I have an older dual processor server that has been happily running FC5 since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2 1G sticks and tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error.
Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially anything over 4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
On Tuesday 03 April 2007, Cokey wrote:
I have an older dual processor server that has been happily running FC5 since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2 1G sticks and tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error.
Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially anything over 4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
There is, in the kernel's .config file, an entry to switch it to handle more than 4GB of memory. I believe you'll have to set that option and rebuild/reinstall the kernel.
In fairly recent kernels, say the last year or so, that option line is:
# CONFIG_HIGHMEM64G is not set
Use a make (x/menu)config to turn that on.
Tnx
Cokey
-- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
Gene Heskett wrote:
On Tuesday 03 April 2007, Cokey wrote:
I have an older dual processor server that has been happily running FC5 since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2 1G sticks and tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error.
Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially anything over 4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
There is, in the kernel's .config file, an entry to switch it to handle more than 4GB of memory. I believe you'll have to set that option and rebuild/reinstall the kernel.
In fairly recent kernels, say the last year or so, that option line is:
# CONFIG_HIGHMEM64G is not set
Use a make (x/menu)config to turn that on.
Tnx
Cokey
I'm using the stock SMP kernel (2.6.19-1.2288) and it comes with the large memory model turned on (CONFIG_HIGHMEM64G=y). Thanks for the idea. Wish it were that easy.
Cokey
Cokey wrote:
I have an older dual processor server that has been happily running FC5 since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2 1G sticks and tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error. Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially anything over 4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Mikkel L. Ellertson wrote:
Cokey wrote:
I have an older dual processor server that has been happily running FC5 since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2 1G sticks and tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error. Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially anything over 4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
Tnx
Cokey
Mikkel L. Ellertson wrote:
Cokey wrote:
I have an older dual processor server that has been
happily running FC5
since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2
1G sticks and
tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error. Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially
anything over
4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
Mikkel L. Ellertson wrote:
Cokey wrote:
I have an older dual processor server that has been
happily running FC5
since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2
1G sticks and
tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error. Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially
anything over
4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
PAE.
Tony Nelson wrote:
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
Mikkel L. Ellertson wrote:
Cokey wrote:
I have an older dual processor server that has been
happily running FC5
since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2
1G sticks and
tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error. Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially
anything over
4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
PAE.
Ok, so? PAE is enabled in the kernel and this box has dual Pentium III which are PAE enabled. Am I missing something?
Cokey
Ok, so? PAE is enabled in the kernel and this box has dual Pentium III which are PAE enabled. Am I missing something?
I have never tried what you are doing but are you sure you do not have a RAM compatibility issue? You may want to try all the same exact RAM modules and see if you still run into this.
Wish I could be more help.
At 5:12 PM -0400 4/5/07, Cokey wrote:
Tony Nelson wrote:
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
Mikkel L. Ellertson wrote:
Cokey wrote:
I have an older dual processor server that has been
happily running FC5
since it was released. It currently has 4G of memory and has the capacity for 8G using 1G sticks. Recently I was given 2
1G sticks and
tried to upgrade the system. The system recognizes the memory and boots, but then locks up with:
exec of initj(/sbin/init) failed: Exec format error. Removing the memory clears the problem. I have also tried some 512 sticks and they also cause the same problem. Essentially
anything over
4G regardless of the how that's arrived at will cause the lockup.
Looking for any hints/ideas.
Tnx
Cokey
You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
PAE.
Ok, so? PAE is enabled in the kernel and this box has dual Pentium III which are PAE enabled. Am I missing something?
Not you, him, in saying that 4 GB was the max physical memory for a 32 bit kernel.
Tony Nelson wrote:
At 5:12 PM -0400 4/5/07, Cokey wrote:
Tony Nelson wrote:
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
Mikkel L. Ellertson wrote:
Cokey wrote:
> I have an older dual processor server that has been > >
happily running FC5
> since it was released. It currently has 4G of memory and has the > capacity for 8G using 1G sticks. Recently I was given 2 > >
1G sticks and
> tried to upgrade the system. The system recognizes the memory and > boots, but then locks up with: > > exec of initj(/sbin/init) failed: Exec format error. > Removing the memory clears the problem. I have also tried some 512 > sticks and they also cause the same problem. Essentially > >
anything over
> 4G regardless of the how that's arrived at will cause the lockup. > > Looking for any hints/ideas. > > Tnx > > Cokey > > > > > You may want to check how your system handles more then 4G of memory. I seam to recall a thread a while ago about a system leaving a hole in memory and the BIOS producing an incorrect memory map when the extra memory was added. (I think it was with more then 2G in that thread.) You should also check to see if there is BIOS update for your machine.
Mikkel
Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
PAE.
Ok, so? PAE is enabled in the kernel and this box has dual Pentium III which are PAE enabled. Am I missing something?
Not you, him, in saying that 4 GB was the max physical memory for a 32 bit kernel.
Ok, now in case anyone's interested, I stuffed 7 G in the box (6 X 1G and 2 X 512) and Windows 2000 server (install) booted just fine.
Also received a different message when I tried to boot Linux... Kernel panic - not syncing: attempt to kill init
I'm really confused now.
Cokey
On 4/6/07, Cokey fdepercin@sc.rr.com wrote:
Tony Nelson wrote:
At 5:12 PM -0400 4/5/07, Cokey wrote:
Tony Nelson wrote:
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
Mikkel L. Ellertson wrote:
> Cokey wrote: > > > >> I have an older dual processor server that has been >> >> happily running FC5
>> since it was released. It currently has 4G of memory and has the >> capacity for 8G using 1G sticks. Recently I was given 2 >> >> 1G sticks and
>> tried to upgrade the system. The system recognizes the memory and >> boots, but then locks up with: >> >> exec of initj(/sbin/init) failed: Exec format error. >> Removing the memory clears the problem. I have also tried some 512 >> sticks and they also cause the same problem. Essentially >> >> anything over
>> 4G regardless of the how that's arrived at will cause the lockup. >> >> Looking for any hints/ideas. >> >> Tnx >> >> Cokey >> >> >> >> >> > You may want to check how your system handles more then 4G of > memory. I seam to recall a thread a while ago about a system leaving > a hole in memory and the BIOS producing an incorrect memory map when > the extra memory was added. (I think it was with more then 2G in > that thread.) You should also check to see if there is BIOS update > for your machine. > > Mikkel > > > Good idea but unfortunately there's no BIOS updates for this system beyond what's installed. Wish there were... And there's nothing I can change in the BIOS either. Manual doesn't say anything about a memory hole... but then it might not.
You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
PAE.
Ok, so? PAE is enabled in the kernel and this box has dual Pentium III which are PAE enabled. Am I missing something?
Not you, him, in saying that 4 GB was the max physical memory for a 32 bit kernel.
Ok, now in case anyone's interested, I stuffed 7 G in the box (6 X 1G and 2 X 512) and Windows 2000 server (install) booted just fine.
Also received a different message when I tried to boot Linux... Kernel panic - not syncing: attempt to kill init
I'm really confused now.
Cokey
Unless you have Advanced Server the box will only see 4GB. Google search produced this: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEmem.mspx
Kam Leo wrote:
On 4/6/07, Cokey fdepercin@sc.rr.com wrote:
Tony Nelson wrote:
At 5:12 PM -0400 4/5/07, Cokey wrote:
Tony Nelson wrote:
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
> Mikkel L. Ellertson wrote: > > >> Cokey wrote: >> >> >> >>> I have an older dual processor server that has been >>> >>> > happily running FC5 > > >>> since it was released. It currently has 4G of memory and has
the
>>> capacity for 8G using 1G sticks. Recently I was given 2 >>> >>> > 1G sticks and > > >>> tried to upgrade the system. The system recognizes the
memory and
>>> boots, but then locks up with: >>> >>> exec of initj(/sbin/init) failed: Exec format error. >>> Removing the memory clears the problem. I have also tried
some 512
>>> sticks and they also cause the same problem. Essentially >>> >>> > anything over > > >>> 4G regardless of the how that's arrived at will cause the
lockup.
>>> >>> Looking for any hints/ideas. >>> >>> Tnx >>> >>> Cokey >>> >>> >>> >>> >>> >> You may want to check how your system handles more then 4G of >> memory. I seam to recall a thread a while ago about a system
leaving
>> a hole in memory and the BIOS producing an incorrect memory
map when
>> the extra memory was added. (I think it was with more then 2G in >> that thread.) You should also check to see if there is BIOS
update
>> for your machine. >> >> Mikkel >> >> >> > Good idea but unfortunately there's no BIOS updates for this
system
> beyond what's installed. Wish there were... And there's > nothing I can change in the BIOS either. Manual doesn't say
anything
> about a memory hole... but then it might not. > > > You say this is an older system. If it is 32 bit, 4 gig is the maximum amount of memory you can address in 32 bits.
PAE.
Ok, so? PAE is enabled in the kernel and this box has dual
Pentium III
which are PAE enabled. Am I missing something?
Not you, him, in saying that 4 GB was the max physical memory for a
32 bit
kernel.
Ok, now in case anyone's interested, I stuffed 7 G in the box (6 X 1G and 2 X 512) and Windows 2000 server (install) booted just fine.
Also received a different message when I tried to boot Linux... Kernel panic - not syncing: attempt to kill init
I'm really confused now.
Cokey
Unless you have Advanced Server the box will only see 4GB. Google search produced this: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEmem.mspx
Your correct and thanks for pointing that out. Forgot about that. Still, it boots. My Fedora 5 install barfs. I'll have to borrow a copy of advanced server to see if the install will come up. Guess that puts and end to this until next week.
Tnx
Cokey
Cokey wrote:
Kam Leo wrote:
Unless you have Advanced Server the box will only see 4GB. Google search produced this: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEmem.mspx
Your correct and thanks for pointing that out. Forgot about that. Still, it boots. My Fedora 5 install barfs. I'll have to borrow a copy of advanced server to see if the install will come up. Guess that puts and end to this until next week.
Tnx
Cokey
You can also try using the mem= parameter to tell Linux to only use 4G of memory and see if it boots. It probably will. If it does, you may want to look at the memory map from the BIOS that will be in the logs.
Mikkel
Your correct and thanks for pointing that out. Forgot about that. Still, it boots. My Fedora 5 install barfs. I'll have to borrow a copy of advanced server to see if the install will come up. Guess that puts and end to this until next week.
You could try booting from a live CD distro like Knoppix and see if that helps.
Cokey wrote:
Kam Leo wrote:
On 4/6/07, Cokey fdepercin@sc.rr.com wrote:
Tony Nelson wrote:
At 5:12 PM -0400 4/5/07, Cokey wrote:
Tony Nelson wrote:
At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote:
>> Mikkel L. Ellertson wrote: >> >> >>> Cokey wrote: >>> >>> >>> >>>> I have an older dual processor server that has been >>>> >>>> >> happily running FC5 >> >> >>>> since it was released. It currently has 4G of memory and
has the
>>>> capacity for 8G using 1G sticks. Recently I was given 2 >>>> >>>> >> 1G sticks and >> >> >>>> tried to upgrade the system. The system recognizes the
memory and
>>>> boots, but then locks up with: >>>> >>>> exec of initj(/sbin/init) failed: Exec format error. >>>> Removing the memory clears the problem. I have also tried
some 512
>>>> sticks and they also cause the same problem. Essentially >>>> >>>> >> anything over >> >> >>>> 4G regardless of the how that's arrived at will cause the
lockup.
>>>> >>>> Looking for any hints/ideas. >>>> >>>> Tnx >>>> >>>> Cokey >>>> >>>> >>>> >>>> >>>> >>> You may want to check how your system handles more then 4G of >>> memory. I seam to recall a thread a while ago about a system
leaving
>>> a hole in memory and the BIOS producing an incorrect memory
map when
>>> the extra memory was added. (I think it was with more then 2G in >>> that thread.) You should also check to see if there is BIOS
update
>>> for your machine. >>> >>> Mikkel >>> >>> >>> >> Good idea but unfortunately there's no BIOS updates for this
system
>> beyond what's installed. Wish there were... And there's >> nothing I can change in the BIOS either. Manual doesn't say
anything
>> about a memory hole... but then it might not. >> >> >> > You say this is an older system. If it is 32 bit, 4 gig is the > maximum amount of memory you can address in 32 bits. > > PAE.
Ok, so? PAE is enabled in the kernel and this box has dual
Pentium III
which are PAE enabled. Am I missing something?
Not you, him, in saying that 4 GB was the max physical memory for
a 32 bit
kernel.
Ok, now in case anyone's interested, I stuffed 7 G in the box (6 X 1G and 2 X 512) and Windows 2000 server (install) booted just fine.
Also received a different message when I tried to boot Linux... Kernel panic - not syncing: attempt to kill init
I'm really confused now.
Cokey
Unless you have Advanced Server the box will only see 4GB. Google search produced this: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEmem.mspx
Your correct and thanks for pointing that out. Forgot about that. Still, it boots. My Fedora 5 install barfs. I'll have to borrow a copy of advanced server to see if the install will come up. Guess that puts and end to this until next week.
Tnx
Cokey
Still haven't managed to borrow a copy of M$ advanced server... but I did run the memtest86 from the Fedora 6 CD. Recognized 6G and ran for 6 hours with no problems encountered. I really don't think it's a hardware problem. Still looking for suggestions.
Tnx
Cokey
At 2:50 PM -0400 4/15/07, Cokey wrote:
Cokey wrote:
Kam Leo wrote:
On 4/6/07, Cokey fdepercin@sc.rr.com wrote:
Tony Nelson wrote:
At 5:12 PM -0400 4/5/07, Cokey wrote:
Tony Nelson wrote:
> At 3:14 PM -0500 4/3/07, Styma, Robert E (Robert) wrote: > > >>> Mikkel L. Ellertson wrote: >>> >>> >>>> Cokey wrote: >>>> >>>> >>>> >>>>> I have an older dual processor server that has been >>>>> >>>>> >>> happily running FC5 >>> >>> >>>>> since it was released. It currently has 4G of memory and
has the
>>>>> capacity for 8G using 1G sticks. Recently I was given 2 >>>>> >>>>> >>> 1G sticks and >>> >>> >>>>> tried to upgrade the system. The system recognizes the
memory and
>>>>> boots, but then locks up with: >>>>> >>>>> exec of initj(/sbin/init) failed: Exec format error. >>>>> Removing the memory clears the problem. I have also tried
some 512
>>>>> sticks and they also cause the same problem. Essentially >>>>> >>>>> >>> anything over >>> >>> >>>>> 4G regardless of the how that's arrived at will cause the
lockup.
>>>>> >>>>> Looking for any hints/ideas. >>>>> >>>>> Tnx >>>>> >>>>> Cokey >>>>> >>>>> >>>>> >>>>> >>>>> >>>> You may want to check how your system handles more then 4G of >>>> memory. I seam to recall a thread a while ago about a system
leaving
>>>> a hole in memory and the BIOS producing an incorrect memory
map when
>>>> the extra memory was added. (I think it was with more then 2G in >>>> that thread.) You should also check to see if there is BIOS
update
>>>> for your machine. >>>> >>>> Mikkel >>>> >>>> >>>> >>> Good idea but unfortunately there's no BIOS updates for this
system
>>> beyond what's installed. Wish there were... And there's >>> nothing I can change in the BIOS either. Manual doesn't say
anything
>>> about a memory hole... but then it might not. >>> >>> >>> >> You say this is an older system. If it is 32 bit, 4 gig is the >> maximum amount of memory you can address in 32 bits. >> >> > PAE. > > Ok, so? PAE is enabled in the kernel and this box has dual
Pentium III
which are PAE enabled. Am I missing something?
Not you, him, in saying that 4 GB was the max physical memory for
a 32 bit
kernel.
Ok, now in case anyone's interested, I stuffed 7 G in the box (6 X 1G and 2 X 512) and Windows 2000 server (install) booted just fine.
Also received a different message when I tried to boot Linux... Kernel panic - not syncing: attempt to kill init
I'm really confused now.
Cokey
Unless you have Advanced Server the box will only see 4GB. Google search produced this: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEmem.mspx
Your correct and thanks for pointing that out. Forgot about that. Still, it boots. My Fedora 5 install barfs. I'll have to borrow a copy of advanced server to see if the install will come up. Guess that puts and end to this until next week.
Tnx
Cokey
Still haven't managed to borrow a copy of M$ advanced server... but I did run the memtest86 from the Fedora 6 CD. Recognized 6G and ran for 6 hours with no problems encountered. I really don't think it's a hardware problem. Still looking for suggestions.
Start a new thread making it more obviously a kernel issue, so the smart kernel guys notice it.
Try a current FC6 (or even rawhide) kernel first, just in case it's already fixed.
Tony Nelson wrote:
At 2:50 PM -0400 4/15/07, Cokey wrote:
Still haven't managed to borrow a copy of M$ advanced server... but I did run the memtest86 from the Fedora 6 CD. Recognized 6G and ran for 6 hours with no problems encountered. I really don't think it's a hardware problem. Still looking for suggestions.
Start a new thread making it more obviously a kernel issue, so the smart kernel guys notice it.
Try a current FC6 (or even rawhide) kernel first, just in case it's already fixed.
Dumb question - is there still a separate kernel for over 4G of RAM?
Mikkel
Mikkel L. Ellertson wrote:
Tony Nelson wrote:
At 2:50 PM -0400 4/15/07, Cokey wrote:
Still haven't managed to borrow a copy of M$ advanced server... but I did run the memtest86 from the Fedora 6 CD. Recognized 6G and ran for 6 hours with no problems encountered. I really don't think it's a hardware problem. Still looking for suggestions.
Start a new thread making it more obviously a kernel issue, so the smart kernel guys notice it.
Try a current FC6 (or even rawhide) kernel first, just in case it's already fixed.
Dumb question - is there still a separate kernel for over 4G of RAM?
Yeah. Read the release notes or yum info kernel-PAE
Rahul
I have an older dual processor server with 4 Gig of memory that has been running FC5 since it was released. I tried to upgraded to 7 Gig of memory under FC5 but the system locked up in the init phase regardless of what I did. I didn't chase the problem because I intended to upgrade the OS.
Recently I did upgrade the OS to F 7 and tried the changing the memory to 7 Gig again. Using the standard i686 kernel, it boots and runs correctly, but warns that only 4 Gig of memory will be used and that the PAE kernel was necessary to use the full 7 Gig.
I installed the 2.6.22.4-65.fc7PAE kernel and the system boots correctly, but stops at the init phase. The system is not locked up as it will respond to C-A-Del and reboot. The output at the time of lock up looks like this:
Switching to new root and running Init Umounting old /dev /proc /sys exec of init (/sbin/init) failed accessing a corrupted shared library request_module: runaway loop modprobe binfmt-0397 (repeated 5 times)
I ran the memory test on the system for 8 hours or so and it found and tested all 7 Gig. Also I have booted this system with M$ Advanced server and it finds/uses all 7 Gig so I don't think there's anything wrong with the hardware.
Rebooting with the original i686 kernel works correctly.
Any ideas or suggestions welcome.
Cokey
Cokey wrote:
I have an older dual processor server with 4 Gig of memory that has been running FC5 since it was released. I tried to upgraded to 7 Gig of memory under FC5 but the system locked up in the init phase regardless of what I did. I didn't chase the problem because I intended to upgrade the OS.
Recently I did upgrade the OS to F 7 and tried the changing the memory to 7 Gig again. Using the standard i686 kernel, it boots and runs correctly, but warns that only 4 Gig of memory will be used and that the PAE kernel was necessary to use the full 7 Gig.
I installed the 2.6.22.4-65.fc7PAE kernel and the system boots correctly, but stops at the init phase. The system is not locked up as it will respond to C-A-Del and reboot. The output at the time of lock up looks like this:
Switching to new root and running Init Umounting old /dev /proc /sys exec of init (/sbin/init) failed accessing a corrupted shared library request_module: runaway loop modprobe binfmt-0397 (repeated 5 times)
I ran the memory test on the system for 8 hours or so and it found and tested all 7 Gig. Also I have booted this system with M$ Advanced server and it finds/uses all 7 Gig so I don't think there's anything wrong with the hardware.
Rebooting with the original i686 kernel works correctly.
Any ideas or suggestions welcome.
Can you try with 6G of memory? I've not known many MB's or BIOS to work well with odd numbers. :-)
On Wed, 2007-08-29 at 10:00 +0800, Ed Greshko wrote:
Can you try with 6G of memory? I've not known many MB's or BIOS to work well with odd numbers. :-)
While I can see some value in trying a different amount, I'd be surprised at "7 being an odd number," being the issue. It's not going to be exactly 7,000,000,000, it'll be a rounded figure.
Tim wrote:
On Wed, 2007-08-29 at 10:00 +0800, Ed Greshko wrote:
Can you try with 6G of memory? I've not known many MB's or BIOS to work well with odd numbers. :-)
While I can see some value in trying a different amount, I'd be surprised at "7 being an odd number," being the issue. It's not going to be exactly 7,000,000,000, it'll be a rounded figure.
Ahh...that's why the " :-) " at the end of the statement.
Cokey wrote:
I have an older dual processor server with 4 Gig of memory that has been running FC5 since it was released. I tried to upgraded to 7 Gig of memory under FC5 but the system locked up in the init phase regardless of what I did. I didn't chase the problem because I intended to upgrade the OS.
Recently I did upgrade the OS to F 7 and tried the changing the memory to 7 Gig again. Using the standard i686 kernel, it boots and runs correctly, but warns that only 4 Gig of memory will be used and that the PAE kernel was necessary to use the full 7 Gig.
I installed the 2.6.22.4-65.fc7PAE kernel and the system boots correctly, but stops at the init phase. The system is not locked up as it will respond to C-A-Del and reboot. The output at the time of lock up looks like this:
Switching to new root and running Init Umounting old /dev /proc /sys exec of init (/sbin/init) failed accessing a corrupted shared library request_module: runaway loop modprobe binfmt-0397 (repeated 5 times)
I ran the memory test on the system for 8 hours or so and it found and tested all 7 Gig. Also I have booted this system with M$ Advanced server and it finds/uses all 7 Gig so I don't think there's anything wrong with the hardware.
Rebooting with the original i686 kernel works correctly.
Any ideas or suggestions welcome.
Have you verified that your kernel supports that much memory? As I recall, there's a 2GB (or possibly 4GB) cut-off for "vanilla" kernels and if you want to run on a system with more memory than the cut-off allows, you'll need to set the "really large memory" option in the kernel's config file and then rebuild the kernel (or download the "large memory kernel" rpm from you usual RPM supplier).
Hope this helps'idly,
-S
Steve Siegfried wrote:
Cokey wrote:
I have an older dual processor server with 4 Gig of memory that has been running FC5 since it was released. I tried to upgraded to 7 Gig of memory under FC5 but the system locked up in the init phase regardless of what I did. I didn't chase the problem because I intended to upgrade the OS.
Recently I did upgrade the OS to F 7 and tried the changing the memory to 7 Gig again. Using the standard i686 kernel, it boots and runs correctly, but warns that only 4 Gig of memory will be used and that the PAE kernel was necessary to use the full 7 Gig.
I installed the 2.6.22.4-65.fc7PAE kernel and the system boots correctly, but stops at the init phase. The system is not locked up as it will respond to C-A-Del and reboot. The output at the time of lock up looks like this:
Switching to new root and running Init Umounting old /dev /proc /sys exec of init (/sbin/init) failed accessing a corrupted shared library request_module: runaway loop modprobe binfmt-0397 (repeated 5 times)
I ran the memory test on the system for 8 hours or so and it found and tested all 7 Gig. Also I have booted this system with M$ Advanced server and it finds/uses all 7 Gig so I don't think there's anything wrong with the hardware.
Rebooting with the original i686 kernel works correctly.
Any ideas or suggestions welcome.
Have you verified that your kernel supports that much memory? As I recall, there's a 2GB (or possibly 4GB) cut-off for "vanilla" kernels and if you want to run on a system with more memory than the cut-off allows, you'll need to set the "really large memory" option in the kernel's config file and then rebuild the kernel (or download the "large memory kernel" rpm from you usual RPM supplier).
Hope this helps'idly,
-S
The std i686 kernel supports up to 4G of memory. The 'PAE' kernel handles 4G and above.
-
Cokey de Percin -
Ed Greshko -
Gene Heskett -
Kam Leo -
Mikkel -
Rahul Sundaram -
Steve Siegfried -
Styma, Robert E (Robert) -
Tim -
Tony Nelson -
Will Yonker