deliverable/linux.git
19 years ago[IA64] sys_mmap doesn't follow posix.1 when parameter len=0
Zhang Yanmin [Mon, 23 May 2005 00:47:00 +0000 (17:47 -0700)] 
[IA64] sys_mmap doesn't follow posix.1 when parameter len=0

In IA64 kernel, sys_mmap calls do_mmap2 and do_mmap2 returns addr if
len=0, which means the mmap sys call succeeds.

Posix.1 says:
The mmap() function shall fail if:
[EINVAL] The value of len is zero.

Here is a patch to fix it.

Signed-off-by: Zhang Yanmin <yanmin.zhang@intel.com>
Acked-by: David Mosberger <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] initialize spinlock pfm_alt_install_check
Tony Luck [Thu, 19 May 2005 00:09:06 +0000 (17:09 -0700)] 
[IA64] initialize spinlock pfm_alt_install_check

I applied the penultimate version of the perfmon patch, which didn't have
the initialization of the new spinlock that was added.

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years agoSync with Linus - rsync://rsync.kernel.org/pub/scm/linux/kernel/git/torvalds/linux...
Tony Luck [Wed, 18 May 2005 23:28:52 +0000 (16:28 -0700)] 
Sync with Linus - /linux/kernel/git/torvalds/linux-2.6.git

19 years ago[IA64] alternate perfmon handler
Tony Luck [Wed, 18 May 2005 23:06:00 +0000 (16:06 -0700)] 
[IA64] alternate perfmon handler

Patch from Charles Spirakis

Some linux customers want to optimize their applications on the latest
hardware but are not yet willing to upgrade to the latest kernel. This
patch provides a way to plug in an alternate, basic, and GPL'ed PMU
subsystem to help with their monitoring needs or for specialty work. It
can also be used in case of serious unexpected bugs in perfmon. Mutual
exclusion between the two subsystems is guaranteed, hence no conflict
can arise from both subsystem being present.

Acked-by: Stephane Eranian <eranian@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[PATCH] Avoid console spam with ext3 aborted journal.
Stephen Tweedie [Wed, 18 May 2005 15:47:17 +0000 (11:47 -0400)] 
[PATCH] Avoid console spam with ext3 aborted journal.

Avoid console spam with ext3 aborted journal.

ext3 usually reports error conditions that it detects in its environment.
But when its journal gets aborted due to such errors, it can sometimes
continue to report that condition forever, spamming the console to such
an extent that the initial first cause of the journal abort can be lost.

When the journal aborts, we put the filesystem into readonly mode.  Most
subsequent filesystem operations will get rejected immediately by checks
for MS_RDONLY either in the filesystem or in the VFS.  But some paths do
not have such checks --- for example, if we continue to write to a file
handle that was opened before the fs went readonly.  (We only check for
the ROFS condition when the file is first opened.)  In these cases, we
can continue to generate log errors similar to

EXT3-fs error (device $DEV) in start_transaction: Journal has aborted

for each subsequent write.

There is really no point in generating these errors after the initial
error has been fully reported.  Specifically, if we're starting a
completely new filesystem operation, and the filesystem is *already*
readonly (ie. the ext3 layer has already detected and handled the
underlying jbd abort), and we see an EROFS error, then there is simply
no point in reporting it again.

Signed-off-by: Stephen Tweedie <sct@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Fix filp being passed through raw ioctl handler
Stephen Tweedie [Wed, 18 May 2005 15:22:31 +0000 (11:22 -0400)] 
[PATCH] Fix filp being passed through raw ioctl handler

Don't pass meaningless file handles to block device ioctls.

The recent raw IO ioctl-passthrough fix started passing the raw file
handle into the block device ioctl handler.  That's unlikely to be
useful, as the file handle is actually open on a character-mode raw
device, not a block device, so dereferencing it is not going to yield
useful results to a block device ioctl handler.

Previously we just passed NULL; also not a value that can usefully
be dereferenced, but at least if it does happen, we'll oops instead of
silently pretending that the file is a block device, so NULL is the more
defensive option here.  This patch reverts to that behaviour.

Noticed by Al Viro.

Signed-off-by: Stephen Tweedie <sct@redhat.com>
Acked-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] PCI: add MODALIAS to hotplug event for pci devices
Greg KH [Thu, 5 May 2005 18:57:25 +0000 (11:57 -0700)] 
[PATCH] PCI: add MODALIAS to hotplug event for pci devices

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] PCI: add modalias sysfs file for pci devices
Greg KH [Thu, 5 May 2005 18:57:25 +0000 (11:57 -0700)] 
[PATCH] PCI: add modalias sysfs file for pci devices

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] PCI Hotplug: remove pci_visit_dev
Scott Murray [Mon, 9 May 2005 21:36:27 +0000 (17:36 -0400)] 
[PATCH] PCI Hotplug: remove pci_visit_dev

If my CPCI hotplug update patch is applied, then there are no longer any
in tree users of the pci_visit_dev API, and it and its related code can be
removed.

Signed-off-by: Scott Murray <scottm@somanetworks.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] PCI Hotplug: CPCI update
Scott Murray [Mon, 9 May 2005 21:31:50 +0000 (17:31 -0400)] 
[PATCH] PCI Hotplug: CPCI update

[PATCH] CPCI: update

I have finally done some work to update the CompactPCI hotplug driver to
fix some of the outstanding issues in 2.6:
- Added adapter and latch status ops so that those files will get created
  by the current PCI hotplug core.  This used to not be required, but
  seems to be now after some of the sysfs rework in the core.
- Replaced slot list spinlock with a r/w semaphore to avoid any potential
  issues with sleeping.  This quiets all of the runtime warnings.
- Reworked interrupt driven hot extraction handling to remove need for a
  polling operator for ENUM# status.  There are a lot of boards that only
  have an interrupt driven by ENUM#, so this lowers the bar to entry.
- Replaced pci_visit_dev usage with better use of the PCI core functions.
  The new code is functionally equivalent to the previous code, but the
  use of pci_enable_device on insert needs to be investigated further, as
  I need to do some more testing to see if it is still necessary.

Signed-off-by: Scott Murray <scottm@somanetworks.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] PCI Hotplug: get pciehp to work on the downstream port of a switch
Dely Sy [Sat, 7 May 2005 00:19:09 +0000 (17:19 -0700)] 
[PATCH] PCI Hotplug: get pciehp to work on the downstream port of a switch

Here is the updated patch to get pciehp driver to work for downstream
port of a switch and handle the difference in the offset value of PCI
Express capability list item of different ports.

Signed-off-by: Dely Sy <dely.l.sy@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[PATCH] PCI Hotplug: Fix echoing 1 to power file of enabled slot problem with SHPC...
Dely Sy [Thu, 5 May 2005 18:57:25 +0000 (11:57 -0700)] 
[PATCH] PCI Hotplug: Fix echoing 1 to power file of enabled slot problem with SHPC driver

Here is a patch to fix the problem of echoing 1 to "power" file
to enabled slot causing the slot to power down, and echoing 0
to disabled slot causing shpchp_disabled_slot() to be called
twice. This problem was reported by kenji Kaneshige.

Thanks,
Dely

Signed-off-by: Dely Sy <dely.l.sy@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
19 years ago[IS64-SGI] Set Altix error handling features
Russ Anderson [Mon, 16 May 2005 22:30:00 +0000 (15:30 -0700)] 
[IS64-SGI] Set Altix error handling features

The 2.6 kernel has CPE error thresholding.
This patch lets SAL know of this error handling feature.
The changes are SN specific.

Signed-off-by: Russ Anderson (rja@sgi.com)
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Make Altix SAL call to POD reentrant
Russ Anderson [Mon, 16 May 2005 22:19:00 +0000 (15:19 -0700)] 
[IA64-SGI] Make Altix SAL call to POD reentrant

Change the SAL call for POD mode to be reentrant.
This change is SN specific.

Signed-off-by: Russ Anderson (rja@sgi.com)
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] cpe interrupts are not being enabled.
Russ Anderson [Mon, 9 May 2005 22:03:00 +0000 (15:03 -0700)] 
[IA64-SGI] cpe interrupts are not being enabled.

acpi_request_vector() is called in ia64_mca_init() to get the cpe_vector.
The problem is that acpi_request_vector() looks in platform_intr_list[] to
get the vector, but platform_intr_list[] is not initialized with a valid
vector until later (in sn_setup()).  Without a valid vector the code
defaults to polling mode.

This patch moves the call to acpi_request_vector() from ia64_mca_init()
to ia64_mca_late_init(), which is after platform_intr_list[] is initialized.

Signed-off-by: Russ Anderson (rja@sgi.com)
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[PATCH] update sn2 maintainer
Jesse Barnes [Mon, 9 May 2005 21:52:00 +0000 (14:52 -0700)] 
[PATCH] update sn2 maintainer

Now that I'm no longer at SGI and don't have access to Altix equipment,
it's time to pass on the job of patch monkey to someone else.  Greg
Edwards has foolishly^Wkindly volunteered for the job, so here's a
patch to update the MAINTAINERS file with the appropriate information.

Signed-off-by: Jesse Barnes <jesse.barnes@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Correct convert_to_non_syscall()
David Mosberger-Tang [Tue, 10 May 2005 18:35:00 +0000 (11:35 -0700)] 
[IA64] Correct convert_to_non_syscall()

convert_to_non_syscall() has the same problem that unwind_to_user()
used to have.  Fix it likewise.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[PATCH] fix memory scribble in arch/i386/pci/fixup.c
Christoph Lameter [Tue, 17 May 2005 15:48:16 +0000 (08:48 -0700)] 
[PATCH] fix memory scribble in arch/i386/pci/fixup.c

The GET_INDEX() macro should use just the low three bits of the devfn,
otherwise we have a memory scribble in pcie_rootport_aspm_quirk that
overwrites ptype_all

Fix it to be more careful about its arguments while at it.

Acked by Dely Sy <dely.l.sy@intel.com>

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] update CREDITS
randy_dunlap [Tue, 17 May 2005 14:12:56 +0000 (07:12 -0700)] 
[PATCH] update CREDITS

free agent

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc32: enable use of early_param
Paul Mackerras [Tue, 17 May 2005 06:48:39 +0000 (16:48 +1000)] 
[PATCH] ppc32: enable use of early_param

We need to call parse_early_param() early on to allow usage of
early_param() for command line parsing.

Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge with linus
Tony Luck [Tue, 17 May 2005 16:10:20 +0000 (09:10 -0700)] 
Merge with linus

19 years agoAutomatic merge of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata...
Linus Torvalds [Tue, 17 May 2005 15:55:00 +0000 (08:55 -0700)] 
Merge ... /linux/kernel/git/jgarzik/libata-2.6.git

19 years agoMerge of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev.git