move scsi headers

James, please do a

bk pull bk://kernel.bkbits.net/jgarzik/scsi-misc-2.5

This will update the following files:

 drivers/scsi/hosts.h          |  603 ------------------------------------
 drivers/scsi/scsi.h           |  690 -----------------------------------------
 drivers/scsi/scsi_obsolete.h  |  106 ------
 drivers/scsi/hosts.h          |    6 
 drivers/scsi/scsi.h           |    6 
 drivers/scsi/scsi_obsolete.h  |    6 
 drivers/scsi/st.h             |    4 
 include/linux/scsi_defs.h     |  692 +++++++++++++++++++++++++++++++++++++++++-
 include/linux/scsi_hosts.h    |  603 ++++++++++++++++++++++++++++++++++++
 include/linux/scsi_obsolete.h |  106 ++++++
 10 files changed, 1419 insertions(+), 1403 deletions(-)

through these ChangeSets:

<[email protected]> (03/05/26 1.1195)
[scsi] add drivers/scsi/{scsi,hosts,scsi_obsolete}.h wrappers

Also, remove "#ifndef _SCSI_H" from st.h, and include linux/scsi_obsolete.h rather than "scsi_obsolete.h" in linux/scsi_defs.h.

<[email protected]> (03/05/26 1.1194)
[scsi] mv drv/scsi/{scsi,hosts,scsi_obsolete}.h to include/linux

New names:
include/linux/scsi_hosts.h
include/linux/scsi_defs.h
include/linux/scsi_obsolete.h

Also I already told you that the header organisation is crap. include/scsi exists for a reason, the scsi.h/host.h split is broken and scsi_obsolete.h should go away not moved around.

If you want to do something useful you could start removing scsi_obsolete.h, make the scsi core and scsi.h/hosts.h clean of the obsolete typedefs first.

This is too ambitious this close to 2.6-pre, and there is also disagreement about use of include/scsi. I took the minimal patch.

As we are very close to 2.6-pre, your suggestions may even want waiting until 2.7. I'll leave it up to James...

Consider this a first step. The world does not change in a day :)

Jeff

No, it's not. Eihet we do it properly or not at all. Having to use different headers for 2.4,2.6and 2.7/2.8 is just extremly stupid.

Also we have more intrusive stuff in scsi pending, 2.6.pre or not.

I'm going to send my patch to Marcelo too, since it doesn't break anything, and greatly reduces include annoyance for in-tree drivers outside drivers/scsi.

We'll see what akpm says...

Jeff