RE: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

RE: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Donald Maner]

The kernel you're using WAS compiled with CONFIG_HIGHMEM4G=y, correct?

-----Original Message-----
From: Raul Miller [mailto:moth@magenta.com] 
Sent: Wednesday, December 10, 2003 10:52 PM
To: linux-kernel@vger.kernel.org
Subject: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB
ram.


[1.] Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Nick Piggin]

Donald Maner wrote:

>The kernel you're using WAS compiled with CONFIG_HIGHMEM4G=y, correct?
>
>-----Original Message-----
>From: Raul Miller [mailto:moth@magenta.com] 
>Sent: Wednesday, December 10, 2003 10:52 PM
>To: linux-kernel@vger.kernel.org
>Subject: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB
>ram.
>
>
>[1.] Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.
>

Raul Miller wrote:

>
>[7.2.] /proc/cpuinfo says:
>
>processor       : 0
>vendor_id       : AuthenticAMD
>cpu family      : 15
>model           : 5
>model name      : AMD Opteron(tm) Processor 240
>

Or ARCH=x86_64 ?

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Ed Sweetman]

I thought highmem wasn't necesarily needed for memory <=2GB? Highmem 
incurs some performance hits doesn't it and so the urge to move to it 
with only 2GB is not very attractive.  Anyways i'm just interested in if 
that's the case or not since 2GB is easy to get to these days and i had 
heard that highmem could be avoided passed the 1GB barrier.



Nick Piggin wrote:
> 
> 
> Donald Maner wrote:
> 
>> The kernel you're using WAS compiled with CONFIG_HIGHMEM4G=y, correct?
>>
>> -----Original Message-----
>> From: Raul Miller [mailto:moth@magenta.com] Sent: Wednesday, December 
>> 10, 2003 10:52 PM
>> To: linux-kernel@vger.kernel.org
>> Subject: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB
>> ram.
>>
>>
>> [1.] Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.
>>
> 
> Raul Miller wrote:
> 
>>
>> [7.2.] /proc/cpuinfo says:
>>
>> processor       : 0
>> vendor_id       : AuthenticAMD
>> cpu family      : 15
>> model           : 5
>> model name      : AMD Opteron(tm) Processor 240
>>
> 
> Or ARCH=x86_64 ?
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[William Lee Irwin III]

On Thu, Dec 11, 2003 at 12:33:39AM -0500, Ed Sweetman wrote:
> I thought highmem wasn't necesarily needed for memory <=2GB? Highmem 
> incurs some performance hits doesn't it and so the urge to move to it 
> with only 2GB is not very attractive.  Anyways i'm just interested in if 
> that's the case or not since 2GB is easy to get to these days and i had 
> heard that highmem could be avoided passed the 1GB barrier.

You're probably thinking of 2:2 split patches.

2:2 splits are at least technically ABI violations, which is probably
why this isn't merged etc. Applications sensitive to it are uncommon.

Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
top of the process address space.


-- wli

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Roland Dreier]

    William> 2:2 splits are at least technically ABI violations, which
    William> is probably why this isn't merged etc. Applications
    William> sensitive to it are uncommon.

    William> Yes, the SVR4 i386 ELF/ABI spec literally mandates
    William> 0xC0000000 as the top of the process address space.

What about the 4G/4G split stuff for x86 (which is in 2.6 as well as
the RH EL 3 kernel)?  It seems that would be just as big a violation
of the ABI...

 - Roland

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Nick Piggin]

William Lee Irwin III wrote:

>On Thu, Dec 11, 2003 at 12:33:39AM -0500, Ed Sweetman wrote:
>
>>I thought highmem wasn't necesarily needed for memory <=2GB? Highmem 
>>incurs some performance hits doesn't it and so the urge to move to it 
>>with only 2GB is not very attractive.  Anyways i'm just interested in if 
>>that's the case or not since 2GB is easy to get to these days and i had 
>>heard that highmem could be avoided passed the 1GB barrier.
>>
>
>You're probably thinking of 2:2 split patches.
>
>2:2 splits are at least technically ABI violations, which is probably
>why this isn't merged etc. Applications sensitive to it are uncommon.
>
>Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
>top of the process address space.
>

At any rate, Raul, highmem shouldn't hurt your performance significantly
with the 2.6 kernel. If it does then send a note to the list.

Your other options are a different user/kernel split, or a 64-bit kernel,
both of which should have less overhead than highmem.

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[William Lee Irwin III]

William> 2:2 splits are at least technically ABI violations, which
William> is probably why this isn't merged etc. Applications
William> sensitive to it are uncommon.
William> Yes, the SVR4 i386 ELF/ABI spec literally mandates
William> 0xC0000000 as the top of the process address space.

On Wed, Dec 10, 2003 at 09:48:11PM -0800, Roland Dreier wrote:
> What about the 4G/4G split stuff for x86 (which is in 2.6 as well as
> the RH EL 3 kernel)?  It seems that would be just as big a violation
> of the ABI...

I oversimplified it. It only requires it to be at or above 0xC0000000,
so the 4/4 patches are compliant.


-- wli

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Raul Miller]

On Wed, Dec 10, 2003 at 11:06:46PM -0600, Donald Maner wrote:
> The kernel you're using WAS compiled with CONFIG_HIGHMEM4G=y, correct?

No.


On Thu, Dec 11, 2003 at 04:13:25PM +1100, Nick Piggin wrote:
> Or ARCH=x86_64 ?

Yes.  Well, no... I don't see that option in my .config.  I
did specify the amd64 bit archictecture, but I don't know
what that means in .config terms.  Here's what's set under
"# Processor type and features":

CONFIG_X86_PC=y
CONFIG_MK8=y
CONFIG_X86_CMPXCHG=y
CONFIG_X86_XADD=y
CONFIG_X86_L1_CACHE_SHIFT=6
CONFIG_RWSEM_XCHGADD_ALGORITHM=y
CONFIG_X86_WP_WORKS_OK=y
CONFIG_X86_INVLPG=y
CONFIG_X86_BSWAP=y
CONFIG_X86_POPAD_OK=y
CONFIG_X86_GOOD_APIC=y
CONFIG_X86_INTEL_USERCOPY=y
CONFIG_X86_USE_PPRO_CHECKSUM=y
CONFIG_HPET_TIMER=y
CONFIG_HPET_EMULATE_RTC=y
CONFIG_SMP=y
CONFIG_NR_CPUS=2
CONFIG_PREEMPT=y
CONFIG_X86_LOCAL_APIC=y
CONFIG_X86_IO_APIC=y
CONFIG_X86_TSC=y
CONFIG_X86_MCE=y
CONFIG_X86_MSR=y
CONFIG_X86_CPUID=y
CONFIG_NOHIGHMEM=y
CONFIG_MTRR=y
CONFIG_HAVE_DEC_LOCK=y

On Thu, Dec 11, 2003 at 04:48:51PM +1100, Nick Piggin wrote:
> At any rate, Raul, highmem shouldn't hurt your performance significantly
> with the 2.6 kernel. If it does then send a note to the list.

Ok, I guess I'll try that (tomorrow, unless I hear any better suggestions
before then).

[I thought highmem was something completely different -- that it declared
a watermark and memory above that watermark was treated differently.
However, I guess I understand that this might have the side effect of
bumping things around such that I get access to the memory.]

Thanks,

-- 
Raul Miller
moth@magenta.com

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Nick Piggin]

Raul Miller wrote:

>On Wed, Dec 10, 2003 at 11:06:46PM -0600, Donald Maner wrote:
>
>>The kernel you're using WAS compiled with CONFIG_HIGHMEM4G=y, correct?
>>
>
>No.
>
>
>On Thu, Dec 11, 2003 at 04:13:25PM +1100, Nick Piggin wrote:
>
>>Or ARCH=x86_64 ?
>>
>
>Yes.  Well, no... I don't see that option in my .config.  I
>did specify the amd64 bit archictecture, but I don't know
>what that means in .config terms.  Here's what's set under
>"# Processor type and features":
>

This optimises the kernel for your chip when its in 32-bit mode.
make 'ARCH=x64_64' to make a 64-bit kernel, however you would
need a cross compiler.

>
>On Thu, Dec 11, 2003 at 04:48:51PM +1100, Nick Piggin wrote:
>
>>At any rate, Raul, highmem shouldn't hurt your performance significantly
>>with the 2.6 kernel. If it does then send a note to the list.
>>
>
>Ok, I guess I'll try that (tomorrow, unless I hear any better suggestions
>before then).
>
>[I thought highmem was something completely different -- that it declared
>a watermark and memory above that watermark was treated differently.
>However, I guess I understand that this might have the side effect of
>bumping things around such that I get access to the memory.]
>

No you're right, but the kernel tries not to use highmem for data it
accesses a lot. cache and anonymous memory for example.

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[David Lang]

Ed,
highmem isn't nessasary for mem<1G (actually 960M) on x86.

you are useing the opteron CPU, but was your kernel compiled for it
(x86064) or are you running an x86 kernel? (both will run, but the x86-64
kernel will give you better performance (including removing the need for
highmem)

David Lang

On Wed, 10 Dec 2003, William Lee Irwin III wrote:

> Date: Wed, 10 Dec 2003 21:41:11 -0800
> From: William Lee Irwin III <wli@holomorphy.com>
> To: Ed Sweetman <ed.sweetman@wmich.edu>
> Cc: Nick Piggin <piggin@cyberone.com.au>, Donald Maner <donjr@maner.org>,
>      Raul Miller <moth@magenta.com>, linux-kernel@vger.kernel.org
> Subject: Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB
>     ram.
>
> On Thu, Dec 11, 2003 at 12:33:39AM -0500, Ed Sweetman wrote:
> > I thought highmem wasn't necesarily needed for memory <=2GB? Highmem
> > incurs some performance hits doesn't it and so the urge to move to it
> > with only 2GB is not very attractive.  Anyways i'm just interested in if
> > that's the case or not since 2GB is easy to get to these days and i had
> > heard that highmem could be avoided passed the 1GB barrier.
>
> You're probably thinking of 2:2 split patches.
>
> 2:2 splits are at least technically ABI violations, which is probably
> why this isn't merged etc. Applications sensitive to it are uncommon.
>
> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
> top of the process address space.
>
>
> -- wli
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

-- 
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." - Brian W. Kernighan

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Martin J. Bligh]

--William Lee Irwin III <wli@holomorphy.com> wrote (on Wednesday, December 10, 2003 21:41:11 -0800):

> On Thu, Dec 11, 2003 at 12:33:39AM -0500, Ed Sweetman wrote:
>> I thought highmem wasn't necesarily needed for memory <=2GB? Highmem 
>> incurs some performance hits doesn't it and so the urge to move to it 
>> with only 2GB is not very attractive.  Anyways i'm just interested in if 
>> that's the case or not since 2GB is easy to get to these days and i had 
>> heard that highmem could be avoided passed the 1GB barrier.
> 
> You're probably thinking of 2:2 split patches.
> 
> 2:2 splits are at least technically ABI violations, which is probably
> why this isn't merged etc. Applications sensitive to it are uncommon.
> 
> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
> top of the process address space.

You mean like we place the stack in the "ABI compliant place"? 
Yeah, right ;-)

M.

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[William Lee Irwin III]

William Lee Irwin III <wli@holomorphy.com> wrote:
>> You're probably thinking of 2:2 split patches.
>> 2:2 splits are at least technically ABI violations, which is probably
>> why this isn't merged etc. Applications sensitive to it are uncommon.
>> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
>> top of the process address space.

On Wed, Dec 10, 2003 at 11:08:38PM -0800, Martin J. Bligh wrote:
> You mean like we place the stack in the "ABI compliant place"? 
> Yeah, right ;-)

No specific address is ever cited as a requirement for stack placement;
stack immediately below text is merely given as a "typical arrangement".
i.e. "Although applications may control their memory assignments, the
typical arrangement appears below: [diagram and other bits]" It then
goes on to say, "Processes, therefore, shount _not_ depend on finding
their stack at a particular virtual address."

The process address space boundary is, however, stated as a requirement:
"the reserved area shall not consume more than 1GB of the address space."


-- wli

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Martin J. Bligh]

>>> You're probably thinking of 2:2 split patches.
>>> 2:2 splits are at least technically ABI violations, which is probably
>>> why this isn't merged etc. Applications sensitive to it are uncommon.
>>> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
>>> top of the process address space.
> 
> On Wed, Dec 10, 2003 at 11:08:38PM -0800, Martin J. Bligh wrote:
>> You mean like we place the stack in the "ABI compliant place"? 
>> Yeah, right ;-)
> 
> No specific address is ever cited as a requirement for stack placement;
> stack immediately below text is merely given as a "typical arrangement".
> i.e. "Although applications may control their memory assignments, the
> typical arrangement appears below: [diagram and other bits]" It then
> goes on to say, "Processes, therefore, shount _not_ depend on finding
> their stack at a particular virtual address."
> 
> The process address space boundary is, however, stated as a requirement:
> "the reserved area shall not consume more than 1GB of the address space."

;-)

OK, fair enough ... it doesn't actually break anything though ;-)

m.

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[William Lee Irwin III]

William Lee Irwin III <wli@holomorphy.com> wrote:
>> You're probably thinking of 2:2 split patches.
>> 2:2 splits are at least technically ABI violations, which is probably
>> why this isn't merged etc. Applications sensitive to it are uncommon.
>> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
>> top of the process address space.

On Wed, Dec 10, 2003 at 11:08:38PM -0800, Martin J. Bligh wrote:
> You mean like we place the stack in the "ABI compliant place"? 
> Yeah, right ;-)
> M.

Something odd is happening here; this is the second time I've gotten this
message. I suspect something is wrong with smtp-out2.blueyonder.co.uk

In fact, other messages in this thread are getting resent, too.


-- wli

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Raul Miller]

> On Thu, Dec 11, 2003 at 12:33:39AM -0500, Ed Sweetman wrote:
> > I thought highmem wasn't necesarily needed for memory <=2GB? Highmem 
> > incurs some performance hits doesn't it and so the urge to move to it 
> > with only 2GB is not very attractive.  Anyways i'm just interested in if 
> > that's the case or not since 2GB is easy to get to these days and i had 
> > heard that highmem could be avoided passed the 1GB barrier.

On Wed, Dec 10, 2003 at 09:41:11PM -0800, William Lee Irwin III wrote:
> You're probably thinking of 2:2 split patches.
> 
> 2:2 splits are at least technically ABI violations, which is probably
> why this isn't merged etc. Applications sensitive to it are uncommon.
> 
> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
> top of the process address space.

Apologies if I'm asking about the obvious, but... 

[1] isn't 0xC0000000 at 3GB?  
[2] Even if ELF did restrict a user process to 1GB (which I'm pretty
sure it doesn't), wouldn't the kernel still be able to manage 2GB of
user memory?

Probably my real question is: "what's this about 2:2 split patches"?

Basically, I thought "linux supports 2GB ram" had been been the case
since the dark ages.  It's hard for me to comprehend how highmem, or 64
bit cpus, could have much to do with a 1GB limit.

[In my fantasies, I was thinking that the system came up with only 1GB of
the memory easily usable, and that the lack of support for my hardware
meant that it couldn't be properly reconfigured.  But I recognize that
I haven't spent the time researching this to see if in fact this is
the case.]

I am in the process of bringing up an cross compilation environment for
amd64 -- I need to do that anyways -- and I'll try building a real 64
bit kernel to see if that helps any.  If that doesn't, I guess I'll try
a couple 4G highmem kernels (one 64 bit, one 32 bit).  If nothing else,
that will eat up some time...

Thanks,

-- 
Raul Miller
moth@magenta.com

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[Raul Miller]

On Thu, Dec 11, 2003 at 09:41:48AM -0500, I wrote:
> since the dark ages.  It's hard for me to comprehend how highmem, or 64
> bit cpus, could have much to do with a 1GB limit.

[Well, unless my motherboard really has a problem caching memory above
1GB.  But that seems unlikely, since it's the CPU which is acting as
the memory controller.  In any event, if this is actually the issue,
shouldn't the kernel issue a warning message, about why it's ignoring
the rest of the memory?]

Thanks again,

-- 
Raul Miller
moth@magenta.com

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[William Lee Irwin III]

On Wed, Dec 10, 2003 at 09:41:11PM -0800, William Lee Irwin III wrote:
>> You're probably thinking of 2:2 split patches.
>> 2:2 splits are at least technically ABI violations, which is probably
>> why this isn't merged etc. Applications sensitive to it are uncommon.
>> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
>> top of the process address space.

On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> Apologies if I'm asking about the obvious, but... 
> [1] isn't 0xC0000000 at 3GB?  

It is.


On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> [2] Even if ELF did restrict a user process to 1GB (which I'm pretty
> sure it doesn't), wouldn't the kernel still be able to manage 2GB of
> user memory?

You have it backward. The SVR4/i386 ELF ABI specification is requiring
userspace to be granted at least 3GB of address space.

This does not necessarily present a restriction for the kernel;
consider task gates (mingo did it by hand).


On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> Probably my real question is: "what's this about 2:2 split patches"?
> Basically, I thought "linux supports 2GB ram" had been been the case
> since the dark ages.  It's hard for me to comprehend how highmem, or 64
> bit cpus, could have much to do with a 1GB limit.

You should probably ignore this thread. It's probably not relevant to
you.


On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> [In my fantasies, I was thinking that the system came up with only 1GB of
> the memory easily usable, and that the lack of support for my hardware
> meant that it couldn't be properly reconfigured.  But I recognize that
> I haven't spent the time researching this to see if in fact this is
> the case.]

Highmem support gets you this on ia32. Other architectures can support it
with less overhead.


On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> I am in the process of bringing up an cross compilation environment for
> amd64 -- I need to do that anyways -- and I'll try building a real 64
> bit kernel to see if that helps any.  If that doesn't, I guess I'll try
> a couple 4G highmem kernels (one 64 bit, one 32 bit).  If nothing else,
> that will eat up some time...

If you have such a cpu why are you bothering with highmem (or wondering
if > 2GB is supported)?


-- wli

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[moth]

On Thu, Dec 11, 2003 at 07:00:11AM -0800, William Lee Irwin III wrote:
> You should probably ignore this thread. It's probably not relevant to
> you.

Ok, thanks.  My mistake.

> > [In my fantasies, I was thinking that the system came up with only 1GB of
> > the memory easily usable, and that the lack of support for my hardware
> > meant that it couldn't be properly reconfigured.  But I recognize that
> > I haven't spent the time researching this to see if in fact this is
> > the case.]
> 
> Highmem support gets you this on ia32. Other architectures can support it
> with less overhead.

Are there docs on this?

> > I am in the process of bringing up an cross compilation environment for
> > amd64 -- I need to do that anyways -- and I'll try building a real 64
> > bit kernel to see if that helps any.  If that doesn't, I guess I'll try
> > a couple 4G highmem kernels (one 64 bit, one 32 bit).  If nothing else,
> > that will eat up some time...
> 
> If you have such a cpu why are you bothering with highmem (or wondering
> if > 2GB is supported)?

I'm not wondering if > 2GB is supported.  I'm trying to get 2GB
to work (and I'm having a problem -- perhaps because I believe
Documentation/memory.txt doesn't cover the issues I'm facing).

I've not yet bothered with highmem, but I will if building a 64 bit
kernel doesn't get me access to 2GB.

Does that answer your question?

Thanks,

-- 
Raul

Re: PROBLEM: Linux 2.6.0-test11 only lets me use 1GB out of 2GB ram.

[William Lee Irwin III]

On Thu, Dec 11, 2003 at 11:17:41AM -0500, moth@magenta.com wrote:
> I'm not wondering if > 2GB is supported.  I'm trying to get 2GB
> to work (and I'm having a problem -- perhaps because I believe
> Documentation/memory.txt doesn't cover the issues I'm facing).
> I've not yet bothered with highmem, but I will if building a 64 bit
> kernel doesn't get me access to 2GB.
> Does that answer your question?

You should be fine with a 64-bit kernel.


-- wli

Where did the ELF spec go? (SCO website?)

[Rob Landley]

On Thursday 11 December 2003 09:00, William Lee Irwin III wrote:
> On Wed, Dec 10, 2003 at 09:41:11PM -0800, William Lee Irwin III wrote:
> >> You're probably thinking of 2:2 split patches.
> >> 2:2 splits are at least technically ABI violations, which is probably
> >> why this isn't merged etc. Applications sensitive to it are uncommon.
> >> Yes, the SVR4 i386 ELF/ABI spec literally mandates 0xC0000000 as the
> >> top of the process address space.
>
> On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> > Apologies if I'm asking about the obvious, but...
> > [1] isn't 0xC0000000 at 3GB?
>
> It is.
>
> On Thu, Dec 11, 2003 at 09:41:48AM -0500, Raul Miller wrote:
> > [2] Even if ELF did restrict a user process to 1GB (which I'm pretty
> > sure it doesn't), wouldn't the kernel still be able to manage 2GB of
> > user memory?
>
> You have it backward. The SVR4/i386 ELF ABI specification is requiring
> userspace to be granted at least 3GB of address space.

Where does one get a copy of the SVR4 spec these days?  The link I could track 
down went to http://www.sco.com/developer/devspecs/ which just ain't there no 
more.

And no, not because of a "DDOS".  There isn't one.  SCO's website IP moved 
from 216.250.128.13 to 216.250.128.20, and it's up at the new IP right now.  
They didn't get the new DNS record propogated on time.  Rookie mistake...

But looking at http.://216.250.128.20/developer/devspecs redirects you to the 
/developer page.  The devspecs page went away...

Is this mirrored somewhere?

Rob

Re: Where did the ELF spec go? (SCO website?)

[William Lee Irwin III]

On Thursday 11 December 2003 09:00, William Lee Irwin III wrote:
>> You have it backward. The SVR4/i386 ELF ABI specification is requiring
>> userspace to be granted at least 3GB of address space.

On Thu, Dec 11, 2003 at 01:26:32PM -0600, Rob Landley wrote:
> Where does one get a copy of the SVR4 spec these days?  The link I
> could track down went to http://www.sco.com/developer/devspecs/ which
> just ain't there no more.
> And no, not because of a "DDOS".  There isn't one.  SCO's website IP moved 
> from 216.250.128.13 to 216.250.128.20, and it's up at the new IP right now.  
> They didn't get the new DNS record propogated on time.  Rookie mistake...
> But looking at http.://216.250.128.20/developer/devspecs redirects
> you to the /developer page.  The devspecs page went away...
> Is this mirrored somewhere?

I'm looking at a dead tree copy. I have no idea if it's online or not.

Also, it's largely an ELF ABI spec; I'm not sure how/why SVR4 got into
the picture, but its name is on there.


-- wli

Re: Where did the ELF spec go? (SCO website?)

[Richard B. Johnson]

On Thu, 11 Dec 2003, William Lee Irwin III wrote:

> On Thursday 11 December 2003 09:00, William Lee Irwin III wrote:
> >> You have it backward. The SVR4/i386 ELF ABI specification is requiring
> >> userspace to be granted at least 3GB of address space.
>
> On Thu, Dec 11, 2003 at 01:26:32PM -0600, Rob Landley wrote:
> > Where does one get a copy of the SVR4 spec these days?  The link I
> > could track down went to http://www.sco.com/developer/devspecs/ which
> > just ain't there no more.
> > And no, not because of a "DDOS".  There isn't one.  SCO's website IP moved
> > from 216.250.128.13 to 216.250.128.20, and it's up at the new IP right now.
> > They didn't get the new DNS record propogated on time.  Rookie mistake...
> > But looking at http.://216.250.128.20/developer/devspecs redirects
> > you to the /developer page.  The devspecs page went away...
> > Is this mirrored somewhere?
>
> I'm looking at a dead tree copy. I have no idea if it's online or not.
>
> Also, it's largely an ELF ABI spec; I'm not sure how/why SVR4 got into
> the picture, but its name is on there.


http://developers.sun.com/solaris/articles/elf.html


Cheers,
Dick Johnson
Penguin : Linux version 2.4.22 on an i686 machine (797.90 BogoMips).
            Note 96.31% of all statistics are fiction.

Re: Where did the ELF spec go? (SCO website?)

[Johannes Stezenbach]

Richard B. Johnson wrote:
> On Thu, 11 Dec 2003, William Lee Irwin III wrote:
> 
> > On Thursday 11 December 2003 09:00, William Lee Irwin III wrote:
> > >> You have it backward. The SVR4/i386 ELF ABI specification is requiring
> > >> userspace to be granted at least 3GB of address space.
> >
> > On Thu, Dec 11, 2003 at 01:26:32PM -0600, Rob Landley wrote:
> > > Where does one get a copy of the SVR4 spec these days?  The link I
> > > could track down went to http://www.sco.com/developer/devspecs/ which
> > > just ain't there no more.

http://www.caldera.com/developers/devspecs/

-js