Skip to contentSkip to conversation

kernel (64bit) 4GB memory support

DISCUSSION Rudolf Usselmann 11 participants 39 messages Dec 11, 2004 – May 12, 2005 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

Could anybody tell me which of the previous (non 2.6.9) kernels do support 4GB of main memory in 64 bit mode ?

I'd appreciate a cc, as I am not a regular list reader.

Thanks !

rudi

#1 lore raw
Quoted 2 lines · Rudolf Usselmann

Could anybody tell me which of the previous (non 2.6.9) kernels do support 4GB of main memory in 64 bit mode ?

64bit kernels have supported >4GB since their ports inception, AFAIK.

Your platform could limit this artificially, of course.

Jeff

#2 lore raw
Quoted 9 lines · Jeff Garzik
Quoted 2 lines · Rudolf Usselmann

Could anybody tell me which of the previous (non 2.6.9) kernels do support 4GB of main memory in 64 bit mode ?

64bit kernels have supported >4GB since their ports inception, AFAIK.

Your platform could limit this artificially, of course.

Jeff

Yes, but it is currently broken (kernel panics). I wonder if anybody knew which kernel does work for 64 bit and >4GB of memory. I'm sure there must be people out there with similar configurations to mine ....

Thanks,
rudi
(Tiger K8W, dual Opteron)

#3 lore raw
Quoted 6 lines · Rudolf Usselmann

Yes, but it is currently broken (kernel panics). I wonder if anybody knew which kernel does work for 64 bit and >4GB of memory. I'm sure there must be people out there with similar configurations to mine ....

There are, but most I've had discussions with on LKML and x86-4 ML are running recent 2.6 series kernels (successfully).

Best advice is

1) Upgrade your bios to latest. Bios problems seem to account for most >4Gb mem problems on x86_64. Pay careful attention to bios memory settings also.

2) Report problems using lastest 2.4 kernel here for help

If you can use 2.6 kernel, you're likely to get more help. 2.4 is in maintenance mode. Most people here will be running/developing 2.6

Andrew Walrond

#39 lore raw
In reply to Rudolf Usselmann
Quoted 3 lines · Rudolf Usselmann

Could anybody tell me which of the previous (non 2.6.9) kernels do support 4GB of main memory in 64 bit mode ?

All of 2.6 I would think.

Signature · 2 lines
http://www.PowerDNS.com      Open source, database driven DNS Software 
http://lartc.org           Linux Advanced Routing & Traffic Control HOWTO
#4 lore raw
In reply to Rudolf Usselmann
Quoted 7 lines · Rudolf Usselmann
Quoted 5 lines · Jeff Garzik
Quoted 2 lines · Rudolf Usselmann

Could anybody tell me which of the previous (non 2.6.9) kernels do support 4GB of main memory in 64 bit mode ?

Quoted 8 lines
Quoted 3 lines

64bit kernels have supported >4GB since their ports inception, AFAIK.

Your platform could limit this artificially, of course.

Yes, but it is currently broken (kernel panics). I wonder if anybody knew which kernel does work for 64 bit and >4GB of memory. I'm sure there must be people out there with similar configurations to mine ....

All 2.6 kernels work with 64bit and >4GB memory, on my configurations (x86-64, ia64, and alpha).

It is a mistake to assume that all 64bit and/or >4GB is broken.

Quoted 1 line

(Tiger K8W, dual Opteron)

Ok, we finally get a bit of useful information.

Are you CERTAIN that you are booting a 64bit kernel? Is your memory PC1600, PC2100, or PC2700?
Is your memory installed in matched pairs? Is all your memory ECC registered?
Is your BIOS at the latest version?

Once we get through the hardware issues, now it is time to do a binary search of 2.6 kernels, to see which one works for you. If no 2.6 kernels work for you, then give 2.4 kernels a try.

Jeff

#5 lore raw

On Sun, 2004-12-12 at 00:47, Jeff Garzik wrote:
...

Quoted 7 lines

All 2.6 kernels work with 64bit and >4GB memory, on my configurations (x86-64, ia64, and alpha).

It is a mistake to assume that all 64bit and/or >4GB is broken.

Quoted 1 line

(Tiger K8W, dual Opteron)

Quoted 7 lines

Ok, we finally get a bit of useful information.

Are you CERTAIN that you are booting a 64bit kernel? Is your memory PC1600, PC2100, or PC2700?
Is your memory installed in matched pairs? Is all your memory ECC registered?
Is your BIOS at the latest version?

Hi Jeff,

"yes" to all of the above. I am 100% certain this is not a hardware problem. I have paired simms and with a 32 bit kernel can use the entire 4GB. I also have run memtest86 ...

Quoted 5 lines

Once we get through the hardware issues, now it is time to do a binary search of 2.6 kernels, to see which one works for you. If no 2.6 kernels work for you, then give 2.4 kernels a try.

Jeff

I have previously reported this bug to the list (about a week ago). I am greatfull for every response and am willing to investigate everything.

Previously I was running Fedora Core 2 32BIT with the 2.6.9 kernel and never had problems with 4GB. After finally upgrading to 64 bit I can't use 4GB memory anymore.

# ver_linux

Linux cpu10 2.6.9RU1.1 #11 SMP Sun Dec 5 11:42:18 ICT 2004 x86_64 x86_64 x86_64 GNU/Linux

Gnu C                  3.4.2
Gnu make               3.80
binutils               2.15.92.0.2
util-linux             2.12a
mount                  2.12a
module-init-tools      3.1-pre5
e2fsprogs              1.35
reiserfsprogs          line
reiser4progs           line
pcmcia-cs              3.2.7
quota-tools            3.12.
PPP                    2.4.2
nfs-utils              1.0.6
Linux C Library        2.3.3
Dynamic linker (ldd)   2.3.3
Procps                 3.2.3
Net-tools              1.60
Kbd                    1.12
Sh-utils               5.2.1
Modules Loaded         autofs4 nfs lockd sunrpc binfmt_misc dm_mod button battery ac nvidia ipv6 ohci_hcd uhci_hcd ehci_hcd hw_random snd_intel8x0 snd_ac97_codec snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd_page_alloc gameport snd_mpu401_uart snd_rawmidi snd_seq_device snd soundcore floppy

The attached boot log shows the kernel panic ....

Kind Regards,
rudi

#6 lore raw
Quoted 1 line · Rudolf Usselmann

The attached boot log shows the kernel panic ....

Of course you forgot the boot log.

#7 lore raw
Quoted 5 lines · Zwane Mwaikambo
Quoted 1 line · Rudolf Usselmann

The attached boot log shows the kernel panic ....

Of course you forgot the boot log.

Oops, Sorry !

rudi

#8 lore raw
Quoted 8 lines · Rudolf Usselmann
Quoted 5 lines · Zwane Mwaikambo
Quoted 1 line · Rudolf Usselmann

The attached boot log shows the kernel panic ....

Of course you forgot the boot log.

Oops, Sorry !

Please reproduce without the nvidia driver ever loaded.

Thanks,
Zwane

#9 lore raw
Quoted 7 lines · Zwane Mwaikambo
Quoted 5 lines · Rudolf Usselmann
Quoted 4 lines · Zwane Mwaikambo
Quoted 1 line · Rudolf Usselmann

The attached boot log shows the kernel panic ....

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Of course you forgot the boot log.

Oops, Sorry !

Please reproduce without the nvidia driver ever loaded.

Thanks,
Zwane

ok, un-installed nvidia drivers ...

boot3.log shows a kernel panic as it is starting to run fsck without me involving any user commands.

b004.log - I tried again (after doing a fsck with only 2gb) and again it did not make it to a login prompt.

Thanks,
rudi

#10 lore raw
Quoted 7 lines · Rudolf Usselmann

ok, un-installed nvidia drivers ...

boot3.log shows a kernel panic as it is starting to run fsck without me involving any user commands.

b004.log - I tried again (after doing a fsck with only 2gb) and again it did not make it to a login prompt.

Thanks for reproducing that, is there any possible chance you could try 2.6.10-rc2-mm4?

#11 lore raw
Quoted 4 lines · Zwane Mwaikambo
Quoted 2 lines · Rudolf Usselmann

ok, un-installed nvidia drivers ...

Quoted 8 lines
Quoted 5 lines

boot3.log shows a kernel panic as it is starting to run fsck without me involving any user commands.

b004.log - I tried again (after doing a fsck with only 2gb) and again it did not make it to a login prompt.

Thanks for reproducing that, is there any possible chance you could try 2.6.10-rc2-mm4?

Actually I tried "2.6.10-rc3" a few days ago, and had the same results. Do you still want me to try "2.6.10-rc2-mm4" ?

Thanks !
rudi

#12 lore raw
Quoted 5 lines · Rudolf Usselmann
Quoted 2 lines

Thanks for reproducing that, is there any possible chance you could try 2.6.10-rc2-mm4?

Actually I tried "2.6.10-rc3" a few days ago, and had the same results. Do you still want me to try "2.6.10-rc2-mm4" ?

No, just the first oops output and your kernel config.

Thanks,
Zwane

#14 lore raw
In reply to a message not in the archive
Quoted 9 lines · Frank Denis (Jedi/Sector One)
Quoted 2 lines · Rudolf Usselmann

Actually I tried "2.6.10-rc3" a few days ago, and had the same results. Do you still want me to try "2.6.10-rc2-mm4" ?

Yes, try 2.6.10-rc2-mm4 and apply add this patch:

ftp://ftp.c9x.org/linux-kernel/2.6.10-rc2/2.6.10-rc2-mm4/2.6.10-rc2-mm4-jedi4.patch.bz2

Does it work?

Hi Frank,

no unfortunately it did not work.

The "panics" are different, the kernel seems to be able to recover from panics ...

Attached is an archive with ver_linux output, my config file, and a log file. You might find this to be more useful this time, as I was able to do several things to make it "crash" at one point I did a remote login in to this computer and that caused a crash as well. Definitely a lot more crashes I was able to capture ....

Kind Regards,
rudi

#15 lore raw
Quoted 18 lines · Rudolf Usselmann
Quoted 9 lines · Frank Denis (Jedi/Sector One)
Quoted 2 lines · Rudolf Usselmann

Actually I tried "2.6.10-rc3" a few days ago, and had the same results. Do you still want me to try "2.6.10-rc2-mm4" ?

Yes, try 2.6.10-rc2-mm4 and apply add this patch:

ftp://ftp.c9x.org/linux-kernel/2.6.10-rc2/2.6.10-rc2-mm4/2.6.10-rc2-mm4-jedi4.patch.bz2

Does it work?

Hi Frank,

no unfortunately it did not work.

The "panics" are different, the kernel seems to be able to recover from panics ...

Quoted 6 lines

Attached is an archive with ver_linux output, my config file, and a log file. You might find this to be more useful this time, as I was able to do several things to make it "crash" at one point I did a remote login in to this computer and that caused a crash as well. Definitely a lot more crashes I was able to capture ....

Rudolf, thanks for collecting that information. Since you said that 2.6.10-rc3 also had similar results it counts out a number of suspects i had in the -mm tree. Could you work towards isolating a working kernel version? You could try these;

2.6.8.1
2.6.9
2.6.9-mm1

Thanks,
Zwane

#16 lore raw
Quoted 6 lines · Zwane Mwaikambo

Rudolf, thanks for collecting that information. Since you said that 2.6.10-rc3 also had similar results it counts out a number of suspects i had in the -mm tree. Could you work towards isolating a working kernel version? You could try these;

Quoted 6 lines

2.6.8.1
2.6.9
2.6.9-mm1

Thanks,
Zwane

Zwane,

is there a more scientific approach to this problem ?

See my problem is that I have a very expensive motherboard and very expensive simm modules which I already inserted and removed several dozen times. I would like to avoid doing this experimenting as much as possible. Specially if there might be a more contractive way. We don't know if there ever has been a kernel that has worked with this configuration, right ?

memtest86 seems to detect and test all 4gb just fine. So there must be some way to find the problem ...

I mean you guys must have a pretty good idea what is going wrong. Is it possible to add some debugging information ? The latest mm4-jeda kernel seems to be able to "recover" from panics/crashes. It doesn't "die" it just kills the offending app (at least thats my limited understanding).

Kind Regards,
rudi

#17 lore raw
Quoted 5 lines · Rudolf Usselmann

I mean you guys must have a pretty good idea what is going wrong. Is it possible to add some debugging information ? The latest mm4-jeda kernel seems to be able to "recover" from panics/crashes. It doesn't "die" it just kills the offending app (at least thats my limited understanding).

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

Thanks,
Zwane

#18 lore raw
Quoted 6 lines · Zwane Mwaikambo

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

Thanks,
Zwane

Well, if somebody is working on it, I would be happy to very the fixes and assist in providing additional debugging information. Please don't take my previous email the wrong way - I do want to help any way I can ...

Kind Regards,
rudi

#19 lore raw
Quoted 5 lines · Rudolf Usselmann
Quoted 4 lines · Zwane Mwaikambo

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

Quoted 7 lines
Quoted 2 lines

Thanks,
Zwane

Well, if somebody is working on it, I would be happy to very the fixes and assist in providing additional debugging information. Please don't take my previous email the wrong way - I do want to help any way I can ...

Not to worry, if you'd like to be notified when the bug is resolved or a patch is immediately available you may open a bug on bugzilla.kernel.org, upon completion just send me the bug number and i'll assign myself as the owner.

Thanks again,
Zwane

#35 lore raw
In reply to Zwane Mwaikambo
Quoted 8 lines · Zwane Mwaikambo
Quoted 5 lines · Rudolf Usselmann

I mean you guys must have a pretty good idea what is going wrong. Is it possible to add some debugging information ? The latest mm4-jeda kernel seems to be able to "recover" from panics/crashes. It doesn't "die" it just kills the offending app (at least thats my limited understanding).

Quoted 5 lines

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

Thanks,
Zwane

Hi Zwane !

just wondering if there where any fixes submitted with 2.6.10 final release for the memory problem. I did not see anything from bugzilla ....

Kind Regards,
rudi

#36 lore raw
Quoted 6 lines · Rudolf Usselmann
Quoted 2 lines

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

just wondering if there where any fixes submitted with 2.6.10 final release for the memory problem. I did not see anything from bugzilla ....

Hi Rudolf,
I've replied on bugzilla.

Thanks,
Zwane

#37 lore raw
Quoted 9 lines · Zwane Mwaikambo
Quoted 6 lines · Rudolf Usselmann
Quoted 2 lines

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

just wondering if there where any fixes submitted with 2.6.10 final release for the memory problem. I did not see anything from bugzilla ....

Quoted 5 lines

Hi Rudolf,
I've replied on bugzilla.

Thanks,
Zwane

Hi Zwane,

I tried the new kernel, same results. Still can't use the extra memory.

Attached is the "dmesg" output ...

Kind Regards,
rudi

#38 lore raw
Quoted 2 lines · Rudolf Usselmann

I tried the new kernel, same results. Still can't use the extra memory.

Thanks for testing that Rudolf, i'll find a system to reproduce on.

#20 lore raw
In reply to Zwane Mwaikambo
Quoted 15 lines · Zwane Mwaikambo
Quoted 5 lines · Rudolf Usselmann
Quoted 4 lines · Zwane Mwaikambo

Ok don't worry about trying to isolate it, there should be a fix for it by 2.6.10.

Quoted 7 lines
Quoted 2 lines

Thanks,
Zwane

Well, if somebody is working on it, I would be happy to very the fixes and assist in providing additional debugging information. Please don't take my previous email the wrong way - I do want to help any way I can ...

Quoted 7 lines

Not to worry, if you'd like to be notified when the bug is resolved or a patch is immediately available you may open a bug on bugzilla.kernel.org, upon completion just send me the bug number and i'll assign myself as the owner.

Thanks again,
Zwane

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

Thanks,
rudi
=============================================================
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis

#21 lore raw
Quoted 10 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

Does the following patch (against a 2.6.12rc3 kernel) fix your problems?

-Andi

Don't look up struct page * of physical address in iounmap

it could be in a memory hole not mapped in mem_map

arch/x86_64/mm/ioremap.c +1 -1
Index: linux/arch/x86_64/mm/ioremap.c
===================================================================
@@ -272,7 +272,7 @@ void iounmap(volatile void __iomem *addr
 	if ((p->flags >> 20) &&
 		p->phys_addr + p->size - 1 < virt_to_phys(high_memory)) {
 		/* p->size includes the guard page, but cpa doesn't like that */
-		change_page_attr(virt_to_page(__va(p->phys_addr)),
+		change_page_attr_addr(p->phys_addr,
 				 p->size >> PAGE_SHIFT,
 				 PAGE_KERNEL);
 		global_flush_tlb();
#22 lore raw
In reply to Rudolf Usselmann
Quoted 9 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

How much ram do you see with 4GB installed running a 64bit kernel?

What does /proc/meminfo show?

How about the memory map dmesg shows at the start of boot?

Len Sorensen

#32 lore raw
In reply to Rudolf Usselmann
Quoted 9 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

I still am unable to find an associated bug with it, is there an updated BIOS for your board?

#34 lore raw

Re: kernel (64bit) 4GB memory support II

In reply to Rudolf Usselmann
Quoted 10 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

Does the following patch (against a 2.6.12rc3 kernel) fix your problems?

-Andi

[...]

Please use this version instead, previous one was broken.

Don't look up struct page * of physical address in iounmap

it could be in a memory hole not mapped in mem_map

arch/x86_64/mm/ioremap.c +1 -1
Index: linux/arch/x86_64/mm/ioremap.c
===================================================================
@@ -272,7 +272,7 @@ void iounmap(volatile void __iomem *addr
 	if ((p->flags >> 20) &&
 		p->phys_addr + p->size - 1 < virt_to_phys(high_memory)) {
 		/* p->size includes the guard page, but cpa doesn't like that */
-		change_page_attr(virt_to_page(__va(p->phys_addr)),
+		change_page_attr_addr(__va(p->phys_addr),
 				 p->size >> PAGE_SHIFT,
 				 PAGE_KERNEL);
 		global_flush_tlb();
#23 lore raw
In reply to Lennart Sorensen
Quoted 11 lines · Lennart Sorensen
Quoted 9 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

Quoted 7 lines

How much ram do you see with 4GB installed running a 64bit kernel?

What does /proc/meminfo show?

How about the memory map dmesg shows at the start of boot?

Len Sorensen

I do see the full 4G. With Fedora Core 2 32bit, I can use all 4G as well. All my problems started when I "upgraded" to x86_64 ...

Best Regards,
rudi
=============================================================
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis

#33 lore raw
In reply to Zwane Mwaikambo
Quoted 6 lines · Zwane Mwaikambo
Quoted 4 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

Quoted 8 lines
Quoted 5 lines

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

I still am unable to find an associated bug with it, is there an updated BIOS for your board?

No there is not. I already have the latest BIOS. The 32 bit version of Fedora Core 2 seems to work just fine. It's when I "upgraded" to 64 bit when I could not use 4G of memory anymore ...

Best Regards,
rudi
=============================================================
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis

#24 lore raw
In reply to Rudolf Usselmann
Quoted 2 lines · Rudolf Usselmann

All my problems started when I "upgraded" to x86_64 ...

And you are surprised by this?

I don't know why so many users buy an arch that didn't exist two years ago and expect it to be 100% as reliable as 32 bit i386 which has been around for 20 plus years. I see tons of bug reports where people don't bother to mention that "oh, btw this is a mixed 32/64 bit environment" or whatever.

When you make the choice to live on the bleeding edge, these things will happen. If you don't like filling out bug reports stick with a 32 bit machine ;-)

Lee

#26 lore raw
In reply to Rudolf Usselmann

"I do see the full 4G. With Fedora Core 2 32bit, I can use all

4G as well. All my problems started when I "upgraded" to x86_64 ..."

Are you using an old 32bit processor or a new 64bit processor? That would make a difference

Quoted 44 lines · Rudolf Usselmann
Quoted 14 lines · Lennart Sorensen
Quoted 11 lines · Rudolf Usselmann

Just curious, did anybody ever look in to this at all ? I keep on downloading new kernels and trying 4GB of memory - still no luck.

I did file a bug report but didn't get any notifications at all. I don't subscribe to the linux-kernel list so not sure if anything ever came up or not.

Is there a way to get this fixed ?

Quoted 9 lines

How much ram do you see with 4GB installed running a 64bit kernel?

What does /proc/meminfo show?

How about the memory map dmesg shows at the start of boot?

Len Sorensen

I do see the full 4G. With Fedora Core 2 32bit, I can use all 4G as well. All my problems started when I "upgraded" to x86_64 ...

Best Regards,
rudi
=============================================================
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#25 lore raw
In reply to Lee Revell
Quoted 12 lines · Lee Revell
Quoted 2 lines · Rudolf Usselmann

All my problems started when I "upgraded" to x86_64 ...

And you are surprised by this?

I don't know why so many users buy an arch that didn't exist two years ago and expect it to be 100% as reliable as 32 bit i386 which has been around for 20 plus years. I see tons of bug reports where people don't bother to mention that "oh, btw this is a mixed 32/64 bit environment" or whatever.

Quoted 5 lines

When you make the choice to live on the bleeding edge, these things will happen. If you don't like filling out bug reports stick with a 32 bit machine ;-)

Lee

Hi Lee,

no I am not surprised ! But, hey, if nobody tries, we will never debug it, right ?! ;*)

I did fill out a bug report several month ago (and did not mind it), I was just following up because I didn't hear anything about it in quite sometime !

Best Regards,
rudi
=============================================================
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis

#28 lore raw
In reply to Rudolf Usselmann
Quoted 2 lines · Rudolf Usselmann

I do see the full 4G. With Fedora Core 2 32bit, I can use all 4G as well. All my problems started when I "upgraded" to x86_64 ...

In 32bit it probably uses the PSE36 extensions or something, which isn't the same thing as flat 64bit memory access. It could just be a matter of needing a memory hole somewhere for PCI space or something. I only have 1G in my 64bit machine so I haven't got near these problems.

Len Sorensen

#27 lore raw
In reply to Terry Vernon

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 6 lines · Terry Vernon

"I do see the full 4G. With Fedora Core 2 32bit, I can use all

4G as well. All my problems started when I "upgraded" to x86_64 ..."

Are you using an old 32bit processor or a new 64bit processor? That would make a difference

How on earth would he be able to run ANYTHING that's 64bit if his CPU was 32bit?

His system is a 64bit system.

// Stefan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)

iD8DBQFCgtMoBrn2kJu9P78RAlVeAJ9Nn4t4xDptAwPZ9X0quxoE4+epdQCgtoWP cpTQqHIkcwlMcceEMge7WYI=
=j68y
-----END PGP SIGNATURE-----

#29 lore raw
In reply to Lennart Sorensen

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 10 lines · Lennart Sorensen
Quoted 2 lines · Rudolf Usselmann

I do see the full 4G. With Fedora Core 2 32bit, I can use all 4G as well. All my problems started when I "upgraded" to x86_64 ...

In 32bit it probably uses the PSE36 extensions or something, which isn't the same thing as flat 64bit memory access. It could just be a matter of needing a memory hole somewhere for PCI space or something. I only have 1G in my 64bit machine so I haven't got near these problems.

I don't recall him saying he's changed kernel from the default redhat kernel in which case he's running the RedHat 4G/4G split kernel and not using PSE/PAE.

// Stefan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)

iD8DBQFCgtPhBrn2kJu9P78RAvRnAKCnSbeH+4i/vzAfRYfWJXgFad3fpACfVQdp JWUmCke6CvQ1mq5Gj4SvkZM=
=50yt
-----END PGP SIGNATURE-----

#30 lore raw
Quoted 15 lines · Stefan Smietanowski

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 10 lines · Lennart Sorensen
Quoted 2 lines · Rudolf Usselmann

I do see the full 4G. With Fedora Core 2 32bit, I can use all 4G as well. All my problems started when I "upgraded" to x86_64 ...

In 32bit it probably uses the PSE36 extensions or something, which isn't the same thing as flat 64bit memory access. It could just be a matter of needing a memory hole somewhere for PCI space or something. I only have 1G in my 64bit machine so I haven't got near these problems.

Quoted 5 lines

I don't recall him saying he's changed kernel from the default redhat kernel in which case he's running the RedHat 4G/4G split kernel and not using PSE/PAE.

// Stefan

I'm using whatever is available on www.kernerl.org :*)

Sorry I am totally lost what the above terminology means.

Is that a configuration option that I should try, or a totally different branch of the kernel ?

Thanks !
rudi
=============================================================
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis

#31 lore raw
Quoted 4 lines · Stefan Smietanowski
Quoted 4 lines
Quoted 4 lines

In 32bit it probably uses the PSE36 extensions or something, which isn't the same thing as flat 64bit memory access. It could just be a matter of needing a memory hole somewhere for PCI space or something. I only have 1G in my 64bit machine so I haven't got near these problems.

Quoted 3 lines · Rudolf Usselmann
Quoted 3 lines

I don't recall him saying he's changed kernel from the default redhat kernel in which case he's running the RedHat 4G/4G split kernel and not using PSE/PAE.

PSE36 is 4MB pages (no 4KB pages allowed!), 36-bit physical addresses.

-- wli