From 1773e60a816b28f4084810f96e9195aaaa7fb8c3 Mon Sep 17 00:00:00 2001 From: Santosh Shilimkar Date: Fri, 19 Nov 2010 23:01:03 +0530 Subject: omap4: l2x0: Construct the AUXCTRL value using defines This patch removes the hardcoded value of auxctrl value and construct it using bitfields Bit 25 is reserved and is always set to 1. Same value of this bit is retained in this patch Signed-off-by: Santosh Shilimkar Tested-by: Nishanth Menon Signed-off-by: Tony Lindgren --- arch/arm/mach-omap2/omap4-common.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) (limited to 'arch/arm/mach-omap2/omap4-common.c') diff --git a/arch/arm/mach-omap2/omap4-common.c b/arch/arm/mach-omap2/omap4-common.c index 2f895553e6a8..c81460445c4d 100644 --- a/arch/arm/mach-omap2/omap4-common.c +++ b/arch/arm/mach-omap2/omap4-common.c @@ -53,6 +53,8 @@ static void omap4_l2x0_disable(void) static int __init omap_l2_cache_init(void) { + u32 aux_ctrl = 0; + /* * To avoid code running on other OMAPs in * multi-omap builds @@ -72,10 +74,17 @@ static int __init omap_l2_cache_init(void) * Way size - 32KB (es1.0) * Way size - 64KB (es2.0 +) */ + aux_ctrl = ((1 << L2X0_AUX_CTRL_ASSOCIATIVITY_SHIFT) | + (0x1 << 25) | + (0x1 << L2X0_AUX_CTRL_NS_LOCKDOWN_SHIFT) | + (0x1 << L2X0_AUX_CTRL_NS_INT_CTRL_SHIFT)); + if (omap_rev() == OMAP4430_REV_ES1_0) - l2x0_init(l2cache_base, 0x0e050000, 0xc0000fff); + aux_ctrl |= 0x2 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT; else - l2x0_init(l2cache_base, 0x0e070000, 0xc0000fff); + aux_ctrl |= 0x3 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT; + + l2x0_init(l2cache_base, aux_ctrl, L2X0_AUX_CTRL_MASK); /* * Override default outer_cache.disable with a OMAP4 -- cgit v1.2.3 From 11e0264046e00544eb044fafc27125babd105e41 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Fri, 19 Nov 2010 23:01:04 +0530 Subject: omap4: l2x0: enable instruction and data prefetching Enabling L2 prefetching improves performance as shown on Panda ES2.1 board with mem test, and it has measurable impact on performances. I think we should consider it, even though it damages "writes" a bit. (rebased to k.org) Usually the prefetch is used at both levels together L1 + L2, however, to enable the CP15 prefetch engines, these are under security, and on GP devices, we cannot enable it(e.g. on PandaBoard). However, just enabling PL310 prefetch seems to provide performance improvement, as shown in the data below (from Ubuntu) and would be a great thing to pull in. What prefetch does is enable automatic next line prefetching. With this enabled, whenever the PL310 receives a cachable read request, it automatically prefetches the following cache line as well. Measurement Data: == STOCK 10.10 WITHOUT PATCH ======================== ~# ./memspeed size 8388608 8192k 8M offset 8388608, 0 buffers 0x2aaad000 0x2b2ad000 copy libc 133 MB/s copy Android v5 273 MB/s copy Android NEON 235 MB/s copy INT32 116 MB/s copy ASM ARM 187 MB/s copy ASM VLDM 64 204 MB/s copy ASM VLDM 128 173 MB/s copy ASM VLD1 216 MB/s read ASM ARM 286 MB/s read ASM VLDM 242 MB/s read ASM VLD1 286 MB/s write libc 1947 MB/s write ASM ARM 1943 MB/s write ASM VSTM 1942 MB/s write ASM VST1 1935 MB/s 10.10 + PATCH ============= ~# ./memspeed size 8388608 8192k 8M offset 8388608, 0 buffers 0x2ab17000 0x2b317000 copy libc 129 MB/s copy Android v5 256 MB/s copy Android NEON 356 MB/s copy INT32 127 MB/s copy ASM ARM 321 MB/s copy ASM VLDM 64 337 MB/s copy ASM VLDM 128 321 MB/s copy ASM VLD1 350 MB/s read ASM ARM 496 MB/s read ASM VLDM 470 MB/s read ASM VLD1 488 MB/s write libc 1701 MB/s write ASM ARM 1682 MB/s write ASM VSTM 1693 MB/s write ASM VST1 1681 MB/s Signed-off-by: Mans Rullgard Signed-off-by: Santosh Shilimkar Tested-by: Nishanth Menon Signed-off-by: Tony Lindgren --- arch/arm/mach-omap2/omap4-common.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) (limited to 'arch/arm/mach-omap2/omap4-common.c') diff --git a/arch/arm/mach-omap2/omap4-common.c b/arch/arm/mach-omap2/omap4-common.c index c81460445c4d..b3cea78b5f09 100644 --- a/arch/arm/mach-omap2/omap4-common.c +++ b/arch/arm/mach-omap2/omap4-common.c @@ -66,9 +66,6 @@ static int __init omap_l2_cache_init(void) l2cache_base = ioremap(OMAP44XX_L2CACHE_BASE, SZ_4K); BUG_ON(!l2cache_base); - /* Enable PL310 L2 Cache controller */ - omap_smc1(0x102, 0x1); - /* * 16-way associativity, parity disabled * Way size - 32KB (es1.0) @@ -79,10 +76,18 @@ static int __init omap_l2_cache_init(void) (0x1 << L2X0_AUX_CTRL_NS_LOCKDOWN_SHIFT) | (0x1 << L2X0_AUX_CTRL_NS_INT_CTRL_SHIFT)); - if (omap_rev() == OMAP4430_REV_ES1_0) + if (omap_rev() == OMAP4430_REV_ES1_0) { aux_ctrl |= 0x2 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT; - else - aux_ctrl |= 0x3 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT; + } else { + aux_ctrl |= ((0x3 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT) | + (1 << L2X0_AUX_CTRL_DATA_PREFETCH_SHIFT) | + (1 << L2X0_AUX_CTRL_INSTR_PREFETCH_SHIFT)); + } + if (omap_rev() != OMAP4430_REV_ES1_0) + omap_smc1(0x109, aux_ctrl); + + /* Enable PL310 L2 Cache controller */ + omap_smc1(0x102, 0x1); l2x0_init(l2cache_base, aux_ctrl, L2X0_AUX_CTRL_MASK); -- cgit v1.2.3 From b0f20ff9d7e347c284ea7718597c978a2969ad7b Mon Sep 17 00:00:00 2001 From: Santosh Shilimkar Date: Fri, 19 Nov 2010 23:01:05 +0530 Subject: omap4: l2x0: Set share override bit Clearing bit 22 in the PL310 Auxiliary Control register (shared attribute override enable) has the side effect of transforming Normal Shared Non-cacheable reads into Cacheable no-allocate reads. Coherent DMA buffers in Linux always have a Cacheable alias via the kernel linear mapping and the processor can speculatively load cache lines into the PL310 controller. With bit 22 cleared, Non-cacheable reads would unexpectedly hit such cache lines leading to buffer corruption Signed-off-by: Santosh Shilimkar Tested-by: Nishanth Menon Signed-off-by: Tony Lindgren --- arch/arm/mach-omap2/omap4-common.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/arm/mach-omap2/omap4-common.c') diff --git a/arch/arm/mach-omap2/omap4-common.c b/arch/arm/mach-omap2/omap4-common.c index b3cea78b5f09..2006da10f5f5 100644 --- a/arch/arm/mach-omap2/omap4-common.c +++ b/arch/arm/mach-omap2/omap4-common.c @@ -80,6 +80,7 @@ static int __init omap_l2_cache_init(void) aux_ctrl |= 0x2 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT; } else { aux_ctrl |= ((0x3 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT) | + (1 << L2X0_AUX_CTRL_SHARE_OVERRIDE_SHIFT) | (1 << L2X0_AUX_CTRL_DATA_PREFETCH_SHIFT) | (1 << L2X0_AUX_CTRL_INSTR_PREFETCH_SHIFT)); } -- cgit v1.2.3 From b89cd71a159b5edca4c429687e4af01708eb1b26 Mon Sep 17 00:00:00 2001 From: Santosh Shilimkar Date: Fri, 19 Nov 2010 23:01:06 +0530 Subject: omap4: l2x0: Enable early BRESP bit The AXI protocol specifies that the write response can only be sent back to an AXI master when the last write data has been accepted. This optimization enables the PL310 to send the write response of certain write transactions as soon as the store buffer accepts the write address. This behavior is not compatible with the AXI protocol and is disabled by default. You enable this optimization by setting the Early BRESP Enable bit in the Auxiliary Control Register (bit [30]). Signed-off-by: Santosh Shilimkar Signed-off-by: Mans Rullgard Tested-by: Nishanth Menon Signed-off-by: Tony Lindgren --- arch/arm/mach-omap2/omap4-common.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/arm/mach-omap2/omap4-common.c') diff --git a/arch/arm/mach-omap2/omap4-common.c b/arch/arm/mach-omap2/omap4-common.c index 2006da10f5f5..e7a9b7f13b52 100644 --- a/arch/arm/mach-omap2/omap4-common.c +++ b/arch/arm/mach-omap2/omap4-common.c @@ -82,7 +82,8 @@ static int __init omap_l2_cache_init(void) aux_ctrl |= ((0x3 << L2X0_AUX_CTRL_WAY_SIZE_SHIFT) | (1 << L2X0_AUX_CTRL_SHARE_OVERRIDE_SHIFT) | (1 << L2X0_AUX_CTRL_DATA_PREFETCH_SHIFT) | - (1 << L2X0_AUX_CTRL_INSTR_PREFETCH_SHIFT)); + (1 << L2X0_AUX_CTRL_INSTR_PREFETCH_SHIFT) | + (1 << L2X0_AUX_CTRL_EARLY_BRESP_SHIFT)); } if (omap_rev() != OMAP4430_REV_ES1_0) omap_smc1(0x109, aux_ctrl); -- cgit v1.2.3