linux-toradex.git/arch/tile/include, branch v3.4.73

tile: use a more conservative __my_cpu_offset in CONFIG_PREEMPT

2013-10-13T22:42:50+00:00

commit f862eefec0b68e099a9fa58d3761ffb10bad97e1 upstream.

It turns out the kernel relies on barrier() to force a reload of the
percpu offset value.  Since we can't easily modify the definition of
barrier() to include "tp" as an output register, we instead provide a
definition of __my_cpu_offset as extended assembly that includes a fake
stack read to hazard against barrier(), forcing gcc to know that it
must reread "tp" and recompute anything based on "tp" after a barrier.

This fixes observed hangs in the slub allocator when we are looping
on a percpu cmpxchg_double.

A similar fix for ARMv7 was made in June in change 509eb76ebf97.

Signed-off-by: Chris Metcalf 
Signed-off-by: Greg Kroah-Hartman

tile: fix bug where fls(0) was not returning 0

2012-06-01T07:18:27+00:00

commit 9f1d62bed7f015d11b9164078b7fea433b474114 upstream.

This is because __builtin_clz(0) returns 64 for the "undefined" case
of 0, since the builtin just does a right-shift 32 and "clz" instruction.
So, use the alpha approach of casting to u32 and using __builtin_clzll().

Signed-off-by: Chris Metcalf 
Signed-off-by: Greg Kroah-Hartman

arch/tile: fix up some issues in calling do_work_pending()

2012-05-16T20:01:16+00:00

First, we were at risk of handling thread-info flags, in particular
do_signal(), when returning from kernel space.  This could happen
after a failed kernel_execve(), or when forking a kernel thread.
The fix is to test in do_work_pending() for user_mode() and return
immediately if so; we already had this test for one of the flags,
so I just hoisted it to the top of the function.

Second, if a ptraced process updated the callee-saved registers
in the ptregs struct and then processed another thread-info flag, we
would overwrite the modifications with the original callee-saved
registers.  To fix this, we add a register to note if we've already
saved the registers once, and skip doing it on additional passes
through the loop.  To avoid a performance hit from the couple of
extra instructions involved, I modified the GET_THREAD_INFO() macro
to be guaranteed to be one instruction, then bundled it with adjacent
instructions, yielding an overall net savings.

Reported-By: Al Viro 
Signed-off-by: Chris Metcalf

arch/tile: fix a couple of functions that should be __init

2012-04-25T16:45:26+00:00

They were marked __devinit by mistake, causing some warnings at link time.

Signed-off-by: Chris Metcalf

arch/tile: use atomic exchange in arch_write_unlock()

2012-04-02T16:13:49+00:00

This idiom is used elsewhere when we do an unlock by writing a zero,
but I missed it here.  Using an atomic operation avoids waiting
on the write buffer for the unlocking write to be sent to the home cache.

Signed-off-by: Chris Metcalf

arch/tile: work around a hardware issue with the return-address stack

2012-04-02T16:12:48+00:00

In certain circumstances we need to do a bunch of jump-and-link
instructions to fill the hardware return-address stack with nonzero values.

Signed-off-by: Chris Metcalf

arch/tile: various bugs in stack backtracer

2012-04-02T16:12:45+00:00

Fix a long-standing bug in the stack backtracer where we would print
garbage to the console instead of kernel function names, if the kernel
wasn't built with symbol support (e.g. mboot).

Make sure to tag every line of userspace backtrace output if we actually
have the mmap_sem, since that way if there's no tag, we know that it's
because we couldn't trylock the semaphore.

Stop doing a TLB flush and examining page tables during backtrace.
Instead, just trust that __copy_from_user_inatomic() will properly fault
and return a failure, which it should do in all cases.

Fix a latent bug where the backtracer would directly examine a signal
context in user space, rather than copying it safely to kernel memory
first.  This meant that a race with another thread could potentially
have caused a kernel panic.

Guard against unaligned sp when trying to restart backtrace at an
interrupt or signal handler point in the kernel backtracer.

Report kernel symbolic information for the call instruction rather
than for the following instruction.  We still report the actual numeric
address corresponding to the instruction after the call, for the sake
of consistency with the normal expectations for stack backtracers.

Signed-off-by: Chris Metcalf

arch/tile: use 0 for IRQ_RESCHEDULE instead of 1

2012-04-02T16:00:16+00:00

This avoids assigning IRQ 0 to PCI devices, because we've seen that
doesn't always work well.

Signed-off-by: Chris Metcalf

arch/tile: fix gcc 4.6 warnings in

2012-04-02T16:00:15+00:00

Fix some signedness and variable usage warnings in change_bit()
and test_and_change_bit().

Signed-off-by: Chris Metcalf

arch/tile: revert comment for atomic64_add_unless().

2012-04-02T16:00:15+00:00

It still returns whether @v was not @u, not the old value,
unlike __atomic_add_unless().

Signed-off-by: Chris Metcalf 
Acked-by: Arun Sharma