Commits · 8f0bcbcab016324c2a3ba4cc715e8e523c29a578 · nexedi / linux

12 Dec, 2012 4 commits

MN10300: fix SMP synchronization between txdma and serial driver · 8f0bcbca

Mark Salter authored Dec 12, 2012

The SoC serial port driver uses a high priority interrupt to handle tx of
characters in the tx ring buffer. The driver needs to disable/enable this IRQ
from outside of irq context. The original code to do this is not foolproof on
SMP machines because the IRQ running on one core could still access the serial
port for a short time after the driver running on another core disables the
interrupt. This patch adds a flag to tell the IRQ handler that the driver
wants to disable the interrupt. After seeing the flag, the IRQ handler will
immediately disable the interrupt and exit. After setting the flag, the driver
will wait for interrupt to be disabled by the IRQ handler.
Signed-off-by: Mark Salter <msalter@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>

8f0bcbca

MN10300: fix serial port vdma irq setup for SMP · 8d160027

Mark Salter authored Dec 12, 2012

The builtin SoC serial ports have no FIFOs and use a virtual DMA mechanism
based on high priority IRQs to avoid overruns. These high priority interrupts
are pinned to cpu#0 on SMP systems. This patch fixes a problem with SMP where
the set_intr_level() interface is used to set the priority for these IRQs. The
set_intr_level() function sets priority on the local cpu but on SMP systems,
this code may be run on some other cpu than the one handling the interrupts.
Instead of setting interrupt level explicitly, this patch uses a special
irq_chip for these interrupts so that the mask/unmask methods can set the
interrupt level implicitly.
Signed-off-by: Mark Salter <msalter@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>

8d160027

MN10300: cleanup IRQ affinity setting · 7d361cb7

Mark Salter authored Dec 12, 2012

The irq_set_affinity handler for the mn10300 cpu pic had some hard-coded IRQs
which were not to be migrated from one cpu to another. This patch cleans those
up by using a combination of IRQF_NOBALANCING and specialized irq chips with
no irq_set_affinity handler. This maintains the previous behavior by using
generic IRQ interfaces rather than hard coding IRQ numbers in the default
irq_set_affinity handler.
Signed-off-by: Mark Salter <msalter@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>

7d361cb7

MN10300: ttySM: Use memory barriers correctly in circular buffer logic · c98c406e

David Howells authored Dec 12, 2012

Use memory barriers correctly in the circular buffer logic used in the driver,
as documented in Documentation/circular-buffers.txt.
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Mark Salter <msalter@redhat.com>

c98c406e

29 Nov, 2012 36 commits
- do_coredump(): get rid of pt_regs argument · 541880d9
  Al Viro authored Nov 05, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  541880d9
- print_fatal_signal(): get rid of pt_regs argument · 4aaefee5
  Al Viro authored Nov 05, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  4aaefee5
- ptrace_signal(): get rid of unused arguments · 94eb22d5
  Al Viro authored Nov 05, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  94eb22d5
- get rid of ptrace_signal_deliver() arguments · b7f9591c
  Al Viro authored Nov 05, 2012
```
the first one is equal to signal_pt_regs(), the second is never used
(and always NULL, while we are at it).
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  b7f9591c
- new helper: signal_pt_regs() · 22062a96
  Al Viro authored Nov 05, 2012
```
Always equal to task_pt_regs(current); defined only when we are in
signal delivery.  It may be different from current_pt_regs() - e.g.
architectures like m68k may have pt_regs location on exception
different from that on a syscall and signals (just as ptrace handling)
may happen on exceptions as well as on syscalls.

When they are equal, it's often better to have signal_pt_regs
defined (in asm/ptrace.h) as current_pt_regs - that tends to be
optimized better than default would be.  However, optimisation is
the only reason why we might want an arch-specific definition;
if current_pt_regs() and task_pt_regs(current) have different
values, the latter one is right.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  22062a96
- unify default ptrace_signal_deliver · 4f4202fe
  Al Viro authored Nov 05, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  4f4202fe
- flagday: kill pt_regs argument of do_fork() · e80d6661
  Al Viro authored Oct 22, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  e80d6661
- death to idle_regs() · 18c26c27
  Al Viro authored Oct 22, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  18c26c27
- don't pass regs to copy_process() · 62e791c1
  Al Viro authored Oct 22, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  62e791c1
- flagday: don't pass regs to copy_thread() · afa86fc4
  Al Viro authored Oct 22, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  afa86fc4
- bfin: switch to generic vfork, get rid of pointless wrappers · 135c37b8
  Al Viro authored Nov 13, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  135c37b8
- xtensa: switch to generic clone() · 3e41f9ba
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  3e41f9ba
- openrisc: switch to use of generic fork and clone · 39d91a9e
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  39d91a9e
- unicore32: switch to generic clone(2) · cb562173
  Al Viro authored Oct 22, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  cb562173
- score: switch to generic fork/vfork/clone · eda96977
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  eda96977
- c6x: sanitize copy_thread(), get rid of clone(2) wrapper, switch to generic clone() · 951b3961
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  951b3961
- take sys_fork/sys_vfork/sys_clone prototypes to linux/syscalls.h · 24465a40
  Al Viro authored Nov 28, 2012
```
now it can be done...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  24465a40
- mn10300: switch to generic fork/vfork/clone · 1ea2a016
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  1ea2a016
- h8300: switch to generic fork/vfork/clone · 3416e809
  Al Viro authored Oct 27, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  3416e809
- tile: switch to generic clone() · e9eac304
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  e9eac304
- m68k: sanitize copy_thread(), fork/vfork/clone wrappers, switch to generic fork/vfork · 20ecc91c
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  20ecc91c
- alpha: switch to generic fork/vfork/clone · dfe09ae0
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  dfe09ae0
- s390: switch to generic fork/vfork/clone · 87f1ca8f
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  87f1ca8f
- microblaze: switch to generic fork/vfork/clone · f3268edb
  Al Viro authored Oct 27, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
```
  f3268edb
- hexagon: switch to generic clone() · f01aceac
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  f01aceac
- sh: switch to generic fork/vfork/clone · 0ad9513d
  Al Viro authored Oct 27, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  0ad9513d
- parisc: switch to generic fork/vfork/clone · 415bfae9
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  415bfae9
- m32r: switch to generic fork/vfork/clone · 92bbe6cd
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  92bbe6cd
- cris: switch to generic fork/vfork/clone · 27d892fb
  Al Viro authored Oct 26, 2012
```
same braindamage as on s390 - the first two arguments of clone(2) in the
wrong order.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  27d892fb
- avr32: sanitize copy_thread(), switch to generic fork/vfork/clone, kill wrappers · 584271bc
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  584271bc
- arm64: sanitize copy_thread(), switch to generic fork/vfork/clone · 9ac08002
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  9ac08002
- frv: switch to generic fork/vfork/clone · b9763ce3
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  b9763ce3
- powerpc: switch to generic fork/clone/vfork · 0bcfe540
  Al Viro authored Oct 26, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  0bcfe540
- arm: switch to generic fork/vfork/clone · 38a61b6b
  Al Viro authored Oct 21, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  38a61b6b
- x86, um: switch to generic fork/vfork/clone · 1d4b4b29
  Al Viro authored Oct 22, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  1d4b4b29
- get rid of pt_regs argument of ->load_binary() · 71613c3b
  Al Viro authored Oct 20, 2012
```
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
```
  71613c3b