per cpu data section contains two types of data. One set which is
exclusively accessed by the local cpu and the other set which is per cpu,
but also shared by remote cpus. In the current kernel, these two sets are
not clearely separated out. This can potentially cause the same data
cacheline shared between the two sets of data, which will result in
unnecessary bouncing of the cacheline between cpus.
One way to fix the problem is to cacheline align the remotely accessed per
cpu data, both at the beginning and at the end. Because of the padding at
both ends, this will likely cause some memory wastage and also the
interface to achieve this is not clean.
This patch:
Moves the remotely accessed per cpu data (which is currently marked
as ____cacheline_aligned_in_smp) into a different section, where all the data
elements are cacheline aligned. And as such, this differentiates the local
only data and remotely accessed data cleanly.
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Acked-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: <linux-arch@vger.kernel.org>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Identical implementations of PTRACE_POKEDATA go into generic_ptrace_pokedata()
function.
AFAICS, fix bug on xtensa where successful PTRACE_POKEDATA will nevertheless
return EPERM.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If the kernel OOPSed or BUGed then it probably should be considered as
tainted. Thus, all subsequent OOPSes and SysRq dumps will report the
tainted kernel. This saves a lot of time explaining oddities in the
calltraces.
Signed-off-by: Pavel Emelianov <xemul@openvz.org>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
[ Added parisc patch from Matthew Wilson -Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6: (68 commits)
sh: sh-rtc support for SH7709.
sh: Revert __xdiv64_32 size change.
sh: Update r7785rp defconfig.
sh: Export div symbols for GCC 4.2 and ST GCC.
sh: fix race in parallel out-of-tree build
sh: Kill off dead mach.c for hp6xx.
sh: hd64461.h cleanup and added comments.
sh: Update the alignment when 4K stacks are used.
sh: Add a .bss.page_aligned section for 4K stacks.
sh: Don't let SH-4A clobber SH-4 CFLAGS.
sh: Add parport stub for SuperIO ports.
sh: Drop -Wa,-dsp for DSP tuning.
sh: Update dreamcast defconfig.
fb: pvr2fb: A few more __devinit annotations for PCI.
fb: pvr2fb: Fix up section mismatch warnings.
sh: Select IPR-IRQ for SH7091.
sh: Correct __xdiv64_32/div64_32 return value size.
sh: Fix timer-tmu build for SH-3.
sh: Add cpu and mach links to CLEAN_FILES.
sh: Preliminary support for the SH-X3 CPU.
...
The current generic bug implementation has a call to dump_stack() in case a
WARN_ON(whatever) gets hit. Since report_bug(), which calls dump_stack(),
gets called from an exception handler we can do better: just pass the
pt_regs structure to report_bug() and pass it to show_regs() in case of a
warning. This will give more debug informations like register contents,
etc... In addition this avoids some pointless lines that dump_stack()
emits, since it includes a stack backtrace of the exception handler which
is of no interest in case of a warning. E.g. on s390 the following lines
are currently always present in a stack backtrace if dump_stack() gets
called from report_bug():
[<000000000001517a>] show_trace+0x92/0xe8)
[<0000000000015270>] show_stack+0xa0/0xd0
[<00000000000152ce>] dump_stack+0x2e/0x3c
[<0000000000195450>] report_bug+0x98/0xf8
[<0000000000016cc8>] illegal_op+0x1fc/0x21c
[<00000000000227d6>] sysc_return+0x0/0x10
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
Acked-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: Kyle McMartin <kyle@parisc-linux.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
GCC 4.2 can emit integer variants of the FP division routines, so
these need to be exported in order to keep the modules happy.
4.1.x versions of the ST compiler have these things backported,
and so also generate these symbols (whereas vanilla gcc 4.1.x
does not), so handle the __GNUC_STM_RELEASE__ case to accomodate
updated versions of the 4.1.x toolchain.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Use the newly added .bss.page_aligned section for aligning the stacks
rather than THREAD_SIZE.
Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
We need to know the CPU ID in order to calculate the mask and ack
registers effectively. Stub this in for UP.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
The current implementation only handles -ERESTARTNOHAND, whereas we
also need to handle -ERESTART_RESTARTBLOCK in the handle_signal()
case for restartable system calls.
As noted by Carl:
This fixes the LTP test nanosleep03 - the current kernel causes
-ERESTART_RESTARTBLOCK to reach user space rather than the correct
-EINTR.
Reported-by: Carl Shaw <shaw.carl@gmail.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
As Russell helpfully pointed out on linux-arch:
http://marc.info/?l=linux-arch&m=118208089204630&w=2
We were missing the oops_enter/exit() in the sh die() implementation.
As we do support lockdep, it's beneficial to add these calls so lockdep
properly disables itself in the die() case.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
We use R0 as the 5th argument of syscall. When the syscall restarts
after signal handling, we should restore the old value of R0.
The attached patch does it. Without this patch, I've experienced random
failures in the situation which signals are issued frequently.
Signed-off-by: Kaz Kojima <kkojima@rr.iij4u.or.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This patch reworks the ipr code by grouping the offset array together
with the ipr_data structure in a new data structure called ipr_desc.
This new structure also contains the name of the controller in struct
irq_chip. The idea behind putting struct irq_chip in there is that we
can use offsetof() to locate the base addresses in the irq_chip
callbacks. This strategy has much in common with the recently merged
intc2 code.
One logic change has been made - the original ipr code enabled the
interrupts by default but with this patch they are all disabled by
default.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
The shared intc2 code currently contains cpu-specific #ifdefs.
This is a tad unclean and it prevents us from using the shared code
to drive board-specific irqs on the se7780 board.
This patch reworks the intc2 code by moving the base addresses of
the intc2 registers into struct intc2_desc. This new structure also
contains the name of the controller in struct irq_chip. The idea
behind putting struct irq_chip in there is that we can use offsetof()
to locate the base addresses in the irq_chip callbacks.
One logic change has been made - the original shared intc2 code
enabled the interrupts by default but with this patch they are all
disabled by default.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
SH-2 can presently get in to some pretty bogus states, so
we tidy up the dependencies a bit and get it all building
again.
This gets us a bit closer to a functional allyesconfig
and allmodconfig, though there are still a few things to
fix up.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This was added during 2.5.x, but was never moved along. This
can easily be resurrected if someone has one they wish to work
with, but it's not worth keeping around in its current form.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Currently cpu_to_node() is always 0 in the UP case, though
we do want to have the CPU association linked in under sysfs
even in the cases where we're only on a single CPU.
Fix this up, so we have the cpu0 link on all of the available
nodes that don't already have a CPU link of their own.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This adds the URAM block on SH7722 as a separate node.
Sparsemem is required for this, or it can simply be disabled
by explicitly selecting a flatmem model.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
We have to call in to sparse_memory_present_with_active_regions()
earlier in order for sparsemem to be happy. This was being called
too late, and was causing troubles with the platforms that needed
to enable sparsemem.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This implements basic sparsemem support for SH. Presently this only
uses static sparsemem, and we still permit explicit selection of
flatmem. Those boards that want sparsemem can select it as usual.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
.machvec.init can be misaligned with the recent machvec changes,
forcibly align it on the boundary that it expects, as before.
Signed-off-by: Takashi YOSHII <takashi.yoshii.ze@hitachi.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Now that select no longer works for selecting the "closest" CPU,
we have to explicitly reference the precise sub-type in the few
places where it actually matters (presently only setup code and
some legacy sh-sci cruft).
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This kills off the BareCPU board as a "special" machvec, rather,
we leave this as a default for when no other vector is available,
or when we want to use it in combination with other vectors for
testing with generic ops. As sh_mv is copied out anyways (or
overloaded when an alternate vector is explicitly selected), this
doesn't consume any additional memory.
The generic machvec can be forcibly selected with sh_mv=generic,
or by not having any other boards enabled.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
We now throw all of the machvecs in to .machvec.init and either
select one on the command line, or copy out the first (and
usually only) one to sh_mv. The rest are freed as usual.
This gets rid of all of the silly sh_mv aliasing and makes the
selection explicit rather than link-order dependent.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This fixes up much of the machvec handling, allowing for it to be
overloaded on boot. Making practical use of this still requires
some Kconfig munging, however.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This adds in some more __user annotations. These weren't being
handled properly in some of the __get_user and __put_user paths,
so tidy those up.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Split out the CPU topology initialization to a separate file,
and switch it to a percpu type, rather than an NR_CPUS array.
At the same time, switch to only registering present CPUs,
rather than using the possible CPU map.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
With the SH7722 changes, ->set_rate() also takes an algo_id,
SH4-202 was overlooked when this change went in.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
If CONFIG_KGDB_NMI is disabled, we're left with a stray in_nmi
reference that can't be resolved. Move the symbol under the ifdef,
too.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Several errors were spotted during building for custom config (SMP
included). Although SMP still does not compile (no ipi and
__smp_call_function) and does not work, this looks a bit cleaner.
Some other errors obtained via gcc-4.1.0 build.
Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
CC arch/sh/kernel/vsyscall/vsyscall.o
a/arch/sh/kernel/vsyscall/vsyscall.c: In function 'arch_setup_additional_pages':
a/arch/sh/kernel/vsyscall/vsyscall.c:63: error: dereferencing pointer to incomplete type
a/arch/sh/kernel/vsyscall/vsyscall.c:67: error: dereferencing pointer to incomplete type
a/arch/sh/kernel/vsyscall/vsyscall.c:82: error: dereferencing pointer to incomplete type
a/arch/sh/kernel/vsyscall/vsyscall.c:85: error: dereferencing pointer to incomplete type
a/arch/sh/kernel/vsyscall/vsyscall.c: In function 'arch_vma_name':
a/arch/sh/kernel/vsyscall/vsyscall.c:91: error: dereferencing pointer to incomplete type
Signed-off-by: Manuel Lauss <mano@roarinelk.homelinux.net>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Add lost in_nmi definition to solve pcrel too far.
Signed-off-by: Takashi YOSHII <takasi-y@ops.dti.ne.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This fixes up the master clock multiplier and initial rate
propagation for the SH7722 clocks.
Signed-off-by: dmitry pervushin <dimka@nomadgs.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Now that we have the basic kdump support in place, add it in to
die() so we can enter the crash kernel automatically.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
As pointed out by Saito-san, without the sr.bl manipulation we can
occasionally hit delays in the idle loop due to interrupt handling, so
ensure that interrupts are blocked before going to sleep.
At the same time, we throw in TIF_POLLING_NRFLAG for the !hlt_counter
case (primarily used by the ST-40 parts).
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
- setup-sh7750.c only defines the sh7751_ipr_map when building
with SH7751 support.
- 7722 Solution Engine was missing a mach-type entry, causing
the macro in cf-enabler to be undefined.
- arch/sh/mm/init.c needs linux/pagemap.h.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
