With CPU_DADDI_WORKAROUNDS enabled __delay assembles with a macro in a
branch delay slot:

{standard input}: Assembler messages:
{standard input}:18: Warning: Macro instruction expanded into multiple
instructions in a branch delay slot

and broken code results:

0000000000000000 <__delay>:
   0:	1480ffff 	bnez	a0,0 <__delay>
   4:	24010001 	li	at,1
   8:	0081202f 	dsubu	a0,a0,at
   c:	03e00008 	jr	ra
  10:	00000000 	nop
  14:	00000000 	nop

Consequently the function loops indefinitely, showing up prominently as a
hang in the delay loop calibration at bootstrap.

This change corrects the problem by forcing the immediate 1 into a
register while keeping code produced identical where CPU_DADDI_WORKAROUNDS
is disabled.
Signed-off-by: NMaciej W. Rozycki <macro@linux-mips.org>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6669/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

Having received another series of whitespace patches I decided to do this
once and for all rather than dealing with this kind of patches trickling
in forever.
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

When building a 32-bit kernel for RBTX4927 with gcc version 4.1.2 20061115
(prerelease) (Ubuntu 4.1.1-21), I get:

arch/mips/lib/delay.c:24:5: warning: "__SIZEOF_LONG__" is not defined

As a consequence, __delay() always uses the 64-bit "dsubu" instruction.

Replace the check for "__SIZEOF_LONG__ == 4" by "BITS_PER_LONG == 32" to
fix this.

Introduced by commit 5210edcd [MIPS: Make
__{,n,u}delay declarations match definitions and generic delay.h"]
Signed-off-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Patchwork: https://patchwork.linux-mips.org/patch/4678/Acked-by: NDavid Daney <david.daney@cavium.com>
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

At some recent point arch/mips/include/asm/delay.h has started being
included into csrc-octeon.c where the __?delay() functions are defined.
This causes a compile failure due to conflicting declarations and
definitions of the functions.

It turns out that the generic definitions in arch/mips/lib/delay.c also
conflict.

Proposed fix: Declare the functions to take unsigned long parameters
just like asm-generic (and x86) does.  Update __delay to agree
(__ndelay and __udelay need no change).

Bonus: Get rid of 'inline' from __delay() definition, as it is globally
visible, and the compiler should be making this decision itself (it does
in fact inline the function without being told to).
Signed-off-by: NDavid Daney <david.daney@cavium.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/4354/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This may lead to warnings like:

BUG: using smp_processor_id() in preemptible [00000000] code: reboot/1989
caller is __udelay+0x14/0x70
Call Trace:
[<ffffffff8110ad28>] dump_stack+0x8/0x34
[<ffffffff812dde04>] debug_smp_processor_id+0xf4/0x110
[<ffffffff812d90bc>] __udelay+0x14/0x70
[<ffffffff81378274>] md_notify_reboot+0x12c/0x148
[<ffffffff81161054>] notifier_call_chain+0x64/0xc8
[<ffffffff811614dc>] __blocking_notifier_call_chain+0x64/0xc0
[<ffffffff8115566c>] kernel_restart_prepare+0x1c/0x38
[<ffffffff811556cc>] kernel_restart+0x14/0x50
[<ffffffff8115581c>] SyS_reboot+0x10c/0x1f0
[<ffffffff81103684>] handle_sysn32+0x44/0x84
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

Signed-off-by: NAtsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

Outlining fixes the issue were on certain CPUs such as the R10000 family
the delay loop would need an extra cycle if it overlaps a cacheline
boundary.

The rewrite also fixes build errors with GCC 4.4 which was changed in
way incompatible with the kernel's inline assembly.

Relying on pure C for computation of the delay value removes the need for
explicit.  The price we pay is a slight slowdown of the computation - to
be fixed on another day.
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>