[PATCH] ARM: bitops: Fix low-level code to be Thumb-2 compatible

[Dave Martin]

On Wed, Feb 2, 2011 at 9:24 PM, Russell King - ARM Linux
<linux at arm.linux.org.uk> wrote:
> On Wed, Feb 02, 2011 at 02:33:55PM -0500, Nicolas Pitre wrote:
>> On Wed, 2 Feb 2011, Dave Martin wrote:
>>
>> > The new bitops code cunningly str <Rt>, [r1, -r1] to trigger
>> > a fault by attempting to store to address zero.
>> >
>> > This code doesn't assemble in Thumb-2, since Thumb-2 doesn't
>> > allow negative register offsets at all for loads and stores.
>> >
>> > The patch loads 0 into r2 and uses that as a base register
>> > instead for the Thumb-2 case: r2 seems non-live at every
>> > instance of this problem.
>> >
>> > The ARM case is unaffected.
>> >
>> > Signed-off-by: Dave Martin <dave.martin at linaro.org>
>>
>> Russell proposed a better solution already:
>>
>> http://mid.gmane.org/alpine.LFD.2.00.1101181312360.8580@xanadu.home
>>
>> No idea why this wasn't folded in his series yet though.
>
> Now added, and original tested-by's dropped.  It now needs re-testing.
>
> 8<----
> Subject: [PATCH] ARM: bitops: ensure set/clear/change bitops take a word-aligned pointer
>
> Add additional instructions to our assembly bitops functions to ensure
> that they only operate on word-aligned pointers.  This will be necessary
> when we switch these operations to use the word-based exclusive
> operations.
>
> Signed-off-by: Russell King <rmk+kernel at arm.linux.org.uk>
> ---
>  arch/arm/lib/bitops.h |    8 ++++++++
>  1 files changed, 8 insertions(+), 0 deletions(-)
>
> diff --git a/arch/arm/lib/bitops.h b/arch/arm/lib/bitops.h
> index d422529..bd00551 100644
> --- a/arch/arm/lib/bitops.h
> +++ b/arch/arm/lib/bitops.h
> @@ -1,6 +1,8 @@
>
>  #if __LINUX_ARM_ARCH__ >= 6 && defined(CONFIG_CPU_32v6K)
>        .macro  bitop, instr
> +       ands    ip, r1, #3
> +       strneb  r1, [ip]                @ assert word-aligned
>        mov     r2, #1
>        and     r3, r0, #7              @ Get bit offset
>        add     r1, r1, r0, lsr #3      @ Get byte offset
> @@ -14,6 +16,8 @@
>        .endm
>
>        .macro  testop, instr, store
> +       ands    ip, r1, #3
> +       strneb  r1, [ip]                @ assert word-aligned
>        and     r3, r0, #7              @ Get bit offset
>        mov     r2, #1
>        add     r1, r1, r0, lsr #3      @ Get byte offset
> @@ -32,6 +36,8 @@
>        .endm
>  #else
>        .macro  bitop, instr
> +       ands    ip, r1, #3
> +       strneb  r1, [ip]                @ assert word-aligned
>        and     r2, r0, #7
>        mov     r3, #1
>        mov     r3, r3, lsl r2
> @@ -52,6 +58,8 @@
>  * to avoid dirtying the data cache.
>  */
>        .macro  testop, instr, store
> +       ands    ip, r1, #3
> +       strneb  r1, [ip]                @ assert word-aligned
>        add     r1, r1, r0, lsr #3
>        and     r3, r0, #7
>        mov     r0, #1
> --
> 1.6.2.5
>
>

Heh, yes, that's a nicer fix.  I'll give it a try.

Thanks
---Dave