Re: [PATCH] MIPS: Fix ejtag handler on SMP

[James Hogan <jhogan@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1502 bytes --]

On Fri, Mar 30, 2018 at 05:05:15PM +0800, r@hev.cc wrote:
> From: Heiher <r@hev.cc>

Please can you add a proper commit description, explaining the problem
and what your patch does to fix it.

> 
> Signed-off-by: Heiher <r@hev.cc>
> ---
>  arch/mips/kernel/genex.S | 19 +++++++++++++++++++
>  1 file changed, 19 insertions(+)
> 
> diff --git a/arch/mips/kernel/genex.S b/arch/mips/kernel/genex.S
> index 37b9383eacd3..9e0857fbe281 100644
> --- a/arch/mips/kernel/genex.S
> +++ b/arch/mips/kernel/genex.S
> @@ -354,6 +354,17 @@ NESTED(ejtag_debug_handler, PT_SIZE, sp)
>  	sll	k0, k0, 30	# Check for SDBBP.
>  	bgez	k0, ejtag_return
>  
> +#ifdef CONFIG_SMP
> +	PTR_LA	k0, ejtag_debug_buffer
> +1:	sync

Is the sync necessary? Or is that one of those platform specific
workarounds?

> +	ll	k0, LONGSIZE(k0)
> +	bnez	k0, 1b
> +	PTR_LA	k0, ejtag_debug_buffer
> +	sc	k0, LONGSIZE(k0)
> +	beqz	k0, 1b
> +	sync
> +#endif
> +
>  	PTR_LA	k0, ejtag_debug_buffer
>  	LONG_S	k1, 0(k0)
>  	SAVE_ALL
> @@ -363,6 +374,11 @@ NESTED(ejtag_debug_handler, PT_SIZE, sp)
>  	PTR_LA	k0, ejtag_debug_buffer
>  	LONG_L	k1, 0(k0)
>  
> +#ifdef CONFIG_SMP
> +	sw	zero, LONGSIZE(k0)
> +	sync

Same question. Its about to deret anyway which should cover that I
think?

> +#endif
> +
>  ejtag_return:
>  	MFC0	k0, CP0_DESAVE

Not specific to your patch, but I wonder whether there should be a
back_to_back_c0_hazard (ehb on r2+) somewhere before this MFC0.

Cheers
James

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]