into ppc970.osdl.org:/home/torvalds/v2.5/linux

into ppc970.osdl.org:/home/torvalds/v2.5/linux

From: "Randy.Dunlap" <rddunlap@osdl.org>
From: Luiz Fernando Capitulino <lcapitulino@prefeitura.sp.gov.br>

The missing __init was found by reference_init26.pl.

From: Frederik Deweerdt <frederik.deweerdt@laposte.net>

They are identical to non _HOB versions (except IDE_CONTROL_OFFSET_HOB).

Set ide_task_t fields (command_type, handler and prehandler) directly.
Remove unused ide_task_t->posthandler and all ide_cmd_type_parser() logic.

ide_cmd_type_parser() was meant to be used for ioctls but
ended up checking validity of kernel generated requests (doh!).

Rationale for removal:
- it can't be used for existing ioctls (changes the way they work)
- kernel shouldn't check validity of (root only) user-space requests
  (it can and should be done in user-space)
- it wastes CPU cycles on going through parsers
- it makes code harder to understand/follow
  (now info about request is localized)

Many thanks to Promise for their support.
Expect more updates/fixes for Promise PATA support soon.

- fix PIO (auto-)tuning - use pdcnew_new_tune_chipset()
  and always tune PIO mode even if (U)DMA is used
- cleanup cable verification code a bit
- remove leftovers from driver split-up
- remove duplicate DISPLAY_PDC202XX_TIMINGS define

into redhat.com:/spare/repo/libata-2.5

The original submittor of the quirk code noted that the errata
of a lockup creating when the condition "sector % 15 == 1" is
true only on chip revisions <= 1.  This is limitation on chip
revision is apparently not correct, therefore we apply to all
affected drives in the quirk list.

This is a sledgehammer fix, limiting max-sectors to 15.  A much better
fix would split the hardware request into two taskfiles, issuing
an additional command, thereby avoiding "sectors % 15 == 1"

Given the errata is limited to "sil 311x + <these drives>", it is hoped
that some janitor cares enough to do a real fix.

into nuts.davemloft.net:/disk1/BK/sparc-2.6

into nuts.davemloft.net:/disk1/BK/net-2.6

into foobazco.org:/sources/2.5-sparc-todave

into foobazco.org:/sources/2.5-sparc-todave

into foobazco.org:/sources/2.5-sparc-todave

into foobazco.org:/sources/2.5-sparc-todave

into foobazco.org:/sources/2.5-sparc-smp

into foobazco.org:/sources/2.5-sparc-smp

into kernel.bkbits.net:/home/wesolows/sparc32-2.6

into foobazco.org:/sources/2.5-sparc-smp

The low 2 bits of the MID are now used exclusively as the software
CPU ID on 4m.  This eliminates entirely the need for mid_xlate.  We
also eliminate linux_num_cpus and linux_cpus; since the hardware-based
id is now used everywhere we don't need indexed CPU tables.

The device probe is almost identical to sparc64's version.  The one
exception is that 4m only uses the bottom 2 bits of the MID, so we also
have a separate function to get the true hardware MID for the rare times
it's actually needed.

Per-cpu data is now used the same way as in sparc64.  As a side benefit
we now display the same format Bogo and Clock values in /proc/cpuinfo.

into ppc970.osdl.org:/home/torvalds/v2.5/linux

into fys.uio.no:/home/linux/bitkeeper/nfsclient-2.5

The very early memory detection patch broke voyager.

This fixes it again.

Fix a bug which was spotted by Alex Lyashkov <shadow@psoft.net>

The fairly unobvious coding in __kill_pg_info() will cause a zero value to be
incorrectly returned if the second or succeeding call to
group_send_sig_info() returns an error.

From: Olaf Hering <olh@suse.de>

This patch is needed on s390.

From: James Bottomley <James.Bottomley@SteelEye.com>

pci_dev.consistent_dma_mask was introduced to get around problems in the
IA64 Altix machine.

Now, we have a use for it in x86: the aacraid needs coherent memory in a
31 bit address range (2GB).  Unfortunately, x86 is converted to the dma
model, so it can't see the pci_dev by the time coherent memory is
allocated.

The solution to all of this is to move pci_dev.consistent_dma_mask to
dev.coherent_dma_mask and make x86 use it in the dma_alloc_coherent()
calls.

This should allow me to make the aacraid set the coherent mask instead
of using it's current dma_mask juggling.

Updates for 2.6.4 for PARISC:

 - defconfigs (Randolph Chung)
 - copyright updates (Grant Grundler)
 - Fix DOS hole in sys_clone (James Bottomley)
 - missing hardware ID (Andy Walker)
 - disable interrupts during cache-flushes (LaMont Jones)
 - Fix crash on machines with <256MB ram (Randolph Chung)
 - Make SuckyIO IDE work better (Randolph Chung)
 - Align data_start so the extable is writable (Randolph Chung)
 - Extensive rewrite of virtual merging code (James Bottomley)
 - Fix EISA, non-PCI module builds (Matthew Wilcox)
 - Fix Elroy PCI config space byte & word writes (Grant Grundler)
 - Eliminate a warning in parport_gsc (Helge Deller)
 - Fix endian problem with ide mmio macros (Randolph Chung)
 - Delete asm/keyboard.h (Matthew Wilcox)
 - Delete asm/md.h (Grant Grundler)
 - Eliminate a warning in ALSA harmony (Matthew Wilcox)

into ppc970.osdl.org:/home/torvalds/v2.5/linux

into ppc970.osdl.org:/home/torvalds/v2.5/linux

into ppc970.osdl.org:/home/torvalds/v2.5/linux

drivers/scsi/sata_vsc.c: In function `vsc_sata_interrupt':
include/linux/libata.h:414: sorry, unimplemented: inlining failed in call to 'ata_host_intr': function body not available
drivers/scsi/sata_vsc.c:187: sorry, unimplemented: called from here

From: Rusty Russell <rusty@rustcorp.com.au>

From: Andrea Arcangeli <andrea@suse.de>

We should drop module sem before calling mod->exit, for practical reasons:
too many module exit functions oops or hang, resulting in a permenantly held
module sem, which blocks all module ops including lsmod.

From: "David S. Miller" <davem@redhat.com>

Currently, for an existing DMA mapping, there is a way to transfer buffer
ownership back to the cpu, yet there is no way to give it back to the device
again explicitly.  The latter really is needed on platforms where the PCI
subsystem does not snoop the cpu caches, MIPS is one example.

Many drivers were expecting the existing DMA sync interface to handle both
directions, which was wrong.

Now, with this change, we have explicit interfaces for DMA syncing to/from
the device and the cpu.