platform/kernel/linux-rpi.git
17 years ago[PATCH] x86-64: Support -mregparm arguments for signals with SA_SIGINFO in compat...
Albert Cahalan [Thu, 7 Dec 2006 01:14:06 +0000 (02:14 +0100)]
[PATCH] x86-64: Support -mregparm arguments for signals with SA_SIGINFO in compat mode

The recent change to make x86_64 support i386 binaries compiled
with -mregparm=3 only covered signal handlers without SA_SIGINFO.
(the 3-arg "real-time" ones)

To be compatible with i386, both types should be supported.

Signed-off-by: Albert Cahalan <acahalan@gmail.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Try multiple timer variants in check_timer
Andi Kleen [Thu, 7 Dec 2006 01:14:06 +0000 (02:14 +0100)]
[PATCH] x86-64: Try multiple timer variants in check_timer

Instead of adding all kinds of more quirks try various timer
routing variants in check_timer.

In particular this tries to handle quirks from:
- Nvidia NF2-4 reference BIOS: wrong timer override
- Asus: Wrong timer override but no HPET table
- ATI: require timer disabled in 8259
- Some boards: require timer enabled in 8259

We just try many of the the known variants in the hopefully right order
in check_timer.

Trying pin 0/2 on Nvidia suggested by Tim Hockin.

TBD Experimental. Needs a lot of testing

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Use probe_kernel_address instead of __get_user in fault paths
Andi Kleen [Thu, 7 Dec 2006 01:14:06 +0000 (02:14 +0100)]
[PATCH] i386: Use probe_kernel_address instead of __get_user in fault paths

Makes the intention of the code cleaner to read and avoids
a potential deadlock on mmap_sem. Also change the types of
the arguments to not include __user because they're really
not user addresses.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Use probe_kernel_address in arch/x86_64/*
Andi Kleen [Thu, 7 Dec 2006 01:14:06 +0000 (02:14 +0100)]
[PATCH] x86-64: Use probe_kernel_address in arch/x86_64/*

Instead of open coded __get_user

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] Generic: Move __user cast into probe_kernel_address
Andi Kleen [Thu, 7 Dec 2006 01:14:05 +0000 (02:14 +0100)]
[PATCH] Generic: Move __user cast into probe_kernel_address

Caller of probe_kernel_address shouldn't need to know that
pka is internally implemented with __get_user. So move the
__user cast into pka.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Extend clear_irq_vector
Yinghai Lu [Thu, 7 Dec 2006 01:14:05 +0000 (02:14 +0100)]
[PATCH] x86-64: Extend clear_irq_vector

Clear the irq releated entries in irq_vector, irq_domain and vector_irq
instead of clearing irq_vector only. So when new irq is created, it
could reuse that vector. (actually is the second loop scanning from
FIRST_DEVICE_VECTOR+8). This could avoid the vectors are used up
with enough module inserting and removing

Cc: Eric W. Biedierman <ebiederm@xmission.com>
Cc: Muli Ben-Yehuda <muli@il.ibm.com>
Signed-off-By: Yinghai Lu <yinghai.lu@amd.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Use CLFLUSH instead of WBINVD in change_page_attr
Andi Kleen [Thu, 7 Dec 2006 01:14:05 +0000 (02:14 +0100)]
[PATCH] i386: Use CLFLUSH instead of WBINVD in change_page_attr

CLFLUSH is a lot faster than WBINVD so try to use that.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Retrieve CLFLUSH size from CPUID
Andi Kleen [Thu, 7 Dec 2006 01:14:05 +0000 (02:14 +0100)]
[PATCH] i386: Retrieve CLFLUSH size from CPUID

Also report it in /proc/cpuinfo similar to x86-64.

Needed for followon patch

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Speed and clean up cache flushing in change_page_attr
Andi Kleen [Thu, 7 Dec 2006 01:14:05 +0000 (02:14 +0100)]
[PATCH] x86-64: Speed and clean up cache flushing in change_page_attr

CLFLUSH is a lot faster than WBINVD so avoid the later if at all
possible.

Always pass the complete list of pages to other CPUs to cut down
the number of IPIs.

Minor other cleanup and sync with i386 version.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Fix entry.S code with !CONFIG_VM86
Joe Korty [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: Fix entry.S code with !CONFIG_VM86

The entry.S code at work_notifysig is surely wrong.  It drops into unrelated
code if the branch to work_notifysig_v86 is taken, and CONFIG_VM86=n.

[PATCH] Make vm86 support optional
tree 9b5daef5280800a0006343a17f63072658d91a1d
pushed to git Jan 8, 2006, and first appears in 2.6.16

The 'fix' here is to also compile out the vm86 test & branch when
CONFIG_VM86=n.

Signed-off-by: Joe Korty <joe.korty@ccur.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Extract segment descriptor definitions for use outside
Avi Kivity [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] x86-64: Extract segment descriptor definitions for use outside

Code that wants to use struct desc_struct cannot do so on i386 because
desc.h contains other code that will only compile on x86_64.

So extract the structure definitions into a asm-x86_64/desc_defs.h.

Signed-off-by: Avi Kivity <avi@qumranet.com>
Signed-off-by: Andi Kleen <ak@suse.de>
 include/asm-x86_64/desc.h      |   53 -------------------------------
 include/asm-x86_64/desc_defs.h |   69 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 70 insertions(+), 52 deletions(-)

17 years ago[PATCH] i386: Mark CONFIG_RELOCATABLE EXPERIMENTAL
Vivek Goyal [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: Mark CONFIG_RELOCATABLE EXPERIMENTAL

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: extend bzImage protocol for relocatable protected mode kernel
Vivek Goyal [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: extend bzImage protocol for relocatable protected mode kernel

Extend bzImage protocol to enable bootloaders to load a completely relocatable
bzImage.  Now protected mode component of kernel is also relocatable and a
boot-loader can load the protected mode component at a differnt physical
address than 1MB.  (If kernel was built with CONFIG_RELOCATABLE)

Kexec can make use of it to load this kernel at a different physical address
to capture kernel crash dumps.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Implement CONFIG_PHYSICAL_ALIGN
Vivek Goyal [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: Implement CONFIG_PHYSICAL_ALIGN

o Now CONFIG_PHYSICAL_START is being replaced with CONFIG_PHYSICAL_ALIGN.
  Hardcoding the kernel physical start value creates a problem in relocatable
  kernel context due to boot loader limitations. For ex, if somebody
  compiles a relocatable kernel to be run from address 4MB, but this kernel
  will run from location 1MB as grub loads the kernel at physical address
  1MB. Kernel thinks that I am a relocatable kernel and I should run from
  the address I have been loaded at. So somebody wanting to run kernel
  from 4MB alignment location (for improved performance regions) can't do
  that.

o Hence, Eric proposed that probably CONFIG_PHYSICAL_ALIGN will make
  more sense in relocatable kernel context. At run time kernel will move
  itself to a physical addr location which meets user specified alignment
  restrictions.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Warn upon absolute relocations being present
Vivek Goyal [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: Warn upon absolute relocations being present

o Relocations generated w.r.t absolute symbols are not processed as by
  definition, absolute symbols are not to be relocated. Explicitly warn
  user about absolutions relocations present at compile time.

o These relocations get introduced either due to linker optimizations or
  some programming oversights.

o Also create a list of symbols which have been audited to be safe and
  don't emit warnings for these.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Relocatable kernel support
Eric W. Biederman [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: Relocatable kernel support

This patch modifies the i386 kernel so that if CONFIG_RELOCATABLE is
selected it will be able to be loaded at any 4K aligned address below
1G.  The technique used is to compile the decompressor with -fPIC and
modify it so the decompressor is fully relocatable.  For the main
kernel relocations are generated.  Resulting in a kernel that is relocatable
with no runtime overhead and no need to modify the source code.

A reserved 32bit word in the parameters has been assigned
to serve as a stack so we figure out where are running.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] relocatable kernel: Kallsyms generate relocatable symbols
Eric W. Biederman [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] relocatable kernel: Kallsyms generate relocatable symbols

Print the addresses of non-absolute symbols relative to _text
so that ld will generate relocations.  Allowing a relocatable
kernel to relocate them.  We can't actually use the symbol names
because kallsyms includes static symbols that are not exported
from their object files.

Add the _text symbol definitions to the architectures which don't
define it otherwise linker will fail.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: CONFIG_PHYSICAL_START cleanup
Eric W. Biederman [Thu, 7 Dec 2006 01:14:04 +0000 (02:14 +0100)]
[PATCH] i386: CONFIG_PHYSICAL_START cleanup

Defining __PHYSICAL_START and __KERNEL_START in asm-i386/page.h works but
it triggers a full kernel rebuild for the silliest of reasons.  This
modifies the users to directly use CONFIG_PHYSICAL_START and linux/config.h
which prevents the full rebuild problem, which makes the code much
more maintainer and hopefully user friendly.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Reserve kernel memory starting from _text
Eric W. Biederman [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Reserve kernel memory starting from _text

Currently when we are reserving the memory the kernel text
resides in we start at __PHYSICAL_START which happens to be
correct but not very obvious.  In addition when we start relocating
the kernel __PHYSICAL_START is the wrong value, as it is an
absolute symbol that does not get relocated.

By starting the reservation at __pa_symbol(_text)
the code is clearer and will be correct when relocated.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: define __pa_symbol()
Eric W. Biederman [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: define __pa_symbol()

On x86_64 we have to be careful with calculating the physical
address of kernel symbols.  Both because of compiler odditities
and because the symbols live in a different range of the virtual
address space.

Having a defintition of __pa_symbol that works on both x86_64 and
i386 simplifies writing code that works for both x86_64 and
i386 that has these kinds of dependencies.

So this patch adds the trivial i386 __pa_symbol definition.

Added assembly magic similar to RELOC_HIDE as suggested by Andi Kleen.
Just picked it up from x86_64.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Add comment for align to vmlinux.lds
Vivek Goyal [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Add comment for align to vmlinux.lds

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Distinguish absolute symbols
Vivek Goyal [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Distinguish absolute symbols

Ld knows about 2 kinds of symbols,  absolute and section
relative.  Section relative symbols symbols change value
when a section is moved and absolute symbols do not.

Currently in the linker script we have several labels
marking the beginning and ending of sections that
are outside of sections, making them absolute symbols.
Having a mixture of absolute and section relative
symbols refereing to the same data is currently harmless
but it is confusing.

This must be done carefully as newer revs of ld do not place
symbols that appear in sections without data and instead
ld makes those symbols global :(

My ultimate goal is to build a relocatable kernel.  The
safest and least intrusive technique is to generate
relocation entries so the kernel can be relocated at load
time.  The only penalty would be an increase in the size
of the kernel binary.  The problem is that if absolute and
relocatable symbols are not properly specified absolute symbols
will be relocated or section relative symbols won't be, which
is fatal.

The practical motivation is that when generating kernels that
will run from a reserved area for analyzing what caused
a kernel panic, it is simpler if you don't need to hard code
the physical memory location they will run at, especially
for the distributions.

[AK: and merged:]

o Also put a message so that in future people can be aware of it and
  avoid introducing absolute symbols.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Implement compat code for SIOCSIFHWBROADCAST
Andi Kleen [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] x86-64: Implement compat code for SIOCSIFHWBROADCAST

This network ioctl wasn't handled before.

Reported by Alexandra.Kossovsky@oktetlabs.ru (Alexandra Kossovsky)
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: Mention PCI instead of RAM in NMI parity error message
Andi Kleen [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] x86: Mention PCI instead of RAM in NMI parity error message

On modern systems RAM errors don't cause NMIs, but it's usually
caused by PCI SERR. Mention PCI instead of RAM in the printk.

Reported by r_hayashi@ctc-g.co.jp (Ryutaro Hayashi)

Cc: r_hayashi@ctc-g.co.jp
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: remove last two pci_find offenders in the core code
Alan Cox [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] x86: remove last two pci_find offenders in the core code

Resending as I believe the discussion about them established they were
correct.

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86: Don't use nested idle loops
Andi Kleen [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] x86: Don't use nested idle loops

Currently the idle loop has two nested loops -- one high level
in cpu_idle and in some low level idle functions another one.

Looping in the low level idle functions breaks the idle notifiers
because interrupts waking up sleep states need to execute
exit_idle() which is only in cpu_idle().

So don't do that, only loop in cpu_idle(). This only removes
code.

In some cases e.g. poll_idle the idle loop is a little longer
now because cpu_idle checks more things. I hope that isn't a problem
ACPI idle doesn't change behaviour because it never looped anyways.

Cc: len.brown@intel.com
Cc: eranian@hpl.hp.com
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: PDA: Fix math emulator for new pt_regs
Andi Kleen [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: PDA: Fix math emulator for new pt_regs

This patch fixes the math emulator, which had not been adjusted
to match the changed struct pt_regs.

AK: extracted from larger patch by Jeremy.
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Store the interrupt regs pointer in the PDA
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Store the interrupt regs pointer in the PDA

Signed-off-by: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Implement "current" with the PDA
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Implement "current" with the PDA

Use the pcurrent field in the PDA to implement the "current" macro.  This ends
up compiling down to a single instruction to get the current task.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Implement smp_processor_id() with the PDA
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Implement smp_processor_id() with the PDA

Use the cpu_number in the PDA to implement raw_smp_processor_id.  This is a
little simpler than using thread_info, though the cpu field in thread_info
cannot be removed since it is used for things other than getting the current
CPU in common code.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Update sys_vm86 to cope with changed pt_regs and %gs usage
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:03 +0000 (02:14 +0100)]
[PATCH] i386: Update sys_vm86 to cope with changed pt_regs and %gs usage

sys_vm86 uses a struct kernel_vm86_regs, which is identical to pt_regs, but
adds an extra space for all the segment registers.  Previously this structure
was completely independent, so changes in pt_regs had to be reflected in
kernel_vm86_regs.  This changes just embeds pt_regs in kernel_vm86_regs, and
makes the appropriate changes to vm86.c to deal with the new naming.

Also, since %gs is dealt with differently in the kernel, this change adjusts
vm86.c to reflect this.

While making these changes, I also cleaned up some frankly bizarre code which
was added when auditing was added to sys_vm86.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Jason Baron <jbaron@redhat.com>
Cc: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Fix places where using %gs changes the usermode ABI
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: Fix places where using %gs changes the usermode ABI

There are a few places where the change in struct pt_regs and the use of %gs
affect the userspace ABI.  These are primarily debugging interfaces where
thread state can be inspected or extracted.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Use %gs as the PDA base-segment in the kernel
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: Use %gs as the PDA base-segment in the kernel

This patch is the meat of the PDA change.  This patch makes several related
changes:

1: Most significantly, %gs is now used in the kernel.  This means that on
   entry, the old value of %gs is saved away, and it is reloaded with
   __KERNEL_PDA.

2: entry.S constructs the stack in the shape of struct pt_regs, and this
   is passed around the kernel so that the process's saved register
   state can be accessed.

   Unfortunately struct pt_regs doesn't currently have space for %gs
   (or %fs). This patch extends pt_regs to add space for gs (no space
   is allocated for %fs, since it won't be used, and it would just
   complicate the code in entry.S to work around the space).

3: Because %gs is now saved on the stack like %ds, %es and the integer
   registers, there are a number of places where it no longer needs to
   be handled specially; namely context switch, and saving/restoring the
   register state in a signal context.

4: And since kernel threads run in kernel space and call normal kernel
   code, they need to be created with their %gs == __KERNEL_PDA.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Initialize the per-CPU data area
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: Initialize the per-CPU data area

When a CPU is brought up, a PDA and GDT are allocated for it.  The GDT's
__KERNEL_PDA entry is pointed to the allocated PDA memory, so that all
references using this segment descriptor will refer to the PDA.

This patch rearranges CPU initialization a bit, so that the GDT/PDA are set up
as early as possible in cpu_init().  Also for secondary CPUs, GDT+PDA are
preallocated and initialized so all the secondary CPU needs to do is set up
the ldt and load %gs.  This will be important once smp_processor_id() and
current use the PDA.

In all cases, the PDA is set up in head.S, before a CPU starts running C code,
so the PDA is always available.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Cc: James Bottomley <James.Bottomley@SteelEye.com>
Cc: Matt Tolentino <matthew.e.tolentino@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Basic definitions for i386-pda
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: Basic definitions for i386-pda

This patch has the basic definitions of struct i386_pda, and the segment
selector in the GDT.

asm-i386/pda.h is more or less a direct copy of asm-x86_64/pda.h.  The most
interesting difference is the use of _proxy_pda, which is used to give gcc a
model for the actual memory operations on the real pda structure.  No actual
reference is ever made to _proxy_pda, so it is never defined.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Use asm-offsets for the offsets of registers into the pt_regs struct
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: Use asm-offsets for the offsets of registers into the pt_regs struct

Use asm-offsets for the offsets of registers into the pt_regs struct, rather
than having hard-coded constants

I left the constants in the comments of entry.S because they're useful for
reference; the code in entry.S is very dependent on the layout of pt_regs,
even when using asm-offsets.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Keith Owens <kaos@ocs.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86-64: miscellaneous entry.S adjustments
Jan Beulich [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] x86-64: miscellaneous entry.S adjustments

This patch:
- makes ret_from_sys_call no longer global (all external users were
  previously switched to use int_ret_from_sys_call)
- adjusts placement of a CFI_{REMEMBER,RESTORE}_STATE pair to better
  fit logic flow
- eliminates an unnecessary pair of CFI_{REMEMBER,RESTORE}_STATE
- glues together function- and unwinder-wise the previously separate
  system_call and int_ret_from_sys_call function fragments

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: dump_trace() atomicity fix
Andrew Morton [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] x86-64: dump_trace() atomicity fix

Fix

BUG: using smp_processor_id() in preemptible [00000001] code:

in backtracer on preemptible debug kernels.

Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: add Intel Core related PMU MSRs
Stephane Eranian [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: add Intel Core related PMU MSRs

- add Intel Precise-Event Based sampling (PEBS) related MSR
- add Intel Data Save (DS) Area related MSR
- add Intel Core microarchitecure performance counter MSRs

Signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: x86-64 add Intel Core related PMU MSRs definitions
Stephane Eranian [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] x86-64: x86-64 add Intel Core related PMU MSRs definitions

Add o the x86-64 tree a bunch of MSRs related to performance
monitoring for the processors based on Intel Core microarchitecture.
It also adds some architectural MSRs for PEBS. A similar patch for i386 will
follow.

changelog:
- add Intel Precise-Event Based sampling (PEBS) related MSR
- add Intel Data Save (DS) Area related MSR
- add Intel Core microarchitecure performance counter MSRs

Signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: add missing iounmap in i386 hpet clocksource code
Amol Lad [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: add missing iounmap in i386 hpet clocksource code

ioremap must be balanced by an iounmap and failing to do so can result
in a memory leak.

Tested (compilation only):
- using allmodconfig
- making sure the files are compiling without any warning/error due to
new changes

Signed-off-by: Amol Lad <amol@verismonetworks.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Add iounmap in error paths in hpet code
Amol Lad [Thu, 7 Dec 2006 01:14:02 +0000 (02:14 +0100)]
[PATCH] i386: Add iounmap in error paths in hpet code

Signed-off-by: Amol Lad <amol@verismonetworks.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Insert Local and IO APIC(s) into resource map
Aaron Durbin [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] x86-64: Insert Local and IO APIC(s) into resource map

Insert the Local APIC and IO APIC(s) into the resource tree.  It allows the
APIC resources to be visible within /proc/iomem.  The patch also takes into
account IO APIC(s) mapped in the PCI space by deferring the insertion until
after PCI has allocated its necessary resources.

Signed-off-by: Aaron Durbin <adurbin@google.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@muc.de>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: add sleazy FPU optimization
Chuck Ebbert [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] i386: add sleazy FPU optimization

i386 port of the sLeAZY-fpu feature.  Chuck reports that this gives him a +/-
0.4% improvement on his simple benchmark

x86_64 description follows:

Right now the kernel on x86-64 has a 100% lazy fpu behavior: after *every*
context switch a trap is taken for the first FPU use to restore the FPU
context lazily.  This is of course great for applications that have very
sporadic or no FPU use (since then you avoid doing the expensive save/restore
all the time).  However for very frequent FPU users...  you take an extra trap
every context switch.

The patch below adds a simple heuristic to this code: After 5 consecutive
context switches of FPU use, the lazy behavior is disabled and the context
gets restored every context switch.  If the app indeed uses the FPU, the trap
is avoided.  (the chance of the 6th time slice using FPU after the previous 5
having done so are quite high obviously).

After 256 switches, this is reset and lazy behavior is returned (until there
are 5 consecutive ones again).  The reason for this is to give apps that do
longer bursts of FPU use still the lazy behavior back after some time.

Signed-off-by: Chuck Ebbert <76306.1226@compuserve.com>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: espfix cleanup
Stas Sergeev [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] i386: espfix cleanup

Clean up the espfix code:

- Introduced PER_CPU() macro to be used from asm
- Introduced GET_DESC_BASE() macro to be used from asm
- Rewrote the fixup code in asm, as calling a C code with the altered %ss
  appeared to be unsafe
- No longer altering the stack from a .fixup section
- 16bit per-cpu stack is no longer used, instead the stack segment base
  is patched the way so that the high word of the kernel and user %esp
  are the same.
- Added the limit-patching for the espfix segment. (Chuck Ebbert)

[jeremy@goop.org: use the x86 scaling addressing mode rather than shifting]
Signed-off-by: Stas Sergeev <stsp@aknet.ru>
Signed-off-by: Andi Kleen <ak@suse.de>
Acked-by: Zachary Amsden <zach@vmware.com>
Acked-by: Chuck Ebbert <76306.1226@compuserve.com>
Acked-by: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86: all cpu backtrace
Andrew Morton [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] x86: all cpu backtrace

When a spinlock lockup occurs, arrange for the NMI code to emit an all-cpu
backtrace, so we get to see which CPU is holding the lock, and where.

Cc: Andi Kleen <ak@muc.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: remove default_ldt, and simplify ldt-setting.
Jeremy Fitzhardinge [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] i386: remove default_ldt, and simplify ldt-setting.

This patch removes the default_ldt[] array, as it has been unused since
iBCS stopped being supported.  This means it is now possible to actually
set an empty LDT segment.

In order to deal with this, the set_ldt_desc/load_LDT pair has been
replaced with a single set_ldt() operation which is responsible for both
setting up the LDT descriptor in the GDT, and reloading the LDT register.
If there are no LDT entries, the LDT register is loaded with a NULL
descriptor.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Acked-by: Zachary Amsden <zach@vmware.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86-64: use BUILD_BUG_ON in FPU code
Alexey Dobriyan [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] x86-64: use BUILD_BUG_ON in FPU code

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: i386 add X86_FEATURE_PEBS and detection
Stephane Eranian [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] i386: i386 add X86_FEATURE_PEBS and detection

Here is a patch (used by perfmon2) to detect the presence of the Precise Event
Based Sampling (PEBS) feature for i386.  The patch also adds the cpu_has_pebs
macro.

- adds X86_FEATURE_PEBS

- adds cpu_has_pebs to test for X86_FEATURE_PEBS

Signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: i386 rename X86_FEATURE_DTES to X86_FEATURE_DS
Stephane Eranian [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] i386: i386 rename X86_FEATURE_DTES to X86_FEATURE_DS

Here is a patch (used by perfmon2) that renames X86_FEATURE_DTES to
X86_FEATURE_DS to match Intel's documentation for the Debug Store save area on
i386.  The patch also adds cpu_has_ds.

- rename X86_FEATURE_DTES to X86_FEATURE_DS to match documentation

- adds cpu_has_ds to test for X86_FEATURE_DS

Signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86-64: remove duplicated cpu_mask_to_apicid in x86_64 smp.h
Yinghai Lu [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] x86-64: remove duplicated cpu_mask_to_apicid in x86_64 smp.h

inline function cpu_mask_to_apicid in smp.h is duplicated with macro
in mach_apic.h.

Signed-off-by: Yinghai Lu <yinghai.lu@amd.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: add X86_FEATURE_PEBS and detection
Stephane Eranian [Thu, 7 Dec 2006 01:14:01 +0000 (02:14 +0100)]
[PATCH] x86-64: add X86_FEATURE_PEBS and detection

Here is a patch (used by perfmon2) to detect the presence of the
Precise Event Based Sampling (PEBS) feature for Intel 64-bit processors.
The patch also adds the cpu_has_pebs macro.

changelog:
- adds X86_FEATURE_PEBS
- adds cpu_has_pebs to test for X86_FEATURE_PEBS

Signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: x86_64 rename X86_FEATURE_DTES to X86_FEATURE_DS
Stephane Eranian [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] x86-64: x86_64 rename X86_FEATURE_DTES to X86_FEATURE_DS

Here is a patch (used by perfmon2) that renamed X86_FEATURE_DTES
to X86_FEATURE_DS to match Intel's documentation for the Debug Store
save area. The patch also adds cpu_has_ds.

changelog:
- rename X86_FEATURE_DTES to X86_FEATURE_DS to match documentation
- adds cpu_has_ds to test for X86_FEATURE_DS

Signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Don't keep interrupts disabled while spinning in spinlocks
Andi Kleen [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] x86-64: Don't keep interrupts disabled while spinning in spinlocks

Follows i386.

Based on patch from some folks at Google (MikeW, Edward G.?), but
completely redone by AK.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: remove pointless printk from i386 oops output
Dave Jones [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] i386: remove pointless printk from i386 oops output

This just got removed on x86-64, do the same on 32bit.
It always annoyed me when this ate a line of oops output pushing
interesting stuff off the screen.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: fix must_checks
Randy Dunlap [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] i386: fix must_checks

Fix __must_check warnings in i386/math-emu.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: Compress stack unwinder output
Andi Kleen [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] x86: Compress stack unwinder output

The unwinder has some extra newlines, which eat up loads of screen
space when it spews. (See https://bugzilla.redhat.com/bugzilla/attachment.cgi?id=137900
for a nasty example).

warning_symbol-> and warning-> already printk a newline, so don't add one
in the strings passed to them.

[AK: redone for new code]

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: shorten lines in unwinder to be <= 80 characters
Andi Kleen [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] x86: shorten lines in unwinder to be <= 80 characters

Andrew complained about > 80 character lines in the new unwinder.
Fix that.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: fix buggy MTRR address checks
Andreas Mohr [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] i386: fix buggy MTRR address checks

Fix checks that failed to realize that values are 4-kB-unit-sized (note the
format strings in this same diff context which *do* realize the unit size,
via appended "000"!).  Also fix an incorrect below-1MB area check (as
gathered from Jan Beulich's unapplied patch at
http://www.ussg.iu.edu/hypermail/linux/kernel/0411.1/1378.html ) Update
mtrr_add_page() docu to make 4-kB-sized calculation more obvious.

Given several further items mentioned in Jan's patch mail, all in all MTRR
code seems surprisingly buggy, for a surprisingly long period of time (many
years).  Further work/investigation would be useful.

TBD Note that my patch is pretty much UNTESTED, since I can only verify that it
TBD successfully boots my machine, but I cannot test against actual buggy
TBD hardware which would require these (formerly broken) checks.  Long -mm
TBD simmering would make sense, especially since these now-working checks might
TBD turn out to have adverse effects on unaffected hardware.

Signed-off-by: Andreas Mohr <andi@lisas.de>
Signed-off-by: Andi Kleen <ak@suse.de>
Acked-by: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] i386: Update defconfig
Andi Kleen [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] i386: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Update defconfig
Andi Kleen [Thu, 7 Dec 2006 01:14:00 +0000 (02:14 +0100)]
[PATCH] x86-64: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6
Linus Torvalds [Wed, 6 Dec 2006 00:09:46 +0000 (16:09 -0800)]
Merge /pub/scm/linux/kernel/git/jejb/scsi-misc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (73 commits)
  [SCSI] aic79xx: Add ASC-29320LPE ids to driver
  [SCSI] stex: version update
  [SCSI] stex: change wait loop code
  [SCSI] stex: add new device type support
  [SCSI] stex: update device id info
  [SCSI] stex: adjust default queue length
  [SCSI] stex: add value check in hard reset routine
  [SCSI] stex: fix controller_info command handling
  [SCSI] stex: fix biosparam calculation
  [SCSI] megaraid: fix MMIO casts
  [SCSI] tgt: fix undefined flush_dcache_page() problem
  [SCSI] libsas: better error handling in sas_expander.c
  [SCSI] lpfc 8.1.11 : Change version number to 8.1.11
  [SCSI] lpfc 8.1.11 : Misc Fixes
  [SCSI] lpfc 8.1.11 : Add soft_wwnn sysfs attribute, rename soft_wwn_enable
  [SCSI] lpfc 8.1.11 : Removed decoding of PCI Subsystem Id
  [SCSI] lpfc 8.1.11 : Add MSI (Message Signalled Interrupts) support
  [SCSI] lpfc 8.1.11 : Adjust LOG_FCP logging
  [SCSI] lpfc 8.1.11 : Fix Memory leaks
  [SCSI] lpfc 8.1.11 : Fix lpfc_multi_ring_support
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/brodo/pcmcia-2.6
Linus Torvalds [Tue, 5 Dec 2006 23:52:06 +0000 (15:52 -0800)]
Merge /pub/scm/linux/kernel/git/brodo/pcmcia-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/brodo/pcmcia-2.6:
  [PATCH] pcmcia: at91_cf update
  [PATCH] pcmcia: fix m32r_cfc.c compilation
  [PATCH] pcmcia: ds.c debug enhancements
  [PATCH] pcmcia: at91_cf update
  [PATCH] pcmcia: conf.ConfigBase and conf.Present consolidation
  [PATCH] pcmcia: remove prod_id indirection
  [PATCH] pcmcia: remove manf_id and card_id indirection
  [PATCH] pcmcia: IDs for Elan serial PCMCIA devcies
  [PATCH] pcmcia: allow for four multifunction subdevices
  [PATCH] pcmcia: handle __copy_from_user() return value in ioctl
  [PATCH] pcmcia: multifunction card handling fixes
  [PATCH] pcmcia: allow shared IRQs on pd6729 sockets
  [PATCH] pcmcia: start over after CIS override
  [PATCH] cm4000_cs: fix return value check
  [PATCH] pcmcia: yet another IDE ID
  [PATCH] pcmcia: Add an id to ide-cs.c

17 years ago[SCSI] aic79xx: Add ASC-29320LPE ids to driver
Salyzyn, Mark [Tue, 5 Dec 2006 19:37:30 +0000 (14:37 -0500)]
[SCSI] aic79xx: Add ASC-29320LPE ids to driver

Simple patch to add the new PCIe version of the 29320 card.

Signed-off: Mark Salyzyn <Mark_Salyzyn@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: version update
Ed Lin [Tue, 5 Dec 2006 01:49:46 +0000 (17:49 -0800)]
[SCSI] stex: version update

Update version to 3.1.0.1

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: change wait loop code
Ed Lin [Tue, 5 Dec 2006 01:49:42 +0000 (17:49 -0800)]
[SCSI] stex: change wait loop code

The original wait loop may be much longer than intended time.
Use more accurate timer_after for it. Also adjust wait value to
avoid unnecessary long waiting.

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: add new device type support
Ed Lin [Tue, 5 Dec 2006 01:49:39 +0000 (17:49 -0800)]
[SCSI] stex: add new device type support

Add support for st_vsc1 type device (st_vsc is ok because it does not
require extra buffer).

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: update device id info
Ed Lin [Tue, 5 Dec 2006 01:49:36 +0000 (17:49 -0800)]
[SCSI] stex: update device id info

- add comments for various devices
- remove unused device ids(0xf350, 0x4301, 0x8301, 0x8302)
- add new device id(0xe350)
- fix vendor id of st_vsc
- modify Kconfig help info

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: adjust default queue length
Ed Lin [Tue, 5 Dec 2006 01:49:34 +0000 (17:49 -0800)]
[SCSI] stex: adjust default queue length

Firmware of new version may adjust default queue length. It is
backward compatible.

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: add value check in hard reset routine
Ed Lin [Tue, 5 Dec 2006 01:49:31 +0000 (17:49 -0800)]
[SCSI] stex: add value check in hard reset routine

During hard reset, an all-1 value from PCI_COMMAND should be invalid.

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: fix controller_info command handling
Ed Lin [Tue, 5 Dec 2006 01:49:28 +0000 (17:49 -0800)]
[SCSI] stex: fix controller_info command handling

This command needs information from both firmware and driver. First copy
information from firmware to buffer, then fill in driver information.

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] stex: fix biosparam calculation
Ed Lin [Tue, 5 Dec 2006 01:49:24 +0000 (17:49 -0800)]
[SCSI] stex: fix biosparam calculation

Fix biosparam calculation.

Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] megaraid: fix MMIO casts
Jeff Garzik [Mon, 4 Dec 2006 01:49:23 +0000 (20:49 -0500)]
[SCSI] megaraid: fix MMIO casts

megaraid's MMIO RD*/WR* macros directly call readl() and writel() with
an 'unsigned long' argument.  This throws a warning, but is otherwise OK
because the 'unsigned long' is really the result of ioremap().  This
setup is also OK because the variable can hold an ioremap cookie /or/ a
PCI I/O port (PIO).

However, to fix the warning thrown when readl() and writel() are passed
an unsigned long cookie, I introduce 'void __iomem *mmio_base', holding
the same value as 'base'.  This will silence the warnings, and also
cause an oops whenever these MMIO-only functions are ever accidentally
passed an I/O address.

Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] tgt: fix undefined flush_dcache_page() problem
akpm@osdl.org [Tue, 5 Dec 2006 07:06:48 +0000 (23:06 -0800)]
[SCSI] tgt: fix undefined flush_dcache_page() problem

drivers/scsi/scsi_tgt_if.c: In function 'tgt_uspace_send_event':
drivers/scsi/scsi_tgt_if.c:88: warning: implicit declaration of function 'flush_dcache_page'

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[PATCH] Centralise definitions of sector_t and blkcnt_t
Matthew Wilcox [Mon, 4 Dec 2006 10:38:31 +0000 (03:38 -0700)]
[PATCH] Centralise definitions of sector_t and blkcnt_t

CONFIG_LBD and CONFIG_LSF are spread into asm/types.h for no particularly
good reason.

Centralising the definition in linux/types.h means that arch maintainers
don't need to bother adding it, as well as fixing the problem with
x86-64 users being asked to make a decision that has absolutely no
effect.

The H8/300 porters seem particularly confused since I'm not aware of any
microcontrollers that need to support 2TB filesystems.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years agoMerge branch 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Tue, 5 Dec 2006 03:38:30 +0000 (19:38 -0800)]
Merge branch 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block

* 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block:
  [PATCH] blktrace: don't return blktrace_seq from trace_note()
  [PATCH] blktrace: uninline trace_note()

17 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfashe...
Linus Torvalds [Tue, 5 Dec 2006 03:33:45 +0000 (19:33 -0800)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/mfasheh/ocfs2

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2: (31 commits)
  ocfs2: implement i_op->permission
  configfs: make configfs_dirent_exists() static
  ocfs2: update file system paths to set atime
  ocfs2: core atime update functions
  ocfs2: Add splice support
  ocfs2: Remove ocfs2_write_should_remove_suid()
  [PATCH] Export should_remove_suid()
  configfs: mutex_lock_nested() fix
  ocfs2: Remove struct ocfs2_journal_handle in favor of handle_t
  ocfs2: remove handle argument to ocfs2_start_trans()
  ocfs2: remove ocfs2_journal_handle journal field
  ocfs2: pass ocfs2_super * into ocfs2_commit_trans()
  ocfs2: remove unused handle argument from ocfs2_meta_lock_full()
  ocfs2: make ocfs2_alloc_handle() static
  ocfs2: remove unused ocfs2_handle_add_lock()
  ocfs2: remove unused ocfs2_handle_add_inode()
  ocfs2: Don't allocate handle early in ocfs2_rename()
  ocfs2: don't use handle for locking in allocation functions
  ocfs2: don't pass handle to ocfs2_meta_lock in ocfs2_rename()
  ocfs2: don't pass handle to ocfs2_meta_lock in ocfs2_symlink()
  ...

17 years ago[PATCH] remote memory corruptor in ibmtr.c
Al Viro [Mon, 4 Dec 2006 22:05:09 +0000 (22:05 +0000)]
[PATCH] remote memory corruptor in ibmtr.c

ip_summed changes last summer had missed that one.  As the result,
we have ip_summed interpreted as CHECKSUM_PARTIAL now.  IOW,
->csum is interpreted as offset of checksum in the packet.  net/core/*
will both read and modify the value as that offset, with obvious
reasons.  At the very least it's a remote memory corruptor.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years ago[PATCH] selinux endianness annotations
Al Viro [Mon, 4 Dec 2006 22:00:55 +0000 (22:00 +0000)]
[PATCH] selinux endianness annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
17 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 5 Dec 2006 03:23:34 +0000 (19:23 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] Cleanup memory barriers for weakly ordered systems.
  [MIPS] Alchemy: Automatically enable CONFIG_RESOURCES_64BIT for PCI configs.
  [MIPS] Unify csum_partial.S
  [MIPS] SWARM: Fix a typo in #error directives
  [MIPS] Fix atomic.h build errors.
  [MIPS] Use SYSVIPC_COMPAT to fix various problems on N32
  [MIPS] klconfig add missing bracket

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Tue, 5 Dec 2006 03:22:33 +0000 (19:22 -0800)]
Merge /pub/scm/linux/kernel/git/paulus/powerpc

* master.kernel.org:/pub/scm/linux/kernel/git/paulus/powerpc: (194 commits)
  [POWERPC] Add missing EXPORTS for mpc52xx support
  [POWERPC] Remove obsolete PPC_52xx and update CLASSIC32 comment
  [POWERPC] ps3: add a default zImage target
  [POWERPC] Add of_platform_bus support to mpc52xx psc uart driver
  [POWERPC] typo fix and whitespace cleanup on mpc52xx-uart driver
  [POWERPC] Fix debug printks for 32-bit resources in the PCI code
  [POWERPC] Replace kmalloc+memset with kzalloc
  [POWERPC] Linkstation / kurobox support
  [POWERPC] Add the e300c3 core to the CPU table.
  [POWERPC] ppc: m48t35 add missing bracket
  [POWERPC] iSeries: don't build head_64.o unnecessarily
  [POWERPC] iSeries: stop dt_mod.o being rebuilt unnecessarily
  [POWERPC] Fix cputable.h for combined build
  [POWERPC] Allow CONFIG_BOOTX_TEXT on iSeries
  [POWERPC] Allow xmon to build on legacy iSeries
  [POWERPC] Change ppc64_defconfig to use AUTOFS_V4 not V3
  [POWERPC] Tell firmware we can handle POWER6 compatible mode
  [POWERPC] Clean images in arch/powerpc/boot
  [POWERPC] Fix OF pci flags parsing
  [POWERPC] defconfig for lite5200 board
  ...

17 years ago[PATCH] pcmcia: at91_cf update
Andrew Victor [Mon, 4 Dec 2006 10:26:36 +0000 (12:26 +0200)]
[PATCH] pcmcia: at91_cf update

This is an update to the AT91 CompactFlash driver.

We replace the hard-coded "chip select 4" with the chip-select value
passed via platform_data.  The configuration of the EBI memory
controller to enable Compact Flash access is now also handled in the
platform setup code and not in the driver.

Signed-off-by: Andrew Victor <andrew@sanpeople.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: fix m32r_cfc.c compilation
Adrian Bunk [Sat, 2 Dec 2006 17:55:06 +0000 (18:55 +0100)]
[PATCH] pcmcia: fix m32r_cfc.c compilation

More fallout of the post 2.6.19-rc1 IRQ changes...

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: ds.c debug enhancements
Dominik Brodowski [Sat, 2 Dec 2006 20:11:44 +0000 (22:11 +0200)]
[PATCH] pcmcia: ds.c debug enhancements

Add verbose error messages and debug information to ds.c

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: at91_cf update
Andrew Victor [Mon, 6 Nov 2006 13:56:07 +0000 (15:56 +0200)]
[PATCH] pcmcia: at91_cf update

This is an update to the AT91RM9200 CompactFlash driver.

The changes include:
- Use the I/O memory address passed via the platform_device resources
  instead of constant global values.
- The IRQ should not be used as a random'ness source.
- Return errors if ioremap() or request_mem_region() fails.

Signed-off-by: Andrew Victor <andrew@sanpeople.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: conf.ConfigBase and conf.Present consolidation
Dominik Brodowski [Thu, 26 Oct 2006 01:49:27 +0000 (21:49 -0400)]
[PATCH] pcmcia: conf.ConfigBase and conf.Present consolidation

struct pcmcia_device *p_dev->conf.ConfigBase and .Present are set in almost
all PCMICA driver right at the beginning, using the same calls but slightly
different implementations. Unfiy this in the PCMCIA core.

Includes a small bugfix ("drivers/net/pcmcia/xirc2ps_cs.c: remove unused
label") from and Signed-off-by Adrian Bunk <bunk@stusta.de>

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: remove prod_id indirection
Dominik Brodowski [Sun, 4 Jun 2006 16:06:13 +0000 (18:06 +0200)]
[PATCH] pcmcia: remove prod_id indirection

As we read out the product information strings (VERS_1) from the PCMCIA device
in the PCMCIA core, and device drivers can access those reliably in struct
pcmcia_device's fields prod_id[], remove additional product information string
detection logic from PCMCIA device drivers.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: remove manf_id and card_id indirection
Dominik Brodowski [Thu, 26 Oct 2006 01:28:53 +0000 (21:28 -0400)]
[PATCH] pcmcia: remove manf_id and card_id indirection

As we read out the manufactor and card_id from the PCMCIA device in the
PCMCIA core, and device drivers can access those reliably in struct
pcmcia_device's fields manf_id and card_id, remove additional (and partly
broken) manf_id and card_id detection logic from PCMCIA device drivers.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: IDs for Elan serial PCMCIA devcies
Tony Olech [Wed, 29 Nov 2006 08:54:51 +0000 (08:54 +0000)]
[PATCH] pcmcia: IDs for Elan serial PCMCIA devcies

Add IDs for Elan serial PCMCIA devices. In addition, move the
pdaudio_cf driver from matching based on manf_id and card_id to
the more specific prod_id1 and prod_id2 to avoid false
positives.

Signed-off-by: Tony Olech <tony.olech@elandigitalsystems.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: allow for four multifunction subdevices
Dominik Brodowski [Tue, 28 Nov 2006 02:03:48 +0000 (21:03 -0500)]
[PATCH] pcmcia: allow for four multifunction subdevices

Some Elan serial cards allow for four (independent) multifunction
subdevices. Teach the PCMCIA core to deal with such devices.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: handle __copy_from_user() return value in ioctl
Dominik Brodowski [Sun, 19 Nov 2006 16:21:27 +0000 (11:21 -0500)]
[PATCH] pcmcia: handle __copy_from_user() return value in ioctl

Handle __copy_from_user() return value.

drivers/pcmcia/pcmcia_ioctl.c:597: warning: ignoring return value of '__copy_from_user', declared with attribute warn_unused_result

Noticed and first fix by Randy Dunlap.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: multifunction card handling fixes
Dominik Brodowski [Tue, 7 Nov 2006 02:52:16 +0000 (21:52 -0500)]
[PATCH] pcmcia: multifunction card handling fixes

s->functions needs to be initialized earlier, for the "let's see
how high it increases" approach means that pcmcia_request_irq()
(which makes use of this value) is confused, and might request
an exclusive IRQ first even though it is not supposed to.

Also, a CIS override autoloaded using the firmware loader may
allow for the use of more or less functions in a multifunction
card. Therefore, we may need to schedule a call to add this
second function later on, or simply remove the other function
(it's always the first -valid- function which reaches this
codepath).

Many thanks to Fabrice Bellet for debugging and testing patches.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: allow shared IRQs on pd6729 sockets
Komuro [Sat, 4 Nov 2006 01:00:47 +0000 (10:00 +0900)]
[PATCH] pcmcia: allow shared IRQs on pd6729 sockets

Use IRQF_PROBE_SHARED flag for request_irq() to find an unused
interrupt for PCMCIA cards.

Signed-off-by: Komuro <komurojun-mbn@nifty.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: start over after CIS override
Dominik Brodowski [Fri, 3 Nov 2006 15:54:00 +0000 (10:54 -0500)]
[PATCH] pcmcia: start over after CIS override

When overriding the CIS, re-start the configuration of the card from
scratch. Reported and debugged by Fabrice Bellet <fabrice@bellet.info>

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] cm4000_cs: fix return value check
Akinobu Mita [Tue, 17 Oct 2006 06:25:59 +0000 (15:25 +0900)]
[PATCH] cm4000_cs: fix return value check

The return value of class_create() need to be checked with IS_ERR().
And register_chrdev() returns errno on failure.
This patch includes these fixes for cm4000_cs and cm4040_cs.

Signed-off-by: Akinbou Mita <akinobu.mita@gmail.com>
Signed-off-by: Harald Welte <laforge@gnumonks.org>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: yet another IDE ID
Marcin Juszkiewicz [Mon, 30 Oct 2006 21:28:09 +0000 (22:28 +0100)]
[PATCH] pcmcia: yet another IDE ID

Microdrive reported by one of OpenEmbedded developers.

product info: "WEIDA", "TWTTI", ""
manfid: 0x000a, 0x0000
function: 4 (fixed disk)

(equivalent update to pata_pcmcia.c by Dominik Brodowski)

Signed-off-by: Marcin Juszkiewicz <openembedded@hrw.one.pl>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[PATCH] pcmcia: Add an id to ide-cs.c
Matt Reimer [Thu, 26 Oct 2006 22:56:00 +0000 (15:56 -0700)]
[PATCH] pcmcia: Add an id to ide-cs.c

Add an ID entry for:

product info: "TRANSCEND", "TS1GCF80", "", ""
manfid: 0x000a, 0x0000
function: 4 (fixed disk)

Signed-off-by: Matt Reimer <mreimer@vpop.net>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
17 years ago[POWERPC] Add missing EXPORTS for mpc52xx support
Grant Likely [Tue, 5 Dec 2006 00:29:12 +0000 (17:29 -0700)]
[POWERPC] Add missing EXPORTS for mpc52xx support

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Remove obsolete PPC_52xx and update CLASSIC32 comment
Grant Likely [Tue, 5 Dec 2006 00:29:13 +0000 (17:29 -0700)]
[POWERPC] Remove obsolete PPC_52xx and update CLASSIC32 comment

The support for the 52xx-based systems is now included under
CONFIG_CLASSIC32, since the 52xx chips have a 603e-based core.

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] ps3: add a default zImage target
Geoff Levand [Tue, 5 Dec 2006 00:11:38 +0000 (16:11 -0800)]
[POWERPC] ps3: add a default zImage target

Add a powerpc make target that can be loaded by the ps3 bootloader (kboot) and
set this as the default image to build for that platform.

Until the compressed zImage wrapper is made, this arranges for a stripped
vmlinux image to be built.

Signed-off-by: Geoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>