48 bit ATA support in Linux 2.4

48 bit ATA support in Linux 2.4

[Robert Szentmihalyi]

Hi Andre,

can you tell me please when your patch to support the 160 GB Matrox 
HDs got into the 2.4 tree (which version)? It was 2.4.14, right?

I searched the changelog but could not find it.

Also, could you please send me a copy of the patch?
Google didn't find it...

I have a machine running 2.4.10-pre2-xfs with XFS support.
Would the patch apply cleanly or will I have to update the kernel 
in order to support the drive?

Thanks,
 Robert

-- 
Where do you want to be tomorrow?

Entracom. Building Linux systems.
http://www.entracom.de

Re: 48 bit ATA support in Linux 2.4

[Robert Szentmihalyi]

On Tuesday, 7. May 2002 18:32, Dan Chen wrote:
> On Tue, May 07, 2002 at 05:21:55PM +0200, Robert Szentmihalyi wrote:
> > can you tell me please when your patch to support the 160 GB
> > Matrox HDs got into the 2.4 tree (which version)? It was
> > 2.4.14, right?
>
> It will be available in 2.4.19 final (was merged early in the
> 19-pres).
>
> [snip]
>
> > Also, could you please send me a copy of the patch?
> > Google didn't find it...
>
> See http://linuxdiskcert.org
>
> > I have a machine running 2.4.10-pre2-xfs with XFS support.
> > Would the patch apply cleanly or will I have to update the
> > kernel in order to support the drive?
>
> If you're handy w/ diff and patch, you can look at *.rej after
> patching and manually apply the diffs.
>
> The easiest route is probably to upgrade to a newer kernel (say,
> 2.4.18), and deal w/ the fewer rejects there since you'll also be
> applying a diff for XFS.

I have applied Andre's patch from http://linuxdiskcert.org on top of 
linux-2.4.19-pre7. Then I applied the XFS patches for 2.4.18
(unfortunately, the XFS-enabled kernel from the SGI-CVS is at 2.4.18
also, so it doesn't contain Ande's patch).

There is one reject in include/linux/sysctl.h I don't know how to fix:

***************
*** 140,145 ****
  	VM_PAGERDAEMON=8,	/* struct: Control kswapd behaviour */
  	VM_PGT_CACHE=9,		/* struct: Set page table cache parameters */
  	VM_PAGE_CLUSTER=10,	/* int: set number of pages to swap together */
         VM_MIN_READAHEAD=12,    /* Min file readahead */
         VM_MAX_READAHEAD=13     /* Max file readahead */
  };
--- 140,148 ----
  	VM_PAGERDAEMON=8,	/* struct: Control kswapd behaviour */
  	VM_PGT_CACHE=9,		/* struct: Set page table cache parameters */
  	VM_PAGE_CLUSTER=10,	/* int: set number of pages to swap together */
+ #if defined(CONFIG_XFS_FS) || defined(CONFIG_XFS_FS_MODULE)
+ 	VM_PAGEBUF=11,		/* struct: Control pagebuf parameters */
+ #endif
         VM_MIN_READAHEAD=12,    /* Min file readahead */
         VM_MAX_READAHEAD=13     /* Max file readahead */
  };

Can anybody who knows the XFS code please tell what the alias VM_PAGEBUF
in this enumeration is used for and how I could fix this?

The original 2.4.49-pre7 code seems to define another alias for 11 in this enum:
 
/* CTL_VM names: */
enum
{
	VM_SWAPCTL=1,		/* struct: Set vm swapping control */
	VM_SWAPOUT=2,		/* int: Linear or sqrt() swapout for hogs */
	VM_FREEPG=3,		/* struct: Set free page thresholds */
	VM_BDFLUSH=4,		/* struct: Control buffer cache flushing */
	VM_OVERCOMMIT_MEMORY=5,	/* Turn off the virtual memory safety limit */
	VM_BUFFERMEM=6,		/* struct: Set buffer memory thresholds */
	VM_PAGECACHE=7,		/* struct: Set cache memory thresholds */
	VM_PAGERDAEMON=8,	/* struct: Control kswapd behaviour */
	VM_PGT_CACHE=9,		/* struct: Set page table cache parameters */
	VM_PAGE_CLUSTER=10,	/* int: set number of pages to swap together */
	VM_MAX_MAP_COUNT=11,	/* int: Maximum number of active map areas */
	VM_MIN_READAHEAD=12,    /* Min file readahead */
	VM_MAX_READAHEAD=13,    /* Max file readahead */
};

Any hints?

Thanks for your help,
 Robert

>
> Alternately, -aa has XFS in his branch, but Andre's patches will
> clash. You'll need to rip out Jens's highmem-i/o patches from -aa
> if you want Andre's patch to apply cleanly on top of
> 2.4.X-pre-aaY. (They can still be applied manually, however.)

-- 
Where do you want to be tomorrow?

Entracom. Building Linux systems.
http://www.entracom.de

Re: 48 bit ATA support in Linux 2.4

[Steve Lord]

On Wed, 2002-05-08 at 09:02, Robert Szentmihalyi wrote:
> 
> Can anybody who knows the XFS code please tell what the alias VM_PAGEBUF
> in this enumeration is used for and how I could fix this?
> 
> The original 2.4.49-pre7 code seems to define another alias for 11 in this enum:
>  
> /* CTL_VM names: */
> enum
> {
> 	VM_SWAPCTL=1,		/* struct: Set vm swapping control */
> 	VM_SWAPOUT=2,		/* int: Linear or sqrt() swapout for hogs */
> 	VM_FREEPG=3,		/* struct: Set free page thresholds */
> 	VM_BDFLUSH=4,		/* struct: Control buffer cache flushing */
> 	VM_OVERCOMMIT_MEMORY=5,	/* Turn off the virtual memory safety limit */
> 	VM_BUFFERMEM=6,		/* struct: Set buffer memory thresholds */
> 	VM_PAGECACHE=7,		/* struct: Set cache memory thresholds */
> 	VM_PAGERDAEMON=8,	/* struct: Control kswapd behaviour */
> 	VM_PGT_CACHE=9,		/* struct: Set page table cache parameters */
> 	VM_PAGE_CLUSTER=10,	/* int: set number of pages to swap together */
> 	VM_MAX_MAP_COUNT=11,	/* int: Maximum number of active map areas */
> 	VM_MIN_READAHEAD=12,    /* Min file readahead */
> 	VM_MAX_READAHEAD=13,    /* Max file readahead */
> };
> 
> Any hints?
> 


Just push the VM_PAGEBUF number up to an unused value, there are no
user space programs which actually use it, just the /proc interface
is normally used for this.

Steve


-- 

Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@sgi.com