Dear All, can anyone help/put me straight on this. I have a i486 machine running Slackware 8.1. As a christmas present I bought it a 32MB Simm, it currently has 16MB installed (one module). When I replaced the 16MB with the 32MB the bios still only detects 16MB on boot and everything runs as before. No matter I thought - I have read somewhere that Linux will override such settings, but I have been unable to get the kernel (2.4.18) to do so. I read the lilo.conf man page which indicated that adding 'append="mem=32M"' to lilo would do it, but the computer would not boot after adding this. A further search of the internet produced a sort of How-to doc that lists three options (but I think is written for RedHat). The first of these is the same as the lilo.conf recommendation, the second involves giving mem sizes and start locations in hexadecimals e.g. append="mem=exactmap mem=0x9f000@0 mem=0x1f00000@100000" and the third is like the 2nd, but in english.. append="mem=exactmap mem#=640K@0 mem=31M@1M" all these end in the same result - failure to boot. I have tried these messages specifying 16MB as the memory size (incase there really was a problem with the 32MB size) but the result is the same. In the mini how-to it is suggested that white space is needed between the start quotes and the first mem statement - but that doesn't resolve this problem either. So - am I correct in assuming that the Linux kernel can use this additional 16MB of memory , and if so how do I let it know its there?
all the best
Glen
On Sun, Dec 29, 2002 at 02:05:19PM -0000, Glen Tyler wrote:
I have a i486 machine running Slackware 8.1. As a christmas present I bought it a 32MB Simm, it currently has 16MB installed (one module). When I replaced the 16MB with the 32MB the bios still only detects 16MB on boot and everything runs as before. No matter I thought - I have read somewhere that Linux will override such settings, but I have been unable to get the kernel (2.4.18) to do so.
Is there not a bios option to redetect memory? and are you sure the Simm is really a 32MB simm? also does the motherboard of your machine recognise more than 16MB simms etc? how many simm sockets does your machine have? must you install large simms in pairs? is the memory speed, parity etc. etc. correct for this motherboard?
They are all the things that first come to mind with this, you could also perhaps try booting from alternative boot media with the memory options to see what happens to rule out it being something to do with the kernel on the machine.
Adam PS you havn't got the 2 modules mixed up have you and inadvertantly put the 16MB simm back in instead of the new one? ;)
----- Original Message ----- From: "Adam Bower" Sent: Sunday, December 29, 2002 2:49 PM Subject: Re: [Alug] over-riding BIOS memory detection
Is there not a bios option to redetect memory? and are you sure the Simm
is
really a 32MB simm? also does the motherboard of your machine recognise
more
than 16MB simms etc? how many simm sockets does your machine have? must
you
install large simms in pairs? is the memory speed, parity etc. etc.
correct for
this motherboard?
Not sure if the motherboard can recognise more than 16MB (a clue is that the boot sequence registers 016000KB, I guess that the preceeding 0 means that it might recognise more) - but what I wanted to do was bypass the motherboard BIOS. The 16MB was EDO 60NS Parity - same as the 32MB (I've tried the 32 in other machines where it seems OK). The machine (an Amstrad PC9486) has one 72pin socket, and 4 x 30 pin sockets - these can't be used together, and so 72pin SIMMs can't be installed in pairs.
They are all the things that first come to mind with this, you could also perhaps try booting from alternative boot media with the memory options to
see
what happens to rule out it being something to do with the kernel on the machine.
I've tried with TOMSRTBT, but I'll try passing some of the options at boot time.
PS you havn't got the 2 modules mixed up have you and inadvertantly put
the
16MB simm back in instead of the new one? ;)
No I'm pretty sure I've got the correct one, and anyway when I try the append option to address 16MB the machine still fails to boot. I'm sure that there is something wrong with the way I'm editing the lilo.conf script.
Glen
If the BIOS on startup isn't seeing more than 016000b of memory, then you have a BIOS problem, not a kernel problem.
Try flashing the BIOS if it supports it
Adding the mem option AFAIK doesn't *bypass* the BIOS, it simply tells linux what memory you have if linux reads the BIOS wrong.
To my mind, you have a hardware problem, not a software problem (unless a BIOS flash can fix it).
TD
On Sun, 2002-12-29 at 18:49, Glen Tyler wrote:
----- Original Message ----- From: "Adam Bower" Sent: Sunday, December 29, 2002 2:49 PM Subject: Re: [Alug] over-riding BIOS memory detection
Is there not a bios option to redetect memory? and are you sure the Simm
is
really a 32MB simm? also does the motherboard of your machine recognise
more
than 16MB simms etc? how many simm sockets does your machine have? must
you
install large simms in pairs? is the memory speed, parity etc. etc.
correct for
this motherboard?
Not sure if the motherboard can recognise more than 16MB (a clue is that the boot sequence registers 016000KB, I guess that the preceeding 0 means that it might recognise more) - but what I wanted to do was bypass the motherboard BIOS. The 16MB was EDO 60NS Parity - same as the 32MB (I've tried the 32 in other machines where it seems OK). The machine (an Amstrad PC9486) has one 72pin socket, and 4 x 30 pin sockets - these can't be used together, and so 72pin SIMMs can't be installed in pairs.
They are all the things that first come to mind with this, you could also perhaps try booting from alternative boot media with the memory options to
see
what happens to rule out it being something to do with the kernel on the machine.
I've tried with TOMSRTBT, but I'll try passing some of the options at boot time.
PS you havn't got the 2 modules mixed up have you and inadvertantly put
the
16MB simm back in instead of the new one? ;)
No I'm pretty sure I've got the correct one, and anyway when I try the append option to address 16MB the machine still fails to boot. I'm sure that there is something wrong with the way I'm editing the lilo.conf script.
Glen
main@lists.alug.org.uk http://www.alug.org.uk/ http://lists.alug.org.uk/mailman/listinfo/main Unsubscribe? See message headers or the web site above!
To my mind, you have a hardware problem, not a software problem (unless a BIOS flash can fix it).
This is a long shot but there could be a jumper that needs setting. Check the m/b and/or search for a manual online. I once had a 386 that needed a jumper setting to tell the m/b there was more than 4megs fitted.
Syd
Glen Tyler glen.tyler@tesco.net wrote:
can anyone help/put me straight on this.
Others have made very good comments. I'll throw two more suggestions in:
1. Some PCs "steal" from main memory to shadow ROM or provide some devices dedicated memory to work in. Try starting with mem=20M and working upwards until it dies. You can type these on the LILO "boot:" prompt after the kernel name (often "Linux") as long as there's no line in lilo.conf yet. When you find the right one, add it to lilo.conf and rerun lilo.
2. Maybe memtest86's floppy disk image will detect how much is really there and make sure that the memory is working properly. Search the web for it.