Patch from Nicolas Pitre
Since:
if (unlikely(__res || __ex_flag))
produces worse code on ARM than:
if (unlikely(__res | __ex_flag))
I therefore made it more explicit:
__res |= __ex_flag;
if (unlikely(__res != 0))
so it is not seen as a typo again.
Also made everything static inline rather than macros for better readability
(both produce the same code after all).
And finally added missing \t from multi-line assembly code.
Signed-off-by: Nicolas Pitre <[email protected]>
Acked-by: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Russell King <[email protected]>
negative
Patch from Nicolas Pitre
The pre ARMv5 implementation can be aborted if an exception occurs in
the middle of it. Because of that, the ARMv6 implementation doesn't
re-attempt the operation on a failed strex either. Let's make this
transient nature of such a false positive more explicit in the
definition.
Signed-off-by: Nicolas Pitre <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Nicolas Pitre
The cmpxchg emulation on pre-ARMv5 relies on user code executed from a
kernel address. If the operation cannot complete atomically, it is
aborted from the usr_entry macro by clearing the Z flag. This clearing
of the Z flag is done whenever the user pc is above TASK_SIZE.
However this "pc >= TASK_SIZE" test cannot work in the non MMU case.
Worse: the current code will corrupt the Z flag on every entry to the
kernel.
Let's disable it in the non MMU case for now. Using NPTL on non MMU
targets needs to be worked out anyway.
Signed-off-by: Nicolas Pitre <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Nicolas Pitre
struct sockaddr_un loses its padding with EABI. Since the size of the
structure is used as a validation test in unix_mkname(), we need to
change the length argument to 110 whenever it is 112.
Signed-off-by: Nicolas Pitre <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Nicolas Pitre
Although OABI_COMPAT works fine in most cases, it is still experimental
and could be for ever since it is nearly impossible to handle
everything, e.g. ioctls.
Signed-off-by: Nicolas Pitre <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Martin Michlmayr
Minor typographical and spelling fixes in Konfig
Signed-off-by: Martin Michlmayr <[email protected]>
---
Kconfig | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
Signed-off-by: Russell King <[email protected]>
Patch from Martin Michlmayr
Add help descriptions to ARCH config items that don't have one.
Signed-off-by: Martin Michlmayr <[email protected]>
---
Kconfig | 32 ++++++++++++++++++++++++++++++--
mach-clps711x/Kconfig | 2 ++
2 files changed, 32 insertions(+), 2 deletions(-)
Signed-off-by: Russell King <[email protected]>
Patch from Ben Dooks
Move to using an enable count for the shared clocks
and protect the clock system using a mutex instead
of just disabling IRQs during the clock update.
Since there is little more code in the path for
non-shared clocks, the enable and disable calls
use the same code for each.
Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Lennert Buytenhek
Most ixp2000 boards don't actually work if pci=firmware isn't used, so
the defconfig isn't really the right place to specify this. Instead of
specifying it in the defconfigs, make the relevant board code take care
of setting pci=firmware.
Signed-off-by: Lennert Buytenhek <[email protected]>
Signed-off-by: Deepak Saxena <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Lennert Buytenhek
The ixdp2x01_clock is already 50MHz by default, so no need to
override it with 50MHz in the ixdp2801 defconfig as is done now,
which is confusing as well.
Signed-off-by: Lennert Buytenhek <[email protected]>
Signed-off-by: Deepak Saxena <[email protected]>
Signed-off-by: Russell King <[email protected]>
Patch from Lennert Buytenhek
The ixdp2x01 pci init call doesn't check whether it's really running
on an ixdp2x01, making it impossible to compile one kernel that works
on both the ixdp2x01 and another ixp2000 board.
Signed-off-by: Lennert Buytenhek <[email protected]>
Signed-off-by: Deepak Saxena <[email protected]>
Signed-off-by: Russell King <[email protected]>
Prevent SN2 specific code to be executed on non SN2 platforms when
running a generic kernel.
Signed-off-by: Jes Sorensen <[email protected]>
Signed-off-by: Tony Luck <[email protected]>
All the percpu data structure walkers want cpu_possible_map to be
initialized early, but alpha instead populated "hwrpb_cpu_present_mask"
early in setup_smp(), and then initialized cpu_possible_map only much
later.
Thanks go to Heiko Carstens and Dipankar Sarma for noticing.
This fixes it and we can get rid of hwrpb_cpu_present_mask entirely.
Signed-off-by: Linus Torvalds <[email protected]>
fix CONFIG_SLOB=y (when CONFIG_SMP=y): get rid of the 'align' parameter
from its __alloc_percpu() implementation. Boot-tested on x86.
Signed-off-by: Ingo Molnar <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
q->ordcolor must only be flipped on initial queueing of a hardbarrier
request.
Constructing ordered sequence and requeueing used to pass through
__elv_add_request() which flips q->ordcolor when it sees a barrier
request.
This patch separates out elv_insert() from __elv_add_request() and uses
elv_insert() when constructing ordered sequence and requeueing.
elv_insert() inserts the given request at the specified position and
does nothing else.
Signed-off-by: Tejun Heo <[email protected]>
Acked-by: Jens Axboe <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Quoth BenH:
"Ok, looks like I forgot to update the Kconfig for the new i2c driver,
it should select I2C_POWERMAC instead. Do you want a new patch or
can you just fix it there ?"
Signed-off-by: Linus Torvalds <[email protected]>
When issuing an SG_IO ioctl through sd that resulted in an unrecoverable
error, a nearly infinite retry loop was discovered. This is due to the
fact that the block layer SG_IO code is not setting up rq->retries. This
patch also fixes up the sg_scsi_ioctl path.
Signed-off-by: Brian King <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
This patch adds a windfarm module, windfarm_pm112, for the dual core G5s
(both 2 and 4 core models), keeping the machine from getting into
vacuum-cleaner mode ;) For proper credits, the patch was initially
written by Paul Mackerras, and slightly reworked by me to add overtemp
handling among others. The patch also removes the sysfs attributes from
windfarm_pm81 and windfarm_pm91 and instead adds code to the windfarm
core to automagically expose attributes for sensor & controls.
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
* compat_alloc_user_space() returns __user pointer
* copying between two userland areas is copy_in_user(), not copy_from_user()
* dereferencing userland pointers is bad
* so's get_user() from local variables
... plus usual __user annotations
Signed-off-by: Al Viro <[email protected]>
A pile of internal functions use only inside mips io.h has names starting
with mem_... and clashing with names in drivers; renamed to __mem_....
Signed-off-by: Al Viro <[email protected]>
MTD_XIP depends on having working asm/mtd-xip.h; it's not just per-architecture
(arm-only, as current Kconfig would have it), but actually per-subarch as
well. Introduced a new symbol (ARCH_MTD_XIP) set by arch Kconfig; MTD_XIP
depends on it.
Signed-off-by: Al Viro <[email protected]>
The PowerMac sound drivers used to rely on a "bug" of the i2c-keywest
driver that implemented I2C_SMBUS_BLOCK_DATA incorrectly, that is it did
what I2C_SMBUS_I2C_BLOCK_DATA should have done. The new i2c-powermac
driver that replaces keywest has this bug fixed, thus the sound drivers
must be fixed too.
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Do not enable CONFIG_LOCALVERSION_AUTO by default.
When doing kernel development it just leaves a ton
of crap around.
Signed-off-by: David S. Miller <[email protected]>
Also, the Solaris syscall table is sized differrently,
and does not go beyond entry 255, so trim off the excess
entries.
Signed-off-by: David S. Miller <[email protected]>
