Linux to be a major pain with most of the common distributions. I guess it has been a while since I last looked: how do you make it work reliably with updates without effectively rolling your own distribution? I didn't do the grunt work on this one, my friends in Germany built the setup to run xcpu, but it can be adapted just as easily to run the KVM drones. Mind you, the distribution in this case is hardly anything -- just the necessary tools to start lguest/kvm and get tun/tap working properly to get to the underlying network. You should effectively be able to build this by hand with any distro. -eric
Re: [9fans] lisp
if you make it executable, i think you should also find that the resulting executable has its contents read into memory on demand (ie, a page fault causes a read from the executable file), which might suit you.
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 1:15 PM, William Josephson [EMAIL PROTECTED] wrote: I've found setting up diskless boot with Linux to be a major pain with most of the common distributions. yes, they all suck. Try this: onesis.org for a reasonable system, used at sandia on a 4096-node cluster. for even lighter weight, see the xcpu.org tutorial on onesis+xcpu. xcpu2 now works almost the same as the cpu command (yes, cpu as in plan 9 cpu). ron
Re: [9fans] lisp
On Tue, 08 Jul 2008 22:12:10 BST C H Forsyth [EMAIL PROTECTED] wrote: if you make it executable, i think you should also find that the resulting executable has its contents read into memory on demand (ie, a page fault causes a read from the executable file), which might suit you. Right. My questions at this point have more to do with understanding the landscape but if I do this, my initial goal would be to get it up and running in as little time as possible even if the result is slow and hacky. Cleanup optimizations can follow. Thanks to all for their suggestions comments.
Re: [9fans] why not Lvx for Plan 9?
yes, they all suck. Try this: onesis.org Ok, it would be a load of work but has anyone tried building a linux filesystem on a plan9 server (/linux perhaps) and PXE booting a linux cpu server off it? Extrapolating you could even get the server to mount its root filesystem using v9fs rather than nfs. It wouldn't buy you much, except a linux environment thats easier for plan9 users to manage and support. With a following wind you might even get cinap's rather amazing linuxemu to use it as a runtime environment (shared libraries etc). I suspose you could start going mad and build a special gethostbyname() for your linux environment which understands how to talk to /net/dns etc. sort of p9p upsidedown... just silly ideas which I should probably have kept to myself. -Steve
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 5:01 PM, don bailey [EMAIL PROTECTED] wrote: But Linux use symlinks. Is there a way to make symlinks on the Plan 9 filesystem and make them accessible with NFS? The kernel probably doesn't care. Symlinks are just files whose contents are another file's path. As long as the kernel knows how to interpret it I'm sure it'd be fine. Look at the inverse: Plan 9 on Linux simply sees a Linux symlink as what ever the symlink points to. For example, in your 9vx tree do: `ln -s sparc64 v9` and run 9vx.Linux. you would think it would work that way. You would think that the server, upon hitting a symlink, would just indirect through it and all would be well. And it's true, the kernel doesn't care. But userspace does. (http://www.linuxinsight.com/ols2006_why_userspace_sucks_or_101_really_dumb_things_your_app_shouldnt_do.html) is really right. When I first got v9fs working, 1998, I tried mounting file systems over 9p. What a mess. Things just broke in weird ways. There is code that really wants a symlink to be there and readable. I can't even recall all the places, but they're there. And things break if you mount and don't have symlinks. Which is why I put readlink etc. in my v9fs, and why the .U version is in today's linux kernels. thanks ron
Re: [9fans] a question of file and the history of magic^H^H^H^H^HUNIX
On 2008-Jul-6, at 14:59 , Brantley Coile wrote: I remember the day I first saw a file magic file. I welcomed it because for the first time I didn't have access to the source code. Those were the days when you had to have $45k to get the source. Closer to $100K for most people. I had great fun writing nroff (yup, *n*roff) output device tables as binary blobs to interface with the non-source UNIXen of the day. And the Convergent Technologies X.25 binary code was a wonder to configure/tune as an end user :-P Remember kids: UNIX source code (BSD, really) wasn't free until 1994 (give or take a bit). You haven't lived until you've resolved device driver configuration and ordering problems when trying to link a binary version of SunOS 2 or 3 (no, not Solaris :-). Or even better, an NBI VME 68K box pretending it's a UNIBUS VAX. Is alt.folklore.computers still alive? 99% of the list traffic [cs]hould be redirected there. -- (creaky/grumpy olde) lyndon
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 7:13 PM, ron minnich [EMAIL PROTECTED] wrote: When I first got v9fs working, 1998, I tried mounting file systems over 9p. What a mess. Things just broke in weird ways. There is code that really wants a symlink to be there and readable. I can't even recall all the places, but they're there. And things break if you mount and don't have symlinks. Which is why I put readlink etc. in my v9fs, and why the .U version is in today's linux kernels. And now you need extended attributes in order to support SElinux or RedHat behaves weirdly. It really never does end.
Re: [9fans] why not Lvx for Plan 9?
On Tue, Jul 8, 2008 at 8:17 PM, Eric Van Hensbergen [EMAIL PROTECTED] wrote: And now you need extended attributes in order to support SElinux or RedHat behaves weirdly. It really never does end. unless we all get smart and go into banking. ron
Re: [9fans] why not Lvx for Plan 9?
And now you need extended attributes in order to support SElinux or RedHat behaves weirdly. It really never does end. unless we all get smart and go into banking. then you'll have fun chasing a different set of endlessly changing rules. - erik