linux-kernel - Re: [PATCH] arc: arcv2: cache: Explicitly set MSB counterpart of region ops addresses

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <2f6ba6bb-ae07-0ff9-c2ff-0f162eab2ef0@synopsys.com>
Date:   Wed, 2 Aug 2017 09:09:44 +0530
From:   Vineet Gupta <Vineet.Gupta1@...opsys.com>
To:     Alexey Brodkin <Alexey.Brodkin@...opsys.com>,
        "linux-snps-arc@...ts.infradead.org" 
        <linux-snps-arc@...ts.infradead.org>
CC:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "Vladimir Kondratiev" <vladimir.kondratiev@...el.com>
Subject: Re: [PATCH] arc: arcv2: cache: Explicitly set MSB counterpart of
 region ops addresses

On 08/01/2017 03:29 PM, Alexey Brodkin wrote:
> It is necessary to explicitly set both SLC_AUX_RGN_START1 and SLC_AUX_RGN_END1
> which hold MSB bits of the physical address correspondingly of region start
> and end otherwise SLC region operation is executed in unpredictable manner,
> for example on HSDK platform where PAE40 support exists in hardware
> we saw each and every SLC region op to take seconds (sic!).
>
> Signed-off-by: Alexey Brodkin <abrodkin@...opsys.com>
> Reported-by: Vladimir Kondratiev <vladimir.kondratiev@...el.com>
> ---
>   arch/arc/include/asm/cache.h | 2 ++
>   arch/arc/mm/cache.c          | 8 ++++++--
>   2 files changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/arch/arc/include/asm/cache.h b/arch/arc/include/asm/cache.h
> index 35127ad95124..1f3c2f967471 100644
> --- a/arch/arc/include/asm/cache.h
> +++ b/arch/arc/include/asm/cache.h
> @@ -91,7 +91,9 @@ extern unsigned long perip_base, perip_end;
>   #define ARC_REG_SLC_FLUSH	0x904
>   #define ARC_REG_SLC_INVALIDATE	0x905
>   #define ARC_REG_SLC_RGN_START	0x914
> +#define ARC_REG_SLC_RGN_START1	0x915
>   #define ARC_REG_SLC_RGN_END	0x916
> +#define ARC_REG_SLC_RGN_END1	0x917
>   
>   /* Bit val in SLC_CONTROL */
>   #define SLC_CTRL_DIS		0x001
> diff --git a/arch/arc/mm/cache.c b/arch/arc/mm/cache.c
> index b7a1face1584..0b4e2650c5de 100644
> --- a/arch/arc/mm/cache.c
> +++ b/arch/arc/mm/cache.c
> @@ -580,6 +580,7 @@ noinline void slc_op(phys_addr_t paddr, unsigned long sz, const int op)
>   	static DEFINE_SPINLOCK(lock);
>   	unsigned long flags;
>   	unsigned int ctrl;
> +	phys_addr_t end;
>   
>   	spin_lock_irqsave(&lock, flags);
>   
> @@ -609,8 +610,11 @@ noinline void slc_op(phys_addr_t paddr, unsigned long sz, const int op)
>   	 * END needs to be setup before START (latter triggers the operation)
>   	 * END can't be same as START, so add (l2_line_sz - 1) to sz
>   	 */
> -	write_aux_reg(ARC_REG_SLC_RGN_END, (paddr + sz + l2_line_sz - 1));
> -	write_aux_reg(ARC_REG_SLC_RGN_START, paddr);
> +	end = paddr + sz + l2_line_sz - 1;
> +	write_aux_reg(ARC_REG_SLC_RGN_END1, upper_32_bits(end));
> +	write_aux_reg(ARC_REG_SLC_RGN_END, lower_32_bits(end));
> +	write_aux_reg(ARC_REG_SLC_RGN_START1, upper_32_bits(paddr));
> +	write_aux_reg(ARC_REG_SLC_RGN_START, lower_32_bits(paddr));

Are these registers present even if PAE is not configured in hardware ?

>   
>   	/* Make sure "busy" bit reports correct stataus, see STAR 9001165532 */
>   	read_aux_reg(ARC_REG_SLC_CTRL);