In an uncensored copy of code from i386 to x86_64 I wound up
with inline assembly with the wrong constraints. Use input
constraints instead of output constraints.
So I know the assembler will do the right thing specify the size
of the operand lidtq and lgdtq instead of just lidt and lgdt.
Make load_segments use an input constraint, and delete the macro fun.
Without having to reload %cs like I do on i386 this code is noticeably
simpler.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
For some reason I was telling my inline assembly that the
input argument was an output argument.
Playing in the trampoline code I have seen a couple of
instances where lgdt get the wrong size (because the
trampolines run in 16bit mode) so use lgdtl and lidtl to
be explicit.
Additionally gcc-3.3 and gcc-3.4 want's an lvalue for a
memory argument and it doesn't think an array of characters
is an lvalue so use a packed structure instead.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] reboot: remove device_suspend(PMSG_FREEZE) from kernel_kexec
If device_suspend(PMSG_FREEZE) is not ready to be called in
kernel_restart it is definitely not ready to be called in the even more
fickle kernel_kexec.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Russell King [Fri, 29 Jul 2005 15:36:48 +0000 (16:36 +0100)]
[ARM SMP] Ensure secondary CPUs see their pen release
Since the secondary CPUs will not be operating in symetric mode
while they are held in the pen, we need to ensure that the write
to pen_release is visible to them, by flushing the cache.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Stephen Smalley [Fri, 29 Jul 2005 04:16:21 +0000 (21:16 -0700)]
[PATCH] selinux: Fix address length checks in connect hook
This patch fixes the address length checks in the selinux_socket_connect
hook to be no more restrictive than the underlying ipv4 and ipv6 code;
otherwise, this hook can reject valid connect calls. This patch is in
response to a bug report where an application was calling connect on an
INET6 socket with an address that didn't include the optional scope id and
failing due to these checks.
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: James Morris <jmorris@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Dave Hansen [Fri, 29 Jul 2005 04:16:18 +0000 (21:16 -0700)]
[PATCH] re-disable TSC on NUMAQ
Somewhere recently, the TSC got re-enabled for timekeeping on NUMAQ
machines. However, the hardware makes these get unsynchronized quite
badly. So badly, in fact, that the code to fix up the skew can just hang
on boot.
This patch re-disables them. It's nicely confined to the numaq.c file. It
would be great if this could make it into 2.6.13, I think it counts as a
bugfix.
Tested on a 16-proc 4-node NUMAQ.
Signed-off-by: Dave Hansen <haveblue@us.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
George Anzinger [Fri, 29 Jul 2005 04:16:16 +0000 (21:16 -0700)]
[PATCH] posix timers: fix normalization problem
(We found this (after a customer complained) and it is in the kernel.org
kernel. Seems that for CLOCK_MONOTONIC absolute timers and clock_nanosleep
calls both the request time and wall_to_monotonic are subtracted prior to
the normalize resulting in an overflow in the existing normalize test.
This causes the result to be shifted ~4 seconds ahead instead of ~2 seconds
back in time.)
The normalize code in posix-timers.c fails when the tv_nsec member is ~1.2
seconds negative. This can happen on absolute timers (and
clock_nanosleeps) requested on CLOCK_MONOTONIC (both the request time and
wall_to_monotonic are subtracted resulting in the possibility of a number
close to -2 seconds.)
This fix uses the set_normalized_timespec() (which does not have an
overflow problem) to fix the problem and as a side effect makes the code
cleaner.
Signed-off-by: George Anzinger <george@mvista.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Christophe Lucas [Fri, 29 Jul 2005 04:16:12 +0000 (21:16 -0700)]
[PATCH] uml: Clean up prink calls
printk() calls should include appropriate KERN_* constant.
Signed-off-by: Christophe Lucas <clucas@rotomalug.org> Signed-off-by: Domen Puncer <domen@coderock.org> Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Fri, 29 Jul 2005 04:16:11 +0000 (21:16 -0700)]
[PATCH] uml: Fix typo
Fix a typo in wait_stub_done.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Fri, 29 Jul 2005 04:16:09 +0000 (21:16 -0700)]
[PATCH] uml: Fix load average >=1
update_process_times was missing its irq_enter/irq_exit wrapper. This caused
ksoftirqd to be scheduled on every clock tick.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Fri, 29 Jul 2005 04:16:08 +0000 (21:16 -0700)]
[PATCH] uml: Fix redundant assignment
By this point, .is_user has already been set, so this assignment is useless.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Bodo Stroesser [Fri, 29 Jul 2005 04:16:06 +0000 (21:16 -0700)]
[PATCH] uml: Fix skas0 stub return
It's wrong to pop a fixed number of words from stack before calling sigreturn,
as the number depends on what code is generated by the compiler for the start
of stub_segv_handler(). What we need is esp containing the address of
sigcontext. So we explicitly load that pointer into esp.
Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Fri, 29 Jul 2005 04:16:05 +0000 (21:16 -0700)]
[PATCH] uml: vm86 compile fix
We added an include of asm/vm86.h in include/asm-i386/ptrace.h. Since UML
includes the underlying arch's ptrace.h, it needs an asm/vm86.h in order to
build.
Signed-off-by: Jeff Dike <jdike@addtoit.com> Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Just a Kbuild subtlety, not listing a target file inside targets causes it
to be rebuilt each time, and as a consequence everything depending on it is
rebuilt.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jeff Dike [Fri, 29 Jul 2005 04:16:01 +0000 (21:16 -0700)]
[PATCH] uml: fix TT mode by reverting "use fork instead of clone"
With Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Revert the following patch, because of miscompilation problems in different
environments leading to UML not working *at all* in TT mode; it was merged
lately in 2.6 development cycle, a little after being written, and has
caused problems to lots of people; I know it's a bit too long, but it
shouldn't have been merged in first place, so I still apply for inclusion
in the -stable tree. Anyone using this feature currently is either using
some older kernel (some reports even used 2.6.12-rc4-mm2) or using this
patch, as included in my -bs patchset.
For now there's not yet a fix for this patch, so for now the best thing is
to drop it (which was widely reported to give a working kernel, and as such
was even merged in -stable tree).
"Convert the boot-time host ptrace testing from clone to fork. They were
essentially doing fork anyway. This cleans up the code a bit, and makes
valgrind a bit happier about grinding it."
Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] device-mapper: fix md->lock deadlocks in core
This patch is an attempt to fix deadlocks discovered in the core dm.
The problems boil down to md->lock having to be held in too many places, so
I've split it into two: md->suspend_lock and md->io_lock.
suspend_lock is now held throughout dm_suspended() as well as dm_resume()
and dm_swap_table() so that these functions cannot run concurrently:
there's no requirement for that and it added complexity.
[PATCH] x86_64: fix cpu_to_node setup for sparse apic_ids
While booting with SMT disabled in bios, when using acpi srat to setup
cpu_to_node[], sparse apic_ids create problems.
Without this patch, intel x86_64 boxes with hyperthreading disabled in the
bios (and which rely on srat for numa setup) endup having incorrect values in
cpu_to_node[] arrays, causing sched domains to be built incorrectly etc.
[PATCH] x86_64: Fix gcc 4 warning in sched_find_first_bit
This patch eliminates the GCC4 warning on the x86_64 platform:
kernel/sched.c:1824: warning: control may reach end of non-void function
'sched_find_first_bit' being inlined.
The change follows the lead of others, i.e. it is guaranteed that at least
one of b[0], b[1], or b[2] will have a bit set and evaluate to true. That
being said, GCC4.0.0 notices that the code flow does not return anything if
b[0], b[1] and b[2] are not true. Since we know better, if it's not b[0] or
b[1], it has to be b[2].
Signed-off-by: Jesse Millan <jessem@cs.pdx.edu> Signed-off-by: Domen Puncer <domen@coderock.org> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] x86_64: Icecream has no way of detecting assembler-level includes
Icecream preprocesses c sources locally, and sends the result off to a remote
host for compiling. It does not recognize includes at assembler level. The
fix is to put the assemberincludes an a separate .s file, which will always be
assembled locally.
Signed-off-by: Andreas Gruenbacher <agruen@suse.de> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] x86_64: Create per CPU machine check sysfs directories
This patch will create machinecheck sysdev directories per CPU. All of the
cpus still share the same ctl banks. When compiled with CONFIG_HOTPLUG_CPU,
it will also bring up/down sysdev directories as cpus go up/down. I have
tested the patch along with CONFIG_HOTPLUG_CPU option on in 2.6.13-rc1 kernel.
Minor changes by AK: remove useless unload function
Signed-off-by: Jacob Shin <jacob.shin@amd.com> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Various code needs this information now before the actual SMP bootup. Instead
of computing it on the fly while booting the other CPUs set it up now while
initial MPtable/MADT parsing.
Alexander Nyberg [Fri, 29 Jul 2005 04:15:20 +0000 (21:15 -0700)]
[PATCH] x86_64: cpu hotplug changes kills nmi watchdog
When the x86_64 cpu hotplug changes went in it added a check in
default_do_nmi() which kills NMI delivery on any CPU but the BSP.
The NMI watchdog is brought up quite some time before the online bit is set
in num_online_cpus so this won't work very well. The nmi watchdogs on cpus
that are not BSP will never be reprogrammed and no NMIs.
Why was this check added? How does an offlined cpu receive an NMI?
Signed-off-by: Alexander Nyberg <alexn@telia.com> Cc: Andi Kleen <ak@suse.de> Cc: Andrew Morton <akpm@osdl.org> Cc: Zwane Mwaikambo <zwane@arm.linux.org.uk> Cc: <ashok.raj@intel.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Joachim Nilsson [Thu, 28 Jul 2005 22:18:03 +0000 (00:18 +0200)]
[PATCH] fix gconfig crash
I ran glade-2 on the glade file, fixed two missing stock icons and
cleaned up the C code that inserts the single/split/full modes. The
rest of the patch is minor cleanups only. I refrained from using all
the included xpm icons in images.c (like qconf.cc does) in favour of
using the stock Gtk+ icons instead. Oh, yes there was a "back" bug
in split mode that I also removed, oh well...
It has been tested with success by several people, including
Jesper Juhl, Randy Dunlap and myself.
Signed-off-by: Sam Ravnborg <sam@ravnborg.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Greg Felix [Thu, 28 Jul 2005 19:54:15 +0000 (15:54 -0400)]
libata: Check PCI sub-class code before disabling AHCI
This patch adds functionality to check the PCI sub-class code of an
AHCI capable device before disabling AHCI. It fixes a bug where an
ICH7 sata controller is being setup by the BIOS as sub-class 1 (ide)
and the AHCI control registers weren't being initialized, thus causing
an IO error in piix_disable_ahci().
Signed-off-by: Gregory Felix <greg.felix@gmail.com>
Russell King [Thu, 28 Jul 2005 19:36:26 +0000 (20:36 +0100)]
[ARM SMP] Fix another ARMv6 bitop problem
We sometimes forgot to check whether the exclusive store succeeded.
Ensure that we always check. Also ensure that we always use the
out of line versions, since the inline versions are not SMP safe.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell King [Thu, 28 Jul 2005 19:32:47 +0000 (20:32 +0100)]
[ARM SMP] Ensure secondary CPUs have a clean TLB
Since ARMv6 CPUs will not flush the TLB on context switches, it is
possible that we may end up with some global TLB entries remaining
present, eventually upsetting userspace. Explicitly flush the
entire TLB on secondary CPUs as they startup, after we have switched
to the init_mm page tables.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Revert broken "statement with no effect" warning fix
It may shut up gcc, but it also incorrectly changes the semantics of the
smp_call_function() helpers.
You can fix the warning other ways if you are interested (create another
inline function that takes no arguments and returns zero), but
preferably gcc just shouldn't complain about unused return values from
statement expressions in the first place.
Dave Jones [Thu, 28 Jul 2005 16:43:56 +0000 (09:43 -0700)]
Here are two possible cleanups in cpufreq.c:
* ret has no need to be unsigned in cpufreq_driver_target()
* ret has no need to be initialized in __cpufreq_governor()
Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Dave Jones <davej@redhat.com>
Dave Jones [Thu, 28 Jul 2005 16:40:04 +0000 (09:40 -0700)]
Opteron revision F will support higher frequencies than
can be encoded in the current driver's 4 bit frequency
field. This patch updates the driver to support Rev F
including 6 bit FIDs and processor ID updates.
This should apply cleanly whether or not the dual-core
bugfix I sent out last week is applied. I'd prefer
that both get applied, of course.
Signed-off-by: David Keck <david.keck@amd.com> Signed-off-by: Mark Langsdorf <mark.langsdorf@amd.com> Signed-off-by: Dave Jones <davej@redhat.com>
Dave Jones [Thu, 28 Jul 2005 16:38:21 +0000 (09:38 -0700)]
powernow-k8 requires that a data structure for
each core be created in the _cpu_init function
call. The cpufreq infrastructure doesn't call
_cpu_init for the second core in each processor.
Some systems crashed when _get was called with
an odd-numbered core because it tried to
dereference a NULL pointer since the data
structure had not been created.
The attached patch solves the problem by
initializing data structures for all shared
cores in the _cpu_init function. It should
apply to 2.6.12-rc6 and has been tested by
AMD and Sun.
Signed-off-by: Mark Langsdorf <mark.langsdorf@amd.com> Signed-off-by: Dave Jones <davej@redhat.com>
[PATCH] alpha: fix "statement with no effect" warnings
Apparently gcc 4.0 complains about "({ 0; });", which leads to -Werror
breakage in one of the alpha oprofile modules.
One might could argue that this is a gcc bug, in that statement-expressions
should be considered to be function-like rather than statement-like for the
purposes of this warning. But it's just as easy to use an inline function
in the first place, side-stepping the issue.
Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Tony Luck [Thu, 28 Jul 2005 08:07:38 +0000 (01:07 -0700)]
[PATCH] e1000: no need for reboot notifier
sys_reboot() now calls device_suspend(), so it is no longer necessary for
the e1000 driver to register a reboot notifier [in fact doing so results in
e1000_suspend() getting called twice].
(akpm: we need to fast-track this. It's causing ia64 to oops on shutdown)
Signed-off-by: Tony Luck <tony.luck@intel.com> Cc: <cramerj@intel.com> Cc: <john.ronciak@intel.com> Cc: <ganesh.venkatesan@intel.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
James Morris [Thu, 28 Jul 2005 08:07:37 +0000 (01:07 -0700)]
[PATCH] SELinux: default labeling of MLS field
Implement kernel labeling of the MLS (multilevel security) field of
security contexts for files which have no existing MLS field. This is to
enable upgrades of a system from non-MLS to MLS without performing a full
filesystem relabel including all of the mountpoints, which would be quite
painful for users.
With this patch, with MLS enabled, if a file has no MLS field, the kernel
internally adds an MLS field to the in-core inode (but not to the on-disk
file). This MLS field added is the default for the superblock, allowing
per-mountpoint control over the values via fixed policy or mount options.
This patch has been tested by enabling MLS without relabeling its
filesystem, and seems to be working correctly.
Signed-off-by: James Morris <jmorris@redhat.com> Signed-off-by: Stephen Smalley <sds@epoch.ncsc.mil> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] Fix incorrect Asus k7m irq router detection
This patch:
http://marc.theaimsgroup.com/?l=bk-commits-head&m=111955644929114&w=2
uncovered a k7m bios bug, where the VT82C686A router is reported as
being "586-compatible". The two chips have different pirq mapping, so
this leads to "irq routing conflict" on many pci devices.
The suggested fix was discussed with Aleksey Gorelov, who helped me
to identify the problem as a probable bios bug.
Signed-off-by: Giancarlo Formicuccia <giancarlo.formicuccia@gmail.com> Cc: Dave Jones <davej@codemonkey.org.uk> Cc: Greg KH <greg@kroah.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Mike Miller [Thu, 28 Jul 2005 08:07:31 +0000 (01:07 -0700)]
[PATCH] cciss per disk queue
This patch adds per disk queue functionality to cciss. Sometime back I
submitted a patch but it looks like only part of what I needed. In the 2.6
kernel if we have more than one logical volume the driver will Oops during
rmmod. It seems all of the queues actually point back to the same queue.
So after deleting the first volume you hit a null pointer on the second
one.
This has been tested in our labs. There is no difference in performance,
it just fixes the Oops.
Signed-off-by: Mike Miller <mike.miller@hp.com> Cc: Jens Axboe <axboe@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Daniel Ritz [Thu, 28 Jul 2005 08:07:30 +0000 (01:07 -0700)]
[PATCH] pcmcia: disable read prefetch/write burst on old O2Micro bridges
Older O2Micro bridges have problems with both read prefetch and write burst
depending on the combination of the chipset, bridge, cardbus card. safest is
to disable read prefetch and write burst on those old bridges.
Signed-off-by: Daniel Ritz <daniel.ritz@gmx.ch> Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] pcmcia: fix sharing IRQs and request_irq without IRQ_HANDLE_PRESENT
Debugging and description from: Noah Misch <noah@cs.caltech.edu>
When a driver calls pcmcia_request_irq with IRQ_HANDLE_PRESENT unset, it looks
for an open IRQ by request_irq()ing with a dummy handler and NULL dev_info.
free_irq uses dev_info as a key for identifying the handler to free among
those sharing an IRQ, so request_irq returns -EINVAL if dev_info is NULL and
the IRQ may be shared. That unknown error code is the -EINVAL.
It looks like only pcnet_cs and axnet_cs are affected. Most other drivers let
pcmcia_request_irq install their interrupt handlers. sym53c500_cs requests
its IRQ manually, but it cannot share an IRQ.
The appended patch changes pcmcia_request_irq to pass an arbitrary, unique,
non-NULL dev_info with the dummy handler.
Pavel Roskin [Thu, 28 Jul 2005 08:07:19 +0000 (01:07 -0700)]
[PATCH] pcmcia: ide-cs id_table update
SanDisk ConnectPlus has two functions. Function 0 is prism2 card, currently
only supported by HostAP (not in the kernel). Function 1 is 128M flash,
supported by ide-cs. This patch adds an entry for function 1 to ide-cs.c.
Signed-off-by: Pavel Roskin <proski@gnu.org> Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
projects / linux-2.6 / log