Hi Ye,

On 07/01/19 10:29, Ye Li wrote:

Apply errata based on PL310 version instead of compile time. Also set Prefetch offset to 15, since it improves memcpy performance by 35%. Don't enable Incr double Linefill enable since it adversely affects memcpy performance by about 32MB/s and reads by 90MB/s. Tested with 4K to 16MB sized src and dst aligned buffer.

Signed-off-by: Nitin Garg nitin.garg@freescale.com Signed-off-by: Ye Li ye.li@nxp.com

---

arch/arm/include/asm/pl310.h | 5 +++++ arch/arm/mach-imx/cache.c | 18 ++++++++++-------- 2 files changed, 15 insertions(+), 8 deletions(-)

diff --git a/arch/arm/include/asm/pl310.h b/arch/arm/include/asm/pl310.h index 3624362..b83978b 100644 --- a/arch/arm/include/asm/pl310.h +++ b/arch/arm/include/asm/pl310.h @@ -19,6 +19,11 @@ #define L310_AUX_CTRL_DATA_PREFETCH_MASK (1 << 28) #define L310_AUX_CTRL_INST_PREFETCH_MASK (1 << 29)

+#define L2X0_CACHE_ID_PART_MASK (0xf << 6) +#define L2X0_CACHE_ID_PART_L310 (3 << 6) +#define L2X0_CACHE_ID_RTL_MASK 0x3f +#define L2X0_CACHE_ID_RTL_R3P2 0x8

struct pl310_regs { u32 pl310_cache_id; u32 pl310_cache_type; diff --git a/arch/arm/mach-imx/cache.c b/arch/arm/mach-imx/cache.c index 82257f3..75e1f54 100644 --- a/arch/arm/mach-imx/cache.c +++ b/arch/arm/mach-imx/cache.c @@ -82,7 +82,7 @@ void v7_outer_cache_enable(void) { struct pl310_regs *const pl310 = (struct pl310_regs *)L2_PL310_BASE; struct iomuxc *iomux = (struct iomuxc *)IOMUXC_BASE_ADDR;

• unsigned int val;

• unsigned int val, cache_id;

  /*

@@ -112,22 +112,24 @@ void v7_outer_cache_enable(void)

val = readl(&pl310->pl310_prefetch_ctrl);

• /* Turn on the L2 I/D prefetch */
• val |= 0x30000000;

• /* Turn on the L2 I/D prefetch, double linefill */

• /* Set prefetch offset with any value except 23 as per errata 765569 */

• val |= 0x7000000f;

  /*

  • The L2 cache controller(PL310) version on the i.MX6D/Q is r3p1-50rel0

• * The L2 cache controller(PL310) version on the i.MX6DL/SOLO/SL is r3p2
  

• * The L2 cache controller(PL310) version on the i.MX6DL/SOLO/SL/SX/DQP
  

• * is r3p2.
  

  • But according to ARM PL310 errata: 752271
  • ID: 752271: Double linefill feature can cause data corruption
  • Fault Status: Present in: r3p0, r3p1, r3p1-50rel0. Fixed in r3p2
  • Workaround: The only workaround to this erratum is to disable the
  • double linefill feature. This is the default behavior.
  */

-#ifndef CONFIG_MX6Q

• val |= 0x40800000;

-#endif

• cache_id = readl(&pl310->pl310_cache_id);
• if (((cache_id & L2X0_CACHE_ID_PART_MASK) == L2X0_CACHE_ID_PART_L310)

•    && ((cache_id & L2X0_CACHE_ID_RTL_MASK) < L2X0_CACHE_ID_RTL_R3P2))
  

• val &= ~(1 << 30);
  

Very good idea !

Reviewd-by: Stefano Babic sbabic@denc.de

Best regards, Stefano Babic

writel(val, &pl310->pl310_prefetch_ctrl);

val = readl(&pl310->pl310_power_ctrl);