The mptctl module has had sparc64 32bit emulation support for some
time. This patch generalizes it to use the generic compat infrastructure.
This makes it work on x86-64 too.

I also removed one special case for sparc in the locking that didn't
make any sense to me.

 - DEFINE_PER_CPU works in modules now, use it
 - rename done_q to scsi_done_q - in the kernel we prefer descriptive
   prefixes even for statics..

I got underflow warnings and failures when sending an INQUIRY page 0x83
via the SG_IO REQ_BLOCK_PC interface.

The underflow field should only be set if we know for certain the size of
the data being transfered, so fix this by leaving it at zero for scsi
REQ_BLOCK_PC cases.

And while at it also fix the scsi_error_handler() prototype.

Looks like the merge error still isn't fixed..

We should try to keep all constants in include/scsi/scsi.h instead
of drivers/scsi/scsi.h.  Also tidy up include/scsi/scsi.h a bit.

 - move the struct parisc_driver down like in most drivers, avoids
   two forward-declarations
 - give lasi700_scsi_tbl and lasi700_driver_callback sensible names
 - properly return -ENOMEM instead of 1 on kmalloc failure
 - check scsi_add_host return value

everyone else uses struct scsi_lun already.

This patch adds shost_for_each_device().  It's used to abstract out
scsi_host.my_devices traversal.  The next step will be to replace
the current simple implementation with one that's fully locked down
an reference counted.

Rename zalon_scsi_callback to the more sensible zalon_probe and fix
a few tiny indentation issues.

This was only used by the old error handling, the new code uses .eh_state
and .eh_eflags instead.

The flag isn't checked ever as drivers using the block layer tcq
code do so by themselves, without any interaction to the scsi midlayer.

Jens' missed this in his cleanup.

For various reasons (e.g. the .my_devices lockdown) it's undesirable
to have scsi_device_{get,put} blocking.  Change .access_count to an
atomic_t to avoid the driver model r/w semaphore.

Sideeffects:

 - possible module reference leak in scsi_device_get when get_device
   fails fixes.
 - the access_count attribute is gone.  We already have this exposed
   to userspace in too many places given that this whole second
   refcount will go away once Al fixes the block layer.

into raven.il.steeleye.com:/home/jejb/BK/scsi-misc-2.5

From: Daniele Bellucci <bellucda@tiscali.it>

handle register error, call unregister;

 linux-260-test5-kj-rddunlap/drivers/scsi/sg.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletion(-)

From: Randy Hron <rwhron@earthlink.net>


 linux-260-test5-kj-rddunlap/drivers/scsi/53c700.c   |    1 -
 linux-260-test5-kj-rddunlap/drivers/scsi/BusLogic.c |    1 -
 linux-260-test5-kj-rddunlap/drivers/scsi/aha152x.c  |    1 -
 linux-260-test5-kj-rddunlap/drivers/scsi/dmx3191d.c |    1 -
 linux-260-test5-kj-rddunlap/drivers/scsi/i60uscsi.c |    1 -
 5 files changed, 5 deletions(-)

The updated SCSI proc directory code that I sent you a few months ago has
a small mistake -- the error path fails to release a semaphore.  This
patch fixes it; please apply.

The attached patch cleans up the ide-scsi driver model code and fixes
the oops described below which is caused when doing a rmmod on the
driver.

I don't actually have a zalon machine to test these on, but they seem
right to me, and compile fine.

Some cleanups for ncr53c8xx & zalon:

 - Inline zalon.h into zalon.c
 - Rationalise (a little) ncr53c8xx.c's includes
 - Remove all the version checks
 - Stop using remap_pci_mem & unmap_pci_mem & delete their definitions.
 - Use mb() instead of custom inline asm for MEMORY_BARRIER.

NCR5380: `phases' is used inside the #ifdef NDEBUG only, so move its definition
inside as well.

Passing the address of a `long' as the `flags' parameter of check_setup_args()
causes a warning. But apparently this parameter isn't used at all, so remove
it.

This patch fixes two bugs in sym 2.1.18.

 - 40-bit addressing doesn't fit into 32-bits, so PciDmaMask needs
   to be an ULL.
 - Add terminating entries to pci_device_id tables.  Bug spotted by
   Matt Domsch.

patch_name:	aha152x_debugon.patch
patch_version:	2003-09-10.11:02:09
author:		Randy.Dunlap <rddunlap@osdl.org>
description:	enable aha152x to build when AHA152X_DEBUG is defined;
product:	Linux
product_versions: 260-test5
maintainer:	James Bottomley
diffstat:	=
 drivers/scsi/aha152x.c |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

Change scsi_scan_mutex from global to per host, which is needed for
parallel SCSI probe.


 gr16_work-anton/drivers/scsi/hosts.c     |    2 ++
 gr16_work-anton/drivers/scsi/scsi_scan.c |   15 ++++-----------
 gr16_work-anton/include/scsi/scsi_host.h |    6 ++++++
 3 files changed, 12 insertions(+), 11 deletions(-)

From: Adrian Bunk <bunk@fs.tum.de>

Earlier patch wasn't correct especially in the !CONFIG_SCSI_IZIP_EPP16
case, reading all uses of this array (IMM_MODE_STRING is used to print the
corresponding string in printks).

If I'm not misunderstanding it, CONFIG_SCSI_IZIP_EPP16 means "use 16bit
even when 32bit is requested".

 drivers/scsi/imm.h |    3 ++-
 1 files changed, 2 insertions(+), 1 deletion(-)

Yes, there's a lot of changes here.  I've been busy :-P

 - Delete linux/version.h include as it's not necessary
 - Delete compatibility with ncr53c8xx configuration options
 - Convert to C99 initialisers
 - Delete the options: pci_parity, scsi_parity, min_sync, max_wide,
   max_offs, max_lun and pci_fix_up.  Some were unused, some are now
   unused.
 - Delete all the default settings from sym_conf.h that were unused.
 - Delete host_lock abstractions.
 - Replace custom jiffy handling with standard ones.
 - Replace u_long with unsigned long
 - Inline scsi_data_direction()
 - Use ioremap directly instead of indirecting through pci_map_mem.
 - Delete typedefs: bus_addr_t, ucmd_p, vm_offset_t, stcb_p,
   slcb_p, sccb_p, shcb_p, sym_device and sym_slot.
 - Use min() instead of MIN()
 - Use memcpy() instead of bcopy().
 - Delete code dependent on SYM_CONF_PCIQ_BROKEN_INTR as it's never defined.
 - Simplify the locking in sym_eh_handler().
 - Make the eh handlers static.
 - Use ctype.h instead of custom variants.
 - Use SYM_CONF_IOMAPPED and SYM_OPT_NO_BUS_MEMORY_MAPPING in a more
   consistent manner.
 - Remove check_region() and use pci_request_region instead.
 - Mark sym_detach() as __devexit.
 - Move many includes from sym_glue.h to sym_glue.c
 - Stop using Scsi_Cmnd.
 - Stop using ../hosts.h.
 - Simplify *_raw definitions.
 - Move PA-RISC specific code into its own function.

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

Update the SA11x0 serial driver for the removal of system devices
from the driver model.  The driver becomes a platform device driver.
The driver is now known to the driver model and resource subsytems
as "sa11x0-uart".

From: Oleg Drokin <green@namesys.com>

Without this, if elv_try_last_merge() returns nonzero, we do not
initialise *req.  That results in a subsequent BUG_ON() in
__make_request() because req is NULL (or is just uninitialised).

Last round of changes broke sdla build if not a module.

drivers whose microcode load ioctl's used types with sizes that
overflow the ioctl number. Mark them _IOx_BAD() to document the fact.

This new version works with all compiler versions, and also catches the
case where somebody tries to pass in an array as an argument to the
ioctl macros.  Thus we verify that only proper types are used.

I have checked now that the object code for arch/s390/kernel/compat_ioctl.o
remains identical and that the whole kernel compiles for s390 and i386,
after fixing the broken ioctl numbers.

The kernel invokes "modprobe" on modules which might not exist:
rightfully, modprobe complains by default when this happens.  So the
correct response is to invoke "modprobe -q", which is silent on such
errors (but still reports other errors such as config errors). 

Also, use MODULE_NAME_LEN from module.h instead of inventing our own.

Someone pointed out that -ffunction-sections can cause a function called
"init<something>" to be put in the init section, and discarded.  This
hurts PARISC badly.  Get more fussy with identifying them.

This adds CONFIG_64BIT to allow us to disable non-64-bit safe drivers on
64-bit platforms and converts the I2O driver to use it.

Nothing too exciting, just some minor new features and bug fixes.
The IOMMU code should now not BUG again without CONFIG_IOMMU_DEBUG/iommu=force

 - Compiles now with -funit-at-a-time
 - Fix up ioctl 32bit tables following Andrew's change.
 - Fix __SI_POLL siginfo_t passing to user space
 - Add 32bit emulation for fadvise64_64
 - Remove unneeded sys32_utimes
 - Various merges with i386 (ACPI, APIC etc.)
 - Port cpuid patching infrastructure from i386
 - Use it to provide better copy_*_user/memcpy/memset/clear/copy_page for
   C stepping K8. String instructions are faster now that unrolled loops.
 - Don't try to merge unforced PCI mappings that don't exceed the device's
   DMA mask.
 - Discard .exit.data/.exit.text at runtime like i386
 - Don't use NTI stores for clear_user
 - Convert bitops.h bitmap functions to use unsigned long * pointers instead
   of void *
 - Fix some warnings in kernel headers.
 - Fix PDA comments