headers
mickey | 1 Apr 2008 20:00
Gravatar

HIMEM.SYS strikes back!

re
80s are back and so is HIMEM.SYS now for the modern 32bit
architecture known as i386 or "that intel crap."

yes as you can have guessed already it gives your >4G
memory another chance at life reincarnated as scsi disk.

	http://mickey.lucifier.net/himem.sys

you can read it write it and damn swap on it if you want!
credit shall be given to mpf <at>  for a productive discussion
where this most atypical concept had been born few months
ago and later became stations creation in about 30h (and
a bottle of port) and later debugged/tested in a few more
other more sober hours...

few caveats:
- man page coming soon!
- booting is not supported (yet);
- one needs to disklabel+newfs(or dd(2) some image) it before use;
- it will get some performance improvement;
- not stress-tested much (yet).

here is samples:
himem0 at root: size 768MB
scsibus2 at himem0: 1 targets
sd2 at scsibus2 targ 0 lun 0: <McIkye, HIMEM drive, 1.0> SCSI2 0/direct fixed
sd2: 768MB, 12 cyl, 255 head, 63 sec, 4096 bytes/sec, 196608 sec total

# dd if=/dev/rsd2c of=/dev/null bs=256k
(Continue reading)

Permalink | Reply | 19 comments |
headers
Marco Peereboom | 1 Apr 2008 20:18
Picon
Favicon

Re: HIMEM.SYS strikes back!

I suggest you add read/write/read capacity(16) for those extra super
large memory machines.

On Tue, Apr 01, 2008 at 06:00:38PM +0000, mickey wrote:
> re
> 80s are back and so is HIMEM.SYS now for the modern 32bit
> architecture known as i386 or "that intel crap."
> 
> yes as you can have guessed already it gives your >4G
> memory another chance at life reincarnated as scsi disk.
> 
> 	http://mickey.lucifier.net/himem.sys
> 
> you can read it write it and damn swap on it if you want!
> credit shall be given to mpf <at>  for a productive discussion
> where this most atypical concept had been born few months
> ago and later became stations creation in about 30h (and
> a bottle of port) and later debugged/tested in a few more
> other more sober hours...
> 
> few caveats:
> - man page coming soon!
> - booting is not supported (yet);
> - one needs to disklabel+newfs(or dd(2) some image) it before use;
> - it will get some performance improvement;
> - not stress-tested much (yet).
> 
> here is samples:
> himem0 at root: size 768MB
> scsibus2 at himem0: 1 targets
(Continue reading)

Permalink | Reply | 6 comments |
headers
Eichert, Diana | 1 Apr 2008 20:31
Picon

Re: HIMEM.SYS strikes back!

does it work with the lb(4) devices?
Permalink | Reply | 3 comments |
headers
Christopher Linn | 1 Apr 2008 20:41
Picon
Favicon

Re: HIMEM.SYS strikes back!

re.

On Tue, Apr 01, 2008 at 06:00:38PM +0000, mickey wrote:
> re
> 80s are back and so is HIMEM.SYS now for the modern 32bit
> architecture known as i386 or "that intel crap."
> 
> yes as you can have guessed already it gives your >4G
> memory another chance at life reincarnated as scsi disk.
> 
> 	http://mickey.lucifier.net/himem.sys
> 
> you can read it write it and damn swap on it if you want!
> credit shall be given to mpf <at>  for a productive discussion
> where this most atypical concept had been born few months
> ago and later became stations creation in about 30h (and
> a bottle of port) and later debugged/tested in a few more
> other more sober hours...

you gotta be freekin kidding me.  just ONE bottle of port!?!

cel

> few caveats:
> - man page coming soon!
> - booting is not supported (yet);
> - one needs to disklabel+newfs(or dd(2) some image) it before use;
> - it will get some performance improvement;
> - not stress-tested much (yet).
>
(Continue reading)

Permalink | Reply | 1 comment |
headers
Matthew Dempsky | 1 Apr 2008 22:27
Favicon

Re: HIMEM.SYS strikes back!

+	if (cf->cf_unit > 1)
+		return 0;

Shouldn't this be >= 1, or would you ever want himem1?
Permalink | Reply | 1 comment |
headers
mickey | 1 Apr 2008 23:29
Gravatar

Re: HIMEM.SYS strikes back!

On Tue, Apr 01, 2008 at 12:31:09PM -0600, Eichert, Diana wrote:
> does it work with the lb(4) devices?

should you dare to read the mail it would become obvious
that it exposes itself as sd(4)... in all my ignorance
i was not aware of this lb(4) innovation in openbsd.
cu
--

-- 
    paranoic mickey       (my employers have changed but, the name has remained)
Permalink | Reply | 2 comments |
headers
mickey | 1 Apr 2008 23:33
Gravatar

Re: HIMEM.SYS strikes back!

On Tue, Apr 01, 2008 at 01:27:17PM -0700, Matthew Dempsky wrote:
> +	if (cf->cf_unit > 1)
> +		return 0;
> 
> Shouldn't this be >= 1, or would you ever want himem1?

indoubitably!
should've been >0 but those digits are so close it must've
been a bit flip of sorts!
although perhaps not a problem w/ the diff supplied (;
anyway.
10x!
cu
--

-- 
    paranoic mickey       (my employers have changed but, the name has remained)
Permalink | Reply | 0 comments |
headers
mickey | 1 Apr 2008 23:31
Gravatar

Re: HIMEM.SYS strikes back!

On Tue, Apr 01, 2008 at 02:41:20PM -0400, Christopher Linn wrote:
> re.
> 
> On Tue, Apr 01, 2008 at 06:00:38PM +0000, mickey wrote:
> > re
> > 80s are back and so is HIMEM.SYS now for the modern 32bit
> > architecture known as i386 or "that intel crap."
> > 
> > yes as you can have guessed already it gives your >4G
> > memory another chance at life reincarnated as scsi disk.
> > 
> > 	http://mickey.lucifier.net/himem.sys
> > 
> > you can read it write it and damn swap on it if you want!
> > credit shall be given to mpf <at>  for a productive discussion
> > where this most atypical concept had been born few months
> > ago and later became stations creation in about 30h (and
> > a bottle of port) and later debugged/tested in a few more
> > other more sober hours...
> 
> you gotta be freekin kidding me.  just ONE bottle of port!?!

i had no more. if you feel inclined supply decent liquids
for more of high quality amusement software for the masses!
cu

--

-- 
    paranoic mickey       (my employers have changed but, the name has remained)
(Continue reading)

Permalink | Reply | 0 comments |
headers
mickey | 1 Apr 2008 23:37
Gravatar

Re: HIMEM.SYS strikes back!

On Tue, Apr 01, 2008 at 01:18:36PM -0500, Marco Peereboom wrote:
> I suggest you add read/write/read capacity(16) for those extra super
> large memory machines.

i do understand what you mean but perhaps you can
elaborate with a diff?
cu

> On Tue, Apr 01, 2008 at 06:00:38PM +0000, mickey wrote:
> > re
> > 80s are back and so is HIMEM.SYS now for the modern 32bit
> > architecture known as i386 or "that intel crap."
> > 
> > yes as you can have guessed already it gives your >4G
> > memory another chance at life reincarnated as scsi disk.
> > 
> > 	http://mickey.lucifier.net/himem.sys
> > 
> > you can read it write it and damn swap on it if you want!
> > credit shall be given to mpf <at>  for a productive discussion
> > where this most atypical concept had been born few months
> > ago and later became stations creation in about 30h (and
> > a bottle of port) and later debugged/tested in a few more
> > other more sober hours...
> > 
> > few caveats:
> > - man page coming soon!
> > - booting is not supported (yet);
> > - one needs to disklabel+newfs(or dd(2) some image) it before use;
> > - it will get some performance improvement;
(Continue reading)

Permalink | Reply | 1 comment |
headers
Eichert, Diana | 2 Apr 2008 00:07
Picon

Re: HIMEM.SYS strikes back!

sigh

lb(4)

NAME
        lb - Lightning Bug transport

SYNOPSIS
        lb* at scsibus?

DESCRIPTION
        The lb driver provides support for extending SCSI buses across
        open space using a mix of optical and mechanical transport me-
        thods.  The primary transport is visible light pulses emana-
        ting from the posterior of the Lightning Bug, with a fallback
        mechanical direct delivery system via wing motion.

> -----Original Message-----
> From: mickey [mailto:mickey <at> lucifier.net]
> Sent: Tuesday, April 01, 2008 3:30 PM
> To: Eichert, Diana
> Cc: 'tech <at> openbsd.org'
> Subject: Re: HIMEM.SYS strikes back!
>
> On Tue, Apr 01, 2008 at 12:31:09PM -0600, Eichert, Diana wrote:
> > does it work with the lb(4) devices?
>
> should you dare to read the mail it would become obvious that
> it exposes itself as sd(4)... in all my ignorance i was not
> aware of this lb(4) innovation in openbsd.
(Continue reading)

Permalink | Reply | 1 comment |

Gmane