openwrt/staging/blogic.git
19 years ago[PATCH] pcmcia: id_table for bluecard_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:35 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for bluecard_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for avm_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:35 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for avm_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for avma1_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:34 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for avma1_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for atmel_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:34 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for atmel_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for vxpocket.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:33 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for vxpocket.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for pcmciamtd.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:32 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for pcmciamtd.c

Add pcmcia_device_id table to pcmciamtd.  The binding of anonymus cards (i.e.
those who do neither report MANFID, CARDID, FUNCID nor product strings) is
protected by a new config option.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: add a few more IDs for pcnet_cs
Jun Komuro [Mon, 27 Jun 2005 23:28:30 +0000 (16:28 -0700)]
[PATCH] pcmcia: add a few more IDs for pcnet_cs

Add new pcmcia id_table for pcnet_cs.  (Allied, Laneed, NextCom and Telecom
Device)

Signed-off-by: Jun Komuro <komurojun-mbn@nifty.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for pcnet_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:30 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for pcnet_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for parport_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:29 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for parport_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: more IDs for ide_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:29 +0000 (16:28 -0700)]
[PATCH] pcmcia: more IDs for ide_cs

Add another ID for ide-cs

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for ide_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:28 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for ide_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for xirc2ps_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:28 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for xirc2ps_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for orinoco_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:27 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for orinoco_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for smc91c92_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:27 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for smc91c92_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for qlogic_stub.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:26 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for qlogic_stub.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for sym53c500_cs.c
Dominik Brodowski [Mon, 27 Jun 2005 23:28:25 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for sym53c500_cs.c

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for wavelan_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:25 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for wavelan_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for ray_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:24 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for ray_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for teles_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:24 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for teles_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for nmclan_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:23 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for nmclan_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for netwave_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:23 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for netwave_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for ibmtr_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:22 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for ibmtr_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for fmvj18x_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:22 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for fmvj18x_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for fdomain_stub
Dominik Brodowski [Mon, 27 Jun 2005 23:28:21 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for fdomain_stub

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for axnet_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:20 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for axnet_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for airo_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:20 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for airo_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for aha152x
Dominik Brodowski [Mon, 27 Jun 2005 23:28:19 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for aha152x

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for 3c589_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:19 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for 3c589_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for 3c574_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:18 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for 3c574_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: id_table for serial_cs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:18 +0000 (16:28 -0700)]
[PATCH] pcmcia: id_table for serial_cs

Add pcmcia_device_id table.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: rescan bus always upon echoing into setup_done
Dominik Brodowski [Mon, 27 Jun 2005 23:28:17 +0000 (16:28 -0700)]
[PATCH] pcmcia: rescan bus always upon echoing into setup_done

Always rescan the devices upon echo'ing something to
available_resources_setup_done.  This is needed for proper "coldplug" support.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: cleanups
Dominik Brodowski [Mon, 27 Jun 2005 23:28:15 +0000 (16:28 -0700)]
[PATCH] pcmcia: cleanups

From: Adrian Bunk <bunk@stusta.de>

This patch contains the following cleanups:
- make needlessly global code static
- remove the following unneeded EXPORT_SYMBOL's:
  - ds.c: pcmcia_report_error
  - ds.c: pcmcia_bus_type

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: request CIS via firmware interface
Dominik Brodowski [Mon, 27 Jun 2005 23:28:14 +0000 (16:28 -0700)]
[PATCH] pcmcia: request CIS via firmware interface

Use the firmware method to load replacement CIS tables.  It is recommended
that the /lib/firmware/cis/ points to /etc/pcmcia/cis or the other way round
so that both old-style cardmgr and new-style hotplug/firmware can access these
"overwrite" files

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: file2alias
Dominik Brodowski [Mon, 27 Jun 2005 23:28:12 +0000 (16:28 -0700)]
[PATCH] pcmcia: file2alias

Create PCMCIA entries in modules.alias

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: allow function-ID based match
Dominik Brodowski [Mon, 27 Jun 2005 23:28:10 +0000 (16:28 -0700)]
[PATCH] pcmcia: allow function-ID based match

The "func_id"-based matching is very fuzzy and can lead to false positives.
Therefore, it should be tried to avoid relying on these matches.  Until
most/all existing func_id-based matches are replaced by
manf_id/card_id/prod_id matches (a patch which will ask to send the
appropriate card information to the PCMCIA mailing list will be added once
other, more pressing issues are adressed), we need to emulate cardmgr
behaviour by allowing func_id matches if no manf_id/card_id/prod_id match
occurs.  This can only be done in userspace because of modules possibly loaded
with long delays.  So, add a per-device sysfs file for this purpose.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Brice Goglin <Brice.Goglin@ens-lyon.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: match "anonymous" cards
Dominik Brodowski [Mon, 27 Jun 2005 23:28:09 +0000 (16:28 -0700)]
[PATCH] pcmcia: match "anonymous" cards

If a card doesn't provide _any_ information about itself, assume it is a
so-called "anonymous" card.  pcmciamtd will bind to it if it is configured to
do so.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: CIS overrid via sysfs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:09 +0000 (16:28 -0700)]
[PATCH] pcmcia: CIS overrid via sysfs

The one thing which surprises me in this patch that cis->Length needs to be
set to count+1.  Without it, it doesn't work, but with it, it doesn't make
sense to me.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: export CIS in sysfs
Dominik Brodowski [Mon, 27 Jun 2005 23:28:08 +0000 (16:28 -0700)]
[PATCH] pcmcia: export CIS in sysfs

Export the CIS to userspace using a sysfs binary file in
/sys/class/pcmcia_socket/pcmcia_scoket%n/

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: match for fake CIS
Dominik Brodowski [Mon, 27 Jun 2005 23:28:07 +0000 (16:28 -0700)]
[PATCH] pcmcia: match for fake CIS

Add another match flag for devices needing a CIS override.  The driver will
only probe/attach if the CIS has been replaced before.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: check for invalid crc32 hashes in id_tables
Dominik Brodowski [Mon, 27 Jun 2005 23:28:07 +0000 (16:28 -0700)]
[PATCH] pcmcia: check for invalid crc32 hashes in id_tables

Check for invalid crc32 hashes in drivers' id_tables if CONFIG_PCMCIA_DEBUG is
set.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: device and driver matching
Dominik Brodowski [Mon, 27 Jun 2005 23:28:06 +0000 (16:28 -0700)]
[PATCH] pcmcia: device and driver matching

The actual matching of pcmcia drivers and pcmcia devices.  The original
version of this was written by David Woodhouse.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: hotplug event for PCMCIA socket devices
Dominik Brodowski [Mon, 27 Jun 2005 23:28:05 +0000 (16:28 -0700)]
[PATCH] pcmcia: hotplug event for PCMCIA socket devices

Generate hotplug event for PCMCIA sockets being registered.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pcmcia: hotplug event for PCMCIA devices
Dominik Brodowski [Mon, 27 Jun 2005 23:28:04 +0000 (16:28 -0700)]
[PATCH] pcmcia: hotplug event for PCMCIA devices

Export information to /sbin/hotplug for PCMCIA devices: card_id, manf_id,
func_id, bus_id (like pcmcia1.0) and crc32-hashes of the prod_id strings.

Why not the prod_id strings themselves?

a) They may contain all sorts of strange and difficult to handle characters,
   like " ".

b) It's impossible to pass multiple strings to userspace.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pci: yenta cardbus fix
Ivan Kokshaysky [Mon, 27 Jun 2005 23:28:02 +0000 (16:28 -0700)]
[PATCH] pci: yenta cardbus fix

On Mon, Jun 20, 2005 at 07:15:34PM +1000, Grant Coady wrote:
> Yenta: CardBus bridge found at 0000:00:0b.0 [1179:0001]
> yenta 0000:00:0b.0: Preassigned resource 0 busy, reconfiguring...

In -mm1 the cardbus resources might be assigned in
pci_assign_unassigned_resources() pass. From your dmesg:
PCI: Bus 2, cardbus bridge: 0000:00:0b.0
  IO window: 00002000-00002fff
  IO window: 00003000-00003fff
  PREFETCH window: 12000000-13ffffff
  MEM window: 14000000-15ffffff

Then yenta_allocate_res() tries to assign these resources again and,
naturally, fails.

This adds check for already assigned cardbus resources.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoCommit the manual part of the input layer merge.
Linus Torvalds [Tue, 28 Jun 2005 00:49:45 +0000 (17:49 -0700)]
Commit the manual part of the input layer merge.

git did actually warn me about the fact that I hadn't actually done an
"update-cache" on these two files, but the warning was at the bottom of
a list of all the files that _did_ change in the merge, so I never
noticed.  My bad.

19 years ago[PATCH] ide: sensible probing for PCI systems
Alan Cox [Mon, 27 Jun 2005 22:24:31 +0000 (15:24 -0700)]
[PATCH] ide: sensible probing for PCI systems

Old ISA/VESA systems sometimes put tertiary IDE controllers at addresses
0x1e8, 0x168, 0x1e0 or 0x160.  Linux thus probes these addresses on x86
systems.  Unfortunately some PCI systems now use these addresses for other
purposes which leads to users seeing minute plus hangs during boot or even
crashes.

The following patch (again has been in Fedora for a while) only probes the
obscure legacy ISA ports on machinea that are pre-PCI.  This seems to keep
everyone happy and if there is someone with that utterly weird corner case
the ide= command line still provides a get out of jail card.
Unsurprisingly we've not found anyone so affected.

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: it8212 backport for Bartlomiej IDE
Alan Cox [Mon, 27 Jun 2005 22:24:30 +0000 (15:24 -0700)]
[PATCH] ide: it8212 backport for Bartlomiej IDE

This lets you throw out the iteraid stuff that has ended up back in due
to stupid goings on in the IDE world. Its the same heavily tested code
shipped in Fedora/Red Hat products but without the other dependancies on
the Bartlomiej IDE layer.

Pre-requisite: the ide-disk patch I sent to handle pure LBA devices.

Obviously you lose things like hot unplug with the Bartlomiej IDE layer
at the moment but that won't matter to most users.

The patch does the following
- Add IT8211/12 to pci_ids.h
- Add Makefile/Kconfig entry
- Add it8212 driver

No core IDE code is touched by this diff

Embedded system testing and the ability to force raid mode off by David
Howells

Made possible by the ite reference code, documentation and also several
clarifications and pieces of assistance provided by ITE themselves

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: fix crashes with hotplug serverworks
Alan Cox [Mon, 27 Jun 2005 22:24:29 +0000 (15:24 -0700)]
[PATCH] ide: fix crashes with hotplug serverworks

You can't install the base kernel on a Stratus box because of the overuse of
__init.  Affects both IDE layers identically.  It isn't the only misuser of
__init so more review of other drivers (or fixing ide_register code to know
about hotplug v non-hotplug chipsets) would be good.

Original issue found by Stratus and their patch was the inspiration for this
trivial one.

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: fix the HPT366 driver layer
Alan Cox [Mon, 27 Jun 2005 22:24:27 +0000 (15:24 -0700)]
[PATCH] ide: fix the HPT366 driver layer

The highpoint driver is unreadable, buggy and crashes on some chipsets.  The
-ac one is more readable (but not ideal) and doesn't crash all over the place.
 Been in Fedora for some time.

Backported from the Fedora one to the old Bartlomiej IDE core.  No other
dependencies.

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: ide-generic, allow for capture of other unsupported devices
Alan Cox [Mon, 27 Jun 2005 22:24:26 +0000 (15:24 -0700)]
[PATCH] ide: ide-generic, allow for capture of other unsupported devices

The ide-generic driver gives you DMA at bios tuned speed so can actually run a
lot of unsupported devices quite well.  It has a pci table so that it doesn't
grab disks owned by other drivers but no way to override this.  The patch adds
an option ide-generic-all which makes the driver grab everything going that is
IDE class.

The diff is messy because I put the special case as case 0 to make the if
conditional and long term maintenance easier.

This has been in Fedora for some time.

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: IDE timing violation on reset
Alan Cox [Mon, 27 Jun 2005 22:24:25 +0000 (15:24 -0700)]
[PATCH] ide: IDE timing violation on reset

Pretty much theoretical for non MMIO thankfully.  We _must_ use OUTBSYNC for
commands or they may be posted and thus ruin the 400nS required delay.

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: samsung SN-124 works perfectly well with DMA
Alan Cox [Mon, 27 Jun 2005 22:24:24 +0000 (15:24 -0700)]
[PATCH] ide: samsung SN-124 works perfectly well with DMA

Been in Red Hat products for ages

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ide: fix ide-disk inability to handle LBA only devices.
Alan Cox [Mon, 27 Jun 2005 22:24:22 +0000 (15:24 -0700)]
[PATCH] ide: fix ide-disk inability to handle LBA only devices.

Years old bug, has to be fixed for it8212 to work

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kprobes/ia64: refuse kprobe on ivt code
Keshavamurthy Anil S [Mon, 27 Jun 2005 22:17:16 +0000 (15:17 -0700)]
[PATCH] kprobes/ia64: refuse kprobe on ivt code

Not safe to insert kprobes on IVT code.

This patch checks to see if the address on which Kprobes is being inserted is
in ivt code and if it is in ivt code then refuse to register kprobe.

Signed-off-by: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Acked-by: David Mosberger <davidm@napali.hpl.hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kprobes/ia64: refuse inserting kprobe on slot 1
Rusty Lynch [Mon, 27 Jun 2005 22:17:15 +0000 (15:17 -0700)]
[PATCH] kprobes/ia64: refuse inserting kprobe on slot 1

Without the ability to atomically write 16 bytes, we can not update the
middle slot of a bundle, slot 1, unless we stop the machine first.  This
patch will ensure the ability to robustly insert and remove a kprobe by
refusing to insert a kprobe on slot 1 until a mechanism is in place to
safely handle this case.

Signed-off-by: Rusty Lynch <rusty.lynch@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Return probe redesign: ppc64 specific implementation
Rusty Lynch [Mon, 27 Jun 2005 22:17:15 +0000 (15:17 -0700)]
[PATCH] Return probe redesign: ppc64 specific implementation

The following is a patch provided by Ananth Mavinakayanahalli that implements
the new PPC64 specific parts of the new function return probe design.

NOTE: Since getting Ananth's patch, I changed trampoline_probe_handler()
      to consume each of the outstanding return probem instances (feedback
      on my original RFC after Ananth cut a patch), and also added the
      arch_init() function (adding arch specific initialization.) I have
      cross compiled but have not testing this on a PPC64 machine.

Changes include:
 * Addition of kretprobe_trampoline to act as a dummy function for instrumented
   functions to return to, and for the return probe infrastructure to place
   a kprobe on on, gaining control so that the return probe handler
   can be called, and so that the instruction pointer can be moved back
   to the original return address.
 * Addition of arch_init(), allowing a kprobe to be registered on
   kretprobe_trampoline
 * Addition of trampoline_probe_handler() which is used as the pre_handler
   for the kprobe inserted on kretprobe_implementation.  This is the function
   that handles the details for calling the return probe handler function
   and returning control back at the original return address
 * Addition of arch_prepare_kretprobe() which is setup as the pre_handler
   for a kprobe registered at the beginning of the target function by
   kernel/kprobes.c so that a return probe instance can be setup when
   a caller enters the target function.  (A return probe instance contains
   all the needed information for trampoline_probe_handler to do it's job.)
 * Hooks added to the exit path of a task so that we can cleanup any left-over
   return probe instances (i.e. if a task dies while inside a targeted function
   then the return probe instance was reserved at the beginning of the function
   but the function never returns so we need to mark the instance as unused.)

Signed-off-by: Rusty Lynch <rusty.lynch@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Return probe redesign: ia64 specific implementation
Rusty Lynch [Mon, 27 Jun 2005 22:17:12 +0000 (15:17 -0700)]
[PATCH] Return probe redesign: ia64 specific implementation

The following patch implements function return probes for ia64 using
the revised design.  With this new design we no longer need to do some
of the odd hacks previous required on the last ia64 return probe port
that I sent out for comments.

Note that this new implementation still does not resolve the problem noted
by Keith Owens where backtrace data is lost after a return probe is hit.

Changes include:
 * Addition of kretprobe_trampoline to act as a dummy function for instrumented
   functions to return to, and for the return probe infrastructure to place
   a kprobe on on, gaining control so that the return probe handler
   can be called, and so that the instruction pointer can be moved back
   to the original return address.
 * Addition of arch_init(), allowing a kprobe to be registered on
   kretprobe_trampoline
 * Addition of trampoline_probe_handler() which is used as the pre_handler
   for the kprobe inserted on kretprobe_implementation.  This is the function
   that handles the details for calling the return probe handler function
   and returning control back at the original return address
 * Addition of arch_prepare_kretprobe() which is setup as the pre_handler
   for a kprobe registered at the beginning of the target function by
   kernel/kprobes.c so that a return probe instance can be setup when
   a caller enters the target function.  (A return probe instance contains
   all the needed information for trampoline_probe_handler to do it's job.)
 * Hooks added to the exit path of a task so that we can cleanup any left-over
   return probe instances (i.e. if a task dies while inside a targeted function
   then the return probe instance was reserved at the beginning of the function
   but the function never returns so we need to mark the instance as unused.)

Signed-off-by: Rusty Lynch <rusty.lynch@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Return probe redesign: x86_64 specific changes
Rusty Lynch [Mon, 27 Jun 2005 22:17:10 +0000 (15:17 -0700)]
[PATCH] Return probe redesign: x86_64 specific changes

The following patch contains the x86_64 specific changes for the new
return probe design.  Changes include:
 * Removing the architecture specific functions for querying a return probe
   instance off a stack address
 * Complete rework onf arch_prepare_kretprobe() and trampoline_probe_handler()
 * Removing trampoline_post_handler()
 * Adding arch_init() so that now we handle registering the return probe
   trampoline instead of kernel/kprobes.c doing it

NOTE:
Note that with this new design, the dependency on calculating a pointer to
the task off the stack pointer no longer exist (resolving the problem of
interruption stacks as pointed out in the original feedback to this port.)

Signed-off-by: Rusty Lynch <rusty.lynch@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Return probe redesign: i386 specific changes
Rusty Lynch [Mon, 27 Jun 2005 22:17:09 +0000 (15:17 -0700)]
[PATCH] Return probe redesign: i386 specific changes

The following patch contains the i386 specific changes for the new
return probe design.  Changes include:

 * Removing the architecture specific functions for querying a return probe
   instance off a stack address
 * Complete rework onf arch_prepare_kretprobe() and trampoline_probe_handler()
 * Removing trampoline_post_handler()
 * Adding arch_init() so that now we handle registering the return probe
   trampoline instead of kernel/kprobes.c doing it

Signed-off-by: Rusty Lynch <rusty.lynch@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Return probe redesign: architecture independent changes
Rusty Lynch [Mon, 27 Jun 2005 22:17:08 +0000 (15:17 -0700)]
[PATCH] Return probe redesign: architecture independent changes

The following is the second version of the function return probe patches
I sent out earlier this week.  Changes since my last submission include:

* Fix in ppc64 code removing an unneeded call to re-enable preemption
* Fix a build problem in ia64 when kprobes was turned off
* Added another BUG_ON check to each of the architecture trampoline
  handlers

My initial patch description ==>

 From my experiences with adding return probes to x86_64 and ia64, and the
feedback on LKML to those patches, I think we can simplify the design
for return probes.

The following patch tweaks the original design such that:

* Instead of storing the stack address in the return probe instance, the
  task pointer is stored.  This gives us all we need in order to:
    - find the correct return probe instance when we enter the trampoline
      (even if we are recursing)
    - find all left-over return probe instances when the task is going away

  This has the side effect of simplifying the implementation since more
  work can be done in kernel/kprobes.c since architecture specific knowledge
  of the stack layout is no longer required.  Specifically, we no longer have:
- arch_get_kprobe_task()
- arch_kprobe_flush_task()
- get_rp_inst_tsk()
- get_rp_inst()
- trampoline_post_handler() <see next bullet>

* Instead of splitting the return probe handling and cleanup logic across
  the pre and post trampoline handlers, all the work is pushed into the
  pre function (trampoline_probe_handler), and then we skip single stepping
  the original function.  In this case the original instruction to be single
  stepped was just a NOP, and we can do without the extra interruption.

The new flow of events to having a return probe handler execute when a target
function exits is:

* At system initialization time, a kprobe is inserted at the beginning of
  kretprobe_trampoline.  kernel/kprobes.c use to handle this on it's own,
  but ia64 needed to do this a little differently (i.e. a function pointer
  is really a pointer to a structure containing the instruction pointer and
  a global pointer), so I added the notion of arch_init(), so that
  kernel/kprobes.c:init_kprobes() now allows architecture specific
  initialization by calling arch_init() before exiting.  Each architecture
  now registers a kprobe on it's own trampoline function.

* register_kretprobe() will insert a kprobe at the beginning of the targeted
  function with the kprobe pre_handler set to arch_prepare_kretprobe
  (still no change)

* When the target function is entered, the kprobe is fired, calling
  arch_prepare_kretprobe (still no change)

* In arch_prepare_kretprobe() we try to get a free instance and if one is
  available then we fill out the instance with a pointer to the return probe,
  the original return address, and a pointer to the task structure (instead
  of the stack address.)  Just like before we change the return address
  to the trampoline function and mark the instance as used.

  If multiple return probes are registered for a given target function,
  then arch_prepare_kretprobe() will get called multiple times for the same
  task (since our kprobe implementation is able to handle multiple kprobes
  at the same address.)  Past the first call to arch_prepare_kretprobe,
  we end up with the original address stored in the return probe instance
  pointing to our trampoline function. (This is a significant difference
  from the original arch_prepare_kretprobe design.)

* Target function executes like normal and then returns to kretprobe_trampoline.

* kprobe inserted on the first instruction of kretprobe_trampoline is fired
  and calls trampoline_probe_handler() (no change here)

* trampoline_probe_handler() consumes each of the instances associated with
  the current task by calling the registered handler function and marking
  the instance as unused until an instance is found that has a return address
  different then the trampoline function.

  (change similar to my previous ia64 RFC)

* If the task is killed with some left-over return probe instances (meaning
  that a target function was entered, but never returned), then we just
  free any instances associated with the task.  (Not much different other
  then we can handle this without calling architecture specific functions.)

  There is a known problem that this patch does not yet solve where
  registering a return probe flush_old_exec or flush_thread will put us
  in a bad state.  Most likely the best way to handle this is to not allow
  registering return probes on these two functions.

  (Significant change)

This patch series applies to the 2.6.12-rc6-mm1 kernel, and provides:
  * kernel/kprobes.c changes
  * i386 patch of existing return probes implementation
  * x86_64 patch of existing return probe implementation
  * ia64 implementation
  * ppc64 implementation (provided by Ananth)

This patch implements the architecture independant changes for a reworking
of the kprobes based function return probes design. Changes include:

  * Removing functions for querying a return probe instance off a stack address
  * Removing the stack_addr field from the kretprobe_instance definition,
    and adding a task pointer
  * Adding architecture specific initialization via arch_init()
  * Removing extern definitions for the architecture trampoline functions
    (this isn't needed anymore since the architecture handles the
     initialization of the kprobe in the return probe trampoline function.)

Signed-off-by: Rusty Lynch <rusty.lynch@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kprobes: fix single-step out of line - take2
Ananth N Mavinakayanahalli [Mon, 27 Jun 2005 22:17:01 +0000 (15:17 -0700)]
[PATCH] kprobes: fix single-step out of line - take2

Now that PPC64 has no-execute support, here is a second try to fix the
single step out of line during kprobe execution.  Kprobes on x86_64 already
solved this problem by allocating an executable page and using it as the
scratch area for stepping out of line.  Reuse that.

Signed-off-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Mon, 27 Jun 2005 22:13:26 +0000 (15:13 -0700)]
Merge /pub/scm/linux/kernel/git/gregkh/usb-2.6

19 years ago[PATCH] cciss: remove partition info from CCISS_GETLUNINFO
Mike Miller [Mon, 27 Jun 2005 21:36:50 +0000 (14:36 -0700)]
[PATCH] cciss: remove partition info from CCISS_GETLUNINFO

This patch fulfills a promise I made to Christoph sometime back.  I am
removing the partition info from the CCISS_GETLUNINFO ioctl as I was informed
my "driver had no damn business reading that structure." ;)

The application folks are to use /proc or /sys for partition info from now on.
 I am only aware of a few apps that use this ioctl and I'm not sure they ever
used the partition info.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cciss: pci domain info pass 2
Mike Miller [Mon, 27 Jun 2005 21:36:49 +0000 (14:36 -0700)]
[PATCH] cciss: pci domain info pass 2

This is pass 2 of my patch to add pci domain info to an existing ioctl.  This
time I insert the domain between dev_fn and board_id as Willy suggested and
change the var to unsigned short to ease Christoph's concerns.  Although I
thought unsigned int was the correct var type for this.  I also thought it
didn't matter where I inserted it in the structure.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] cciss: pci id fix
Mike Miller [Mon, 27 Jun 2005 21:36:48 +0000 (14:36 -0700)]
[PATCH] cciss: pci id fix

This patch fixes a PCI ID I got wrong before.  It also adds support for
another new SAS controller due out this summer.  I didn't have a marketing
name prior to my last submission.  Also modifies the copyright date range.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] MAINTAINERS: Update Roland Dreier's email
Roland Dreier [Mon, 27 Jun 2005 21:36:47 +0000 (14:36 -0700)]
[PATCH] MAINTAINERS: Update Roland Dreier's email

Cisco bought Topspin, so I'm now a shiny happy Cisco employee.  Update my
entry in MAINTAINERS.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB: Fix pack/unpack when size_bits == 64
Roland Dreier [Mon, 27 Jun 2005 21:36:47 +0000 (14:36 -0700)]
[PATCH] IB: Fix pack/unpack when size_bits == 64

Fix handling of fields with size_bits == 64.  Pointed out by Hal Rosenstock.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB: Fix race in sa_query
Roland Dreier [Mon, 27 Jun 2005 21:36:46 +0000 (14:36 -0700)]
[PATCH] IB: Fix race in sa_query

Use a copy of the id we'll return to the consumer so that we don't
dereference query->sa_query after calling send_mad().  A completion may
occur very quickly and end up freeing the query before we get to do
anything after send_mad().

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Bump version
Roland Dreier [Mon, 27 Jun 2005 21:36:46 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Bump version

It's about time for a version bump.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Align FW command mailboxes to 4K
Roland Dreier [Mon, 27 Jun 2005 21:36:45 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Align FW command mailboxes to 4K

Future versions of Mellanox HCA firmware will require command mailboxes to be
aligned to 4K.  Support this by using a pci_pool to allocate all mailboxes.
This has the added benefit of shrinking the source and text of mthca.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Encapsulate command interface init
Roland Dreier [Mon, 27 Jun 2005 21:36:45 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Encapsulate command interface init

Encapsulate mthca command interface initialization/cleanup.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Fix memory leak on error path
Roland Dreier [Mon, 27 Jun 2005 21:36:44 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Fix memory leak on error path

Free page_list buffer on error path of mthca_reg_phys_mr().

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Split off MTT allocation
Roland Dreier [Mon, 27 Jun 2005 21:36:43 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Split off MTT allocation

Split allocation of MTT range from creation of MR.  This will be useful for
implementing shared memory regions and userspace verbs.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Move mthca_is_memfree checks
Roland Dreier [Mon, 27 Jun 2005 21:36:43 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Move mthca_is_memfree checks

Make mthca_table_put() and mthca_table_put_range() NOPs if the device is not
mem-free, so that we don't have to have "if (mthca_is_memfree())" tests in the
callers of these functions.  This makes our code more readable and
maintainable, and saves a couple dozen bytes of text in ib_mthca.ko as well.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Fix memset size
Roland Dreier [Mon, 27 Jun 2005 21:36:42 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Fix memset size

Fix memset to use sizeof *props instead of just sizeof props.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Enable unreliable connected transport
Roland Dreier [Mon, 27 Jun 2005 21:36:42 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Enable unreliable connected transport

Add support for unreliable connected (UC) transport to mthca driver:
 - Add attributes for UC to modify QP table.
 - Add support for posting UC work requests.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Set RDMA/atomic capabilities correctly
Roland Dreier [Mon, 27 Jun 2005 21:36:41 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Set RDMA/atomic capabilities correctly

mthca apparently had the meanings of the max_rd_atomic and max_dest_rd_atomic
QP attributes backwards.  max_rd_atomic limits the maximum number of
outstanding RDMA/atomic requests as an initiator (on a send queue), and
max_dest_rd_atomic specifies the resources allocated to handle RMDA/atomic
requests from the remote end of the connection.  We were programming our QP
context with these values swapped.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Set QP static rate correctly
Roland Dreier [Mon, 27 Jun 2005 21:36:40 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Set QP static rate correctly

Fix offset of static_rate in QP context.  Pointed out by Dror Goldenberg.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Use dma_alloc_coherent instead of pci_alloc_consistent
Roland Dreier [Mon, 27 Jun 2005 21:36:40 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Use dma_alloc_coherent instead of pci_alloc_consistent

Switch all allocations of coherent memory from pci_alloc_consistent() to
dma_alloc_coherent(), so that we can pass GFP_KERNEL.  This should help when
the system is low on memory.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Clean up CQ debug
Roland Dreier [Mon, 27 Jun 2005 21:36:39 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Clean up CQ debug

Clean up CQ debugging code: make dump_cqe print on one line, and only dump
error CQ entries for local operation errors.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Clean up error messages
Bernhard Fischer [Mon, 27 Jun 2005 21:36:39 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Clean up error messages

- Fix incorrect cut-n-paste in error messages.
- Add missing newlines in error messages.
- Use DRV_NAME instead of "ib_mthca" in a couple of places.

Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] IB/mthca: Add Sun copyright notice
Tom Duffy [Mon, 27 Jun 2005 21:36:37 +0000 (14:36 -0700)]
[PATCH] IB/mthca: Add Sun copyright notice

Add Sun copyright to files modified by Tom Duffy.

Signed-off-by: Tom Duffy <tduffy@sun.com>
Signed-off-by: Roland Dreier <roland@topspin.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] seccomp: tsc disable
Andrea Arcangeli [Mon, 27 Jun 2005 21:36:36 +0000 (14:36 -0700)]
[PATCH] seccomp: tsc disable

I believe at least for seccomp it's worth to turn off the tsc, not just for
HT but for the L2 cache too.  So it's up to you, either you turn it off
completely (which isn't very nice IMHO) or I recommend to apply this below
patch.

This has been tested successfully on x86-64 against current cogito
repository (i686 compiles so I didn't bother testing ;).  People selling
the cpu through cpushare may appreciate this bit for a peace of mind.

There's no way to get any timing info anymore with this applied
(gettimeofday is forbidden of course).  The seccomp environment is
completely deterministic so it can't be allowed to get timing info, it has
to be deterministic so in the future I can enable a computing mode that
does a parallel computing for each task with server side transparent
checkpointing and verification that the output is the same from all the 2/3
seller computers for each task, without the buyer even noticing (for now
the verification is left to the buyer client side and there's no
checkpointing, since that would require more kernel changes to track the
dirty bits but it'll be easy to extend once the basic mode is finished).

Eliminating a cold-cache read of the cr4 global variable will save one
cacheline during the tlb flush while making the code per-cpu-safe at the
same time.  Thanks to Mikael Pettersson for noticing the tlb flush wasn't
per-cpu-safe.

The global tlb flush can run from irq (IPI calling do_flush_tlb_all) but
it'll be transparent to the switch_to code since the IPI won't make any
change to the cr4 contents from the point of view of the interrupted code
and since it's now all per-cpu stuff, it will not race.  So no need to
disable irqs in switch_to slow path.

Signed-off-by: Andrea Arcangeli <andrea@cpushare.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: Add missing exports
Benjamin Herrenschmidt [Mon, 27 Jun 2005 21:36:35 +0000 (14:36 -0700)]
[PATCH] ppc64: Add missing exports

This patch adds a couple of missing symbol exports.  flush_dcache_page is
used by the AGP driver and rtc_lock by the RTC driver.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: Remove CONFIG_PMAC_PBOOK
Benjamin Herrenschmidt [Mon, 27 Jun 2005 21:36:34 +0000 (14:36 -0700)]
[PATCH] ppc32: Remove CONFIG_PMAC_PBOOK

This patch removes CONFIG_PMAC_PBOOK (PowerBook support).  This is now
split into CONFIG_PMAC_MEDIABAY for the actual hotswap bay that some
powerbooks have, CONFIG_PM for power management related code, and just left
out of any CONFIG_* option for some generally useful stuff that can be used
on non-laptops as well.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: remove obsolete macserial driver
Benjamin Herrenschmidt [Mon, 27 Jun 2005 21:36:33 +0000 (14:36 -0700)]
[PATCH] ppc32: remove obsolete macserial driver

The macserial driver has been obsoleted by the new pmac_zilog driver for a
while now and probably doesn't even work anymore on recent kernels.  This
patch removes it.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: Bump PMU interrupt priority
Benjamin Herrenschmidt [Mon, 27 Jun 2005 21:36:32 +0000 (14:36 -0700)]
[PATCH] ppc32: Bump PMU interrupt priority

The Power Management Unit on PowerMacs is very sensitive to timeouts during
async message exchanges.  It uses rather crude protocol based on a shift
register with an interrupt and is almost continuously exchanging messages with
the host CPU on laptops.

This patch adds a routine to the open_pic driver to be able to select a PMU
driver so that it bumps it's interrupt priority to above the normal level.

This will allow PMU interrupts to occur while another interrupt is pending,
and thus reduce the risk of machine beeing abruptly shutdown by the PMU due to
a timeout in PMU communication caused by excessive interrupt latency.  The
problem is very rare, and usually just doesn't happen, but it is still useful
to make things even more robust.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] 8xx: avoid "dcbst" misbehaviour with unpopulated TLB
Marcelo Tosatti [Mon, 27 Jun 2005 16:09:00 +0000 (13:09 -0300)]
[PATCH] 8xx: avoid "dcbst" misbehaviour with unpopulated TLB

The proposed _tlbie call at update_mmu_cache() is safe because:

Addresses for which update_mmu_cache() gets invocated are never inside the
static kernel virtual mapping, meaning that there is no risk for the
_tlbie() here to be thrashing the pinned entry, as Dan suspected.

The intermediate TLB state in which this bug can be triggered is not
visible by userspace or any other contexts, except the page fault handling
path.  So there is no need to worry about userspace dcbxxx users.

The other solution to this is to avoid dcbst misbehaviour in the first
place, which involves changing in-kernel "dcbst" callers to use 8xx
specific SPR's.

Summary:

On 8xx, cache control instructions (particularly "dcbst" from
flush_dcache_icache) fault as write operation if there is an unpopulated
TLB entry for the address in question.  To workaround that, we invalidate
the TLB here, thus avoiding dcbst misbehaviour.

Signed-off-by: Marcelo Tosatti <marcelo.tosatti@cyclades.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] mips: fixed try_to_freeze build error
Yoichi Yuasa [Mon, 27 Jun 2005 21:36:30 +0000 (14:36 -0700)]
[PATCH] mips: fixed try_to_freeze build error

arch/mips/kernel/signal.c: In function 'do_signal':
arch/mips/kernel/signal.c:460: error: too many arguments to function 'try_to_freeze'

Signed-off-by: Yoichi Yuasa <yuasa@hh.iij4u.or.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] fix WANT_PAGE_VIRTUAL in memmap_init
Bob Picco [Mon, 27 Jun 2005 21:36:28 +0000 (14:36 -0700)]
[PATCH] fix WANT_PAGE_VIRTUAL in memmap_init

I spotted this issue while in memmap_init last week.  I can't say the
change has any test coverage by me.  start_pfn was formerly used in main
"for" loop.  The fix is replace start_pfn with pfn.

Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: Fix compiling of sandpoint platform
Kumar Gala [Mon, 27 Jun 2005 21:36:16 +0000 (14:36 -0700)]
[PATCH] ppc32: Fix compiling of sandpoint platform

Lost a curly brace in translation.  Everything is better now.

Signed-off-by: Matt McClintock <msm@freescale.com>
Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-serial
Linus Torvalds [Mon, 27 Jun 2005 22:04:08 +0000 (15:04 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-serial

19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 27 Jun 2005 22:00:10 +0000 (15:00 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

19 years agoMerge 'upstream' branch of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Mon, 27 Jun 2005 21:55:50 +0000 (14:55 -0700)]
Merge 'upstream' branch of /linux/kernel/git/jgarzik/netdev-2.6

19 years agoMerge 'upstream' branch of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Mon, 27 Jun 2005 21:54:50 +0000 (14:54 -0700)]
Merge 'upstream' branch of /linux/kernel/git/jgarzik/libata-dev

19 years agoMerge 'upstream' branch of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Mon, 27 Jun 2005 21:53:48 +0000 (14:53 -0700)]
Merge 'upstream' branch of /linux/kernel/git/jgarzik/misc-2.6

19 years agoMerge rsync://rsync.kernel.org/pub/scm/linux/kernel/git/dtor/input.git manually
Linus Torvalds [Mon, 27 Jun 2005 21:47:31 +0000 (14:47 -0700)]
Merge /linux/kernel/git/dtor/input.git manually

Some manual fixups required due to clashes with the PF_FREEZE cleanups.

19 years ago[PATCH] USB: usbcore: inverted test for resuming interfaces
Alan Stern [Wed, 15 Jun 2005 19:49:48 +0000 (15:49 -0400)]
[PATCH] USB: usbcore: inverted test for resuming interfaces

This one-liner fixes a test for interfaces that are already resumed.

It would be nice if this could get into 2.6.12, but it's not critical
since it only affects people doing selective (runtime) suspend/resume.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: resolve ethernet gadget build glitch on pxa
David Brownell [Mon, 13 Jun 2005 13:55:03 +0000 (06:55 -0700)]
[PATCH] USB: resolve ethernet gadget build glitch on pxa

This fixes a build error on pxa25x processes with pxa2xx_udc and

        CONFIG_USB_ETH=m
        # CONFIG_USB_ETH_RNDIS is not set

The error is because on that CPU there's no status transfer support
except with RNDIS.  Workaround, enable the RNDIS support too.

Signed-off-by: Ian Campbell <icampbell@arcom.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] USB: usbnet debug message fix
David Brownell [Wed, 15 Jun 2005 15:04:30 +0000 (08:04 -0700)]
[PATCH] USB: usbnet debug message fix

One debug message won't print the right value; OSDL bugid 4545.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>