That's an oversight by Linux programmers. Linux uses an old BIOS call
that can only report up to 64M, but a new BIOS call has existed for some
time now that can return up to 4G of memory, I believe.
The new development kernels use this new call, but current kernels and
old ones (like 1.2.13 in Slackware 3.0) use the old call.
The solution is to add this to your /etc/lilo.conf:
append="mem=128M"
Then, re-run LILO to write the changes to disk.
Please recommend high-end PCI motherboards which can
support Linux Slackware 3.0 with more than 128 Mbytes Memory.
My current PCI motherboard (Intel Atlantis) has 128 Mbytes memory,
but Linux can only recongize 64M.
Thanks in advance!
------
Chih-chang Lin
San Jose
Just specify to LILO how much memoty you have
i think its just
lilo: <image name> mem=128
something like that, see the lilo man page
Its not your motherboard im sure...
-- Cliff Wakefield
**********************************************************************
* Masters Student(Info Tech) * http://zulu.cs.rmit.edu.au/~cliff/ *
* WeBTESt Admin/Programmer * http://zulu.cs.rmit.edu.au/webtest *
* Distributed Lab Admin(Zulu)* Redhat Linux v4.2, Kernel 2.0.30 *
* CS109 Tutor, CS280 Staff * "Reboots are for upgrades!" *
**********************************************************************
* Office 251.2.33 Phone 9407-6131 Mobile 0411110687 *
**********************************************************************
: i think its just
: lilo: <image name> mem=128
Hi cliff !! I actually have this much memory now ;-)
the command is:
mem=131072
--
-- mark heath - Netspace Online Systems. http://www.netspace.net.au/
:wq
>Please recommend high-end PCI motherboards which can
>support Linux Slackware 3.0 with more than 128 Mbytes Memory.
>My current PCI motherboard (Intel Atlantis) has 128 Mbytes memory,
>but Linux can only recongize 64M.
In /etc/lilo.conf:
append="mem=128M"
If you want to put up to 384MB on a motherboard, try ASUS P/I-P55T2P4S.
Nick.
--
Kralizec / Zeta Microcomputer Software Fax: +61-2-9233-6545 Voice: 9837-1397
G.P.O. Box 3400, Sydney NSW 1043 http://www.zeta.org.au/
: In /etc/lilo.conf:
: append="mem=128M"
If you want to put up to 512MB go with the Tyan Tomcat 3 or 4....
if you want to put up to 1024MB go with the Tyan Titan AT or ATX for Pent Pro
Todd Anderson
The Asus P55T2P4 will run 512Mb, The P65UP5 will run 1Gb.
--
Donovan Ready,
Lindsay Computer Systems
http://www.jumpnet.com/~lcs
: Please recommend high-end PCI motherboards which can
: support Linux Slackware 3.0 with more than 128 Mbytes Memory.
: My current PCI motherboard (Intel Atlantis) has 128 Mbytes memory,
: but Linux can only recongize 64M.
I am running a Tyan Tomcat which supports up to 512M on board
The problem is with lilo not linux. Linux gets several bits of
information
from lilo when booting hd specs, ram, etc.
Does DOS see >64MB I think so.
--
Anti spam email address == ro...@127.0.0.1
Email: q9722124~mail/connect/usq/edu/au
In comp.os.linux Michael James <ro...@127.0.0.1> wrote:
:Josh Yelon wrote:
:> Does MS-DOS automatically recognize more than 64meg of ram? If so,
:> how? And could linux do the same thing?
:The problem is with lilo not linux. Linux gets several bits of
:information
:from lilo when booting hd specs, ram, etc.
:Does DOS see >64MB I think so.
ms-dos has a hard time with more than 640K! ;-)
the problem is that the BIOS doesn't see more than 64M.
just tack on that `append "mem=128M"' or whatever it is for your
system. while autodetection would be preferable, this workaround
isn't so bad. after all, how often do you change the amount of ram
you have?
--
Johan Kullstam [joh...@idt.net]
: the problem is that the BIOS doesn't see more than 64M.
well, not quite: the bios call that linux uses to query memory size
simply cannot express >64M. later kernels improve on this.
regards, mark hahn.
--
operator may differ from spokesperson. ha...@neurocog.lrdc.pitt.edu
http://neurocog.lrdc.pitt.edu/~hahn/
> Dear All,
>
> Please recommend high-end PCI motherboards which can
> support Linux Slackware 3.0 with more than 128 Mbytes Memory.
> My current PCI motherboard (Intel Atlantis) has 128 Mbytes memory,
> but Linux can only recongize 64M.
>
> Thanks in advance!
> ------
> Chih-chang Lin
> San Jose
If you are using lilo to boot, edit your lilo.conf, and add the
following line in the global section:
append="mem=128M"
Then reinstall lilo.
May be this will solve your problem. (To check , call top)
Good luck
: : the problem is that the BIOS doesn't see more than 64M.
: well, not quite: the bios call that linux uses to query memory size
: simply cannot express >64M. later kernels improve on this.
This is the problem with the old BIOS calls, could only pass max 64meg.
Of course new ones had to written when this limit was broken. Older
linux kernels still use the old bios calls.
FreeBSD has the same problem and the amount of memory has to be compiled
into the kernel. The FreeBSD manual says If the memory detection
routines used the new bios calls the freebsd loader wouldn't fit in the
boot block.
--
-- mark heath - Netspace Online Systems. http://www.netspace.net.au/
Obnoxious Usenet Habits #10. Reacquaint the readers of aus.jokes with the
"why did the chicken cross the road" joke.
:wq
Todd Anderson (tan...@news.ori.net) wrote:
: if you want to put up to 1024MB go with the Tyan Titan AT or ATX for Pent Pro
Im just about to purchase a tyan 1662 (dual PPro) it has 8 simms sockets.
Im not aware that 128M 72pin simms exist. Yet Tyan claims 1024M... 128M
simms must exist? correct?
--
-- mark heath - Netspace Online Systems. http://www.netspace.net.au/
Obnoxious Usenet Habits #9. Ask readers of rec.music.misc to post their
favorite Zeppelin tune "for a poll".
:wq
On Tue, 16 Sep 1997 21:28:41 -0700, ilya@home <il...@tamri.com> wrote:
>Chih-chang Lin wrote:
>
>If you are using lilo to boot, edit your lilo.conf, and add the
>following line in the global section:
>append="mem=128M"
>Then reinstall lilo.
>May be this will solve your problem. (To check , call top)
On the other hand, after you do this your system could fail to boot and
you could cry. I should know. I did what you said above and that's what
happened to me. So, the above just could be some rather 'incomplete' advice
(to put it politely).
The exact number one should put in lilo.conf is the number one's bios reports
at bootup. In my case, this is 130688k. So, the line append="mem=130688k"
works for me. append="mem=128M" caused lilo to report that the kernel image
was corrupt and would not decompress, followed by a HALT. It did that for
my backup kernel too and for all kernels. I was truely screwed. I drove
across town and got a standalone boot floppy from my brother, booted up
with it, and removed the append line from my lilo.conf. Ah, that was better.
- Dan
> Im not aware that 128M 72pin simms exist. Yet Tyan claims 1024M... 128M
> simms must exist? correct?
I've seen them advertised in the classified section at the back
of Australian Personal Computer. Price per bit is a lot more
than for 32M simms, but that might change in a year or two ...
Bradley Broom
This topic has been discussed at length in most Linux news
groups.
Please read the BootPrompt-HOWTO where in section 3.3 (html
version), to quote:
NOTE NOTE NOTE: some machines might use the top of memory for BIOS
cacheing or whatever, so you might not
actually have up to the full 96MB addressable. The reverse is also true:
some chipsets will map the physical memory
that is covered by the BIOS area into the area just past the top of
memory, so the top-of-mem might actually be 96MB
+ 384kB for example. If you tell linux that it has more memory than it
actually does have, bad things will happen: maybe
not at once, but surely eventually.''
End of quote.
That is, Dan is correct to raise the note of caution.
We are all guilty at times of "when all else fails read the
instruction manual" -- or in this case the HOWTO.
There are many different configurations of Personel Computers
and it is difficult to cover all cases. That is why we need
these newsgroups. But when it is covered in the available
documentation...
By the way that documentation is available from the Linux
Documentation Project (LDP) site is at:
http://sunsite.unc.edu/LDP/
with mirror sites listed at:
http://sunsite.unc.edu/LDP/hmirrors.html
From the LDP page there is:
Installation and Getting Started Guide, version 2.3 by
Matt Welsh, is available in HTML (online), HTML (tared and
gziped), plain text, DVI, PostScript and PDF.
Note: version 2.2.2 and 2.3 are identical except for the
copyright.
The Linux HOWTO documents are at:
http://sunsite.unc.edu/LDP/HOWTO/HOWTO-INDEX-3.html
Just replace the sunsite.unc.edu with your mirror site. For example:
http://www.localnet.com.au/LDP/HOWTO/HOWTO-INDEX-3.html
Followup set to: comp.os.linux.setup
--
Ian Cowell
ico...@mpce.mq.edu.au
----------------------
Nope. Look in arch/i386/boot/setup.S; the kernel does an
int15,ah=088h call to get memory size, and that call returns the
memory size in kbytes in a 16-bit register, thus no more than 64mb.
2.1.x (x>23) supports a better memory detection scheme that uses a
modern bios call that works on most of the hardware out there and
returns the correct amount of memory (except for Compaq EISA
machines, where it LIES through its teeth) and you can get patches to
put this memory detection scheme into earlier Linuxes from (formerly
linuxhq :-() and from http://www.pell.chi.il.us/~orc/memory.html
____
david parsons \bi/ lots of machines with >64mb memory.
\/
>FreeBSD has the same problem and the amount of memory has to be compiled
>into the kernel. The FreeBSD manual says If the memory detection
>routines used the new bios calls the freebsd loader wouldn't fit in the
>boot block.
Which is somewhat odd, because it looks like with a little bit of
tightening up of the various prompts it gives, you could pretty easily
fit the additional code in (I count about 30 bytes additional in this
patch to /usr/src/sys/i386/boot/biosboot; you could get that back by
pulling the `use wd(...' line from the boot prompt; it doesn't seem
like that much of a sacrifice to have for getting memory detection.
I don't run FreeBSD, alas (my last freeBSD machine snuffed it when
it complained bitterly about being a test machine on the inside of
a test WebShield box; it is not Yet Another Orcware Linux machine),
but I did do a memory patch for it, which some bold person might
want to patch into their system and see what happens:
*** bios.S~ Thu Sep 18 23:09:59 1997
--- bios.S Thu Sep 18 23:15:26 1997
***************
*** 308,315 ****
push %ebp
mov %esp, %ebp
push %ebx
! mov 8(%ebp), %ebx
call EXT(prot_to_real) /* enter real mode */
--- 308,317 ----
push %ebp
mov %esp, %ebp
push %ebx
+ push %ecx
+ push %edx
! mov 16(%ebp), %ebx
call EXT(prot_to_real) /* enter real mode */
***************
*** 323,340 ****
data32
jmp xdone
! xext: movb $0x88, %ah
sti
! int $0x15
cli
xdone:
mov %eax, %ebx
!
data32
call EXT(real_to_prot)
mov %ebx, %eax
pop %ebx
pop %ebp
ret
--- 325,359 ----
data32
jmp xdone
! xext:
!
! mov $0xe801, %ax
sti
! int $0x15
cli
+ jc xoldstylemem
+
+ and $0xffff, %ebx ! clear sign extend
+ shl $6, %ebx ! and go from 64k to 1k chunks
+
+ and $0xffff, %eax ! clear sign extend
+ add %eax, %ebx
+ jmp xfinis
+ xoldstylemem:
+ movb $0x88, %ah
+ sti
+ int $0x15
+ cli
xdone:
mov %eax, %ebx
! xfinis:
data32
call EXT(real_to_prot)
mov %ebx, %eax
+ pop %edx
+ pop %ecx
pop %ebx
pop %ebp
ret
____
david parsons \bi/ clean and simple horrific kludges 'R us.
\/
I previously had a Gigabyte motherboard with 128M of RAM. Linux would
only use 64M unless I used the LILO trick. Then I upgraded to a Tycan
Tomcat III dual CPU motherboard. Image my surprise that Linux now
managed to detect the 128M all on it's own with any LILO help. This
was with Kernel 2.0.27. Possibly the "new" BIOS call was broken on
the Gigabyte?
> append="mem=128M"
Tip 'o the day: While searching for this I also found append="panic=30"
which is handy to use for unattended servers. Causes kernel panics to
auto-reboot after 30 seconds.
Cheers.
-- +------------------------------------------------------------+
. | John Saunders mailto:John.S...@scitec.com.au (Work) |
,--_|\ | mailto:jo...@nlc.net.au (Home) |
/ Oz \ | http://www.nlc.net.au/~john/ |
\_,--\_/ | SCITEC LIMITED Phone +61 2 9428 9563 Fax +61 2 9428 9933 |
v | "By the time you make ends meet, they move the ends." |
+------------------------------------------------------------+
I have a Brother MFC 4550 FAX printer.
It works fine as a printer on window95.
However, I have trouble to install it on Linux.
I can type lpr and lpq without seeing error message.
But nothing comes out from the printer.
Please advise if you have a solution!
Thanks in advance!
------
Chih-chang Lin
San Jose
--
Chih-chang Lin
The Triton HX is the only one that caches more than 64MB that I know of.
At least it's the only Intel chipset that does.
This isn't a problem with PPro and P2 processors, since the L2 cache is
on the CPU itself, independent of the motherboard's control.
The Intel TX and, I believe, VX, chipsets do *NOT* support cacheing more
than 64MB of RAM. The Intel HX chipset and VIA Apollo VP-2 *DO* support
it, though I don't know precisely what their limits are. Also, boards
using the Intel HX and VIA VP-2 may or may not IMPLEMENT this feature; a
specific board can cripple itself in this respect, so it's wise to check
with the board manufacturer or with other users of a specific model
before buying, if this is important to you.
--
--Rod Smith
rods...@linc.cis.upenn.edu
--
I have a Tyan Tomcat board with 128MB and want to have the second 64 cached as
well. All I need to do this is the second tag ram.
Any hints as to where to get this from?????
>mac...@s03-pm09.snwestsac.campus.mci.net wrote:
>
> > What is the story on memory caching above the 64 meg memory
>threshold?
> > I have heard this is supported by some PCI chipsets.
>
>The Triton HX is the only one that caches more than 64MB that I know of.
>At least it's the only Intel chipset that does.
>
>This isn't a problem with PPro and P2 processors, since the L2 cache is
>on the CPU itself, independent of the motherboard's control.
Most of the chipsets from VIA, SIS, and AMD also cache more than 64
Meg. The VIA VP2/97 (which is them same as the AMD 640) is supposed to
be very high in performance, caches all the memory on the mother
board, and supports up to 1 meg of cache ram. Might want to check out
VIA's web site. Have used mother boards with their chipset before, and
the work very well.
You should be able to get this from any competent mobo dealer, it is
simply an SRAM DIP chip, 15ns or faster, often 32kx8. Check with
www.aberdeeninc.com to see what they can do for you.
.B.
> mac...@s03-pm09.snwestsac.campus.mci.net wrote:
>
> > What is the story on memory caching above the 64 meg memory
> threshold?
> > I have heard this is supported by some PCI chipsets.
>
> The Triton HX is the only one that caches more than 64MB that I know of.
> At least it's the only Intel chipset that does.
The upcoming VIA/AMD chipset will also cache more than 64mb (with 2mb L2
cache I think it can take 256mb, maybe more)...I'm pretty sure Micron's
new Samurai chipset will as well....
It's a shame that Asus is dropping their best HX board, the T2P4...I
can't say I like the TX's RAM limitations...
Tom
-----------------------------------------------------------------------------
Recedite, plebes! Gero rem imperialem. Tom Kyle
http://www.inlink.com/~rkyle/tomsweb/tom.html tom...@umsl.edu
-----------------------------------------------------------------------------
--------------3EAA758129BD8DC94B1D05FD
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Hi!
There is a parameter you can pass to lilo or use kernel 2.0.30 eg RH4.2.
I have 96MB of ram and a Tyan TomcatII. works fine.
Grahame
Andrew Fidel wrote:
--
"Do or do not; There is no try"
Yoda
--------------3EAA758129BD8DC94B1D05FD
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit
<HTML>
Hi!
<P>There is a parameter you can pass to lilo or use kernel 2.0.30
eg RH4.2.
<BR>I have 96MB of ram and a Tyan TomcatII. works fine.
<P>Grahame
<P>Andrew Fidel wrote:
<BLOCKQUOTE TYPE=CITE>Graham Philip Crowe wrote:
<BR>>
<BR>> Chih-chang Lin (lin...@violin.ece.ucsb.edu) wrote:
<BR>> : Dear All,
<BR>>
<BR>> : Please recommend high-end PCI motherboards which can
<BR>> : support Linux Slackware 3.0 with more than 128 Mbytes Memory.
<BR>> : My current PCI motherboard (Intel Atlantis) has 128 Mbytes memory,
<BR>> : but Linux can only recongize 64M.
<BR>>
<BR>> I am running a Tyan Tomcat which supports up to 512M on board
<BR>try disabling the memory hole at 64 MB option in your BIOS. some MB's
<BR>support this for OS/2 and it accidentaly get turned on.</BLOCKQUOTE>
<PRE>--
"Do or do not; There is no try"
Yoda</PRE>
</HTML>
--------------3EAA758129BD8DC94B1D05FD--
append="mem=128"
to the /etc/lilo.conf file. (Probably in the section relating to your Linux
partition). If you want, e-mail me, and I'll have a more concerted look for
the info.
Regards,
ROSCO
: Tip 'o the day: While searching for this I also found append="panic=30"
: which is handy to use for unattended servers. Causes kernel panics to
: auto-reboot after 30 seconds.
This is a joke, right?
--
John Lellis (lel...@dmccorp.com)
--
... Our continuing mission: To seek out knowledge of C, to explore
strange UNIX commands, and to boldly code where no one has man page 4.
Nah its a feature.
> I just got Linux up for the first time ever for me, and I put it
> on an Intel dual 200MHZ Pentium pro motherboard with 256 MB of RAM.
>
> What do I type to see if Linux only sees 64MB? (please use full
> detail / explain to clueless newbie mode)
dmesg|grep Memory, or free.
That should be append="mem=128M" otherwise it will try to load the
kernel into 128 bytes!
I did the very same thing myself earlier this week.
Regards
CMS
Holy crap! You've got that much system, and you don't even know if you
like Linux yet?
> What do I type to see if Linux only sees 64MB? (please use full
> detail / explain to clueless newbie mode)
With that system, you're more of a Psycho-newbie on crack :)
"dmesg | more" will show the bootup messages, including a memory count.
"top" will list memory usage. If it adds up to close to 64MB, it's only
seeing 64MB of memory.
"ls -la /proc/kcore" and look at its size. /proc/kcore is a link to main
memory. If it's only 64MB in size, it's only seeing 64MB.
As you can see, there are several ways.
Basically, if you've got more than 64MB of RAM, and you haven't added
the 'append="mem=256M"' line to your /etc/lilo.conf, you are only using
64MB of RAM for Linux.
> Please recommend high-end PCI motherboards which can
> support Linux Slackware 3.0 with more than 128 Mbytes Memory.
> My current PCI motherboard (Intel Atlantis) has 128 Mbytes memory,
> but Linux can only recongize 64M.
I just got Linux up for the first time ever for me, and I put it
on an Intel dual 200MHZ Pentium pro motherboard with 256 MB of RAM.
What do I type to see if Linux only sees 64MB? (please use full
> That should be append="mem=128M" otherwise it will try to load the
> kernel into 128 bytes!
Be careful though that your BIOS doesn't snip off some memory on
top. You might have to specify 128MB - 384KB as xxxK.
FYI see the LINUX Bootprompt HOWTO with an excellent comment
by Linus on this.
regards
--
Holger J. Lenz <holge...@franken1.de> Munich, Germany
voice: +49-89-6379838
lenz@hal: rm -rf /mnt/windows
Check out Tyan. I have one and it will support 1 gig of ram.
Alas, the Linux bootup message reported only 64-Mb.
(grep Memory /var/log/syslog)
The solution was to explicitly tell Linux how much ram was present
using the following entry in /etc/lilo.conf:
append = "mem=127M"
Bootup now reports "126796k/130048k available", which I don't
entirely understand, but it works.
Regards,
Lew Wolfgang
Easy: 127*1024=130048
So, you are using 127Mb of memory. Actually, you should specify
mem=128 instead, to use the full 128Mb. Otherwise, you are wasting a
whole MB of ram.
-My opinions only
This seems an astoundingly perverse way of doing it. The only reason you
need root access is in order to write the output into a place where it's
silly to put it anyway. There's a reason why some directories are
writable only to root. Here's a better way:
$ cat /proc/meminfo
You want to be able to keep it? OK then:
$ cat /proc/meminfo >/tmp/memdata.txt
John
--
John Winters. Wallingford, Oxon, England.
Want to buy Linux CDs cheaply in the UK? Join the Linux Buyers' Consortium.
See <http://www.polo.demon.co.uk/lbc.html>
> Lewis E. Wolfgang (wolf...@cx48762-a.cv1.sdca.home.com) wrote:
> : Bootup now reports "126796k/130048k available", which I don't
> : entirely understand, but it works.
>
> Easy: 127*1024=130048
I did'nt understand the "126796k" portion of the message.
> So, you are using 127Mb of memory. Actually, you should specify
> mem=128 instead, to use the full 128Mb. Otherwise, you are wasting a
> whole MB of ram.
I thought about this, but remain confused about the "baroque" Wintel
architecture. The ROM-BIOS claims to move portions of itself into
RAM (shadowing?) for performance reasons. I was under the impression
that other stuff was reserved between 640-Kb and 1-Mb too. Do
these reservations go away when the x86 is running in protected mode?
Does Linux use the ROM-BIOS? Does it care if it doesn't get mapped?
Not knowing about these matters, I felt that specifying 127-MB of RAM
was the safer thing to do.
Regards,
Lew Wolfgang
Three questions:
Just out of curiousity, how many of those 128M swap partitions can I have?
Is there a limit to how much RAM I can have?
Why do I read about a 16MB swap space limit? Did this change?
I am using Slackware Linux, latest version.
I have a HP laster Jet 5SE.
Please suggest how to setup /etc/printcap for linux.
I add the following line in /etc/printcap.
lp|lp1:lp=/dev/lp1:sd=/usr/spool/lp1:sh
For regular text, it printed the text like a stairway.
For postscript file, it printed nothing but a lot of empty pages
Thanks!
Andy Lin
San Jose
I configure the Linux to have a 500M swap partition.
However, if I type "free" command, I see only 128M is used as swap
space.
Please suggest fix for this .
Thanks!
Andy Lin
San Jose
Swap partitions can only be 128MB each in Linux.
You'll have to split it into several partitions <= 128MB.
//Marcus
--
-------------------------------+-------------------------------------
Marcus Sundberg | WWW: http://www.e.kth.se/~e94_msu/
Royal Institute of Technology | E-Mail:
Stockholm/Sweden | e94_msu-nospam-@-nospam-e.kth.se
> Is there a limit to how much RAM I can have?
8 I believe, though there has been patches floating around the kernel list to
bump it up to 16.
> Why do I read about a 16MB swap space limit? Did this change?
A long time ago (I believe 0.99, but I wasn't using Linux back then). Its
amazing how these old FAQs seem to hang around forever.
--
Michael Meissner, Cygnus Solutions (East Coast)
4th floor, 955 Massachusetts Avenue, Cambridge, MA 02139, USA
meis...@cygnus.com, 617-354-5416 (office), 617-354-7161 (fax)
IIRC, a single swap partition is limited to 128MEg. Try setting up as
four partitions and it should work.
--
Stephen Carville
scar...@earthlink.net
=========================================================
Civilization, as we know it today, owes it's existence to the
engineers. These are the men who, down the long centuries, have learned
to exploit the properties of matter and the sources of power for the
benefit of mankind.
L. Sprague DeCamp
=========================================================