From pedersen at ccsscorp.com Thu Jun 13 00:09:21 2013
From: pedersen at ccsscorp.com (Bill Pedersen)
Date: Wed, 12 Jun 2013 18:09:21 -0400
Subject: [FreeVMS] FreeVMS
Message-ID: <001401ce67b9$7e464c30$7ad2e490$@ccsscorp.com>
Hello, new to the list but interested.
And definitely not new to VMS!
How does one get information on the current state of FreeVMS. I have
visited the freeVMS.net site but find that there is a lot of incompleteness
there, links that go to blank pages and such.
Given the current upheaval in the VMS community I would like to better
understand this environment so as to be able to help promote it and
encourage it.
Thank,
Bill.
Bill Pedersen
CCSS - Computer Consulting System Services, LLC
211 Ruth Drive
Gaffney, SC 29341
Telephone: 864-490-8863
Mobile: 408-892-5204
Facsimile: 206-984-3068
www: www.ccsscorp.com
Skype Name: william.a.pedersen
LinkedIn Profile:
www.linkedin.com/in/billpedersen
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From joel.bertrand at systella.fr Thu Jun 13 07:43:37 2013
From: joel.bertrand at systella.fr (=?UTF-8?Q?BERTRAND_Jo=C3=ABl?=)
Date: Thu, 13 Jun 2013 07:43:37 +0200
Subject: [FreeVMS] FreeVMS
In-Reply-To: <001401ce67b9$7e464c30$7ad2e490$@ccsscorp.com>
References: <001401ce67b9$7e464c30$7ad2e490$@ccsscorp.com>
Message-ID: <6823d292689a27b0003aec2c0ee9d374@systella.fr>
Le 13.06.2013 0:09, Bill Pedersen a ?crit?:
> Hello, new to the list but interested?
Hello,
> And definitely not new to VMS!
>
> How does one get information on the current state of FreeVMS. I have
> visited the freeVMS.net site but find that there is a lot of
> incompleteness there, links that go to blank pages and such.
>
> Given the current upheaval in the VMS community I would like to
> better
> understand this environment so as to be able to help promote it and
> encourage it.
FreeVMS project was created a long time ago. First kernels (until 0.3)
were derivated from Linux kernel (2.4.15 if I remember) and were
monolithic. These kernels were bootable but not easely maintenable.
0.4 is rewritten from scratch on a L4 (like Hurd is written on a Mach
kernel or MacOS X on NVU). L4 is choosen because it deos not contains
unixism. It boots, it contains first memory management subroutines, but
pager is missing.
Now, FreeVMS is slepping because there is no developers. To be alive,
this project need :
- a pager ;
- some drivers (disk I/O), filesystems... Some code have been written
(ODS2/5 for example) for O.3 kernel ;
- a swapper ;
- some libraries (a real POSIX libc et VMS libraries) ;
- some contributers to write documentation and write pages on
www.freevms.net.
Goal of FreeVMS is to propose a clone of OpenVMS with a good POSIX
subsystem. Microkernel (L4) is written in C++ and servers written over
L4 are written in procedural C++ too. First arch is amd64.
Regards,
JKB
From mveety at gmail.com Thu Jun 13 07:54:50 2013
From: mveety at gmail.com (Matthew Veety)
Date: Thu, 13 Jun 2013 01:54:50 -0400
Subject: [FreeVMS] FreeVMS
In-Reply-To: <6823d292689a27b0003aec2c0ee9d374@systella.fr>
References: <001401ce67b9$7e464c30$7ad2e490$@ccsscorp.com>
<6823d292689a27b0003aec2c0ee9d374@systella.fr>
Message-ID:
On Jun 13, 2013, at 1:43, BERTRAND Jo?l wrote:
> Le 13.06.2013 0:09, Bill Pedersen a ?crit :
>> Hello, new to the list but interested?
>
> Hello,
>
>> And definitely not new to VMS!
>>
>> How does one get information on the current state of FreeVMS. I have
>> visited the freeVMS.net site but find that there is a lot of
>> incompleteness there, links that go to blank pages and such.
>>
>> Given the current upheaval in the VMS community I would like to better
>> understand this environment so as to be able to help promote it and
>> encourage it.
>
> FreeVMS project was created a long time ago. First kernels (until 0.3) were derivated from Linux kernel (2.4.15 if I remember) and were monolithic. These kernels were bootable but not easely maintenable.
>
> 0.4 is rewritten from scratch on a L4 (like Hurd is written on a Mach kernel or MacOS X on NVU). L4 is choosen because it deos not contains unixism. It boots, it contains first memory management subroutines, but pager is missing.
>
> Now, FreeVMS is slepping because there is no developers. To be alive, this project need :
> - a pager ;
> - some drivers (disk I/O), filesystems... Some code have been written (ODS2/5 for example) for O.3 kernel ;
> - a swapper ;
> - some libraries (a real POSIX libc et VMS libraries) ;
> - some contributers to write documentation and write pages on www.freevms.net.
>
> Goal of FreeVMS is to propose a clone of OpenVMS with a good POSIX subsystem. Microkernel (L4) is written in C++ and servers written over L4 are written in procedural C++ too. First arch is amd64.
>
> Regards,
>
> JKB
Is development even going on still? The project seems really dead to me.
Also why didn't we write a kernel ourselves?
From joel.bertrand at systella.fr Thu Jun 13 08:11:16 2013
From: joel.bertrand at systella.fr (=?UTF-8?Q?BERTRAND_Jo=C3=ABl?=)
Date: Thu, 13 Jun 2013 08:11:16 +0200
Subject: [FreeVMS] FreeVMS
In-Reply-To:
References: <001401ce67b9$7e464c30$7ad2e490$@ccsscorp.com>
<6823d292689a27b0003aec2c0ee9d374@systella.fr>
Message-ID: <773dee3f7dee8121993e005bd4370213@systella.fr>
Le 13.06.2013 7:54, Matthew Veety a ?crit?:
> On Jun 13, 2013, at 1:43, BERTRAND Jo?l
> wrote:
>
>> Le 13.06.2013 0:09, Bill Pedersen a ?crit :
>>> Hello, new to the list but interested?
>>
>> Hello,
>>
>>> And definitely not new to VMS!
>>>
>>> How does one get information on the current state of FreeVMS. I
>>> have
>>> visited the freeVMS.net site but find that there is a lot of
>>> incompleteness there, links that go to blank pages and such.
>>>
>>> Given the current upheaval in the VMS community I would like to
>>> better
>>> understand this environment so as to be able to help promote it and
>>> encourage it.
>>
>> FreeVMS project was created a long time ago. First kernels (until
>> 0.3) were derivated from Linux kernel (2.4.15 if I remember) and were
>> monolithic. These kernels were bootable but not easely maintenable.
>>
>> 0.4 is rewritten from scratch on a L4 (like Hurd is written on a
>> Mach kernel or MacOS X on NVU). L4 is choosen because it deos not
>> contains unixism. It boots, it contains first memory management
>> subroutines, but pager is missing.
>>
>> Now, FreeVMS is slepping because there is no developers. To be
>> alive, this project need :
>> - a pager ;
>> - some drivers (disk I/O), filesystems... Some code have been
>> written (ODS2/5 for example) for O.3 kernel ;
>> - a swapper ;
>> - some libraries (a real POSIX libc et VMS libraries) ;
>> - some contributers to write documentation and write pages on
>> www.freevms.net.
>>
>> Goal of FreeVMS is to propose a clone of OpenVMS with a good POSIX
>> subsystem. Microkernel (L4) is written in C++ and servers written over
>> L4 are written in procedural C++ too. First arch is amd64.
>>
>> Regards,
>>
>> JKB
>
> Is development even going on still? The project seems really dead to
> me.
Project is sleeping as it need some developers.
> Also why didn't we write a kernel ourselves?
I don't understand your question. Do you want to reuse an existing
kernel or to write a kernel without L4 ?
JKB
From joel.bertrand at systella.fr Thu Jun 13 11:44:57 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 13 Jun 2013 11:44:57 +0200
Subject: [FreeVMS] Wanted
Message-ID: <51B99499.6010106@systella.fr>
Hello,
I have checked this morning and there are more than 1000 subscribers to
this mailing list (even with hp.com mail address). HP says that, today,
it remains only 2500 OpenVMS customers. Thus, a lot of VMS users read
this mailing list.
Development was stalled for two years because I don't want to continue
alone. I'll restart this project if and only if new contributers support
parts of this OS. Haiku, ReactOS, Linux, *BSD are much more complex than
a clone of VMS over microkernel and are actively developed for years. If
we cannot found some developers to participate to this project, I
suppose that OpenVMS will be definitely dead when HP will stop support.
I have shown that it is possible to write a kernel as a sum if L4
servers. 0.4 boots until process management but as pager is missing, it
cannot launch tasks in virtual memory (sigma0 only maps physical memory
in virtual memory without address translation).
Now, HP has announced OpenVMS death. I don't think that OpenVMS will be
opensourced, and even if HP opensources OpenVMS, there will be a lot of
stuff to obtain a usable operating system (Bliss, Macro32 and 64 have to
be ported on new architectures). I'm not sure that rewrite a new OS is
more difficult.
Today, FreeVMS 0.4 requires :
- a pager (FreeVMS/sources/freevms/vm/vm_pager.c)
- some drivers (console, hard disk, network)
- a swapper
- a command line
- some volunteers to keep webpages alive.
If anyone is interested, please answer to this mail. With more than 5
new and active developers, I think we can restart this project.
Regards,
JKB
From rmctwo at gmail.com Thu Jun 13 14:53:03 2013
From: rmctwo at gmail.com (Renee)
Date: Thu, 13 Jun 2013 08:53:03 -0400
Subject: [FreeVMS] Wanted
In-Reply-To: <51B99499.6010106@systella.fr>
References: <51B99499.6010106@systella.fr>
Message-ID: <000401ce6834$f1ead450$d5c07cf0$@gmail.com>
Hi joel,
I have worked in VMS development for about 4.5 years in Nashua and did HBS.
I'd like to help.
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
BERTRAND Jo?l
Sent: Thursday, June 13, 2013 5:45 AM
To: freevms at systella.fr
Subject: [FreeVMS] Wanted
Hello,
I have checked this morning and there are more than 1000 subscribers
to this mailing list (even with hp.com mail address). HP says that, today,
it remains only 2500 OpenVMS customers. Thus, a lot of VMS users read this
mailing list.
Development was stalled for two years because I don't want to
continue alone. I'll restart this project if and only if new contributers
support parts of this OS. Haiku, ReactOS, Linux, *BSD are much more complex
than a clone of VMS over microkernel and are actively developed for years.
If we cannot found some developers to participate to this project, I suppose
that OpenVMS will be definitely dead when HP will stop support.
I have shown that it is possible to write a kernel as a sum if L4
servers. 0.4 boots until process management but as pager is missing, it
cannot launch tasks in virtual memory (sigma0 only maps physical memory in
virtual memory without address translation).
Now, HP has announced OpenVMS death. I don't think that OpenVMS will
be opensourced, and even if HP opensources OpenVMS, there will be a lot of
stuff to obtain a usable operating system (Bliss, Macro32 and 64 have to be
ported on new architectures). I'm not sure that rewrite a new OS is more
difficult.
Today, FreeVMS 0.4 requires :
- a pager (FreeVMS/sources/freevms/vm/vm_pager.c)
- some drivers (console, hard disk, network)
- a swapper
- a command line
- some volunteers to keep webpages alive.
If anyone is interested, please answer to this mail. With more than
5 new and active developers, I think we can restart this project.
Regards,
JKB
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From joel.bertrand at systella.fr Thu Jun 13 17:35:48 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 13 Jun 2013 17:35:48 +0200
Subject: [FreeVMS] Wanted
In-Reply-To: <000401ce6834$f1ead450$d5c07cf0$@gmail.com>
References: <51B99499.6010106@systella.fr>
<000401ce6834$f1ead450$d5c07cf0$@gmail.com>
Message-ID: <51B9E6D4.6040309@systella.fr>
Renee wrote:
> Hi joel,
>
> I have worked in VMS development for about 4.5 years in Nashua and did HBS.
> I'd like to help.
Thanks a lot. You're welcome.
Today, some modules can be directly loaded into memory by kickstart
(FreeVMS/sources/kernel/util/kickstart). To boot, grub loads kickstart
from a disk (including ODS/2 disk) and load into memory several modules :
- x86-kernel (L4)
- sigma0
- freevms.sys (L4 root task)
- pager.sys (root task pager)
- init.exe
This boot procedure works fine. It works fine but pager.sys does not do
anything. Do you think you could write a VMS like pager ? I will put on
freevms.net all documentations about L4/X2 microkernel. I have tried to
rebuild last git this afternoon, I have to patch L4 to build microkernel
with last gcc. I think I'll have more time next week end.
Regards,
JKB
From denis.reichert at laposte.net Thu Jun 13 20:23:11 2013
From: denis.reichert at laposte.net (Denis Reichert)
Date: Thu, 13 Jun 2013 19:23:11 +0100
Subject: [FreeVMS] Wanted
In-Reply-To: <51B99499.6010106@systella.fr>
References: <51B99499.6010106@systella.fr>
Message-ID: <51BA0E0F.3060209@laposte.net>
Une pi?ce jointe HTML a ?t? nettoy?e...
URL:
From rmctwo at gmail.com Thu Jun 13 20:52:42 2013
From: rmctwo at gmail.com (Renee)
Date: Thu, 13 Jun 2013 14:52:42 -0400
Subject: [FreeVMS] Wanted
In-Reply-To: <51BA0E0F.3060209@laposte.net>
References: <51B99499.6010106@systella.fr> <51BA0E0F.3060209@laposte.net>
Message-ID: <000801ce6867$30131440$90393cc0$@gmail.com>
Macro32 was part of VMS. I was not around when 64 bit vms was.
Bliss32 is on the User pack.
Renee
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Denis Reichert
Sent: Thursday, June 13, 2013 2:23 PM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] Wanted
Le 13/06/2013 10:44, BERTRAND Jo?l a ?crit :
Hello,
I have checked this morning and there are more than 1000 subscribers to
this mailing list (even with hp.com mail address). HP says that, today, it
remains only 2500 OpenVMS customers. Thus, a lot of VMS users read this
mailing list.
Development was stalled for two years because I don't want to continue
alone. I'll restart this project if and only if new contributers support
parts of this OS. Haiku, ReactOS, Linux, *BSD are much more complex than a
clone of VMS over microkernel and are actively developed for years. If we
cannot found some developers to participate to this project, I suppose that
OpenVMS will be definitely dead when HP will stop support.
I have shown that it is possible to write a kernel as a sum if L4
servers. 0.4 boots until process management but as pager is missing, it
cannot launch tasks in virtual memory (sigma0 only maps physical memory in
virtual memory without address translation).
Now, HP has announced OpenVMS death. I don't think that OpenVMS will be
opensourced, and even if HP opensources OpenVMS, there will be a lot of
stuff to obtain a usable operating system (Bliss, Macro32 and 64 have to be
ported on new architectures). I'm not sure that rewrite a new OS is more
difficult.
Today, FreeVMS 0.4 requires :
- a pager (FreeVMS/sources/freevms/vm/vm_pager.c)
- some drivers (console, hard disk, network)
- a swapper
- a command line
- some volunteers to keep webpages alive.
If anyone is interested, please answer to this mail. With more than 5
new and active developers, I think we can restart this project.
Regards,
JKB
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
Hello
I worked as a low level system programmer a long time ago but when I look
today at LIB$ sources, F11 assembler manual, all this is still familiar to
me.
I have never understood why ODS-like file versioning is not implemented into
recent filesystems.
About web pages, is it not better to open a WordPress blog which is more
modern, attractive, interactive and collaborative ?
Regards
Denis
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From joel.bertrand at systella.fr Thu Jun 13 20:58:12 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 13 Jun 2013 20:58:12 +0200
Subject: [FreeVMS] Wanted
In-Reply-To: <51BA0E0F.3060209@laposte.net>
References: <51B99499.6010106@systella.fr> <51BA0E0F.3060209@laposte.net>
Message-ID: <51BA1644.4060702@systella.fr>
> Hello
> I worked as a low level system programmer a long time ago but when I
> look today at LIB$ sources, F11 assembler manual, all this is still
> familiar to me.
> I have never understood why ODS-like file versioning is not implemented
> into recent filesystems.
> About web pages, is it not better to open a WordPress blog which is more
> modern, attractive, interactive and collaborative ?
For me, main webpages cannot be a blog or a wiki. Actual webpages are
written with spip, that is a multiuser and multilinguage CRM. If you
want to modify a page, you need an account. On each page, you can add a
public or private space and there are several internal forums.
From joel.bertrand at systella.fr Thu Jun 13 23:53:25 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 13 Jun 2013 23:53:25 +0200
Subject: [FreeVMS] FreeVMS volunteers
Message-ID: <51BA3F55.7020502@systella.fr>
John,
I have receive your mail, but I think you have sent this mail to a
wrong address (to mailman bounce address ?).
You're welcome. If you know internal VMS structures, your knowledge
will be a great help to design the new OS.
I think that all documentations should be written and accessible on web
pages. I can open for you an access. Maybe you don't know spip. It's a
collaborative platform. You can prepare several articles on private
pages, all articles will be read by some reviewers and will be published
after acceptation.
Regards,
JKB
From joel.bertrand at systella.fr Fri Jun 14 11:58:42 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Fri, 14 Jun 2013 11:58:42 +0200
Subject: [FreeVMS] Auto-discard notification
In-Reply-To:
References:
Message-ID: <51BAE952.8010106@systella.fr>
>On Thu, Jun 13, 2013 at 11:44:57AM +0200, BERTRAND Jo?l wrote:
>> Today, FreeVMS 0.4 requires :
>> - a pager (FreeVMS/sources/freevms/vm/vm_pager.c)
>> - some drivers (console, hard disk, network)
>> - a swapper
>> - a command line
>> - some volunteers to keep webpages alive.
>>
>I will be happy to help somehow. But I'm not a great VMS expert,
>and I think I will be only able to start writing only some
>drivers and/or propose a hardware abstraction ala NetBSD (for
>example).
>
>Cheers,
I'm not sure that FreeVMS requires today great VMS experts. I'm
currently trying to update git repository to push all modifications I
have made since 2010. I think that this week-end, you will able to find
a disk image that boots under qemu and you will see if you can help this
project. I will upload L4/X2 documentation too.
Here is console output (with some comments) :
riemann:[~/openvms] > qemu-system-x86_64 -hdc freevms.img -smp 1 -serial
stdio -m 128M -k fr -display none
// Grub is loading kickstart. Kickstart loads into memory several
// modules.
KickStart 0.12.56 for FreeVMS
Detected multiboot compliant loader
Reserving 8MB for kernel memory
L4/X2 kernel (0x0011b000-0x0018c00a) => 0x0095a000
(0x0011c000-0x00175510) -> 0x00800000-0x00859510
(0x00176000-0x0017de70) -> 0x0095a000-0x00961e70
sigma0 (0x0018d000-0x001a6b30) => 0x00080000
(0x0018d0c0-0x00193080) -> 0x00080000-0x00085fc0
vmskernel.sys (0x001a7000-0x001f4775) => 0x01003d10
(0x001a70b0-0x00220f18) -> 0x010000b0-0x01079f18
Launching kernel ...
// Kickstart launchs L4/X2 microkernel (kernel mode)
L4Ka::Pistachio - built on Dec 7 2010 08:23:14 by bertrand at cauchy using
gcc version 4.4.5 20100728 (prerelease)
// starting L4 root server over sigma0 (idempotent memory mapper).
>>> FreeVMS 0.4.0 (R)
%SYSBOOT-I-SYSBOOT, leaving kernel privileges
%SYSBOOT-I-SYSBOOT, launching FreeVMS kernel with executive privileges
%SYSBOOT-I-SYSBOOT, booting main processor
%SYSBOOT-I-SYSBOOT, computing page size: 4096 bytes
%SYSBOOT-I-SYSBOOT, CPU0 EXTFREQ=1001 MHz, INTFREQ=1648 MHz
%SYSBOOT-I-SYSBOOT, parsing command line: /boot/vmskernel.sys
root=dqa0.0 console=opa0 nodbg$virtual_memory dbg$sys_pagefault
dbg$vms_pagefault
%SYSBOOT-I-SYSBOOT, selecting root device: DQA0.0
%SYSBOOT-I-SYSBOOT, selecting console device: OPA0
%SYSBOOT-I-SYSBOOT, initializing memory
%MEM-I-AREA, $0000000000000000 - $000000000007FFFF: physical memory
%MEM-I-AREA, $0000000000086000 - $000000000009FBFF: physical memory
%MEM-I-AREA, $0000000000100000 - $0000000000108FFF: physical memory
%MEM-I-AREA, $000000000010B000 - $00000000001F4FFF: physical memory
%MEM-I-AREA, $0000000000221000 - $00000000007FFFFF: physical memory
%MEM-I-AREA, $000000000095A000 - $0000000000FFFFFF: physical memory
%MEM-I-AREA, $0000000001010000 - $00000000075FFFFF: physical memory
%MEM-I-AREA, $0000000007E00000 - $0000000007FFDFFF: physical memory
%MEM-I-AREA, $0000000000001000 - $FFFFFEFFFFFFFFFF: virtual memory
%MEM-I-AREA, $000000000009FC00 - $00000000000FFFFF: mapped IO
%MEM-I-AREA, $0000000007FFE000 - $FFFFFFFFFFFFFFFF: mapped IO
%MEM-I-AREA, $0000000007600000 - $0000000007DFFFFF: reserved by kernel
%MEM-I-AREA, $0000000000800000 - $0000000000859FFF: reserved by kernel
%MEM-I-AREA, $000000000085A000 - $0000000000959FFF: reserved by kernel
%MEM-I-AREA, $0000000000832000 - $0000000000832FFF: reserved by kernel
%MEM-I-AREA, $FFFFFFFFFFC00000 - $FFFFFFFFFFC00FFF: reserved by kernel
%MEM-I-AREA, $00000000FEC00000 - $00000000FEC00FFF: reserved by kernel
%MEM-I-AREA, $0000000000080000 - $0000000000085FC0: reserved by kernel
%MEM-I-AREA, $0000000000080000 - $0000000000084817: modules
%MEM-I-AREA, $0000000000084818 - $0000000000084837: kernel
%MEM-I-AREA, $00000000010000B0 - $000000000100EA77: modules
%MEM-I-AREA, $000000000100EA78 - $000000000100F257: kernel
%MEM-I-AREA, $00000000001F5000 - $000000000020120A: modules
%MEM-I-AREA, $0000000000202000 - $000000000020B97A: modules
%MEM-I-AREA, $000000000020C000 - $0000000000212C3A: modules
%MEM-I-AREA, $0000000000213000 - $0000000000219C3A: modules
%MEM-I-AREA, $000000000021A000 - $0000000000220C3A: modules
%MEM-I-AREA, $000000000010A000 - $000000000010AFFF: boot information
%MEM-I-AREA, $0000000000109000 - $000000000010933F: boot information
%SYSBOOT-I-SYSBOOT, reserving memory for preloaded objects
%MEM-I-ALLOC, allocating $0000000000080000 - $0000000000084817
%MEM-I-ALLOC, allocating $0000000000084818 - $0000000000084837
%MEM-I-ALLOC, allocating $00000000010000B0 - $000000000100EA77
%MEM-I-ALLOC, allocating $000000000100EA78 - $000000000100F257
%MEM-I-ALLOC, allocating $00000000001F5000 - $000000000020120A
%MEM-I-ALLOC, allocating $0000000000202000 - $000000000020B97A
%MEM-I-ALLOC, allocating $000000000020C000 - $0000000000212C3A
%MEM-I-ALLOC, allocating $0000000000213000 - $0000000000219C3A
%MEM-I-ALLOC, allocating $000000000021A000 - $0000000000220C3A
%MEM-I-ALLOC, allocating $000000000010A000 - $000000000010AFFF
%MEM-I-ALLOC, allocating $0000000000109000 - $000000000010933F
%MEM-I-F_ALLOC, bootstrapping Fpage allocator at virtual addresses
%MEM-I-F_ALLOC, $0000000000001000 - $000000000007FFFF
%MEM-I-S_ALLOC, bootstrapping Slab allocator at physical addresses
%MEM-I-S_ALLOC, $0000000000000000 - $000000000007FFFF
%MEM-I-FREE, freeing region $0000000000086000 - $000000000009EFFF
%MEM-I-FREE, freeing region $0000000000100000 - $0000000000108FFF
%MEM-I-FREE, freeing region $000000000010B000 - $00000000001F4FFF
%MEM-I-FREE, freeing region $0000000000221000 - $00000000007FFFFF
%MEM-I-FREE, freeing region $000000000095A000 - $0000000000FFFFFF
%MEM-I-FREE, freeing region $0000000001010000 - $00000000075FFFFF
%MEM-I-FREE, freeing region $0000000007E00000 - $0000000007FFDFFF
%MEM-I-VM_ALLOC, adding $0000000000085000 - $0000000000108FFF to VM
allocator
%MEM-I-VM_ALLOC, adding $000000000010B000 - $00000000001F4FFF to VM
allocator
%MEM-I-VM_ALLOC, adding $0000000000221000 - $0000000000FFFFFF to VM
allocator
%MEM-I-VM_ALLOC, adding $0000000001010000 - $FFFFFEFFFFFFFFFF to VM
allocator
%JOBCTL-I-MAX_THRDS, setting maximal concurrent threads to 262085
%DEV-I-INIT, initializing devices tree
%NSR-I-INIT, initializing names tree
%SYSBOOT-I-SYSBOOT, spawning PAGER.SYS with executive privileges
%SYSBOOT-I-SYSBOOT, creating PAGER.SYS UTCB pages
%SYSBOOT-I-SYSBOOT, reserving 4096 bytes for 4 pager threads
// PAGER.SYS is loaded by FreeVMS kernel in the same memory space.
// Pagefaults come from memory requests sent by $3C process id
// (PAGER.SYS) to freevms kernel pagefault handler. See
// http://www.freevms.net/article35.html
%SYS-I-PAGEFLT, pagefault request from $3C at $0000000000800200
%SYS-I-PAGEFLT, pagefault request from $3C at $0000000000007FE8
%SYS-I-PAGEFLT, pagefault request from $3C at $0000000000801C50
%SYS-I-PAGEFLT, pagefault request from $3C at $00000000008023C8
%SYS-I-PAGEFLT, pagefault request from $3C at $0000000000300000
%SYS-I-PAGEFLT, pagefault request from $3C at $000000000100C598
// All ELF segments from PAGER.SYS have been loaded, PAGER.SYS is
// started. Problem : PAGER.SYS does not do anything.
%RUN-S-STARTED, PAGER.SYS process started
%PAGER-I-SWPADDR, setting swapper base address at $0000000007FFE000
%PAGER-I-SWPSIZE, setting swap size to 0 byte
// Here, FreeVMS kernel has to start some modules like disk drivers,
// console driver... But to load these modules, PAGER.SYS has to be
// written to offer real memory mapping, not idempotent mapping.
// PAGER.SYS is in sources/freevms/vm
%RUN-S-STOPPED, PAGER.SYS process stopped
qemu: terminating on signal 2
Before trying to write drivers, next step will be to write a minimal
PAGER.SYS. I can help you to do that if you want.
Regards,
JKB
From joel.bertrand at systella.fr Fri Jun 14 13:04:53 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Fri, 14 Jun 2013 13:04:53 +0200
Subject: [FreeVMS] Auto-discard notification
In-Reply-To: <51BAE952.8010106@systella.fr>
References:
<51BAE952.8010106@systella.fr>
Message-ID: <51BAF8D5.6030303@systella.fr>
Bruno, please use your real mail address. You can only send messages
with your mailman's registered address.
That being said, I will patch L4/X2. I have sent some patches to Ka
devs a long time ago I have never pushed into git tree.
For virtual memory, don't worry, FreeVMS only requires in a first time a
pagefault handler that is able to give virtual memory to child process
(and that is able to associate physical to virtual pages and to process ID).
Regards,
JKB
From joel.bertrand at systella.fr Sun Jun 16 17:42:32 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sun, 16 Jun 2013 17:42:32 +0200
Subject: [FreeVMS] FreeVMS disk image update
In-Reply-To: <02c201ce69c7$229d5df0$67d819d0$@ccsscorp.com>
References: <02c201ce69c7$229d5df0$67d819d0$@ccsscorp.com>
Message-ID: <51BDDCE8.9010807@systella.fr>
Hello,
I have uploaded this afternoon a qemu image with a recent L4/X2 kernel
(on http://www.freevms.net), the last vmskernel.sys and some
modifications to boot this system with grub2 instead of grub-lecacy.
This kernel supports until 4 processors. System boots, starts PAGER.SYS
that doesn't do anything and stops.
Regards,
JKB
From mveety at gmail.com Sun Jun 16 17:49:47 2013
From: mveety at gmail.com (Matthew Veety)
Date: Sun, 16 Jun 2013 11:49:47 -0400
Subject: [FreeVMS] FreeVMS disk image update
In-Reply-To: <51BDDCE8.9010807@systella.fr>
References: <02c201ce69c7$229d5df0$67d819d0$@ccsscorp.com>
<51BDDCE8.9010807@systella.fr>
Message-ID:
What exactly needs to be done on the system?
On Jun 16, 2013, at 11:42, BERTRAND Jo?l wrote:
> Hello,
>
> I have uploaded this afternoon a qemu image with a recent L4/X2 kernel (on http://www.freevms.net), the last vmskernel.sys and some modifications to boot this system with grub2 instead of grub-lecacy. This kernel supports until 4 processors. System boots, starts PAGER.SYS that doesn't do anything and stops.
>
> Regards,
>
> JKB
> _______________________________________________
> FreeVMS mailing list
> FreeVMS at rayleigh.systella.fr
> https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From pedersen at ccsscorp.com Sun Jun 16 20:42:10 2013
From: pedersen at ccsscorp.com (Bill Pedersen)
Date: Sun, 16 Jun 2013 14:42:10 -0400
Subject: [FreeVMS] UserMode VMS on *Nix.
In-Reply-To: <51BDFB12.8000406@encompasserve.org>
References: <02c201ce69c7$229d5df0$67d819d0$@ccsscorp.com>
<51BDDCE8.9010807@systella.fr> <51BDEF1E.7040807@sneddon.id.au>
<51BDFB12.8000406@encompasserve.org>
Message-ID: <001501ce6ac1$36d600b0$a4820210$@ccsscorp.com>
As far as a tool for collaboration I think this is decent and it can be
"private":
http://www.google.com/sites/help/intl/en/overview.html
It is not as good as Google Wave as far as interactive collaboration but it
will give us a location where we can have a discussion and also cooperate in
the development of ideas.
Bill.
Bill Pedersen
CCSS - Computer Consulting System Services, LLC
211 Ruth Drive
Gaffney, SC 29341
Telephone: 864-490-8863
Mobile: 408-892-5204
Facsimile: 206-984-3068
www: www.ccsscorp.com
Skype Name: william.a.pedersen
LinkedIn Profile: www.linkedin.com/in/billpedersen
-----Original Message-----
From: John E. Malmberg [mailto:malmberg at Encompasserve.org]
Sent: Sunday, June 16, 2013 1:51 PM
To: Tim Sneddon
Cc: BERTRAND Jo?l; Bill Pedersen; Keith Parris; Fr?d?ric BOYER;
freevms at systella.fr; Bruno Ducrot
Subject: UserMode VMS on *Nix.
On 6/16/2013 12:00 PM, Tim Sneddon wrote:
Tim, I am basically in full agreement with your analysis.
I chose Linux over BSD simply because of perceived market share, and the
knowledge that LAT and DECNET are available in some state.
I would want a version that runs on my current laptop as a VM under
virtualbox, and one that I could run on one of the desktop systems that I
need to repair.
> The real killer will be emulators. Why move to the "new VMS" when the
> "real VMS" can be run on an emulator faster than real hardware anyway?
Exactly. But the emulators are not yet faster than the high end hardware
though. But that is only a matter of time.
Basically the existing VMS commercial users that can use emulators are going
to use them. The ones that can not are already preparing to move off of
VMS.
The ones that move to Linux based emulators or to Linux may adopt standalone
features that augment them from a project like what we are investigating.
So that and hobbyist is probably our target market.
So I think that we need to have a list of deliverable packages that can be
linked together.
I think that we should bundle an open source emulator into it. SimH/VAX is
stable.
But I do not know about ES-40. Is it usable at all?
There are hacks that we can make to the emulator and to VMS running on it to
make it run even better. And you can bet that any hacks that we developed
will be picked up by the commercial vendors.
> Lastly, I think we need to pull together a list of absolute base line
> goals of the project and start working out from that. I made my
> project decisions based on quite a bit of research I did personally
> over a few months (between work and everything else...). I am happy
> to re-visit all this with a new collection of minds.
>
> These are the first two goals I would like to put forward:
>
> * User-mode public API compatibility.
Is the goal to be able to build any program on HybridVMS and
also have it work on a real VMS system?
My take is no. This would be like a 10.x API, where old
programs could be built on it, as a forward migration.
There are some obvious bugs in the system services.
I would prefer to clean them up, but provide a compile time
macro to support the incorrect interface to make them cleaner.
I would tend to implement system services in an object
oriented way with a backward compatibility layer.
>
> * Open source (preferably BSD or GPLv3) licensed source code.
Curl uses a more unrestricted open source license than BSD or
lGPLv3.
* *Nix/VMS interoperability. NFS/NTP/ODS2/ODS5/LAT/DECNET etc.
Is there a google docs tool that could be used to help organize these
options?
Does anyone have a list of existing VMS compatible packages on Linux and BSD
that can be used either as a base or to help decide on which platform?
Regards,
-John
From joel.bertrand at systella.fr Sun Jun 16 23:09:35 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Sun, 16 Jun 2013 23:09:35 +0200
Subject: [FreeVMS] FreeVMS disk image update
In-Reply-To:
References: <02c201ce69c7$229d5df0$67d819d0$@ccsscorp.com>
<51BDDCE8.9010807@systella.fr>
Message-ID: <51BE298F.4070301@systella.fr>
Matthew Veety a ?crit :
> What exactly needs to be done on the system?
In a first time, memory management. The first mapping over L4 is
idempotent (virtual address are directly mapped on pysical ones) and is
only used to start root task (freevms.sys).
With this memory management, system needs IO drivers and process
management. L4/X2 uses threads and address space, but there is no system
to associate address space and thread to process ID ? la VMS.
In a third time, we must write all basic VMS library and port newlib to
provide a real POSIX compatibility.
Of course, we can discuss about a more precise roadmap.
Regards,
JKB
From joel.bertrand at systella.fr Sun Jun 16 23:29:18 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sun, 16 Jun 2013 23:29:18 +0200
Subject: [FreeVMS] FreeVMS disk image update
In-Reply-To: <51BDEF1E.7040807@sneddon.id.au>
References: <02c201ce69c7$229d5df0$67d819d0$@ccsscorp.com>
<51BDDCE8.9010807@systella.fr> <51BDEF1E.7040807@sneddon.id.au>
Message-ID: <51BE2E2E.6010101@systella.fr>
Tim Sneddon a ?crit :
> On 16/06/2013 11:42 PM, BERTRAND Jo?l wrote:
>> Hello,
>>
>> I have uploaded this afternoon a qemu image with a recent L4/X2 kernel
>> (on http://www.freevms.net), the last vmskernel.sys and some
>> modifications to boot this system with grub2 instead of grub-lecacy.
>> This kernel supports until 4 processors. System boots, starts
>> PAGER.SYS that doesn't do anything and stops.
>>
> Hi Guys,
>
> There have been a number of posts going backwards and forwards and
> keeping track of all the different points is a bit difficult, so I
> thought I might just summarize my thoughts regarding some of the points.
>
> Before coming up with my OpenBSD based plan I had looked at L4. There
> are lots of good reasons for going with L4.
>
> 1. VMS on Mach was already done. Some of the high level decisions made
> were quite good are a discussed in a paper submitted to USENIX. This
> would give an implementer a clear direction.
>
> 2. L4 does have a hypervisor. It is part of a related project and I
> believe it could certainly be used. It is called NOVA.
>
> 3. Dresden University of Technology has been putting heaps of effort
> into L4 in the form of Fiasco.OC,
> http://www.inf.tu-dresden.de/index.php?node_id=2697.
> They provide quite
> decent supporting libraries to get projects off the ground.
When we have choosen L4 four years ago, we have choosen L4Ka, not
Fiasco, as L4Ka API was better than L4/Fiasco one. Fiasco comes with
L4RE, but this toolkit needs corba and some other technologies.
If I remember L4Ka was also designed to provide very short IPCs.
> 4. Linux drivers could likely be used in a shell process that presents
> the correct environment to the driver.
With L4, each driver can run in its own address space (and of course in
a separated thread).
> However, while that all sounds great. This was not something that I
> could easily handle on my own and so I made the decisions, related to my
> project that I would only support user mode public interfaces. Don't
> bother with presenting four modes. This is a great design feature of
> VMS, but not critical to get people with user mode applications off the
> ground. VMS engineering have made the same decisions when supporting
> code between VAX, Alpha and I64. So, I did the same. There is also a lot
> less work in writing a bunch of libraries that provide a supported call
> interfaces, rather than an entire operating system.
>
> My feeling is that most people who are looking for somewhere to go are
> likely unconcerned with whether or not the shell runs in supervisor mode
> or if RMS runs in exec mode. I think that modifying something like BSD
> UNIX to support some underlying features, like logicals and ASTs and
> then handling everything else with a compat package will be
> comparatively easier to implement. The other things is that all that
> UNIX compatibility that the masses are begging for is right there!
Sure, but Linux or xBSD kernels are monolithic. Thus, you have to patch
mainstream kernel (we have try until 0.3 FreeVMS kernel). It's a huge
work too.
With a L4 kernel, you only have to know root task's API to write a new
kernel that cannot crash your running system (it runs as a separate task
in a separate address space). Critical tasks are only root task, its
pager and pagefault handler.
> These are the first two goals I would like to put forward:
>
> * User-mode public API compatibility.
>
> * Open source (preferably BSD or GPLv3) licensed source code.
Regards,
JKB
PS: if you send mail to freevms at systella.fr, you must subsribe to the list.
From johann at myrkraverk.com Mon Jun 17 02:27:51 2013
From: johann at myrkraverk.com (Johann 'Myrkraverk' Oskarsson)
Date: Mon, 17 Jun 2013 00:27:51 +0000
Subject: [FreeVMS] Wanted
In-Reply-To: <51B99499.6010106@systella.fr>
References: <51B99499.6010106@systella.fr>
Message-ID:
Hi,
On Thu, Jun 13, 2013 at 9:44 AM, BERTRAND Jo?l
wrote:
> If anyone is interested, please answer to this mail. With more than
> 5 new and active developers, I think we can restart this project.
I have some experience with VMS application programming, mostly in C
but some in Fortran. I also have experience with direct hardware
programming (various platforms) and device driver development for
Illumos/Solaris.
I'd like to help but at the moment the time I can spend on the project
is *extremely* limited.
--
Johann
I'm not from the internet, I just work there.
From rmctwo at gmail.com Mon Jun 17 15:36:03 2013
From: rmctwo at gmail.com (Renee)
Date: Mon, 17 Jun 2013 09:36:03 -0400
Subject: [FreeVMS] DCL
Message-ID: <00cb01ce6b5f$9d013410$d7039c30$@gmail.com>
Has anyone considered DCL or is it too early in the process?
Also being a real DECcie I just realized that I don't know any C. How badly
will this hurt?
Renee
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From joel.bertrand at systella.fr Mon Jun 17 23:42:55 2013
From: joel.bertrand at systella.fr (=?windows-1252?Q?BERTRAND_Jo=EBl?=)
Date: Mon, 17 Jun 2013 23:42:55 +0200
Subject: [FreeVMS] DCL
In-Reply-To: <00cb01ce6b5f$9d013410$d7039c30$@gmail.com>
References: <00cb01ce6b5f$9d013410$d7039c30$@gmail.com>
Message-ID: <51BF82DF.4020804@systella.fr>
Renee a ?crit :
> Has anyone considered DCL or is it too early in the process?
To early in the process. We need to obtain a running kernel with some
IO capabilities before trying to wrtie a shell.
> Also being a real DECcie I just realized that I don?t know any C. How
> badly will this hurt?
Regards,
JKB
From joel.bertrand at systella.fr Mon Jun 17 23:48:45 2013
From: joel.bertrand at systella.fr (=?windows-1252?Q?BERTRAND_Jo=EBl?=)
Date: Mon, 17 Jun 2013 23:48:45 +0200
Subject: [FreeVMS] FW: DCL
In-Reply-To: <01dd01ce6b70$09ed9f40$1dc8ddc0$@ccsscorp.com>
References: <00cb01ce6b5f$9d013410$d7039c30$@gmail.com>
<01dd01ce6b70$09ed9f40$1dc8ddc0$@ccsscorp.com>
Message-ID: <51BF843D.8080307@systella.fr>
Bill Pedersen a ?crit :
> Renee:
>
> While it would be nice to think we as a team could support multiple
> languages right off the bat this is very unlikely.
>
> That said C and or C++ are going to have to be the primary
> implementation languages just because of their availability in general
> and the fact that many folks have them as a skill set.
>
> I am fully aware of your antithetical response to C based upon its
> origins (Unix) but it does not mean it cannot be used in a fashion which
> is consistent with the needs of the project. Just because you can
> terminate a string with a Null does not mean that is the best way to
> define strings for the OS?s usage ? descriptors are probably more
> effective. Likewise not allowing IO to go on unchecked ? validating
> overflow conditions to prevent intrusions is very important.
Null terminated string is only a way to implement string in C.
In FreeVMS, we have defined a string descriptor and all string
manipulations use it:
struct vms$string
{
vms$pointer length;
vms$pointer length_trim;
unsigned char *c;
};
JKB
From remi.chateauneu at gmail.com Mon Jun 17 23:56:11 2013
From: remi.chateauneu at gmail.com (remi.chateauneu at gmail.com)
Date: Mon, 17 Jun 2013 22:56:11 +0100
Subject: [FreeVMS] DCL
In-Reply-To: <51BF82DF.4020804@systella.fr>
References: <00cb01ce6b5f$9d013410$d7039c30$@gmail.com>
<51BF82DF.4020804@systella.fr>
Message-ID: <51BF85FB.1010902@gmail.com>
Le 17.06.2013 22:42, BERTRAND Jo?l a ?crit :
> Renee a ?crit :
>> Has anyone considered DCL or is it too early in the process?
>
> To early in the process. We need to obtain a running kernel with some IO
> capabilities before trying to wrtie a shell.
Indeed. On the other hand it might be possible to write the highest
level part of a DCL interpreter:
* Commands parsing (Options etc...)
* Execution of a commands file.
... etc ...
This is not a small part of the code and could probably be tested on a
Unix-like environment. This would save time and would deliver "something
to show" : This is important for motivation.
This might be also true for tools such as CDD (Common Data Dictionary,
http://h71000.www7.hp.com/commercial/c/docs/5492p036.html ). In itself,
this is a very convenient tool which can be used with or without OpenVMS
running it. For example, the parsing of CDD input files can probably be
done independently.
Also, some part of Record Management Services (
http://en.wikipedia.org/wiki/Record_Management_Services ) could be
written (And used) independently of OpenVMS. At least prototyped.
Remi
From rmctwo at gmail.com Tue Jun 18 00:57:13 2013
From: rmctwo at gmail.com (Renee)
Date: Mon, 17 Jun 2013 18:57:13 -0400
Subject: [FreeVMS] DCL
In-Reply-To: <51BF85FB.1010902@gmail.com>
References: <00cb01ce6b5f$9d013410$d7039c30$@gmail.com>
<51BF82DF.4020804@systella.fr> <51BF85FB.1010902@gmail.com>
Message-ID: <000b01ce6bae$02269e90$0673dbb0$@gmail.com>
Unfortunately I have not had any experience with parsing, I did drivers. Of
note DCL will be the item which will give it a VMS feel. Since it is the UI,
It'll be worth some attention.
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
remi.chateauneu at gmail.com
Sent: Monday, June 17, 2013 5:56 PM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] DCL
Le 17.06.2013 22:42, BERTRAND Jo?l a ?crit :
> Renee a ?crit :
>> Has anyone considered DCL or is it too early in the process?
>
> To early in the process. We need to obtain a running kernel with some
> IO capabilities before trying to wrtie a shell.
Indeed. On the other hand it might be possible to write the highest level
part of a DCL interpreter:
* Commands parsing (Options etc...)
* Execution of a commands file.
... etc ...
This is not a small part of the code and could probably be tested on a
Unix-like environment. This would save time and would deliver "something to
show" : This is important for motivation.
This might be also true for tools such as CDD (Common Data Dictionary,
http://h71000.www7.hp.com/commercial/c/docs/5492p036.html ). In itself, this
is a very convenient tool which can be used with or without OpenVMS running
it. For example, the parsing of CDD input files can probably be done
independently.
Also, some part of Record Management Services (
http://en.wikipedia.org/wiki/Record_Management_Services ) could be written
(And used) independently of OpenVMS. At least prototyped.
Remi
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From joel.bertrand at systella.fr Tue Jun 18 12:05:16 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Tue, 18 Jun 2013 12:05:16 +0200
Subject: [FreeVMS] FreeVMS developers
Message-ID: <51C030DC.3090308@systella.fr>
Hello,
For one week, some new potential contributers have sent mail to this
mailing list. Without particular order :
- Bill Pedersen (write access to freevms.net)
- Renee (I haven't his name)
- Bruno Ducrot (write access to freevms.net)
- Fr?d?ric Boyer
- Johann Oskarsson (with limited time)
- Bertrand Jo?l (write access to freevms.net)
I can open write access on request on both freevms.net and git
repository. I will fix bugzilla and gitweb configurations as soon as
possible.
I have uploaded last bootable image last sunday with some modifications
to boot with grub2 instead of legacy grub. Boot rootfs is now in ext2 as
grub2 cannot boot from ODS2 (legacy grub was patched to boot from ODS2
filesystem). L4/X2 doesn't work if it is built with gcc 4.7 or 4.8. I'm
investigating. Seems to be a regression in gcc 4.7.
If you mount freevms.img, you will find a configuration file
/grub/grub.cfg that contains :
set timeout=10
menuentry 'FreeVMS' {
multiboot /boot/kickstart maxmem=128M kmem=8M
module /boot/x86-kernel
module /boot/sigma0
module /boot/vmskernel.sys vmskernel.sys root=dqa0.0 console=opa0 \
nodbg\$virtual_memory \
dbg\$sys_pagefault dbg\$vms_pagefault
module /boot/pager.sys pager.sys
module /boot/init.exe init.exe
# module /boot/drv/dq\$pxii.sys
# module /boot/drv/dp\$pc99.sys
# module /boot/drv/fs\$ext2.sys
}
Repetions are required to boot with grub2 as multiboot specifications
have changed between legacy grub and grub2. All modules can be preloaded
into memory by grub2, remapped by sigma0 and tested. In a real system,
only pager.sys and basic IO (dq$pxii.sys for IDE, dp$pc99.sys for disk
paritioning and fs$ext2.sys for root filesystem will be required).
When you run this image under qemu, last lines are :
%RUN-S-STARTED, PAGER.SYS process started
%PAGER-I-SWPADDR, setting swapper base address at $....
%PAGER-I-SWPSIZE, setting swap size to 0 byte
%RUN-S-STOPPED, PAGER.SYS process stopped
This output is generated by sources/freevms/vm/vm_pager.c (launched by
sources/freevms/sys/sys_pager.c as a new thread in a separated address
space). %RUN-S-STOPPED is written by return(0) from PAGER.SYS main
function, as this function ends. In real system, it has to contain an
infinite loop that manage memory.
I have written a first algorithm here :
http://www.freevms.net/article35.html
but it doesn't interact with thread ID.
The question is... Who wants to become official maintainer of this pager ?
You can find all L4Ka documentation here :
http://os.ibds.kit.edu/1956.php
http://os.ibds.kit.edu/l4ka/l4-x2-r7.pdf
Regards,
JKB
From rmctwo at gmail.com Tue Jun 18 18:04:13 2013
From: rmctwo at gmail.com (Renee)
Date: Tue, 18 Jun 2013 12:04:13 -0400
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <51C030DC.3090308@systella.fr>
References: <51C030DC.3090308@systella.fr>
Message-ID: <003501ce6c3d$81b94290$852bc7b0$@gmail.com>
Renee Culver is a she.
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
BERTRAND Jo?l
Sent: Tuesday, June 18, 2013 6:05 AM
To: FreeVMS mailing list
Subject: [FreeVMS] FreeVMS developers
Hello,
For one week, some new potential contributers have sent mail to this
mailing list. Without particular order :
- Bill Pedersen (write access to freevms.net)
- Renee (I haven't his name)
- Bruno Ducrot (write access to freevms.net)
- Fr?d?ric Boyer
- Johann Oskarsson (with limited time)
- Bertrand Jo?l (write access to freevms.net)
I can open write access on request on both freevms.net and git
repository. I will fix bugzilla and gitweb configurations as soon as
possible.
I have uploaded last bootable image last sunday with some
modifications to boot with grub2 instead of legacy grub. Boot rootfs is now
in ext2 as
grub2 cannot boot from ODS2 (legacy grub was patched to boot from ODS2
filesystem). L4/X2 doesn't work if it is built with gcc 4.7 or 4.8. I'm
investigating. Seems to be a regression in gcc 4.7.
If you mount freevms.img, you will find a configuration file
/grub/grub.cfg that contains :
set timeout=10
menuentry 'FreeVMS' {
multiboot /boot/kickstart maxmem=128M kmem=8M
module /boot/x86-kernel
module /boot/sigma0
module /boot/vmskernel.sys vmskernel.sys root=dqa0.0 console=opa0 \
nodbg\$virtual_memory \
dbg\$sys_pagefault dbg\$vms_pagefault
module /boot/pager.sys pager.sys
module /boot/init.exe init.exe
# module /boot/drv/dq\$pxii.sys
# module /boot/drv/dp\$pc99.sys
# module /boot/drv/fs\$ext2.sys
}
Repetions are required to boot with grub2 as multiboot specifications have
changed between legacy grub and grub2. All modules can be preloaded into
memory by grub2, remapped by sigma0 and tested. In a real system, only
pager.sys and basic IO (dq$pxii.sys for IDE, dp$pc99.sys for disk
paritioning and fs$ext2.sys for root filesystem will be required).
When you run this image under qemu, last lines are :
%RUN-S-STARTED, PAGER.SYS process started %PAGER-I-SWPADDR, setting swapper
base address at $....
%PAGER-I-SWPSIZE, setting swap size to 0 byte %RUN-S-STOPPED, PAGER.SYS
process stopped
This output is generated by sources/freevms/vm/vm_pager.c (launched
by sources/freevms/sys/sys_pager.c as a new thread in a separated address
space). %RUN-S-STOPPED is written by return(0) from PAGER.SYS main function,
as this function ends. In real system, it has to contain an infinite loop
that manage memory.
I have written a first algorithm here :
http://www.freevms.net/article35.html
but it doesn't interact with thread ID.
The question is... Who wants to become official maintainer of this
pager ?
You can find all L4Ka documentation here :
http://os.ibds.kit.edu/1956.php
http://os.ibds.kit.edu/l4ka/l4-x2-r7.pdf
Regards,
JKB
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From guidoj at users.sourceforge.net Tue Jun 18 21:20:24 2013
From: guidoj at users.sourceforge.net (Guido)
Date: Tue, 18 Jun 2013 21:20:24 +0200
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <51C030DC.3090308@systella.fr>
References: <51C030DC.3090308@systella.fr>
Message-ID: <1551952.QTsGxChNh5@kermit>
On Tuesday 18 June 2013 12:05:16 BERTRAND Jo?l wrote:
[snip]
> This output is generated by sources/freevms/vm/vm_pager.c (launched by
> sources/freevms/sys/sys_pager.c as a new thread in a separated address
> space). %RUN-S-STOPPED is written by return(0) from PAGER.SYS main
> function, as this function ends. In real system, it has to contain an
> infinite loop that manage memory.
>
> I have written a first algorithm here :
> http://www.freevms.net/article35.html
> but it doesn't interact with thread ID.
>
> The question is... Who wants to become official maintainer of this pager ?
>
> You can find all L4Ka documentation here :
> http://os.ibds.kit.edu/1956.php
> http://os.ibds.kit.edu/l4ka/l4-x2-r7.pdf
Shouldn't there be some interface specification first?
Correct me if I'm wrong but as I understand it, the pager should use some
functions provided by the kernel (for instance sys$morecore()). I assume that
the interface between pager and kernel is to be built upon the standard L4 IPC
mechanism. If so, message specifications (label, parameters, etc) would be very
helpful.
Regards,
Guido
From remi.chateauneu at gmail.com Tue Jun 18 22:29:58 2013
From: remi.chateauneu at gmail.com (remi.chateauneu at gmail.com)
Date: Tue, 18 Jun 2013 21:29:58 +0100
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <003501ce6c3d$81b94290$852bc7b0$@gmail.com>
References: <51C030DC.3090308@systella.fr>
<003501ce6c3d$81b94290$852bc7b0$@gmail.com>
Message-ID: <51C0C346.3060002@gmail.com>
And spells with a "?" like Ren?e (The male version is Ren?)
Le 18.06.2013 17:04, Renee a ?crit :
> Renee Culver is a she.
>
> Renee
>
> -----Original Message-----
From joel.bertrand at systella.fr Tue Jun 18 23:19:45 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Tue, 18 Jun 2013 23:19:45 +0200
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <1551952.QTsGxChNh5@kermit>
References: <51C030DC.3090308@systella.fr> <1551952.QTsGxChNh5@kermit>
Message-ID: <51C0CEF1.40703@systella.fr>
Guido a ?crit :
> On Tuesday 18 June 2013 12:05:16 BERTRAND Jo?l wrote:
> [snip]
>> This output is generated by sources/freevms/vm/vm_pager.c (launched by
>> sources/freevms/sys/sys_pager.c as a new thread in a separated address
>> space). %RUN-S-STOPPED is written by return(0) from PAGER.SYS main
>> function, as this function ends. In real system, it has to contain an
>> infinite loop that manage memory.
>>
>> I have written a first algorithm here :
>> http://www.freevms.net/article35.html
>> but it doesn't interact with thread ID.
>>
>> The question is... Who wants to become official maintainer of this pager ?
>>
>> You can find all L4Ka documentation here :
>> http://os.ibds.kit.edu/1956.php
>> http://os.ibds.kit.edu/l4ka/l4-x2-r7.pdf
>
> Shouldn't there be some interface specification first?
Yes, we have to write some specifications. But we need to design a
maintainer that collects all documentations and specifications.
> Correct me if I'm wrong but as I understand it, the pager should use some
> functions provided by the kernel (for instance sys$morecore()). I assume that
> the interface between pager and kernel is to be built upon the standard L4 IPC
> mechanism. If so, message specifications (label, parameters, etc) would be very
> helpful.
Right.
JKB
From joel.bertrand at systella.fr Tue Jun 18 23:28:33 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Tue, 18 Jun 2013 23:28:33 +0200
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <003501ce6c3d$81b94290$852bc7b0$@gmail.com>
References: <51C030DC.3090308@systella.fr>
<003501ce6c3d$81b94290$852bc7b0$@gmail.com>
Message-ID: <51C0D101.2090300@systella.fr>
Renee a ?crit :
> Renee Culver is a she.
Sorry, I beg your pardon...
From pedersen at ccsscorp.com Tue Jun 18 23:37:31 2013
From: pedersen at ccsscorp.com (Bill Pedersen)
Date: Tue, 18 Jun 2013 17:37:31 -0400
Subject: [FreeVMS] Documentation...
Message-ID: <004101ce6c6c$0a28b1f0$1e7a15d0$@ccsscorp.com>
Joel:
You stated:
"Yes, we have to write some specifications. But we need to
design a maintainer that collects all documentations and specifications."
I think you meant "assign a maintainer".
Bill.
Bill Pedersen
CCSS - Computer Consulting System Services, LLC
211 Ruth Drive
Gaffney, SC 29341
Telephone: 864-490-8863
Mobile: 408-892-5204
Facsimile: 206-984-3068
www: www.ccsscorp.com
Skype Name: william.a.pedersen
LinkedIn Profile:
www.linkedin.com/in/billpedersen
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From joel.bertrand at systella.fr Wed Jun 19 07:28:35 2013
From: joel.bertrand at systella.fr (=?windows-1252?Q?BERTRAND_Jo=EBl?=)
Date: Wed, 19 Jun 2013 07:28:35 +0200
Subject: [FreeVMS] Documentation...
In-Reply-To: <004101ce6c6c$0a28b1f0$1e7a15d0$@ccsscorp.com>
References: <004101ce6c6c$0a28b1f0$1e7a15d0$@ccsscorp.com>
Message-ID: <51C14183.7030807@systella.fr>
Bill Pedersen a ?crit :
> Joel:
>
> You stated:
>
> ?Yes, we have to write some specifications. But we need to design a
> maintainer that collects all documentations and specifications.?
>
> I think you meant ?assign a maintainer??
Yes, of course...
From guidoj at users.sourceforge.net Wed Jun 19 21:15:32 2013
From: guidoj at users.sourceforge.net (Guido)
Date: Wed, 19 Jun 2013 21:15:32 +0200
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <51C0CEF1.40703@systella.fr>
References: <51C030DC.3090308@systella.fr> <1551952.QTsGxChNh5@kermit>
<51C0CEF1.40703@systella.fr>
Message-ID: <4764899.auX85Kaz1g@kermit>
Some more observations:
1) mutexes
Mutex code exists in sys_mutex.c/h but should be available in pager (and other
servers like device drivers) too. I guess that IPC overhead is a waste in this
case and that the mutex code should be made available as an internal library.
2) panic
Right now the kernel is able to panic in certain circumstances, although I
don't think an unknown IPC should be one of them. Anyway, the pager (and other
servers) should have some similar functionality. The pager is a special case
since if the pager panics, nothing will be able to run, so effectively the
entire system should hang. If a device driver panics, it could probably be
restarted.
Regards,
Guido
From joel.bertrand at systella.fr Wed Jun 19 22:40:29 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Wed, 19 Jun 2013 22:40:29 +0200
Subject: [FreeVMS] FreeVMS developers
In-Reply-To: <4764899.auX85Kaz1g@kermit>
References: <51C030DC.3090308@systella.fr> <1551952.QTsGxChNh5@kermit>
<51C0CEF1.40703@systella.fr> <4764899.auX85Kaz1g@kermit>
Message-ID: <51C2173D.7020200@systella.fr>
Guido a ?crit :
> Some more observations:
>
> 1) mutexes
> Mutex code exists in sys_mutex.c/h but should be available in pager (and other
> servers like device drivers) too. I guess that IPC overhead is a waste in this
> case and that the mutex code should be made available as an internal library.
Right. But sys_mutex.c does not contain any real mutexes. Code is
missing and, if I remember, I have to add in sys/sys_mutex.c some pieces
of code to obtain active wait (? la linux spinlock due to
sys/sys_mutex_amd64.S implementation). Posix mutex has to written in
external library.
> 2) panic
> Right now the kernel is able to panic in certain circumstances, although I
> don't think an unknown IPC should be one of them. Anyway, the pager (and other
> servers) should have some similar functionality. The pager is a special case
> since if the pager panics, nothing will be able to run, so effectively the
> entire system should hang. If a device driver panics, it could probably be
> restarted.
I agree. As kernel is in a very early stage of development, PANIC() is
only added to stop kernel with a backtrace. In a real system, some
PANIC() have to be replaced with some code to catch errors.
Regards,
JKB
From remi.chateauneu at gmail.com Thu Jun 20 00:17:06 2013
From: remi.chateauneu at gmail.com (remi.chateauneu at gmail.com)
Date: Wed, 19 Jun 2013 23:17:06 +0100
Subject: [FreeVMS] PDP-11 Still Working In Nuclear Plants - For 37 More Years
Message-ID: <51C22DE2.6080406@gmail.com>
"Most of the younger /. readers never heard of the PDP-11, while we
geezers have to retrieve bits and pieces of our affairs with PDP-11 from
the vast warehouse inside our memory lanes." From the article: "HP might
have nuked OpenVMS, but its parent, PDP-11, is still spry and powering
GE nuclear power-plant robots and will do for another 37 years. That's
right: PDP-11 assembler programmers are hard to find, but the nuclear
industry is planning on keeping them until 2050 ? long enough for a
couple of generations of programmers to come and go." Not sure about the
OpenVMS vs PDP comparison, but it's still amusing that a PDP might
outlast all of the VAX machines.
http://developers.slashdot.org/story/13/06/19/1227244/pdp-11-still-working-in-nuclear-plants---for-37-more-years
From joel.bertrand at systella.fr Thu Jun 20 18:13:15 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 20 Jun 2013 18:13:15 +0200
Subject: [FreeVMS] Bugzilla
Message-ID: <51C32A1B.40104@systella.fr>
Hello,
I have fixed FreeVMS bugzilla. It's now available at
https://www.systella.fr/bugzilla3
Regards,
JKB
From guidoj at users.sourceforge.net Thu Jun 20 21:33:51 2013
From: guidoj at users.sourceforge.net (Guido)
Date: Thu, 20 Jun 2013 21:33:51 +0200
Subject: [FreeVMS] rtl$print
Message-ID: <3805141.VSqOSIrj85@kermit>
Hi,
Please tell me there is also a simple one liner to print a message in stead of
this abomination (of which I even doubt that it does what is intended):
L4_Word_t arg$print1;
const char * arg$print2;
rtl$strcpy(&message, IPC_F_FAILED "IPC failed (error %ld: %s)");
L4_Word_t arg$print1 = L4_ErrorCode();
const char * arg$print2 = L4_ErrorCode_String(L4_ErrorCode());
void* arg$print[] = {&arg$print1, &arg$print2};
rtl$print(&message, arg$print);
Regards,
Guido
From joel.bertrand at systella.fr Thu Jun 20 23:39:25 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 20 Jun 2013 23:39:25 +0200
Subject: [FreeVMS] rtl$print
In-Reply-To: <3805141.VSqOSIrj85@kermit>
References: <3805141.VSqOSIrj85@kermit>
Message-ID: <51C3768D.2080605@systella.fr>
Guido a ?crit :
> Hi,
>
> Please tell me there is also a simple one liner to print a message in stead of
> this abomination (of which I even doubt that it does what is intended):
>
> L4_Word_t arg$print1;
> const char * arg$print2;
>
> rtl$strcpy(&message, IPC_F_FAILED "IPC failed (error %ld: %s)");
> L4_Word_t arg$print1 = L4_ErrorCode();
> const char * arg$print2 = L4_ErrorCode_String(L4_ErrorCode());
> void* arg$print[] = {&arg$print1,&arg$print2};
> rtl$print(&message, arg$print);
I'll check as soon as possible. I think there is another solution, but
now, it's midnight in Europe, time to go to sleep... ;-)
Regards,
JKB
From joel.bertrand at systella.fr Fri Jun 21 11:51:05 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Fri, 21 Jun 2013 11:51:05 +0200
Subject: [FreeVMS] rtl$print
In-Reply-To: <3805141.VSqOSIrj85@kermit>
References: <3805141.VSqOSIrj85@kermit>
Message-ID: <51C42209.8040808@systella.fr>
Guido wrote:
> Hi,
Hello,
librtl is a static library that contains all convenient functions
called from FreeVMS tasks (modules, drivers, but not kernel itself).
> Please tell me there is also a simple one liner to print a message in stead of
> this abomination (of which I even doubt that it does what is intended):
>
> L4_Word_t arg$print1;
> const char * arg$print2;
vms$string_initializer(message, 80);
> rtl$strcpy(&message, IPC_F_FAILED "IPC failed (error %ld: %s)");
> L4_Word_t arg$print1 = L4_ErrorCode();
> const char * arg$print2 = L4_ErrorCode_String(L4_ErrorCode());
> void* arg$print[] = {&arg$print1,&arg$print2};
> rtl$print(&message, arg$print);
Today, this abomination is the solution as there is not mechanism to
fill arg$print[]. I think we have to write a macro like rtl$args() to
fill this list.
// Initialization of a static string descriptor
vms$string_initializer(message, 80);
rtl$strcpy(&message, IPC_F_FAILED "IPC failed (error %ld: %s)");
rtl$args(arg$print, L4_ErrorCode(),
L4_ErrorCode_String(L4_ErrorCode()));
rtl$print(&message, arg$print);
I'm not sure that I'll have time to write it this week-end as I'm
trying to debug strange bugs in L4 built by gcc 4.7 and 4.8.
Regards,
JKB
From joel.bertrand at systella.fr Sat Jun 22 14:32:00 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Sat, 22 Jun 2013 14:32:00 +0200
Subject: [FreeVMS] L4Ka. Regression with 4.7 and 4.8 ?
In-Reply-To: <51C376A5.4060107@systella.fr>
References: <51C1AB3A.5080907@systella.fr> <2138987.2tPF7NEjmS@localhost>
<51C376A5.4060107@systella.fr>
Message-ID: <51C59940.3020009@systella.fr>
Hello,
I have bissected the code and found the faulty .o. I have attached
assembly output. idt.gcc46.s is the good one. If L4 is linked with
idt.gcc.47.s, it only reboots.
L4 sources are available here : git://github.com/l4ka/pistachio.git
Regards,
JKB
-------------- section suivante --------------
Un texte encapsul? et encod? dans un jeu de caract?res inconnu a ?t? nettoy?...
Nom : idt.gcc46.s
URL :
-------------- section suivante --------------
Un texte encapsul? et encod? dans un jeu de caract?res inconnu a ?t? nettoy?...
Nom : idt.gcc47.s
URL :
-------------- section suivante --------------
Une pi?ce jointe autre que texte a ?t? nettoy?e...
Nom: idt.cc
Type: text/x-c++src
Taille: 4178 octets
Desc: non disponible
URL:
-------------- section suivante --------------
Une pi?ce jointe autre que texte a ?t? nettoy?e...
Nom: idt.h
Type: text/x-chdr
Taille: 2415 octets
Desc: non disponible
URL:
From joel.bertrand at systella.fr Sat Jun 22 14:38:10 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Sat, 22 Jun 2013 14:38:10 +0200
Subject: [FreeVMS] L4Ka. Regression with 4.7 and 4.8 ?
In-Reply-To: <51C59940.3020009@systella.fr>
References: <51C1AB3A.5080907@systella.fr> <2138987.2tPF7NEjmS@localhost>
<51C376A5.4060107@systella.fr> <51C59940.3020009@systella.fr>
Message-ID: <51C59AB2.9030208@systella.fr>
I have forgotten... Compilation is done with :
gcc -nostdinc -I/home/bertrand/openvms/l4/gcc-4.7/include
-I/home/bertrand/openvms/l4/pistachio/kernel/src
-I/home/bertrand/openvms/l4/pistachio/kernel/src/generic
-I/usr/lib/gcc/x86_64-linux-gnu/4.7/include/ -Ux64 -Ux86 -Uk8 -Upc99
-Uv4 -Uhs -D__SUBARCH__=x64 -D__ARCH__=x86 -D__CPU__=k8
-D__PLATFORM__=pc99 -D__API__=v4 -D__SCHED__=hs -imacros
/home/bertrand/openvms/l4/gcc-4.7/config/config.h -imacros
/home/bertrand/openvms/l4/pistachio/kernel/src/generic/macros.h -include
/home/bertrand/openvms/l4/pistachio/kernel/src/generic/config.h
-include /home/bertrand/openvms/l4/pistachio/kernel/src/generic/types.h
-fno-rtti -fno-builtin -fomit-frame-pointer -fno-exceptions -Wall
-Wno-non-virtual-dtor -Wno-format -O2 -m64 -mcmodel=kernel -mno-red-zone
-mno-mmx -mno-sse -mno-sse2 -mno-sse3 -march=k8 -Wno-conversion
-fno-stack-protector -c
/home/bertrand/openvms/l4/pistachio/kernel/src/glue/v4-x86/idt.cc
From joel.bertrand at systella.fr Sat Jun 22 19:54:19 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Sat, 22 Jun 2013 19:54:19 +0200
Subject: [FreeVMS] L4Ka. Regression with 4.7 and 4.8 ?
In-Reply-To: <1371906201.3742.53.camel@yam-132-YW-E178-FTW>
References: <51C1AB3A.5080907@systella.fr> <2138987.2tPF7NEjmS@localhost>
<51C376A5.4060107@systella.fr> <51C59940.3020009@systella.fr>
<1371906201.3742.53.camel@yam-132-YW-E178-FTW>
Message-ID: <51C5E4CB.40205@systella.fr>
Oleg Endo a ?crit :
> On Sat, 2013-06-22 at 14:32 +0200, BERTRAND Jo?l wrote:
>> Hello,
>>
>> I have bissected the code and found the faulty .o. I have attached
>> assembly output. idt.gcc46.s is the good one. If L4 is linked with
>> idt.gcc.47.s, it only reboots.
>
> The code of 4.6 and 4.7 seems to be pretty much the same, except for
> differences in the register allocations.
> (I removed all .cfi lines before doing a diff)
>
> However, maybe the following difference ...
>
> 4.6:
> .section .ctors.10003,"aw", at progbits
> .align 8
> .quad _GLOBAL__sub_I.55532_idt
> ...
>
> 4.7:
> .section .init_array.55532,"aw"
> .align 8
> .quad _GLOBAL__sub_I.55532_idt
>
> (different sections being used)
>
> ... triggers another problem somewhere else? Maybe .init_array is not
> handled by the startup code or something like that? I'm just
> guessing...
Maybe. But if I build all but idt.c with gcc 4.7 and only idt.c with
gcc 4.6, I obtain a running kernel.
Regards,
JKB
From joel.bertrand at systella.fr Sat Jun 22 21:38:36 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Sat, 22 Jun 2013 21:38:36 +0200
Subject: [FreeVMS] L4Ka. Regression with 4.7 and 4.8 ?
In-Reply-To: <1371924700.3742.108.camel@yam-132-YW-E178-FTW>
References: <51C1AB3A.5080907@systella.fr> <2138987.2tPF7NEjmS@localhost>
<51C376A5.4060107@systella.fr> <51C59940.3020009@systella.fr>
<1371906201.3742.53.camel@yam-132-YW-E178-FTW> <51C5E4CB.40205@systella.fr>
<1371924700.3742.108.camel@yam-132-YW-E178-FTW>
Message-ID: <51C5FD3C.3070609@systella.fr>
Oleg Endo a ?crit :
> The code in idt.c compiled with 4.6 puts static initialization functions
> into the .ctors section, while 4.7 puts them into the .init_array
> section. Probably this happens only for this single file in the whole
> kernel. My guess is that the .init_array section is not handled
> properly by the startup code of the kernel. Or maybe it's even stripped
> out completely (missing in the linker script). Either way, it seems
> that static initialization for idt.c is not being done properly and thus
> the code crashes.
I don't think that ctors are stripped by linker script. This script is
built by makefile. Here is my script :
OUTPUT_FORMAT("elf64-x86-64")
OUTPUT_ARCH("i386:x86-64")
BOOTMEM_SIZE = 1024K;
_start_text_phys = 0x600000;
_start_text = _start_text_phys + 0xFFFFFFFFC0000000;
SECTIONS
{
. = ALIGN(4k);
.text _start_text : AT (ADDR(.text) - 0xFFFFFFFFC0000000)
{
*(.text);
*(.gnu.linkonce.*);
*(.spinlock);
}
. = ALIGN(4k);
.rodata . : AT (ADDR(.rodata) - 0xFFFFFFFFC0000000)
{
*(.rodata);
*(.rodata.*);
}
. = ALIGN(4K);
.kip . : AT (ADDR(.kip) - 0xFFFFFFFFC0000000)
{
*(.data.kip)
*(.data.kip.pdesc)
*(.data.kip.kdesc)
*(.data.kip.versionparts)
*(.data.kip.features)
*(.data.kip.features.end)
. = ALIGN(16);
*(.data.kip.mdesc)
_memory_descriptors_raw = ABSOLUTE((_memory_descriptors_offset << 32)
+ _memory_descriptors_size);
}
_memory_descriptors_offset = memory_descriptors - kip;
_memory_descriptors_size =
(4K - ((memory_descriptors - kip) & (4K-1))) / 16;
. = ALIGN(((1 << 21)));
_start_syscalls = .;
.syscalls . : AT (ADDR(.syscalls) - 0xFFFFFFFFC0000000)
{
*(.user.syscall.ipc);
. = ALIGN((0x100));
*(.user.syscall.lipc);
. = ALIGN((0x100));
*(.user.syscall.exregs);
. = ALIGN((0x100));
*(.user.syscall.threadctrl);
. = ALIGN((0x100));
*(.user.syscall.sysclock);
. = ALIGN((0x100));
*(.user.syscall.threadswtch);
. = ALIGN((0x100));
*(.user.syscall.schedule);
. = ALIGN((0x100));
*(.user.syscall.unmap);
. = ALIGN((0x100));
*(.user.syscall.spacectrl);
. = ALIGN((0x100));
*(.user.syscall.procctrl);
. = ALIGN((0x100));
*(.user.syscall.memctrl);
}
. = ALIGN(((1 << 21)));
_end_syscalls = .;
. = ALIGN(((1 << 21)));
_start_cpu_local = .;
.cpulocal . : AT (ADDR(.cpulocal) - 0xFFFFFFFFC0000000)
{
*(.data.cpulocal)
*(.data.x86.cpulocal)
}
. = ALIGN(((1 << 21)));
_end_cpu_local = .;
.data . : AT (ADDR(.data) - 0xFFFFFFFFC0000000)
{
*(.sdata);
*(.data);
*(.data.x86.idt);
*(.data.x86.exc_all);
*(.data.x86.exc_common);
*(.data.*);
_start_bss = .;
*(.bss);
_end_bss = .;
}
. = ALIGN(4K);
.kdebug . : AT(ADDR(.kdebug) - 0xFFFFFFFFC0000000)
{
*(.kdebug);
*(.kdebug-bss);
*(.comment);
}
. = ALIGN(4K);
.sets . : AT(ADDR(.sets) - 0xFFFFFFFFC0000000)
{
. = ALIGN(16);
_start_setlist = .;
*(.setlist)
_end_setlist = .;
. = ALIGN(16);
_start_sets = .;
*(SORT(set_*))
_end_sets = .;
}
. = ALIGN(4K);
_end_text = .;
. = ALIGN(4K);
_start_bootmem = .;
. = . + BOOTMEM_SIZE;
_end_bootmem = .;
. = ALIGN(4K);
_start_init = . - 0xFFFFFFFFC0000000;
.init (. - 0xFFFFFFFFC0000000) : AT(ADDR(.init))
{
*(.init.startup32);
. = ALIGN(1K);
*(.init.init32) FILL(0x90909090);
. = . + 1K;
. = ALIGN(1K);
*(.init.init64);
*(.init);
*(.roinit);
*(.init.data);
*(.init.memory);
*(.init.cpulocal);
*(.init.*);
. = ALIGN(16);
_start_mdb_funcs = .;
*(SORT(.mdb_funcs*))
_end_mdb_funcs = .;
. = ALIGN(16);
/* cpu-local static initializers */
__ctors_CPU__ = .;
*(SORT(.ctors.3*))
QUAD(0)
/* node-local static initializers */
__ctors_NODE__ = .;
*(SORT(.ctors.2*))
QUAD(0)
/* global static initializers */
__ctors_GLOBAL__ = .;
*(SORT(.ctors.1*))
*(SORT(.ctors.*))
*(.ctors)
QUAD(0)
}
_end_init = .;
/* special section that is discarded during linking
all unwanted sections should go here */
. = ALIGN(4K);
/DISCARD/ :
{
*(.eh_frame);
*(.note);
*(.comment);
}
}
/* physical addresses */
_end_text_phys = _end_text - 0xFFFFFFFFC0000000;
_start_syscalls_phys = _start_syscalls - 0xFFFFFFFFC0000000;
_end_syscalls_phys = _end_syscalls - 0xFFFFFFFFC0000000;
_start_bootmem_phys = _start_bootmem - 0xFFFFFFFFC0000000;
_end_bootmem_phys = _end_bootmem - 0xFFFFFFFFC0000000;
Regards,
JKB
From rmctwo at gmail.com Sat Jun 22 22:30:09 2013
From: rmctwo at gmail.com (Renee)
Date: Sat, 22 Jun 2013 16:30:09 -0400
Subject: [FreeVMS] Question
Message-ID: <000f01ce6f87$4a81c610$df855230$@gmail.com>
And the embarrassing thing is that I may have asked it before. Obviously we
are not building a VAX. We can make this thing "source compatible" to be
recompiled. And I don't think a privileged or even a CMKRNL task would work.
With that in mind, I'll ask again..what are we building? Certainly some form
of DCL can and must be built.
Renee
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From johann at myrkraverk.com Sun Jun 23 06:43:24 2013
From: johann at myrkraverk.com (Johann 'Myrkraverk' Oskarsson)
Date: Sun, 23 Jun 2013 04:43:24 +0000
Subject: [FreeVMS] Question
In-Reply-To: <000f01ce6f87$4a81c610$df855230$@gmail.com>
References: <000f01ce6f87$4a81c610$df855230$@gmail.com>
Message-ID:
On Sat, Jun 22, 2013 at 8:30 PM, Renee wrote:
> And the embarrassing thing is that I may have asked it before. Obviously we
> are not building a VAX. We can make this thing ?source compatible? to be
> recompiled. And I don?t think a privileged or even a CMKRNL task would work.
> With that in mind, I?ll ask again?.what are we building? Certainly some form
> of DCL can and must be built.
And that task is quite independent of the kernel. DCL is a shell and
can be written in C on top of Posix interfaces. I'm not advocating
that, only pointing out it's an option.
In between DCL and the kernel are the LIB$ and other VMS interfaces.
These interfaces can be written on top of pretty much any kernel. At
least when we have the option to modify the kernel too.
The kernel itself is then a layer between the hardware and system
interfaces and provides services such as ODS.
That said, I'm not up to speed on the details of what's been done
kernel or interface wise so far.
--
Johann
I'm not from the internet, I just work there.
From joel.bertrand at systella.fr Sun Jun 23 11:56:43 2013
From: joel.bertrand at systella.fr (=?UTF-8?B?QkVSVFJBTkQgSm/Dq2w=?=)
Date: Sun, 23 Jun 2013 11:56:43 +0200
Subject: [FreeVMS] L4Ka. Regression with 4.7 and 4.8 ?
In-Reply-To: <1371932623.3742.132.camel@yam-132-YW-E178-FTW>
References: <51C1AB3A.5080907@systella.fr> <2138987.2tPF7NEjmS@localhost>
<51C376A5.4060107@systella.fr> <51C59940.3020009@systella.fr>
<1371906201.3742.53.camel@yam-132-YW-E178-FTW> <51C5E4CB.40205@systella.fr>
<1371924700.3742.108.camel@yam-132-YW-E178-FTW>
<51C5FD3C.3070609@systella.fr>
<1371932623.3742.132.camel@yam-132-YW-E178-FTW>
Message-ID: <51C6C65B.50403@systella.fr>
Oleg Endo a ?crit :
> On Sat, 2013-06-22 at 21:38 +0200, BERTRAND Jo?l wrote:
>> Oleg Endo a ?crit :
>>> The code in idt.c compiled with 4.6 puts static initialization functions
>>> into the .ctors section, while 4.7 puts them into the .init_array
>>> section. Probably this happens only for this single file in the whole
>>> kernel. My guess is that the .init_array section is not handled
>>> properly by the startup code of the kernel. Or maybe it's even stripped
>>> out completely (missing in the linker script). Either way, it seems
>>> that static initialization for idt.c is not being done properly and thus
>>> the code crashes.
>>
>> I don't think that ctors are stripped by linker script. This script is
>> built by makefile. Here is my script :
>>
>>
>> OUTPUT_FORMAT("elf64-x86-64")
>> OUTPUT_ARCH("i386:x86-64")
>> BOOTMEM_SIZE = 1024K;
>> ...
>
> As I initially assumed, the .init_array section seems to be missing in
> the linker script, so it gets stripped. The .init_array section has to
> be handled for code generated by GCC 4.7. I would say this is an issue
> of the L4 software, not GCC.
OK. I have done some test since yesterday. I have tried to include
.init_array in kernel.
If I modify my linker script like this :
SECTIONS
{
. = ALIGN(4k);
.text _start_text : AT (ADDR(.text) - KERNEL_OFFSET)
{
*(.text);
*(.gnu.linkonce.*);
*(.spinlock);
}
. = ALIGN(4k);
.init_array . : AT (ADDR(.init_array) - KERNEL_OFFSET)
{
PROVIDE_HIDDEN (__init_array_start = .);
KEEP (*(SORT(.init_array.*)))
KEEP (*(.init_array))
PROVIDE_HIDDEN (__init_array_end = .);
}
. = ALIGN(4k);
.rodata . : AT (ADDR(.rodata) - KERNEL_OFFSET)
...
I obtain a kernel with an .init_array section. But result is the same as
I suppose that kickstart does not load this new section in memory.
I have tried to write .init_array in .text section with :
.text _start_text : AT (ADDR(.text) - KERNEL_OFFSET)
{
*(.text);
*(.gnu.linkonce.*);
*(.spinlock);
PROVIDE_HIDDEN (__init_array_start = .);
KEEP (*(SORT(.init_array.*)))
KEEP (*(.init_array))
PROVIDE_HIDDEN (__init_array_end = .);
}
I believe that .init_array is included in text section and loaded by
kickstart into memory, but kernel only reboots too.
Regards,
JKB
From joel.bertrand at systella.fr Sun Jun 23 12:29:47 2013
From: joel.bertrand at systella.fr (=?windows-1252?Q?BERTRAND_Jo=EBl?=)
Date: Sun, 23 Jun 2013 12:29:47 +0200
Subject: [FreeVMS] Question
In-Reply-To:
References: <000f01ce6f87$4a81c610$df855230$@gmail.com>
Message-ID: <51C6CE1B.5090904@systella.fr>
Johann 'Myrkraverk' Oskarsson a ?crit :
> On Sat, Jun 22, 2013 at 8:30 PM, Renee wrote:
>> And the embarrassing thing is that I may have asked it before. Obviously we
>> are not building a VAX. We can make this thing ?source compatible? to be
>> recompiled. And I don?t think a privileged or even a CMKRNL task would work.
>> With that in mind, I?ll ask again?.what are we building? Certainly some form
>> of DCL can and must be built.
>
> And that task is quite independent of the kernel. DCL is a shell and
> can be written in C on top of Posix interfaces. I'm not advocating
> that, only pointing out it's an option.
DCL will be written when we'll have a running kernel. There are a lot
of stuff to do before starting DCL.
If you want to start to write some userland code (not kernel, executive
or system ones), in a first time you have to implemant
sources/userland/rtl and sources/userland/starlet to obtain a VMS API or
to port newlib on FreeVMS to obtain a real POSIX layer. But to write
STARLET/LIB$ or port newlib, we have to define some SYSCALLs. Today,
only SYSCALL$PRINT, SYSCALL$EXIT_VALUE and SYSCALL$KILL_THREAD have been
written (sys/sys_loop.c).
Regards,
JKB
From joel.bertrand at systella.fr Sun Jun 23 12:56:24 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sun, 23 Jun 2013 12:56:24 +0200
Subject: [FreeVMS] L4Ka. Regression with 4.7 and 4.8 ?
In-Reply-To: <1371983504.3742.157.camel@yam-132-YW-E178-FTW>
References: <51C1AB3A.5080907@systella.fr> <2138987.2tPF7NEjmS@localhost>
<51C376A5.4060107@systella.fr> <51C59940.3020009@systella.fr>
<1371906201.3742.53.camel@yam-132-YW-E178-FTW> <51C5E4CB.40205@systella.fr>
<1371924700.3742.108.camel@yam-132-YW-E178-FTW>
<51C5FD3C.3070609@systella.fr>
<1371932623.3742.132.camel@yam-132-YW-E178-FTW> <51C6C65B.50403@systella.fr>
<1371983504.3742.157.camel@yam-132-YW-E178-FTW>
Message-ID: <51C6D458.1000704@systella.fr>
Oleg Endo a ?crit :
> Simply merging .init_array into .text is not going to produce anything
> but some unused data in .text.
> .init_array is a vector of function pointers. These functions are
> supposed to be invoked by the startup code before invoking 'main'.
> You could try merging .init_array into the existing .ctors and see where
> it goes.
Thanks a lot for your explanations. With :
. = ALIGN(16);
/* cpu-local static initializers */
__ctors_CPU__ = .;
*(SORT(.ctors.3*))
QUAD(0)
/* node-local static initializers */
__ctors_NODE__ = .;
*(SORT(.ctors.2*))
QUAD(0)
/* global static initializers */
__ctors_GLOBAL__ = .;
*(SORT(.ctors.1*))
*(.init_array)
*(SORT(.init_array.*))
*(SORT(.ctors.*))
*(.ctors)
QUAD(0)
L4 can be built with gcc-4.7.
Regards,
JKB
From rmctwo at gmail.com Sun Jun 23 20:33:38 2013
From: rmctwo at gmail.com (Renee)
Date: Sun, 23 Jun 2013 14:33:38 -0400
Subject: [FreeVMS] Question
In-Reply-To: <51C6CE1B.5090904@systella.fr>
References: <000f01ce6f87$4a81c610$df855230$@gmail.com>
<51C6CE1B.5090904@systella.fr>
Message-ID: <003b01ce7040$2e438d80$8acaa880$@gmail.com>
Posix? Why write posix at all?
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
BERTRAND Jo?l
Sent: Sunday, June 23, 2013 6:30 AM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] Question
Johann 'Myrkraverk' Oskarsson a ?crit :
> On Sat, Jun 22, 2013 at 8:30 PM, Renee wrote:
>> And the embarrassing thing is that I may have asked it before.
>> Obviously we are not building a VAX. We can make this thing ?source
>> compatible? to be recompiled. And I don?t think a privileged or even a
CMKRNL task would work.
>> With that in mind, I?ll ask again
.what are we building? Certainly
>> some form of DCL can and must be built.
>
> And that task is quite independent of the kernel. DCL is a shell and
> can be written in C on top of Posix interfaces. I'm not advocating
> that, only pointing out it's an option.
DCL will be written when we'll have a running kernel. There are a
lot of stuff to do before starting DCL.
If you want to start to write some userland code (not kernel,
executive or system ones), in a first time you have to implemant
sources/userland/rtl and sources/userland/starlet to obtain a VMS API or to
port newlib on FreeVMS to obtain a real POSIX layer. But to write
STARLET/LIB$ or port newlib, we have to define some SYSCALLs. Today, only
SYSCALL$PRINT, SYSCALL$EXIT_VALUE and SYSCALL$KILL_THREAD have been written
(sys/sys_loop.c).
Regards,
JKB
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From johann at myrkraverk.com Sun Jun 23 23:23:58 2013
From: johann at myrkraverk.com (Johann 'Myrkraverk' Oskarsson)
Date: Sun, 23 Jun 2013 21:23:58 +0000
Subject: [FreeVMS] Question
In-Reply-To: <003b01ce7040$2e438d80$8acaa880$@gmail.com>
References: <000f01ce6f87$4a81c610$df855230$@gmail.com>
<51C6CE1B.5090904@systella.fr>
<003b01ce7040$2e438d80$8acaa880$@gmail.com>
Message-ID:
Hi,
On Sun, Jun 23, 2013 at 6:33 PM, Renee wrote:
> Posix? Why write posix at all?
> Renee
Posix is an important part of any server environment today; at least
the C RTL parts of it. It's better to think about it from the start
rather than try to wedge it on the system after the fact.
> -----Original Message-----
> From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
> BERTRAND Jo?l
> DCL will be written when we'll have a running kernel. There are a
> lot of stuff to do before starting DCL.
True.
> If you want to start to write some userland code (not kernel,
> executive or system ones), in a first time you have to implemant
> sources/userland/rtl and sources/userland/starlet to obtain a VMS API or to
> port newlib on FreeVMS to obtain a real POSIX layer. But to write
> STARLET/LIB$ or port newlib, we have to define some SYSCALLs. Today, only
> SYSCALL$PRINT, SYSCALL$EXIT_VALUE and SYSCALL$KILL_THREAD have been written
> (sys/sys_loop.c).
What exactly is newlib?
It's not my call, but I would build the kernel to provide both VMS and
Posix system calls, similar what's been done in Haiku where they
implement both BeOS and a subset of Posix system calls.
Is there a tutorial on how to build and run FreeVMS? Something that
demonstrates how to go from pure source code to running it in a
(virtual) machine?
--
Johann
I'm not from the internet, I just work there.
From joel.bertrand at systella.fr Sun Jun 23 23:52:56 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sun, 23 Jun 2013 23:52:56 +0200
Subject: [FreeVMS] Question
In-Reply-To:
References: <000f01ce6f87$4a81c610$df855230$@gmail.com>
<51C6CE1B.5090904@systella.fr> <003b01ce7040$2e438d80$8acaa880$@gmail.com>
Message-ID: <51C76E38.6030200@systella.fr>
Johann 'Myrkraverk' Oskarsson a ?crit :
> Hi,
>
> On Sun, Jun 23, 2013 at 6:33 PM, Renee wrote:
>> Posix? Why write posix at all?
>> Renee
>
> Posix is an important part of any server environment today; at least
> the C RTL parts of it. It's better to think about it from the start
> rather than try to wedge it on the system after the fact.
>
>> -----Original Message-----
>> From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
>> BERTRAND Jo?l
>
>> DCL will be written when we'll have a running kernel. There are a
>> lot of stuff to do before starting DCL.
>
> True.
>
>> If you want to start to write some userland code (not kernel,
>> executive or system ones), in a first time you have to implemant
>> sources/userland/rtl and sources/userland/starlet to obtain a VMS API or to
>> port newlib on FreeVMS to obtain a real POSIX layer. But to write
>> STARLET/LIB$ or port newlib, we have to define some SYSCALLs. Today, only
>> SYSCALL$PRINT, SYSCALL$EXIT_VALUE and SYSCALL$KILL_THREAD have been written
>> (sys/sys_loop.c).
>
> What exactly is newlib?
>
> It's not my call, but I would build the kernel to provide both VMS and
> Posix system calls, similar what's been done in Haiku where they
> implement both BeOS and a subset of Posix system calls.
Newlib is a POSIX libc that only uses a few syscalls.
> Is there a tutorial on how to build and run FreeVMS? Something that
> demonstrates how to go from pure source code to running it in a
> (virtual) machine?
You have an howto in sources to build a qemu disk image. Today, you
cannot run any code as FreeVMS requires a pager (and sometimes, kernel
freezes during initialization as recursive mutexes are broken. I think I
will fix these mutexes this monday and push modifications in git tree).
Regards,
JKB
From joel.bertrand at systella.fr Sun Jun 23 23:57:10 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sun, 23 Jun 2013 23:57:10 +0200
Subject: [FreeVMS] Question
In-Reply-To: <51C76E38.6030200@systella.fr>
References: <000f01ce6f87$4a81c610$df855230$@gmail.com>
<51C6CE1B.5090904@systella.fr> <003b01ce7040$2e438d80$8acaa880$@gmail.com>
<51C76E38.6030200@systella.fr>
Message-ID: <51C76F36.2000203@systella.fr>
BERTRAND Jo?l a ?crit :
> Johann 'Myrkraverk' Oskarsson a ?crit :
>> Hi,
>>
>> On Sun, Jun 23, 2013 at 6:33 PM, Renee wrote:
>>> Posix? Why write posix at all?
>>> Renee
>>
>> Posix is an important part of any server environment today; at least
>> the C RTL parts of it. It's better to think about it from the start
>> rather than try to wedge it on the system after the fact.
>>
>>> -----Original Message-----
>>> From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
>>> BERTRAND Jo?l
>>
>>> DCL will be written when we'll have a running kernel. There are a
>>> lot of stuff to do before starting DCL.
>>
>> True.
>>
>>> If you want to start to write some userland code (not kernel,
>>> executive or system ones), in a first time you have to implemant
>>> sources/userland/rtl and sources/userland/starlet to obtain a VMS API
>>> or to
>>> port newlib on FreeVMS to obtain a real POSIX layer. But to write
>>> STARLET/LIB$ or port newlib, we have to define some SYSCALLs. Today,
>>> only
>>> SYSCALL$PRINT, SYSCALL$EXIT_VALUE and SYSCALL$KILL_THREAD have been
>>> written
>>> (sys/sys_loop.c).
>>
>> What exactly is newlib?
>>
>> It's not my call, but I would build the kernel to provide both VMS and
>> Posix system calls, similar what's been done in Haiku where they
>> implement both BeOS and a subset of Posix system calls.
>
> Newlib is a POSIX libc that only uses a few syscalls.
Sorry, I have forgotten to add its url :
http://sourceware.org/newlib/
JKB
From alex22_7 at hotmail.com Thu Jun 27 12:36:13 2013
From: alex22_7 at hotmail.com (Slex Sangiuliano)
Date: Thu, 27 Jun 2013 10:36:13 +0000
Subject: [FreeVMS] New, L4 and OS development
Message-ID:
Hello,
I'm new to this mailing list, and very new to the FreeVMS project, this is why I have some questions.
1) Do you have an IRC channel?
2) Do you accept code from more permissive licenses like BSD and MIT?
3) Are new comers that want to learn OS development, and that can help in someway, accepted?
4) Is this project registerd to ohloh?
I'm very interested in ?kernel based OS development, but I have to say I'm very new to the VMS world, so my intention is to learn and help where I can.
I'm a computer science student and going to finish mi Bachelor.
You can find me on ohloh:
https://www.ohloh.net/people?q=Alessandro%20Sangiuliano
Do you think and IRC channel can be usefull?
Yes, I would join this project but woth humility and learning spirit.
Thank you,
Cheers
Alex.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From joel.bertrand at systella.fr Thu Jun 27 21:48:50 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Thu, 27 Jun 2013 21:48:50 +0200
Subject: [FreeVMS] New, L4 and OS development
In-Reply-To:
References:
Message-ID: <51CC9722.3020707@systella.fr>
Slex Sangiuliano a ?crit :
> Hello,
Hello,
> I'm new to this mailing list, and very new to the FreeVMS project, this
> is why I have some questions.
>
> 1) Do you have an IRC channel?
No, we haven't, but I can open a jabber server. I prefer mailing list
as mailman archives all discussions.
> 2) Do you accept code from more permissive licenses like BSD and MIT?
L4ka is given with a BSD license, but I prefer GPL (or CeCCILv2).
> 3) Are new comers that want to learn OS development, and that can help
> in someway, accepted?
Why not ?
> 4) Is this project registerd to ohloh?
No. On Freecode.
> I'm very interested in ?kernel based OS development, but I have to say
> I'm very new to the VMS world, so my intention is to learn and help
> where I can.
Today, we need to write a pager that is not related to VMS itself.
> I'm a computer science student and going to finish mi Bachelor.
> You can find me on ohloh:
>
> https://www.ohloh.net/people?q=Alessandro%20Sangiuliano
>
> Do you think and IRC channel can be usefull?
I don't think. I prefer open some new and more specialized mailing lists.
> Yes, I would join this project but woth humility and learning spirit.
>
> Thank you,
You're welcome,
JKB
From alex22_7 at hotmail.com Fri Jun 28 12:29:30 2013
From: alex22_7 at hotmail.com (Slex Sangiuliano)
Date: Fri, 28 Jun 2013 10:29:30 +0000
Subject: [FreeVMS] New, L4 and OS development
In-Reply-To: <51CC9722.3020707@systella.fr>
References: ,
<51CC9722.3020707@systella.fr>
Message-ID:
> Date: Thu, 27 Jun 2013 21:48:50 +0200
> From: joel.bertrand at systella.fr
> To: freevms at rayleigh.systella.fr
> Subject: Re: [FreeVMS] New, L4 and OS development
>
> Slex Sangiuliano a ?crit :
> > Hello,
>
> Hello,
>
> > I'm new to this mailing list, and very new to the FreeVMS project, this
> > is why I have some questions.
> >
> > 1) Do you have an IRC channel?
>
> No, we haven't, but I can open a jabber server. I prefer mailing list
> as mailman archives all discussions.
A jabber server would be fine to me, so I can do lot of learning questions. Consider me like a n00b
Thank You for your patience.
Cheers,
Alex
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From herb.nowell at gmail.com Fri Jun 28 15:12:24 2013
From: herb.nowell at gmail.com (Herb Nowell)
Date: Fri, 28 Jun 2013 09:12:24 -0400
Subject: [FreeVMS] New, L4 and OS development
In-Reply-To:
References:
<51CC9722.3020707@systella.fr>
Message-ID:
As another newbie with little OS programming experience I'm also at a loss
as where to start in helping move the project forward. My interest in VMS
is driven mostly by the file system so I'd like to start there but I get
the impression that's well in the future.
On Fri, Jun 28, 2013 at 6:29 AM, Slex Sangiuliano wrote:
>
>
> > Date: Thu, 27 Jun 2013 21:48:50 +0200
> > From: joel.bertrand at systella.fr
> > To: freevms at rayleigh.systella.fr
> > Subject: Re: [FreeVMS] New, L4 and OS development
>
> >
> > Slex Sangiuliano a ?crit :
> > > Hello,
> >
> > Hello,
> >
> > > I'm new to this mailing list, and very new to the FreeVMS project, this
> > > is why I have some questions.
> > >
> > > 1) Do you have an IRC channel?
> >
> > No, we haven't, but I can open a jabber server. I prefer mailing list
> > as mailman archives all discussions.
>
> A jabber server would be fine to me, so I can do lot of learning
> questions. Consider me like a n00b
>
> Thank You for your patience.
>
> Cheers,
> Alex
>
> _______________________________________________
> FreeVMS mailing list
> FreeVMS at rayleigh.systella.fr
> https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
>
>
--
Herbert H. Nowell
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
From guidoj at users.sourceforge.net Fri Jun 28 20:59:45 2013
From: guidoj at users.sourceforge.net (Guido)
Date: Fri, 28 Jun 2013 20:59:45 +0200
Subject: [FreeVMS] pager question
Message-ID: <2017497.RiuaGyRr4O@kermit>
Hi all,
Looking at the pager algorithm I noticed that a page fault in user space could
lead to allocation of a new page for the process that generated the page
fault. Maybe I'm missing something, but it seems to me that that is not
desired behavior. Buggy or malicious software in user space should not be able
to acquire resources simply by accessing invalid memory addresses. In stead
this should lead to an access violation and termination of the originating
process.
I'd suggest that page fault handling and mapping/granting memory are entirely
separate functions. A page fault on a page that is not mapped is an access
violation and should lead to termination of the originating process. If the
page is mapped, the pager could (like OpenVMS does) distinguish between soft
and hard page faults. A soft page fault means that the page still resides
somewhere in cache, while a hard page fault means it is swapped out and has to
be loaded from disk. This depends on the paging algorithm.
Memory allocation and mapping/granting pages is something completely different.
My suggestion would be that allocation requests by user tasks are handled by
the task pager. If the task pager runs out of memory it can request one or
more pages from the main pager through a page request system call. The main
pager can map/grant a page from the free page pool it maintains or, if the
pool is empty, it can request a new page from the kernel. If the kernel runs
out of free pages, the pager should free up some pages by swapping out some
pages to disk. Note that this could (should?) be optimised by initially
mapping/granting all free pages to the main pager, so that the pager does not
have to use system calls to request memory from the kernel once the system is
up and running.
I could be wrong, of course ...
Regards,
Guido
From joel.bertrand at systella.fr Sat Jun 29 10:34:19 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sat, 29 Jun 2013 10:34:19 +0200
Subject: [FreeVMS] Mutexes are broken
Message-ID: <51CE9C0B.4080202@systella.fr>
Hello,
Don't use git tree to build FreeVMS from scratch. Mutexes are currently
broken and roottask allocator enters in deadlock. I'm investigating.
Regards,
JKB
From joel.bertrand at systella.fr Sat Jun 29 12:44:23 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sat, 29 Jun 2013 12:44:23 +0200
Subject: [FreeVMS] pager question
In-Reply-To: <2017497.RiuaGyRr4O@kermit>
References: <2017497.RiuaGyRr4O@kermit>
Message-ID: <51CEBA87.3050107@systella.fr>
Guido a ?crit :
> Hi all,
Hello,
> Looking at the pager algorithm I noticed that a page fault in user space could
> lead to allocation of a new page for the process that generated the page
> fault.
Right.
> Maybe I'm missing something, but it seems to me that that is not
> desired behavior. Buggy or malicious software in user space should not be able
> to acquire resources simply by accessing invalid memory addresses. In stead
> this should lead to an access violation and termination of the originating
> process.
I agree. For kernel itself, sys_pagefault always returns new
ressources. For an unprivilegied process, pager has to save a list of
allocated pages for this process. Pagefault handler only has to return
new pages when process is loaded in memory, when stack is growing or
when process uses something like mmap() or brk(). In all other cases,
pagefault has to raise a new syscall to kill process.
> I'd suggest that page fault handling and mapping/granting memory are entirely
> separate functions. A page fault on a page that is not mapped is an access
> violation and should lead to termination of the originating process. If the
> page is mapped, the pager could (like OpenVMS does) distinguish between soft
> and hard page faults. A soft page fault means that the page still resides
> somewhere in cache, while a hard page fault means it is swapped out and has to
> be loaded from disk. This depends on the paging algorithm.
I agree.
> Memory allocation and mapping/granting pages is something completely different.
> My suggestion would be that allocation requests by user tasks are handled by
> the task pager. If the task pager runs out of memory it can request one or
> more pages from the main pager through a page request system call. The main
> pager can map/grant a page from the free page pool it maintains or, if the
> pool is empty, it can request a new page from the kernel. If the kernel runs
> out of free pages, the pager should free up some pages by swapping out some
> pages to disk. Note that this could (should?) be optimised by initially
> mapping/granting all free pages to the main pager, so that the pager does not
> have to use system calls to request memory from the kernel once the system is
> up and running.
>
> I could be wrong, of course ...
No, I don't think. In a first time, I propose that task pager doesn't
use swapped memory as we need disk I/O drivers to handle swap and as
these drivers require task pager.
Kickstart loads L4/X2, sigma0 and several modules :
- vmskernel.sys (uses sigma0 pager) ;
- pager.sys (main pager that use sys_pagefault) that is launched
_without_ swapped memory area as disk I/O are not available ;
- dq driver + partionning to add swapping capabilities.
Regards,
JKB
From guidoj at users.sourceforge.net Sat Jun 29 14:59:23 2013
From: guidoj at users.sourceforge.net (Guido)
Date: Sat, 29 Jun 2013 14:59:23 +0200
Subject: [FreeVMS] pager question
In-Reply-To: <51CEBA87.3050107@systella.fr>
References: <2017497.RiuaGyRr4O@kermit> <51CEBA87.3050107@systella.fr>
Message-ID: <3734761.O7y1JblQ8u@kermit>
On Saturday 29 June 2013 12:44:23 BERTRAND Jo?l wrote:
> I agree. For kernel itself, sys_pagefault always returns new
> ressources. For an unprivilegied process, pager has to save a list of
> allocated pages for this process. Pagefault handler only has to return
> new pages when process is loaded in memory, when stack is growing or
> when process uses something like mmap() or brk(). In all other cases,
> pagefault has to raise a new syscall to kill process.
OK, but could you explain why a page fault (essentially a hardware generated
error condition) and a memory allocation request (software generated, on
purpose) are handled by the same function (the pagefault handler) when they
are to be handled differently? Wouldn't it be easier to have separate IPC calls
and handlers?
> No, I don't think. In a first time, I propose that task pager doesn't
> use swapped memory as we need disk I/O drivers to handle swap and as
> these drivers require task pager.
Agreed. My suggestion was more an ultimate goal. At first don't bother with
swap and disk I/O, physical memory should be more than adequate for now.
Regards,
Guido
From joel.bertrand at systella.fr Sat Jun 29 18:10:16 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sat, 29 Jun 2013 18:10:16 +0200
Subject: [FreeVMS] pager question
In-Reply-To: <3734761.O7y1JblQ8u@kermit>
References: <2017497.RiuaGyRr4O@kermit> <51CEBA87.3050107@systella.fr>
<3734761.O7y1JblQ8u@kermit>
Message-ID: <51CF06E8.4060208@systella.fr>
Guido a ?crit :
> On Saturday 29 June 2013 12:44:23 BERTRAND Jo?l wrote:
>> I agree. For kernel itself, sys_pagefault always returns new
>> ressources. For an unprivilegied process, pager has to save a list of
>> allocated pages for this process. Pagefault handler only has to return
>> new pages when process is loaded in memory, when stack is growing or
>> when process uses something like mmap() or brk(). In all other cases,
>> pagefault has to raise a new syscall to kill process.
>
> OK, but could you explain why a page fault (essentially a hardware generated
> error condition) and a memory allocation request (software generated, on
> purpose) are handled by the same function (the pagefault handler) when they
> are to be handled differently? Wouldn't it be easier to have separate IPC calls
> and handlers?
When a process try to access to a unmapped page, it generates a
pagefault IPC. This IPC is received by an event handler loop in its
parent (for example, PAGER.SYS sends L4_PAGEFAULT IPC to VMSKERNEL.SYS
and all IPC are caught in sys/sys_loop.c).
Child process does not know if its request can or will be done. It only
sends L4_PAGEFAULT IPC to its parent and waits for an answer. This
answer can be a new L4_MapItem() or a L4_AbortIpc_and_stop(). Thus, I
don't see why we need to separate L4_PAGEFAULT IPC in two IPC calls.
Regards,
JKB
From guidoj at users.sourceforge.net Sat Jun 29 21:02:34 2013
From: guidoj at users.sourceforge.net (Guido)
Date: Sat, 29 Jun 2013 21:02:34 +0200
Subject: [FreeVMS] pager question
In-Reply-To: <51CF06E8.4060208@systella.fr>
References: <2017497.RiuaGyRr4O@kermit> <3734761.O7y1JblQ8u@kermit>
<51CF06E8.4060208@systella.fr>
Message-ID: <5517291.U0PMCm2LmT@kermit>
On Saturday 29 June 2013 18:10:16 BERTRAND Jo?l wrote:
> When a process try to access to a unmapped page, it generates a
> pagefault IPC. This IPC is received by an event handler loop in its
> parent (for example, PAGER.SYS sends L4_PAGEFAULT IPC to VMSKERNEL.SYS
> and all IPC are caught in sys/sys_loop.c).
>
> Child process does not know if its request can or will be done. It only
> sends L4_PAGEFAULT IPC to its parent and waits for an answer. This
> answer can be a new L4_MapItem() or a L4_AbortIpc_and_stop(). Thus, I
> don't see why we need to separate L4_PAGEFAULT IPC in two IPC calls.
The question is if the the process should try to access an unmapped page if it
actually tries to allocate memory. Shouldn't the process try to request more
pages first and wait for them to be mapped into its address space before it
tries to access them?
It seems to me that this makes the pagefault handling algorithm needlessly
complex. The algorithm could as simple as "if the page is mapped, load it from
cache or swap, otherwise kill the process". A pagefault should be an exception
not a function.
IMHO a page fault and a memory request are completely different and should not
be combined into one IPC call. I'm afraid that combining them will in de long
run this only hurt reliability, stability and maintainability of the OS.
Regards,
Guido
From joel.bertrand at systella.fr Sat Jun 29 22:21:38 2013
From: joel.bertrand at systella.fr (=?ISO-8859-1?Q?BERTRAND_Jo=EBl?=)
Date: Sat, 29 Jun 2013 22:21:38 +0200
Subject: [FreeVMS] pager question
In-Reply-To: <5517291.U0PMCm2LmT@kermit>
References: <2017497.RiuaGyRr4O@kermit> <3734761.O7y1JblQ8u@kermit>
<51CF06E8.4060208@systella.fr> <5517291.U0PMCm2LmT@kermit>
Message-ID: <51CF41D2.8000901@systella.fr>
Guido a ?crit :
> On Saturday 29 June 2013 18:10:16 BERTRAND Jo?l wrote:
>> When a process try to access to a unmapped page, it generates a
>> pagefault IPC. This IPC is received by an event handler loop in its
>> parent (for example, PAGER.SYS sends L4_PAGEFAULT IPC to VMSKERNEL.SYS
>> and all IPC are caught in sys/sys_loop.c).
>>
>> Child process does not know if its request can or will be done. It only
>> sends L4_PAGEFAULT IPC to its parent and waits for an answer. This
>> answer can be a new L4_MapItem() or a L4_AbortIpc_and_stop(). Thus, I
>> don't see why we need to separate L4_PAGEFAULT IPC in two IPC calls.
>
> The question is if the the process should try to access an unmapped page if it
> actually tries to allocate memory. Shouldn't the process try to request more
> pages first and wait for them to be mapped into its address space before it
> tries to access them?
With L4, you only map a page in process address space when you try to
access to this page.
> It seems to me that this makes the pagefault handling algorithm needlessly
> complex. The algorithm could as simple as "if the page is mapped, load it from
> cache or swap, otherwise kill the process". A pagefault should be an exception
> not a function.
L4_PAGEFAULT IPC is sent when a process tries to access to unmapped
memory. Process sends this IPC when it is loaded in memory too. Thus, I
think we have to use a second syscall to indicate to pager two modes :
1/ process can request memory (initialization or brk()) ;
2/ process has to be stopped (access violation).
> IMHO a page fault and a memory request are completely different and should not
> be combined into one IPC call. I'm afraid that combining them will in de long
> run this only hurt reliability, stability and maintainability of the OS.
On L4, you only have one IPC. But you can add other IPC's to switch
between more than one functions.
Regards,
JKB
From rmctwo at gmail.com Sat Jun 29 22:36:01 2013
From: rmctwo at gmail.com (Renee)
Date: Sat, 29 Jun 2013 16:36:01 -0400
Subject: [FreeVMS] pager question
In-Reply-To: <51CEBA87.3050107@systella.fr>
References: <2017497.RiuaGyRr4O@kermit> <51CEBA87.3050107@systella.fr>
Message-ID: <000e01ce7508$45303b20$cf90b160$@gmail.com>
This process is good for me because it brings my memory back.
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
BERTRAND Jo?l
Sent: Saturday, June 29, 2013 6:44 AM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] pager question
Guido a ?crit :
> Hi all,
Hello,
> Looking at the pager algorithm I noticed that a page fault in user
> space could lead to allocation of a new page for the process that
> generated the page fault.
Right.
> Maybe I'm missing something, but it seems to me that that is not
> desired behavior. Buggy or malicious software in user space should not
> be able to acquire resources simply by accessing invalid memory
> addresses. In stead this should lead to an access violation and
> termination of the originating process.
O It seems this process is devoid of acknowledgement that sysgen existed and
was an important part of resource allocation and limiting said allocation.
I agree. For kernel itself, sys_pagefault always returns new
ressources. For an unprivilegied process, pager has to save a list of
allocated pages for this process.
O I should add that the database in memory was complex.
Pagefault handler only has to return new pages when process is loaded in
memory, when stack is growing or when process uses something like mmap() or
brk(). In all other cases, pagefault has to raise a new syscall to kill
process.
Again process defaults are set by sysgen.
> I'd suggest that page fault handling and mapping/granting memory are
> entirely separate functions. A page fault on a page that is not mapped
> is an access violation and should lead to termination of the
> originating process. If the page is mapped, the pager could (like
> OpenVMS does) distinguish between soft and hard page faults. A soft
> page fault means that the page still resides somewhere in cache, while
> a hard page fault means it is swapped out and has to be loaded from disk.
This depends on the paging algorithm.
I agree.
> Memory allocation and mapping/granting pages is something completely
different.
> My suggestion would be that allocation requests by user tasks are
> handled by the task pager. If the task pager runs out of memory it can
> request one or more pages from the main pager through a page request
> system call. The main pager can map/grant a page from the free page
> pool it maintains or, if the pool is empty, it can request a new page
> from the kernel. If the kernel runs out of free pages, the pager
> should free up some pages by swapping out some pages to disk. Note
> that this could (should?) be optimised by initially mapping/granting
> all free pages to the main pager, so that the pager does not have to
> use system calls to request memory from the kernel once the system is up
and running.
>
> I could be wrong, of course ...
No, I don't think. In a first time, I propose that task pager
doesn't use swapped memory as we need disk I/O drivers to handle swap and as
these drivers require task pager.
Kickstart loads L4/X2, sigma0 and several modules :
- vmskernel.sys (uses sigma0 pager) ;
- pager.sys (main pager that use sys_pagefault) that is launched _without_
swapped memory area as disk I/O are not available ;
- dq driver + partionning to add swapping capabilities.
Renee
From rmctwo at gmail.com Sat Jun 29 22:38:28 2013
From: rmctwo at gmail.com (Renee)
Date: Sat, 29 Jun 2013 16:38:28 -0400
Subject: [FreeVMS] pager question
In-Reply-To: <5517291.U0PMCm2LmT@kermit>
References: <2017497.RiuaGyRr4O@kermit> <3734761.O7y1JblQ8u@kermit>
<51CF06E8.4060208@systella.fr> <5517291.U0PMCm2LmT@kermit>
Message-ID: <000f01ce7508$9ccfdf20$d66f9d60$@gmail.com>
Guido... I hate doing this,,,but what does IPC stand for?
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Guido
Sent: Saturday, June 29, 2013 3:03 PM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] pager question
On Saturday 29 June 2013 18:10:16 BERTRAND Jo?l wrote:
> When a process try to access to a unmapped page, it generates a
> pagefault IPC. This IPC is received by an event handler loop in its
> parent (for example, PAGER.SYS sends L4_PAGEFAULT IPC to VMSKERNEL.SYS
> and all IPC are caught in sys/sys_loop.c).
>
> Child process does not know if its request can or will be done. It
> only sends L4_PAGEFAULT IPC to its parent and waits for an answer.
> This answer can be a new L4_MapItem() or a L4_AbortIpc_and_stop().
> Thus, I don't see why we need to separate L4_PAGEFAULT IPC in two IPC
calls.
The question is if the the process should try to access an unmapped page if
it actually tries to allocate memory. Shouldn't the process try to request
more pages first and wait for them to be mapped into its address space
before it tries to access them?
It seems to me that this makes the pagefault handling algorithm needlessly
complex. The algorithm could as simple as "if the page is mapped, load it
from cache or swap, otherwise kill the process". A pagefault should be an
exception not a function.
IMHO a page fault and a memory request are completely different and should
not be combined into one IPC call. I'm afraid that combining them will in de
long run this only hurt reliability, stability and maintainability of the
OS.
Regards,
Guido
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From pedersen at ccsscorp.com Sat Jun 29 23:04:17 2013
From: pedersen at ccsscorp.com (Bill Pedersen)
Date: Sat, 29 Jun 2013 17:04:17 -0400
Subject: [FreeVMS] pager question
In-Reply-To: <000f01ce7508$9ccfdf20$d66f9d60$@gmail.com>
References: <2017497.RiuaGyRr4O@kermit> <3734761.O7y1JblQ8u@kermit>
<51CF06E8.4060208@systella.fr> <5517291.U0PMCm2LmT@kermit>
<000f01ce7508$9ccfdf20$d66f9d60$@gmail.com>
Message-ID: <061201ce750c$388942a0$a99bc7e0$@ccsscorp.com>
Inter-Process Communications...
This might be a good starting point:
http://en.wikipedia.org/wiki/Microkernel
Bill Pedersen
CCSS - Computer Consulting System Services, LLC
211 Ruth Drive
Gaffney, SC 29341
Telephone: 864-490-8863
Mobile: 408-892-5204
Facsimile: 206-984-3068
www: www.ccsscorp.com
Skype Name: william.a.pedersen
LinkedIn Profile: www.linkedin.com/in/billpedersen
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Renee
Sent: Saturday, June 29, 2013 4:38 PM
To: 'FreeVMS mailing list'
Subject: Re: [FreeVMS] pager question
Guido... I hate doing this,,,but what does IPC stand for?
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Guido
Sent: Saturday, June 29, 2013 3:03 PM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] pager question
On Saturday 29 June 2013 18:10:16 BERTRAND Jo?l wrote:
> When a process try to access to a unmapped page, it generates a
> pagefault IPC. This IPC is received by an event handler loop in its
> parent (for example, PAGER.SYS sends L4_PAGEFAULT IPC to VMSKERNEL.SYS
> and all IPC are caught in sys/sys_loop.c).
>
> Child process does not know if its request can or will be done. It
> only sends L4_PAGEFAULT IPC to its parent and waits for an answer.
> This answer can be a new L4_MapItem() or a L4_AbortIpc_and_stop().
> Thus, I don't see why we need to separate L4_PAGEFAULT IPC in two IPC
calls.
The question is if the the process should try to access an unmapped page if
it actually tries to allocate memory. Shouldn't the process try to request
more pages first and wait for them to be mapped into its address space
before it tries to access them?
It seems to me that this makes the pagefault handling algorithm needlessly
complex. The algorithm could as simple as "if the page is mapped, load it
from cache or swap, otherwise kill the process". A pagefault should be an
exception not a function.
IMHO a page fault and a memory request are completely different and should
not be combined into one IPC call. I'm afraid that combining them will in de
long run this only hurt reliability, stability and maintainability of the
OS.
Regards,
Guido
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
From rmctwo at gmail.com Sun Jun 30 01:56:40 2013
From: rmctwo at gmail.com (Renee)
Date: Sat, 29 Jun 2013 19:56:40 -0400
Subject: [FreeVMS] pager question
In-Reply-To: <061201ce750c$388942a0$a99bc7e0$@ccsscorp.com>
References: <2017497.RiuaGyRr4O@kermit> <3734761.O7y1JblQ8u@kermit>
<51CF06E8.4060208@systella.fr> <5517291.U0PMCm2LmT@kermit>
<000f01ce7508$9ccfdf20$d66f9d60$@gmail.com>
<061201ce750c$388942a0$a99bc7e0$@ccsscorp.com>
Message-ID: <000201ce7524$4d0c5790$e72506b0$@gmail.com>
There is a difference. When we did inter-process communication at DEC, we
called it " inter-process communication" we didn't abbreviate it to IPC. Of
course, communication was generally oral and you'd be amazed at how little
inter-process communication is done when your writing an exec.
You'll probably find me asking other questions like this especially around
acronyms.
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Bill Pedersen
Sent: Saturday, June 29, 2013 5:04 PM
To: 'FreeVMS mailing list'
Subject: Re: [FreeVMS] pager question
Inter-Process Communications...
This might be a good starting point:
http://en.wikipedia.org/wiki/Microkernel
Bill Pedersen
CCSS - Computer Consulting System Services, LLC
211 Ruth Drive
Gaffney, SC 29341
Telephone: 864-490-8863
Mobile: 408-892-5204
Facsimile: 206-984-3068
www: www.ccsscorp.com
Skype Name: william.a.pedersen
LinkedIn Profile: www.linkedin.com/in/billpedersen
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Renee
Sent: Saturday, June 29, 2013 4:38 PM
To: 'FreeVMS mailing list'
Subject: Re: [FreeVMS] pager question
Guido... I hate doing this,,,but what does IPC stand for?
Renee
-----Original Message-----
From: FreeVMS [mailto:freevms-bounces at rayleigh.systella.fr] On Behalf Of
Guido
Sent: Saturday, June 29, 2013 3:03 PM
To: FreeVMS mailing list
Subject: Re: [FreeVMS] pager question
On Saturday 29 June 2013 18:10:16 BERTRAND Jo?l wrote:
> When a process try to access to a unmapped page, it generates a
> pagefault IPC. This IPC is received by an event handler loop in its
> parent (for example, PAGER.SYS sends L4_PAGEFAULT IPC to VMSKERNEL.SYS
> and all IPC are caught in sys/sys_loop.c).
>
> Child process does not know if its request can or will be done. It
> only sends L4_PAGEFAULT IPC to its parent and waits for an answer.
> This answer can be a new L4_MapItem() or a L4_AbortIpc_and_stop().
> Thus, I don't see why we need to separate L4_PAGEFAULT IPC in two IPC
calls.
The question is if the the process should try to access an unmapped page if
it actually tries to allocate memory. Shouldn't the process try to request
more pages first and wait for them to be mapped into its address space
before it tries to access them?
It seems to me that this makes the pagefault handling algorithm needlessly
complex. The algorithm could as simple as "if the page is mapped, load it
from cache or swap, otherwise kill the process". A pagefault should be an
exception not a function.
IMHO a page fault and a memory request are completely different and should
not be combined into one IPC call. I'm afraid that combining them will in de
long run this only hurt reliability, stability and maintainability of the
OS.
Regards,
Guido
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms
_______________________________________________
FreeVMS mailing list
FreeVMS at rayleigh.systella.fr
https://www.systella.fr/cgi-bin/mailman/listinfo/freevms