Prev: Firefox and Java no longer work.
Next: linux and sysctl -a all ... comment
From: J G Miller on 28 Jun 2010 08:41
On Monday, June 28th, 2010 at 09:45:54h +0200,
HeHeHe the Laughing Idiot suggested:

> so i dont think that i need "XFS Realtime Enabled" in kernel...

And you are indeed correct.

So what web searches have you done to investigate the problem with
appropriate keywords, viz

xfs "Function not implemented"

because one match reveals this posting to the SGI XFS mailing list in 2007

<http://oss.sgi.COM/archives/xfs/2004-07/msg00524.html>

QUOTE
Anthony Biacco wrote:
ok, elaborate please. It's a valid parameter, yes? From what I
understand XFS can do up to 64k blocksizes. Am I mistaken?

Not on linux, it can only do filesystem blocksize upto pagesize.
So on ia32 it maxes out at 4K.

Steve
UNQUOTE

Then a web search with the key words

xfs linux "block size"

reveals this pertinent article at SGI on XFS

<http://oss.sgi.COM/projects/xfs/>

QUOTE
Filesystem Block Size

The maximum filesystem block size is the page size
of the kernel, which is 4K on x86 architecture
UNQUOTE

What is the architecture on which you are trying to mount the
XFS file system with block size greater than 4k?
From: hehehe on 29 Jun 2010 05:15

Użytkownik "J G Miller" <miller(a)yoyo.ORG> napisał w wiadomości
news:i0a597$e0g$3(a)news.eternal-september.org...
> On Monday, June 28th, 2010 at 09:45:54h +0200,
> HeHeHe the Laughing Idiot suggested:
>
>> so i dont think that i need "XFS Realtime Enabled" in kernel...
>
> And you are indeed correct.
>
> So what web searches have you done to investigate the problem with
> appropriate keywords, viz
>
> xfs "Function not implemented"
>
> because one match reveals this posting to the SGI XFS mailing list in 2007
>
> <http://oss.sgi.COM/archives/xfs/2004-07/msg00524.html>
>
> QUOTE
> Anthony Biacco wrote:
> ok, elaborate please. It's a valid parameter, yes? From what I
> understand XFS can do up to 64k blocksizes. Am I mistaken?
>
> Not on linux, it can only do filesystem blocksize upto pagesize.
> So on ia32 it maxes out at 4K.
>
> Steve
> UNQUOTE
>
> Then a web search with the key words
>
> xfs linux "block size"
>
> reveals this pertinent article at SGI on XFS
>
> <http://oss.sgi.COM/projects/xfs/>
>
> QUOTE
> Filesystem Block Size
>
> The maximum filesystem block size is the page size
> of the kernel, which is 4K on x86 architecture
> UNQUOTE
>
> What is the architecture on which you are trying to mount the
> XFS file system with block size greater than 4k?


64bit.

From: J G Miller on 29 Jun 2010 08:50
On Tuesday, June 29th, 2010 at 11:15:51h +0200,
HeHeHe the Laughing Idiot explained:

> 64bit.

Thus, compile and run this program to show you what page size you have
and what is the maximum filesystem block size available for XFS on your
system.

#include <stdio.h>
#include <unistd.h>

int main()
{
int pageSize = getpagesize();

printf("The Page size on this system = %i bytes\n", pageSize);

return 0;
}


Further details in man 2 getpagesize - get memory page size


Perhaps now it is clear why the mount command is failing with the
"Function not implemented" message, but of course the warning message
could have been worded more appropriately.
From: hehehe on 29 Jun 2010 09:42
4kb .... how can i change this value?
From: Aragorn on 29 Jun 2010 10:50
On Tuesday 29 June 2010 15:42 in comp.os.linux.misc, somebody
identifying as hehehe wrote...

> 4kb .... how can i change this value?

By either rewriting the XFS driver for the Linux kernel, by rewriting
the memory allocation logic for x86-32 and x86-64 in the Linux kernel
insofar that the hardware poses no limitations for doing so, or by
writing an entirely different kernel altogether and while you're at it,
develop another kind of microprocessor for your machine that will work
with all of the other components in it. <grin>

Quoting from J.G. Miller's reply to you...:

>> >> QUOTE
>> >>   Anthony Biacco wrote:
>> >>  ok, elaborate please. It's a valid parameter,
>> >> yes? From what I understand XFS can do up to
>> >> 64k blocksizes. Am I mistaken?
>> >>
>> >>   Not on linux, it can only do filesystem blocksize upto
>> >> pagesize. So on ia32 it maxes out at 4K.
>> >>
>> >>   Steve
>> >> UNQUOTE

Also, for the record, you were asked what architecture you were trying
this on. "64-bit" is not an architecture. It is a property of *many*
architectures. A proper architecture name could be x86-32 (alias
IA32), x86-64 (alias AMD64, alias Intel EM64T), IA64 (alias Intel
Itanium), SPARC/UltraSPARC, PPC, MIPS, et al.

As you may (or may not) know, XFS is the default filesystem in use in
SGI's IRIX operating system. IRIX is a UNIX system designed to run on
SGI's MIPS processor architecture, and if my knowledge is correct, then
no port of it to x86-32/x86-64 was ever released, but SGI does sell
machines based upon the Intel Itanium processor, and although the only
ones I've read about were running GNU/Linux, it is possible that
they've ported IRIX to that architecture as well.

So please try being more specific when you are asked for information.
The better the information you provide, the better people will be able
to help you or explain things to you.

P.S.: Your original post and the follow-up thread is not the only one
people reply to, so please also leave a minimum of quoted (and
relevant) content from the post you are replying to in the message body
of your own post, so that people know what you are talking about
without having to read through the whole thread again. ;-)

--
*Aragorn*
(registered GNU/Linux user #223157)
First  |  Prev  |  Next  |  Last
Pages: 1 2 3
Prev: Firefox and Java no longer work.
Next: linux and sysctl -a all ... comment