17 Oct, 2007
3 commits
-
[apw@shadowen.org: style fixups]
[apw@shadowen.org: vmemmap sparc64: convert to new config options]
Signed-off-by: Andy Whitcroft
Acked-by: Mel Gorman
Acked-by: Christoph Lameter
Cc: KAMEZAWA Hiroyuki
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
x86(-64) are the last architectures still using the page fault notifier
cruft for the kprobes page fault hook. This patch converts them to the
proper direct calls, and removes the now unused pagefault notifier bits
aswell as the cruft in kprobes.c that was related to this mess.I know Andi didn't really like this, but all other architecture maintainers
agreed the direct calls are much better and besides the obvious cruft
removal a common way of dealing with kprobes across architectures is
important aswell.[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: fix sparc64]
Signed-off-by: Christoph Hellwig
Cc: Andi Kleen
Cc:
Cc: Prasanna S Panchamukhi
Cc: Ananth N Mavinakayanahalli
Cc: Anil S Keshavamurthy
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Convert cpu_sibling_map from a static array sized by NR_CPUS to a per_cpu
variable. This saves sizeof(cpumask_t) * NR unused cpus. Access is mostly
from startup and CPU HOTPLUG functions.Signed-off-by: Mike Travis
Cc: Andi Kleen
Cc: Christoph Lameter
Cc: "Siddha, Suresh B"
Cc: "David S. Miller"
Cc: Paul Mackerras
Cc: Benjamin Herrenschmidt
Cc: "Luck, Tony"
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
14 Oct, 2007
6 commits
-
We used to need this to compute virt_irq --> ino, but that
is no longer necessary.Signed-off-by: David S. Miller
-
We have a place to stick INO information in the
virt_to_real_irq_table[], which is currently only used for VIRQs.
And that is readily accessible from the one __irq_ino() call site.Signed-off-by: David S. Miller
-
Signed-off-by: David S. Miller
-
Currently we chain IVEC entries using 32-bit "pointers"
because we know that the ivector_table is in the main
kernel image, thus below 4GB.This uses proper 64-bit pointers instead.
Whilst this bloats up the kernel image size, this sets
the infrastructure necessary to significantly shrink the
kernel size by using physical addresses and dynamically
allocating the ivector table.Signed-off-by: David S. Miller
-
This also makes us use the MSI queues correctly.
Each MSI queue is serviced by a normal sun4u/sun4v INO interrupt
handler. This handler runs the MSI queue and dispatches the
virtual interrupts indicated by arriving MSIs in that MSI queue.All of the common logic is placed in pci_msi.c, with callbacks to
handle the PCI controller specific aspects of the operations.This common infrastructure will make it much easier to add MSG
support.Signed-off-by: David S. Miller
-
The support code is identical to the hypervisor sun4v stuff,
just replacing the hypervisor calls with register reads and
writes in the Fire controller.Signed-off-by: David S. Miller
17 Sep, 2007
1 commit
-
As noted by Al Viro, when we try to call prom_set_trap_table()
in the SMP trampoline code we try to take the PROM call spinlock
which doesn't work because the current thread pointer isn't
valid yet and lockdep depends upon that being correct.Furthermore, we cannot set the current thread pointer register
because it can't be properly dereferenced until we return from
prom_set_trap_table(). Kernel TLB misses only work after that
call.So do the PROM call to set the trap table directly instead of
going through the OBP library C code, and thus avoid the lock
altogether.These calls are guarenteed to be serialized fully.
Since there are now no calls to the prom_set_trap_table{_sun4v}()
library functions, they can be deleted.Signed-off-by: David S. Miller
31 Aug, 2007
2 commits
-
1) sun4{u,v}_build_msi() have improper return value handling.
We should always return negative error codes, instead of
using the magic value "0" which could in fact be a valid
MSI number.2) sun4{u,v}_build_msi() should return -ENOMEM instead of
calling prom_prom() halt with kzalloc() of the interrupt
data fails.3) We 'remembered' the MSI number using a singleton in the
struct device archdata area, this doesn't work for MSI-X
which can cause multiple MSIs assosciated with one device.Delete that archdata member, and instead store the MSI
number in the IRQ chip data area.Signed-off-by: David S. Miller
-
Sometimes we were using 32-bit values and the top bits were
getting inadvertantly chopped off. This will matter for the
forthcoming Fire controller MSI support.Signed-off-by: David S. Miller
21 Aug, 2007
1 commit
-
Otherwise the compiler can't see the things like the
per-cpu area base register are changing.Signed-off-by: David S. Miller
17 Aug, 2007
2 commits
-
The underflow exception cases were wrong.
This is one weird area of ieee1754 handling in that the underflow
behavior changes based upon whether underflow is enabled in the trap
enable mask of the FPU control register. As a specific case the Sparc
V9 manual gives us the following description:--------------------
If UFM = 0: Underflow occurs if a nonzero result is tiny and a
loss of accuracy occurs. Tininess may be detected
before or after rounding. Loss of accuracy may be
either a denormalization loss or an inexact result.If UFM = 1: Underflow occurs if a nonzero result is tiny.
Tininess may be detected before or after rounding.
--------------------What this amounts to in the packing case is if we go subnormal,
we set underflow if any of the following are true:1) rounding sets inexact
2) we ended up rounding back up to normal (this is the case where
we set the exponent to 1 and set the fraction to zero), this
should set inexact too
3) underflow is set in FPU control register trap-enable maskThe initially discovered example was "DBL_MIN / 16.0" which
incorrectly generated an underflow. It should not, unless underflow
is set in the trap-enable mask of the FPU csr.Another example, "0x0.0000000000001p-1022 / 16.0", should signal both
inexact and underflow. The cpu implementations and ieee1754
literature is very clear about this. This is case #2 above.However, if underflow is set in the trap enable mask, only underflow
should be set and reported as a trap. That is handled properly by the
prioritization logic inarch/sparc{,64}/math-emu/math.c:record_exception().
Based upon a report and test case from Jakub Jelinek.
Signed-off-by: David S. Miller
-
Signed-off-by: David S. Miller
09 Aug, 2007
1 commit
-
Check the cpu type in the OBP device tree before committing to
using the optimized Niagara memcpy and memset implementation.If we don't recognize the cpu type, use a completely generic
version.Signed-off-by: David S. Miller
08 Aug, 2007
1 commit
-
Signed-off-by: David S. Miller
04 Aug, 2007
2 commits
-
The one choosen by asm-generic/fcntl.h is not appropriate
for this platform.Noticed by Ulrich Drepper.
Signed-off-by: David S. Miller
-
Reported by Andrew Morton.
Signed-off-by: David S. Miller
31 Jul, 2007
1 commit
-
drivers/infiniband/hw/mthca/mthca_main.c: In function `mthca_init_icm':
drivers/infiniband/hw/mthca/mthca_main.c:468: error: implicit declaration of function `dma_get_cache_alignment'Pinch the one from asm-generic/dma-mapping.h
Signed-off-by: Andrew Morton
Signed-off-by: David S. Miller
30 Jul, 2007
3 commits
-
They are handled in a ->compat_ioctl() handler, so it's just noise
when compat_ioctl.c warns which occurs when they are used on non-SBUS
framebuffer devices.Signed-off-by: David S. Miller
-
It uses pci_dev, calls pci_map_*(), etc.
Signed-off-by: David S. Miller
-
Fully unify all of the DMA ops so that subordinate bus types to
the DMA operation providers (such as ebus, isa, of_device) can
work transparently.Basically, we just make sure that for every system device we
create, the dev->archdata 'iommu' and 'stc' fields are filled
in.Then we have two platform variants of the DMA ops, one for SUN4U which
actually programs the real hardware, and one for SUN4V which makes
hypervisor calls.This also fixes the crashes in parport_pc on sparc64, reported by
Meelis Roos.Signed-off-by: David S. Miller
21 Jul, 2007
4 commits
-
Signed-off-by: David S. Miller
-
Signed-off-by: David S. Miller
-
The current scheme works on static interpretation of text names, which
is wrong.The output-device setting, for example, must be resolved via an alias
or similar to a full path name to the console device.Paths also contain an optional set of 'options', which starts with a
colon at the end of the path. The option area is used to specify
which of two serial ports ('a' or 'b') the path refers to when a
device node drives multiple ports. 'a' is assumed if the option
specification is missing.This was caught by the UltraSPARC-T1 simulator. The 'output-device'
property was set to 'ttya' and we didn't pick upon the fact that this
is an OBP alias set to '/virtual-devices/console'. Instead we saw it
as the first serial console device, instead of the hypervisor console.The infrastructure is now there to take advantage of this to resolve
the console correctly even in multi-head situations in fbcon too.Thanks to Greg Onufer for the bug report.
Signed-off-by: David S. Miller
-
* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/sfr/ofcons:
Create drivers/of/platform.c
Create linux/of_platorm.h
[SPARC/64] Rename some functions like PowerPC
Begin consolidation of of_device.h
Begin to consolidate of_device.c
Consolidate of_find_node_by routines
Consolidate of_get_next_child
Consolidate of_get_parent
Consolidate of_find_property
Consolidate of_device_is_compatible
Start split out of common open firmware code
Split out common parts of prom.h
20 Jul, 2007
13 commits
-
The "id" property in vdc-port nodes are not unique, they
are all zero. Therefore assign ID's using the parent's
"cfg-handle" property which will be unique.Signed-off-by: David S. Miller
-
Signed-off-by: David S. Miller
-
Signed-off-by: David S. Miller
-
Move common stuff from asm-powerpc/of_platform.h to here and
move the common bits from asm-sparc*/of_device.h here as well.Create asm-sparc*/of_platform.h and move appropriate parts of
of_device.h to them.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
This is to make the of merge easier. Also rename of_bus_type.
Signed-off-by: Stephen Rothwell
Acked-by: David S. Miller -
This just moves the common stuff from the arch of_device.h files to
linux/of_device.h.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
This consolidates the routines of_find_node_by_path, of_find_node_by_name,
of_find_node_by_type and of_find_compatible_device. Again, the comparison
of strings are done differently by Sparc and PowerPC and also these add
read_locks around the iterations.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
This requires creating dummy of_node_{get,put} routines for sparc and
sparc64. It also adds a read_lock around the parent accesses.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
The only change here is that a readlock is taken while the property list
is being traversed on Sparc where it was not taken previously.Also, Sparc uses strcasecmp to compare property names while PowerPC
uses strcmp.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
The only difference here is that Sparc uses strncmp to match compatibility
names while PowerPC uses strncasecmp.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
This creates drivers/of/base.c (depending on CONFIG_OF) and puts
the first trivially common bits from the prom.c files into it.Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
This creates linux/of.h and includes asm/prom.h from it.
We also include linux/of.h from asm/prom.h while we transition.
Signed-off-by: Stephen Rothwell
Acked-by: Paul Mackerras
Acked-by: David S. Miller -
Since Ingo's recent scheduler rewrite which was merged as commit
0437e109e1841607f2988891eaa36c531c6aa6ac sched_cacheflush is unused.Signed-off-by: Ralf Baechle
Signed-off-by: Ingo Molnar
