* OMAP3: DVFS core patch set
@ 2008-12-11 15:52 Tero Kristo
2008-12-11 15:52 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tero Kristo
` (2 more replies)
0 siblings, 3 replies; 32+ messages in thread
From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw)
To: linux-omap
This patch set provides DVFS support for OMAP3. First 16 patches contain
clock framework modifications from Paul Walmsley, which are needed for VDD2
OPP clock control to work properly. Rest of the patches are misc fixes for
DVFS control, providing locking interface for OPPs and providing support
for DPLL3 M2 rates higher than 2.
^ permalink raw reply [flat|nested] 32+ messages in thread* [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type 2008-12-11 15:52 OMAP3: DVFS core patch set Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tero Kristo 2008-12-15 19:06 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tony Lindgren 2008-12-12 1:59 ` OMAP3: DVFS core patch set Paul Walmsley 2008-12-15 21:48 ` Kevin Hilman 2 siblings, 2 replies; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Add a Non-cacheable Normal ARM executable memory type, MT_MEMORY_NONCACHED. This is needed for the OMAP3 SDRAM clock change code, which must run from SRAM. The SRAM must be marked as non-cacheable memory to avoid dirty cache line writebacks to SDRAM while the SDRAM controller is paused. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/include/asm/mach/map.h | 1 + arch/arm/mm/mmu.c | 23 +++++++++++++++++++++++ 2 files changed, 24 insertions(+), 0 deletions(-) diff --git a/arch/arm/include/asm/mach/map.h b/arch/arm/include/asm/mach/map.h index 39d949b..58cf91f 100644 --- a/arch/arm/include/asm/mach/map.h +++ b/arch/arm/include/asm/mach/map.h @@ -26,6 +26,7 @@ struct map_desc { #define MT_HIGH_VECTORS 8 #define MT_MEMORY 9 #define MT_ROM 10 +#define MT_MEMORY_NONCACHED 11 #ifdef CONFIG_MMU extern void iotable_init(struct map_desc *, int); diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c index 7f36c82..9ad6413 100644 --- a/arch/arm/mm/mmu.c +++ b/arch/arm/mm/mmu.c @@ -242,6 +242,10 @@ static struct mem_type mem_types[] = { .prot_sect = PMD_TYPE_SECT, .domain = DOMAIN_KERNEL, }, + [MT_MEMORY_NONCACHED] = { + .prot_sect = PMD_TYPE_SECT | PMD_SECT_AP_WRITE, + .domain = DOMAIN_KERNEL, + }, }; const struct mem_type *get_mem_type(unsigned int type) @@ -405,9 +409,28 @@ static void __init build_mem_type_table(void) kern_pgprot |= L_PTE_SHARED; vecs_pgprot |= L_PTE_SHARED; mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S; + mem_types[MT_MEMORY_NONCACHED].prot_sect |= PMD_SECT_S; #endif } + /* + * Non-cacheable Normal - intended for memory areas that must + * not cause cache line evictions when used + */ + if (cpu_arch >= CPU_ARCH_ARMv6) { + if (cpu_arch >= CPU_ARCH_ARMv7 && (cr & CR_TRE)) { + /* Non-cacheable Normal is XCB = 001 */ + mem_types[MT_MEMORY_NONCACHED].prot_sect |= + PMD_SECT_BUFFERED; + } else { + /* For both ARMv6 and non-TEX-remapping ARMv7 */ + mem_types[MT_MEMORY_NONCACHED].prot_sect |= + PMD_SECT_TEX(1); + } + } else { + mem_types[MT_MEMORY_NONCACHED].prot_sect |= PMD_SECT_BUFFERABLE; + } + for (i = 0; i < 16; i++) { unsigned long v = pgprot_val(protection_map[i]); protection_map[i] = __pgprot(v | user_pgprot); -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory 2008-12-11 15:52 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 03/23] OMAP3 SRAM: add ARM barriers to omap3_sram_configure_core_dpll Tero Kristo 2008-12-15 19:07 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tony Lindgren 2008-12-15 19:06 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tony Lindgren 1 sibling, 2 replies; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Mark the SRAM (aka OCM RAM) as Non-cacheable Normal memory[1]. This is to prevent the ARM from evicting existing cache lines to SDRAM while code is executing from the SRAM. Necessary since one of the primary uses for the SRAM is to hold the code and data for the CORE DPLL M2 divider reprogramming code, which must execute while the SDRC is idled. If the ARM attempts to write cache lines back to the while the SRAM code is running, the ARM will stall[2]. TI deals with this problem in the CDP kernel by marking the SRAM as Strongly-ordered memory. Tero Kristo <tero.kristo@nokia.com> caught a bug in an earlier version of this patch - thanks Tero. ... 1. ARMv7 ARM (DDI 0406A) pp. A3-30, A3-31, B3-32. 2. Private communication with Richard Woodruff <r-woodruff2@ti.com> Signed-off-by: Paul Walmsley <paul@pwsan.com> Cc: Tero Kristo <tero.kristo@nokia.com> Cc: Richard Woodruff <r-woodruff2@ti.com> --- arch/arm/plat-omap/sram.c | 9 +++++++++ 1 files changed, 9 insertions(+), 0 deletions(-) mode change 100755 => 100644 arch/arm/plat-omap/sram.c diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c old mode 100755 new mode 100644 index abcc05b..04214e1 --- a/arch/arm/plat-omap/sram.c +++ b/arch/arm/plat-omap/sram.c @@ -207,6 +207,15 @@ void __init omap_map_sram(void) base = OMAP3_SRAM_PA; base = ROUND_DOWN(base, PAGE_SIZE); omap_sram_io_desc[0].pfn = __phys_to_pfn(base); + + /* + * SRAM must be marked as non-cached on OMAP3 since the + * CORE DPLL M2 divider change code (in SRAM) runs with the + * SDRAM controller disabled, and if it is marked cached, + * the ARM may attempt to write cache lines back to SDRAM + * which will cause the system to hang. + */ + omap_sram_io_desc[0].type = MT_MEMORY_NONCACHED; } omap_sram_io_desc[0].length = 1024 * 1024; /* Use section desc */ -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 03/23] OMAP3 SRAM: add ARM barriers to omap3_sram_configure_core_dpll 2008-12-11 15:52 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 04/23] OMAP3 clock: add interconnect barriers to CORE DPLL M2 change Tero Kristo 2008-12-15 19:07 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tony Lindgren 1 sibling, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Add more barriers in the SRAM CORE DPLL M2 divider change code. - Add a DSB SY after the function's entry point to flush all cached and buffered writes and wait for the interconnect to claim that they have completed[1]. The idea here is to force all delayed write traffic going to the SDRAM to at least post to the L3 interconnect before continuing. If these writes are allowed to occur after the SDRC is idled, the writes will not be acknowledged and the ARM will stall. Note that in this case, it does not matter if the writes actually complete to the SDRAM - it is only necessary for the writes to leave the ARM itself. If the writes are posted by the interconnect when the SDRC goes into idle, the writes will be delayed until the SDRC returns from idle[2]. If the SDRC is in the middle of a write when it is requested to enter idle, the SDRC will not acknowledge the idle request until the writes complete to the SDRAM.[3] The old-style DMB in sdram_in_selfrefresh is now superfluous, so, remove it. - Add an ISB before the function's exit point to prevent the ARM from speculatively executing into SDRAM before the SDRAM is enabled[4]. ... 1. ARMv7 ARM (DDI 0406A) A3-47, A3-48. 2. Private communication with Richard Woodruff <r-woodruff2@ti.com>. 3. Private communication with Richard Woodruff <r-woodruff2@ti.com>. 4. ARMv7 ARM (DDI 0406A) A3-48. Signed-off-by: Paul Walmsley <paul@pwsan.com> Cc: Richard Woodruff <r-woodruff2@ti.com> --- arch/arm/mach-omap2/sram34xx.S | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 2c71461..f4a356d 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -43,6 +43,7 @@ */ ENTRY(omap3_sram_configure_core_dpll) stmfd sp!, {r1-r12, lr} @ store regs to stack + dsb @ flush buffered writes to interconnect cmp r3, #0x2 blne configure_sdrc cmp r3, #0x2 @@ -58,6 +59,7 @@ ENTRY(omap3_sram_configure_core_dpll) blne wait_dll_lock cmp r3, #0x1 blne configure_sdrc + isb @ prevent speculative exec past here mov r0, #0 @ return value ldmfd sp!, {r1-r12, pc} @ restore regs and return unlock_dll: @@ -73,8 +75,6 @@ lock_dll: str r5, [r4] bx lr sdram_in_selfrefresh: - mov r5, #0x0 @ Move 0 to R5 - mcr p15, 0, r5, c7, c10, 5 @ memory barrier ldr r4, omap3_sdrc_power @ read the SDRC_POWER register ldr r5, [r4] @ read the contents of SDRC_POWER orr r5, r5, #0x40 @ enable self refresh on idle req -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 04/23] OMAP3 clock: add interconnect barriers to CORE DPLL M2 change 2008-12-11 15:52 ` [PATCH 03/23] OMAP3 SRAM: add ARM barriers to omap3_sram_configure_core_dpll Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 05/23] OMAP3 SRAM: clear the SDRC PWRENA bit during SDRC frequency change Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Where necessary, add interconnect barriers to force posted writes to complete before continuing. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sram34xx.S | 9 ++++++--- 1 files changed, 6 insertions(+), 3 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index f4a356d..8d524f3 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -66,22 +66,23 @@ unlock_dll: ldr r4, omap3_sdrc_dlla_ctrl ldr r5, [r4] orr r5, r5, #0x4 - str r5, [r4] + str r5, [r4] @ (no OCP barrier needed) bx lr lock_dll: ldr r4, omap3_sdrc_dlla_ctrl ldr r5, [r4] bic r5, r5, #0x4 - str r5, [r4] + str r5, [r4] @ (no OCP barrier needed) bx lr sdram_in_selfrefresh: ldr r4, omap3_sdrc_power @ read the SDRC_POWER register ldr r5, [r4] @ read the contents of SDRC_POWER orr r5, r5, #0x40 @ enable self refresh on idle req str r5, [r4] @ write back to SDRC_POWER register + ldr r5, [r4] @ posted-write barrier for SDRC ldr r4, omap3_cm_iclken1_core @ read the CM_ICLKEN1_CORE reg ldr r5, [r4] - bic r5, r5, #0x2 @ disable iclk bit for SRDC + bic r5, r5, #0x2 @ disable iclk bit for SDRC str r5, [r4] wait_sdrc_idle: ldr r4, omap3_cm_idlest1_core @@ -97,6 +98,7 @@ configure_core_dpll: and r5, r5, r6 orr r5, r5, r3, lsl #0x1B @ r3 contains the M2 val str r5, [r4] + ldr r5, [r4] @ posted-write barrier for CM mov r5, #0x800 @ wait for the clock to stabilise cmp r3, #2 bne wait_clk_stable @@ -152,6 +154,7 @@ configure_sdrc: str r1, [r4] ldr r4, omap3_sdrc_actim_ctrlb str r2, [r4] + ldr r2, [r4] @ posted-write barrier for SDRC bx lr omap3_sdrc_power: -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 05/23] OMAP3 SRAM: clear the SDRC PWRENA bit during SDRC frequency change 2008-12-11 15:52 ` [PATCH 04/23] OMAP3 clock: add interconnect barriers to CORE DPLL M2 change Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 06/23] OMAP3 SDRC: Add 166MHz, 83MHz SDRC settings for the BeagleBoard Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Clear the SDRC_POWER.PWRENA bit before putting the SDRAM into self-refresh mode. This prevents the SDRC from attempting to power off the SDRAM, which can cause the system to hang. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sram34xx.S | 7 ++++--- 1 files changed, 4 insertions(+), 3 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 8d524f3..9a45415 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -77,7 +77,9 @@ lock_dll: sdram_in_selfrefresh: ldr r4, omap3_sdrc_power @ read the SDRC_POWER register ldr r5, [r4] @ read the contents of SDRC_POWER + mov r9, r5 @ keep a copy of SDRC_POWER bits orr r5, r5, #0x40 @ enable self refresh on idle req + bic r5, r5, #0x4 @ clear PWDENA str r5, [r4] @ write back to SDRC_POWER register ldr r5, [r4] @ posted-write barrier for SDRC ldr r4, omap3_cm_iclken1_core @ read the CM_ICLKEN1_CORE reg @@ -128,10 +130,9 @@ wait_sdrc_idle1: and r5, r5, #0x2 cmp r5, #0 bne wait_sdrc_idle1 +restore_sdrc_power_val: ldr r4, omap3_sdrc_power - ldr r5, [r4] - bic r5, r5, #0x40 - str r5, [r4] + str r9, [r4] @ restore SDRC_POWER, no barrier needed bx lr wait_dll_lock: ldr r4, omap3_sdrc_dlla_status -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 06/23] OMAP3 SDRC: Add 166MHz, 83MHz SDRC settings for the BeagleBoard 2008-12-11 15:52 ` [PATCH 05/23] OMAP3 SRAM: clear the SDRC PWRENA bit during SDRC frequency change Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 07/23] OMAP3 SDRC: initialize SDRC_POWER at boot Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> The BeagleBoard u-boot uses DPLL3 settings that result in 83000000 / 166000000 Hz clock rates for the SDRC, rather than the derated DPLL3 settings used by earlier bootloaders. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sdram-micron-mt46h32m32lf-6.h | 22 +++++++++++++++++--- 1 files changed, 18 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-omap2/sdram-micron-mt46h32m32lf-6.h b/arch/arm/mach-omap2/sdram-micron-mt46h32m32lf-6.h index ef35415..b6c1db3 100644 --- a/arch/arm/mach-omap2/sdram-micron-mt46h32m32lf-6.h +++ b/arch/arm/mach-omap2/sdram-micron-mt46h32m32lf-6.h @@ -20,34 +20,48 @@ /* XXX Using ARE = 0x1 (no autorefresh burst) -- can this be changed? */ static struct omap_sdrc_params mt46h32m32lf6_sdrc_params[] = { [0] = { - .rate = 165941176, + .rate = 166000000, .actim_ctrla = 0x9a9db4c6, .actim_ctrlb = 0x00011217, .rfr_ctrl = 0x0004dc01, .mr = 0x00000032, }, [1] = { + .rate = 165941176, + .actim_ctrla = 0x9a9db4c6, + .actim_ctrlb = 0x00011217, + .rfr_ctrl = 0x0004dc01, + .mr = 0x00000032, + }, + [2] = { .rate = 133333333, .actim_ctrla = 0x7a19b485, .actim_ctrlb = 0x00011213, .rfr_ctrl = 0x0003de01, .mr = 0x00000032, }, - [2] = { + [3] = { + .rate = 83000000, + .actim_ctrla = 0x51512283, + .actim_ctrlb = 0x0001120c, + .rfr_ctrl = 0x00025501, + .mr = 0x00000032, + }, + [4] = { .rate = 82970588, .actim_ctrla = 0x51512283, .actim_ctrlb = 0x0001120c, .rfr_ctrl = 0x00025501, .mr = 0x00000032, }, - [3] = { + [5] = { .rate = 66666666, .actim_ctrla = 0x410d2243, .actim_ctrlb = 0x0001120a, .rfr_ctrl = 0x0001d601, .mr = 0x00000032, }, - [4] = { + [6] = { .rate = 0 }, }; -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 07/23] OMAP3 SDRC: initialize SDRC_POWER at boot 2008-12-11 15:52 ` [PATCH 06/23] OMAP3 SDRC: Add 166MHz, 83MHz SDRC settings for the BeagleBoard Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 08/23] OMAP3 SRAM: renumber registers to make space for argument passing Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Initialize SDRC_POWER to a known-good setting when the kernel boots. Necessary since some bootloaders don't initialize SDRC_POWER properly. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sdrc.c | 19 ++++++++++++++++++- 1 files changed, 18 insertions(+), 1 deletions(-) diff --git a/arch/arm/mach-omap2/sdrc.c b/arch/arm/mach-omap2/sdrc.c index 2a30060..d62e4e1 100644 --- a/arch/arm/mach-omap2/sdrc.c +++ b/arch/arm/mach-omap2/sdrc.c @@ -37,6 +37,10 @@ static struct omap_sdrc_params *sdrc_init_params; void __iomem *omap2_sdrc_base; void __iomem *omap2_sms_base; +/* SDRC_POWER register bits */ +#define SDRC_POWER_EXTCLKDIS_SHIFT 3 +#define SDRC_POWER_PWDENA_SHIFT 2 +#define SDRC_POWER_PAGEPOLICY_SHIFT 0 /** * omap2_sdrc_get_params - return SDRC register values for a given clock rate @@ -74,7 +78,14 @@ void __init omap2_set_globals_sdrc(struct omap_globals *omap2_globals) omap2_sms_base = omap2_globals->sms; } -/* turn on smart idle modes for SDRAM scheduler and controller */ +/** + * omap2_sdrc_init - initialize SMS, SDRC devices on boot + * @sp: pointer to a null-terminated list of struct omap_sdrc_params + * + * Turn on smart idle modes for SDRAM scheduler and controller. + * Program a known-good configuration for the SDRC to deal with buggy + * bootloaders. + */ void __init omap2_sdrc_init(struct omap_sdrc_params *sp) { u32 l; @@ -90,4 +101,10 @@ void __init omap2_sdrc_init(struct omap_sdrc_params *sp) sdrc_write_reg(l, SDRC_SYSCONFIG); sdrc_init_params = sp; + + /* XXX Enable SRFRONIDLEREQ here also? */ + l = (1 << SDRC_POWER_EXTCLKDIS_SHIFT) | + (1 << SDRC_POWER_PWDENA_SHIFT) | + (1 << SDRC_POWER_PAGEPOLICY_SHIFT); + sdrc_write_reg(l, SDRC_POWER); } -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 08/23] OMAP3 SRAM: renumber registers to make space for argument passing 2008-12-11 15:52 ` [PATCH 07/23] OMAP3 SDRC: initialize SDRC_POWER at boot Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 09/23] OMAP3 clock: only unlock SDRC DLL if SDRC clk < 83MHz Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Renumber registers in omap3_sram_configure_core_dpll() assembly code to make space for additional parameters. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sram34xx.S | 114 ++++++++++++++++++++-------------------- 1 files changed, 57 insertions(+), 57 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 9a45415..35131e5 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -63,50 +63,50 @@ ENTRY(omap3_sram_configure_core_dpll) mov r0, #0 @ return value ldmfd sp!, {r1-r12, pc} @ restore regs and return unlock_dll: - ldr r4, omap3_sdrc_dlla_ctrl - ldr r5, [r4] - orr r5, r5, #0x4 - str r5, [r4] @ (no OCP barrier needed) + ldr r11, omap3_sdrc_dlla_ctrl + ldr r12, [r11] + orr r12, r12, #0x4 + str r12, [r11] @ (no OCP barrier needed) bx lr lock_dll: - ldr r4, omap3_sdrc_dlla_ctrl - ldr r5, [r4] - bic r5, r5, #0x4 - str r5, [r4] @ (no OCP barrier needed) + ldr r11, omap3_sdrc_dlla_ctrl + ldr r12, [r11] + bic r12, r12, #0x4 + str r12, [r11] @ (no OCP barrier needed) bx lr sdram_in_selfrefresh: - ldr r4, omap3_sdrc_power @ read the SDRC_POWER register - ldr r5, [r4] @ read the contents of SDRC_POWER - mov r9, r5 @ keep a copy of SDRC_POWER bits - orr r5, r5, #0x40 @ enable self refresh on idle req - bic r5, r5, #0x4 @ clear PWDENA - str r5, [r4] @ write back to SDRC_POWER register - ldr r5, [r4] @ posted-write barrier for SDRC - ldr r4, omap3_cm_iclken1_core @ read the CM_ICLKEN1_CORE reg - ldr r5, [r4] - bic r5, r5, #0x2 @ disable iclk bit for SDRC - str r5, [r4] + ldr r11, omap3_sdrc_power @ read the SDRC_POWER register + ldr r12, [r11] @ read the contents of SDRC_POWER + mov r9, r12 @ keep a copy of SDRC_POWER bits + orr r12, r12, #0x40 @ enable self refresh on idle req + bic r12, r12, #0x4 @ clear PWDENA + str r12, [r11] @ write back to SDRC_POWER register + ldr r12, [r11] @ posted-write barrier for SDRC + ldr r11, omap3_cm_iclken1_core @ read the CM_ICLKEN1_CORE reg + ldr r12, [r11] + bic r12, r12, #0x2 @ disable iclk bit for SDRC + str r12, [r11] wait_sdrc_idle: - ldr r4, omap3_cm_idlest1_core - ldr r5, [r4] - and r5, r5, #0x2 @ check for SDRC idle - cmp r5, #2 + ldr r11, omap3_cm_idlest1_core + ldr r12, [r11] + and r12, r12, #0x2 @ check for SDRC idle + cmp r12, #2 bne wait_sdrc_idle bx lr configure_core_dpll: - ldr r4, omap3_cm_clksel1_pll - ldr r5, [r4] - ldr r6, core_m2_mask_val @ modify m2 for core dpll - and r5, r5, r6 - orr r5, r5, r3, lsl #0x1B @ r3 contains the M2 val - str r5, [r4] - ldr r5, [r4] @ posted-write barrier for CM - mov r5, #0x800 @ wait for the clock to stabilise + ldr r11, omap3_cm_clksel1_pll + ldr r12, [r11] + ldr r10, core_m2_mask_val @ modify m2 for core dpll + and r12, r12, r10 + orr r12, r12, r3, lsl #0x1B @ r3 contains the M2 val + str r12, [r11] + ldr r12, [r11] @ posted-write barrier for CM + mov r12, #0x800 @ wait for the clock to stabilise cmp r3, #2 bne wait_clk_stable bx lr wait_clk_stable: - subs r5, r5, #1 + subs r12, r12, #1 bne wait_clk_stable nop nop @@ -120,42 +120,42 @@ wait_clk_stable: nop bx lr enable_sdrc: - ldr r4, omap3_cm_iclken1_core - ldr r5, [r4] - orr r5, r5, #0x2 @ enable iclk bit for SDRC - str r5, [r4] + ldr r11, omap3_cm_iclken1_core + ldr r12, [r11] + orr r12, r12, #0x2 @ enable iclk bit for SDRC + str r12, [r11] wait_sdrc_idle1: - ldr r4, omap3_cm_idlest1_core - ldr r5, [r4] - and r5, r5, #0x2 - cmp r5, #0 + ldr r11, omap3_cm_idlest1_core + ldr r12, [r11] + and r12, r12, #0x2 + cmp r12, #0 bne wait_sdrc_idle1 restore_sdrc_power_val: - ldr r4, omap3_sdrc_power - str r9, [r4] @ restore SDRC_POWER, no barrier needed + ldr r11, omap3_sdrc_power + str r9, [r11] @ restore SDRC_POWER, no barrier needed bx lr wait_dll_lock: - ldr r4, omap3_sdrc_dlla_status - ldr r5, [r4] - and r5, r5, #0x4 - cmp r5, #0x4 + ldr r11, omap3_sdrc_dlla_status + ldr r12, [r11] + and r12, r12, #0x4 + cmp r12, #0x4 bne wait_dll_lock bx lr wait_dll_unlock: - ldr r4, omap3_sdrc_dlla_status - ldr r5, [r4] - and r5, r5, #0x4 - cmp r5, #0x0 + ldr r11, omap3_sdrc_dlla_status + ldr r12, [r11] + and r12, r12, #0x4 + cmp r12, #0x0 bne wait_dll_unlock bx lr configure_sdrc: - ldr r4, omap3_sdrc_rfr_ctrl - str r0, [r4] - ldr r4, omap3_sdrc_actim_ctrla - str r1, [r4] - ldr r4, omap3_sdrc_actim_ctrlb - str r2, [r4] - ldr r2, [r4] @ posted-write barrier for SDRC + ldr r11, omap3_sdrc_rfr_ctrl + str r0, [r11] + ldr r11, omap3_sdrc_actim_ctrla + str r1, [r11] + ldr r11, omap3_sdrc_actim_ctrlb + str r2, [r11] + ldr r2, [r11] @ posted-write barrier for SDRC bx lr omap3_sdrc_power: -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 09/23] OMAP3 clock: only unlock SDRC DLL if SDRC clk < 83MHz 2008-12-11 15:52 ` [PATCH 08/23] OMAP3 SRAM: renumber registers to make space for argument passing Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 10/23] OMAP3 clock: use pr_debug() rather than pr_info() in some clock change code Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> According to the 34xx TRM Rev. K section 11.2.4.4.11.1 "Purpose of the DLL/CDL Module," the SDRC delay-locked-loop can be locked at any SDRC clock frequency from 83MHz to 166MHz. CDP code unconditionally unlocked the DLL whenever shifting to a lower SDRC speed, but this seems unnecessary and error-prone, as the DLL is no longer able to compensate for process, voltage, and temperature variations. Instead, only unlock the DLL when the SDRC clock rate would be less than 83MHz. In situations where the DLL is unlocked, we should probably be changing the FIXEDDELAY field, but currently we have no information on how to calculate it. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/clock34xx.c | 11 ++++++++++- arch/arm/mach-omap2/sram34xx.S | 13 +++++++------ arch/arm/plat-omap/include/mach/sram.h | 3 ++- arch/arm/plat-omap/sram.c | 7 ++++--- 4 files changed, 23 insertions(+), 11 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index 3000c33..efeb9c9 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -48,6 +48,9 @@ #define MAX_DPLL_WAIT_TRIES 1000000 +/* Minimum SDRC clock frequency that the SDRC DLL can lock at */ +#define MIN_SDRC_DLL_LOCK_FREQ 83000000 + struct omap_opp *curr_vdd1_prcm_set; struct omap_opp *curr_vdd2_prcm_set; static struct clk *dpll1_clk, *dpll2_clk, *dpll3_clk; @@ -482,6 +485,7 @@ static int omap3_noncore_dpll_set_rate(struct clk *clk, unsigned long rate) static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) { u32 new_div = 0; + u32 unlock_dll = 0; unsigned long validrate, sdrcrate; struct omap_sdrc_params *sp; @@ -508,6 +512,11 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) if (!sp) return -EINVAL; + if (sdrcrate < MIN_SDRC_DLL_LOCK_FREQ) { + pr_debug("clock: will unlock SDRC DLL\n"); + unlock_dll = 1; + } + pr_info("clock: changing CORE DPLL rate from %lu to %lu\n", clk->rate, validrate); pr_info("clock: SDRC timing params used: %08x %08x %08x\n", @@ -519,7 +528,7 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) /* REVISIT: Add SDRC_MR changing to this code also */ local_irq_disable(); omap3_configure_core_dpll(sp->rfr_ctrl, sp->actim_ctrla, - sp->actim_ctrlb, new_div); + sp->actim_ctrlb, new_div, unlock_dll); local_irq_enable(); omap2_clksel_recalc(clk); diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 35131e5..c080c82 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -40,22 +40,23 @@ /* * Change frequency of core dpll * r0 = sdrc_rfr_ctrl r1 = sdrc_actim_ctrla r2 = sdrc_actim_ctrlb r3 = M2 + * r4 = Unlock SDRC DLL? (1 = yes, 0 = no) -- only unlock DLL for + * SDRC rates < 83MHz */ ENTRY(omap3_sram_configure_core_dpll) stmfd sp!, {r1-r12, lr} @ store regs to stack + ldr r4, [sp, #52] @ pull extra args off the stack dsb @ flush buffered writes to interconnect cmp r3, #0x2 blne configure_sdrc - cmp r3, #0x2 + cmp r4, #0x1 + bleq unlock_dll blne lock_dll - cmp r3, #0x1 - blne unlock_dll bl sdram_in_selfrefresh @ put the SDRAM in self refresh bl configure_core_dpll bl enable_sdrc - cmp r3, #0x1 - blne wait_dll_unlock - cmp r3, #0x2 + cmp r4, #0x1 + bleq wait_dll_unlock blne wait_dll_lock cmp r3, #0x1 blne configure_sdrc diff --git a/arch/arm/plat-omap/include/mach/sram.h b/arch/arm/plat-omap/include/mach/sram.h index 0c0b45f..8ce41e0 100644 --- a/arch/arm/plat-omap/include/mach/sram.h +++ b/arch/arm/plat-omap/include/mach/sram.h @@ -23,7 +23,8 @@ extern u32 omap2_set_prcm(u32 dpll_ctrl_val, u32 sdrc_rfr_val, int bypass); extern u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, - u32 sdrc_actim_ctrlb, u32 m2); + u32 sdrc_actim_ctrlb, u32 m2, + u32 unlock_dll); extern void omap3_sram_restore_context(void); /* Do not use these */ diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c index 04214e1..65fcfdf 100644 --- a/arch/arm/plat-omap/sram.c +++ b/arch/arm/plat-omap/sram.c @@ -371,16 +371,17 @@ static inline int omap243x_sram_init(void) static u32 (*_omap3_sram_configure_core_dpll)(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, - u32 m2); + u32 m2, u32 unlock_dll); u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, - u32 sdrc_actim_ctrlb, u32 m2) + u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll) { if (!_omap3_sram_configure_core_dpll) omap_sram_error(); return _omap3_sram_configure_core_dpll(sdrc_rfr_ctrl, sdrc_actim_ctrla, - sdrc_actim_ctrlb, m2); + sdrc_actim_ctrlb, m2, + unlock_dll); } #ifdef CONFIG_PM -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 10/23] OMAP3 clock: use pr_debug() rather than pr_info() in some clock change code 2008-12-11 15:52 ` [PATCH 09/23] OMAP3 clock: only unlock SDRC DLL if SDRC clk < 83MHz Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 11/23] OMAP3 clock: remove wait for DPLL3 M2 clock to stabilize Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> The CORE DPLL M2 frequency change code should use pr_debug(), not pr_info(), for its debug messages. Same with omap2_clksel_round_rate_div(). While here, convert a few printk(KERN_ERR .. into pr_err(). Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/clock.c | 12 ++++++------ arch/arm/mach-omap2/clock34xx.c | 8 ++++---- 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/arch/arm/mach-omap2/clock.c b/arch/arm/mach-omap2/clock.c index b4ef3a3..7c7e60e 100644 --- a/arch/arm/mach-omap2/clock.c +++ b/arch/arm/mach-omap2/clock.c @@ -564,8 +564,8 @@ u32 omap2_clksel_round_rate_div(struct clk *clk, unsigned long target_rate, const struct clksel_rate *clkr; u32 last_div = 0; - printk(KERN_INFO "clock: clksel_round_rate_div: %s target_rate %ld\n", - clk->name, target_rate); + pr_debug("clock: clksel_round_rate_div: %s target_rate %ld\n", + clk->name, target_rate); *new_div = 1; @@ -579,7 +579,7 @@ u32 omap2_clksel_round_rate_div(struct clk *clk, unsigned long target_rate, /* Sanity check */ if (clkr->div <= last_div) - printk(KERN_ERR "clock: clksel_rate table not sorted " + pr_err("clock: clksel_rate table not sorted " "for clock %s", clk->name); last_div = clkr->div; @@ -591,7 +591,7 @@ u32 omap2_clksel_round_rate_div(struct clk *clk, unsigned long target_rate, } if (!clkr->div) { - printk(KERN_ERR "clock: Could not find divisor for target " + pr_err("clock: Could not find divisor for target " "rate %ld for clock %s parent %s\n", target_rate, clk->name, clk->parent->name); return ~0; @@ -599,8 +599,8 @@ u32 omap2_clksel_round_rate_div(struct clk *clk, unsigned long target_rate, *new_div = clkr->div; - printk(KERN_INFO "clock: new_div = %d, new_rate = %ld\n", *new_div, - (clk->parent->rate / clkr->div)); + pr_debug("clock: new_div = %d, new_rate = %ld\n", *new_div, + (clk->parent->rate / clkr->div)); return (clk->parent->rate / clkr->div); } diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index efeb9c9..adf0be3 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -517,10 +517,10 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) unlock_dll = 1; } - pr_info("clock: changing CORE DPLL rate from %lu to %lu\n", clk->rate, - validrate); - pr_info("clock: SDRC timing params used: %08x %08x %08x\n", - sp->rfr_ctrl, sp->actim_ctrla, sp->actim_ctrlb); + pr_debug("clock: changing CORE DPLL rate from %lu to %lu\n", clk->rate, + validrate); + pr_debug("clock: SDRC timing params used: %08x %08x %08x\n", + sp->rfr_ctrl, sp->actim_ctrla, sp->actim_ctrlb); /* REVISIT: SRAM code doesn't support other M2 divisors yet */ WARN_ON(new_div != 1 && new_div != 2); -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 11/23] OMAP3 clock: remove wait for DPLL3 M2 clock to stabilize 2008-12-11 15:52 ` [PATCH 10/23] OMAP3 clock: use pr_debug() rather than pr_info() in some clock change code Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 12/23] OMAP3 clock: initialize SDRC timings at kernel start Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> The original CDP kernel that this code comes from waited for 0x800 loops after switching the CORE DPLL M2 divider. This does not appear to be necessary. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sram34xx.S | 3 --- 1 files changed, 0 insertions(+), 3 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index c080c82..84781a6 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -102,9 +102,6 @@ configure_core_dpll: orr r12, r12, r3, lsl #0x1B @ r3 contains the M2 val str r12, [r11] ldr r12, [r11] @ posted-write barrier for CM - mov r12, #0x800 @ wait for the clock to stabilise - cmp r3, #2 - bne wait_clk_stable bx lr wait_clk_stable: subs r12, r12, #1 -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 12/23] OMAP3 clock: initialize SDRC timings at kernel start 2008-12-11 15:52 ` [PATCH 11/23] OMAP3 clock: remove wait for DPLL3 M2 clock to stabilize Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 13/23] OMAP3 clock: add a short delay when lowering CORE clk rate Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> On the OMAP3, initialize SDRC timings when the kernel boots. This ensures that the kernel is running with known, optimized SDRC timings, rather than whatever was configured by the bootloader. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/clock34xx.c | 3 --- arch/arm/mach-omap2/io.c | 36 ++++++++++++++++++++++++++++++++++++ 2 files changed, 36 insertions(+), 3 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index adf0be3..cea3840 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -495,9 +495,6 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) if (clk != &dpll3_m2_ck) return -EINVAL; - if (rate == clk->rate) - return 0; - validrate = omap2_clksel_round_rate_div(clk, rate, &new_div); if (validrate != rate) return -EINVAL; diff --git a/arch/arm/mach-omap2/io.c b/arch/arm/mach-omap2/io.c index 5b529ae..6fb6c29 100644 --- a/arch/arm/mach-omap2/io.c +++ b/arch/arm/mach-omap2/io.c @@ -19,6 +19,7 @@ #include <linux/kernel.h> #include <linux/init.h> #include <linux/io.h> +#include <linux/clk.h> #include <asm/tlb.h> @@ -198,6 +199,38 @@ void __init omap2_map_common_io(void) omapfb_reserve_sdram(); } +/* + * omap2_init_reprogram_sdrc - reprogram SDRC timing parameters + * + * Sets the CORE DPLL3 M2 divider to the same value that it's at + * currently. This has the effect of setting the SDRC SDRAM AC timing + * registers to the values currently defined by the kernel. Currently + * only defined for OMAP3; will return 0 if called on OMAP2. Returns + * -EINVAL if the dpll3_m2_ck cannot be found, 0 if called on OMAP2, + * or passes along the return value of clk_set_rate(). + */ +static int __init _omap2_init_reprogram_sdrc(void) +{ + struct clk *dpll3_m2_ck; + int v = -EINVAL; + + if (!cpu_is_omap34xx()) + return 0; + + dpll3_m2_ck = clk_get(NULL, "dpll3_m2_ck"); + if (!dpll3_m2_ck) + return -EINVAL; + + pr_info("Reprogramming SDRC\n"); + v = clk_set_rate(dpll3_m2_ck, clk_get_rate(dpll3_m2_ck)); + if (v) + pr_err("dpll3_m2_clk rate change failed: %d\n", v); + + clk_put(dpll3_m2_ck); + + return v; +} + void __init omap2_init_common_hw(struct omap_sdrc_params *sp, struct omap_opp *mpu_opps, struct omap_opp *dsp_opps, @@ -212,5 +245,8 @@ void __init omap2_init_common_hw(struct omap_sdrc_params *sp, omap2_clk_init(); omap_pm_if_init(); omap2_sdrc_init(sp); + + _omap2_init_reprogram_sdrc(); + gpmc_init(); } -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 13/23] OMAP3 clock: add a short delay when lowering CORE clk rate 2008-12-11 15:52 ` [PATCH 12/23] OMAP3 clock: initialize SDRC timings at kernel start Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 14/23] OMAP3 clock/SDRC: program SDRC_MR register during SDRC clock change Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> When changing the SDRAM clock from 166MHz to 83MHz via the CORE DPLL M2 divider, add a short delay before returning to SDRAM to allow the SDRC time to stabilize. Without this delay, the system is prone to random panics upon re-entering SDRAM. This time delay varies based on MPU frequency. At 500MHz MPU frequency at room temperature, 64 loops seems to work okay; so add another 32 loops for environmental and process variation. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/clock34xx.c | 30 ++++++++++++++++++++++++++++-- arch/arm/mach-omap2/sram34xx.S | 20 +++++++++----------- arch/arm/plat-omap/include/mach/sram.h | 2 +- arch/arm/plat-omap/sram.c | 8 +++++--- 4 files changed, 43 insertions(+), 17 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index cea3840..8ef0186 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -51,6 +51,20 @@ /* Minimum SDRC clock frequency that the SDRC DLL can lock at */ #define MIN_SDRC_DLL_LOCK_FREQ 83000000 +#define CYCLES_PER_MHZ 1000000 + +/* Scale factor for fixed-point arith in omap3_core_dpll_m2_set_rate() */ +#define SDRC_MPURATE_SCALE 8 + +/* 2^SDRC_MPURATE_BASE_SHIFT: MPU MHz that SDRC_MPURATE_LOOPS is defined for */ +#define SDRC_MPURATE_BASE_SHIFT 9 + +/* + * SDRC_MPURATE_LOOPS: Number of MPU loops to execute at + * 2^MPURATE_BASE_SHIFT MHz for SDRC to stabilize + */ +#define SDRC_MPURATE_LOOPS 96 + struct omap_opp *curr_vdd1_prcm_set; struct omap_opp *curr_vdd2_prcm_set; static struct clk *dpll1_clk, *dpll2_clk, *dpll3_clk; @@ -486,7 +500,8 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) { u32 new_div = 0; u32 unlock_dll = 0; - unsigned long validrate, sdrcrate; + u32 c; + unsigned long validrate, sdrcrate, mpurate; struct omap_sdrc_params *sp; if (!clk || !rate) @@ -514,6 +529,17 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) unlock_dll = 1; } + /* + * XXX This only needs to be done when the CPU frequency changes + */ + mpurate = arm_fck.rate / CYCLES_PER_MHZ; + c = (mpurate << SDRC_MPURATE_SCALE) >> SDRC_MPURATE_BASE_SHIFT; + c += 1; /* for safety */ + c *= SDRC_MPURATE_LOOPS; + c >>= SDRC_MPURATE_SCALE; + if (c == 0) + c = 1; + pr_debug("clock: changing CORE DPLL rate from %lu to %lu\n", clk->rate, validrate); pr_debug("clock: SDRC timing params used: %08x %08x %08x\n", @@ -525,7 +551,7 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) /* REVISIT: Add SDRC_MR changing to this code also */ local_irq_disable(); omap3_configure_core_dpll(sp->rfr_ctrl, sp->actim_ctrla, - sp->actim_ctrlb, new_div, unlock_dll); + sp->actim_ctrlb, new_div, unlock_dll, c); local_irq_enable(); omap2_clksel_recalc(clk); diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 84781a6..8d4a88c 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -42,10 +42,14 @@ * r0 = sdrc_rfr_ctrl r1 = sdrc_actim_ctrla r2 = sdrc_actim_ctrlb r3 = M2 * r4 = Unlock SDRC DLL? (1 = yes, 0 = no) -- only unlock DLL for * SDRC rates < 83MHz + * r5 = number of MPU cycles to wait for SDRC to stabilize after + * reprogramming the SDRC when switching to a slower MPU speed + * */ ENTRY(omap3_sram_configure_core_dpll) stmfd sp!, {r1-r12, lr} @ store regs to stack ldr r4, [sp, #52] @ pull extra args off the stack + ldr r5, [sp, #56] @ load extra args from the stack dsb @ flush buffered writes to interconnect cmp r3, #0x2 blne configure_sdrc @@ -59,7 +63,11 @@ ENTRY(omap3_sram_configure_core_dpll) bleq wait_dll_unlock blne wait_dll_lock cmp r3, #0x1 - blne configure_sdrc + beq return_to_sdram + bl configure_sdrc + mov r12, r5 @ if slowing, wait for SDRC to stabilize + bl wait_clk_stable +return_to_sdram: isb @ prevent speculative exec past here mov r0, #0 @ return value ldmfd sp!, {r1-r12, pc} @ restore regs and return @@ -106,16 +114,6 @@ configure_core_dpll: wait_clk_stable: subs r12, r12, #1 bne wait_clk_stable - nop - nop - nop - nop - nop - nop - nop - nop - nop - nop bx lr enable_sdrc: ldr r11, omap3_cm_iclken1_core diff --git a/arch/arm/plat-omap/include/mach/sram.h b/arch/arm/plat-omap/include/mach/sram.h index 8ce41e0..54913fc 100644 --- a/arch/arm/plat-omap/include/mach/sram.h +++ b/arch/arm/plat-omap/include/mach/sram.h @@ -24,7 +24,7 @@ extern u32 omap2_set_prcm(u32 dpll_ctrl_val, u32 sdrc_rfr_val, int bypass); extern u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, - u32 unlock_dll); + u32 unlock_dll, u32 f); extern void omap3_sram_restore_context(void); /* Do not use these */ diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c index 65fcfdf..8d27e91 100644 --- a/arch/arm/plat-omap/sram.c +++ b/arch/arm/plat-omap/sram.c @@ -371,9 +371,11 @@ static inline int omap243x_sram_init(void) static u32 (*_omap3_sram_configure_core_dpll)(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, - u32 m2, u32 unlock_dll); + u32 m2, u32 unlock_dll, + u32 f); u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, - u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll) + u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll, + u32 f) { if (!_omap3_sram_configure_core_dpll) omap_sram_error(); @@ -381,7 +383,7 @@ u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, return _omap3_sram_configure_core_dpll(sdrc_rfr_ctrl, sdrc_actim_ctrla, sdrc_actim_ctrlb, m2, - unlock_dll); + unlock_dll, f); } #ifdef CONFIG_PM -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 14/23] OMAP3 clock/SDRC: program SDRC_MR register during SDRC clock change 2008-12-11 15:52 ` [PATCH 13/23] OMAP3 clock: add a short delay when lowering CORE clk rate Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 15/23] OMAP3 SRAM: add more comments on the SRAM code Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Program the SDRC_MR_0 register as well during SDRC clock changes. This register allows selection of the memory CAS latency. Some SDRAM chips, such as the Qimonda HYB18M512160AF6, have a lower CAS latency at lower clock rates. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/clock34xx.c | 4 ++-- arch/arm/mach-omap2/sram34xx.S | 8 +++++++- arch/arm/plat-omap/include/mach/sram.h | 2 +- arch/arm/plat-omap/sram.c | 6 +++--- 4 files changed, 13 insertions(+), 7 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index 8ef0186..47a8465 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -548,10 +548,10 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) /* REVISIT: SRAM code doesn't support other M2 divisors yet */ WARN_ON(new_div != 1 && new_div != 2); - /* REVISIT: Add SDRC_MR changing to this code also */ local_irq_disable(); omap3_configure_core_dpll(sp->rfr_ctrl, sp->actim_ctrla, - sp->actim_ctrlb, new_div, unlock_dll, c); + sp->actim_ctrlb, new_div, unlock_dll, c, + sp->mr); local_irq_enable(); omap2_clksel_recalc(clk); diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 8d4a88c..d13f1cc 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -44,12 +44,14 @@ * SDRC rates < 83MHz * r5 = number of MPU cycles to wait for SDRC to stabilize after * reprogramming the SDRC when switching to a slower MPU speed + * r6 = SDRC_MR_0 register value * */ ENTRY(omap3_sram_configure_core_dpll) stmfd sp!, {r1-r12, lr} @ store regs to stack ldr r4, [sp, #52] @ pull extra args off the stack ldr r5, [sp, #56] @ load extra args from the stack + ldr r6, [sp, #60] @ load extra args from the stack dsb @ flush buffered writes to interconnect cmp r3, #0x2 blne configure_sdrc @@ -151,7 +153,9 @@ configure_sdrc: str r1, [r11] ldr r11, omap3_sdrc_actim_ctrlb str r2, [r11] - ldr r2, [r11] @ posted-write barrier for SDRC + ldr r11, omap3_sdrc_mr_0 + str r6, [r11] + ldr r6, [r11] @ posted-write barrier for SDRC bx lr omap3_sdrc_power: @@ -168,6 +172,8 @@ omap3_sdrc_actim_ctrla: .word OMAP34XX_SDRC_REGADDR(SDRC_ACTIM_CTRL_A_0) omap3_sdrc_actim_ctrlb: .word OMAP34XX_SDRC_REGADDR(SDRC_ACTIM_CTRL_B_0) +omap3_sdrc_mr_0: + .word OMAP34XX_SDRC_REGADDR(SDRC_MR_0) omap3_sdrc_dlla_status: .word OMAP34XX_SDRC_REGADDR(SDRC_DLLA_STATUS) omap3_sdrc_dlla_ctrl: diff --git a/arch/arm/plat-omap/include/mach/sram.h b/arch/arm/plat-omap/include/mach/sram.h index 54913fc..d07da3b 100644 --- a/arch/arm/plat-omap/include/mach/sram.h +++ b/arch/arm/plat-omap/include/mach/sram.h @@ -24,7 +24,7 @@ extern u32 omap2_set_prcm(u32 dpll_ctrl_val, u32 sdrc_rfr_val, int bypass); extern u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, - u32 unlock_dll, u32 f); + u32 unlock_dll, u32 f, u32 sdrc_mr); extern void omap3_sram_restore_context(void); /* Do not use these */ diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c index 8d27e91..68a1f69 100644 --- a/arch/arm/plat-omap/sram.c +++ b/arch/arm/plat-omap/sram.c @@ -372,10 +372,10 @@ static u32 (*_omap3_sram_configure_core_dpll)(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll, - u32 f); + u32 f, u32 sdrc_mr); u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll, - u32 f) + u32 f, u32 sdrc_mr) { if (!_omap3_sram_configure_core_dpll) omap_sram_error(); @@ -383,7 +383,7 @@ u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, return _omap3_sram_configure_core_dpll(sdrc_rfr_ctrl, sdrc_actim_ctrla, sdrc_actim_ctrlb, m2, - unlock_dll, f); + unlock_dll, f, sdrc_mr); } #ifdef CONFIG_PM -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 15/23] OMAP3 SRAM: add more comments on the SRAM code 2008-12-11 15:52 ` [PATCH 14/23] OMAP3 clock/SDRC: program SDRC_MR register during SDRC clock change Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 16/23] OMAP3 SRAM: convert SRAM code to use macros rather than magic numbers Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Clean up comments and copyrights on the CORE DPLL3 M2 divider change code. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sram34xx.S | 45 +++++++++++++++++++++------------------ 1 files changed, 24 insertions(+), 21 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index d13f1cc..37a1e1f 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -3,13 +3,12 @@ * * Omap3 specific functions that need to be run in internal SRAM * - * (C) Copyright 2007 - * Texas Instruments Inc. - * Rajendra Nayak <rnayak@ti.com> + * Copyright (C) 2004, 2007, 2008 Texas Instruments, Inc. + * Copyright (C) 2008 Nokia Corporation * - * (C) Copyright 2004 - * Texas Instruments, <www.ti.com> + * Rajendra Nayak <rnayak@ti.com> * Richard Woodruff <r-woodruff2@ti.com> + * Paul Walmsley * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License as @@ -38,13 +37,16 @@ .text /* - * Change frequency of core dpll - * r0 = sdrc_rfr_ctrl r1 = sdrc_actim_ctrla r2 = sdrc_actim_ctrlb r3 = M2 - * r4 = Unlock SDRC DLL? (1 = yes, 0 = no) -- only unlock DLL for + * omap3_sram_configure_core_dpll - change DPLL3 M2 divider + * r0 = new SDRC_RFR_CTRL register contents + * r1 = new SDRC_ACTIM_CTRLA register contents + * r2 = new SDRC_ACTIM_CTRLB register contents + * r3 = new M2 divider setting (only 1 and 2 supported right now) + * r4 = unlock SDRC DLL? (1 = yes, 0 = no). Only unlock DLL for * SDRC rates < 83MHz * r5 = number of MPU cycles to wait for SDRC to stabilize after * reprogramming the SDRC when switching to a slower MPU speed - * r6 = SDRC_MR_0 register value + * r6 = new SDRC_MR_0 register value * */ ENTRY(omap3_sram_configure_core_dpll) @@ -53,22 +55,22 @@ ENTRY(omap3_sram_configure_core_dpll) ldr r5, [sp, #56] @ load extra args from the stack ldr r6, [sp, #60] @ load extra args from the stack dsb @ flush buffered writes to interconnect - cmp r3, #0x2 - blne configure_sdrc - cmp r4, #0x1 + cmp r3, #0x2 @ if increasing SDRC clk rate, + blne configure_sdrc @ program the SDRC regs early (for RFR) + cmp r4, #0x1 @ set the intended DLL state bleq unlock_dll blne lock_dll - bl sdram_in_selfrefresh @ put the SDRAM in self refresh - bl configure_core_dpll - bl enable_sdrc - cmp r4, #0x1 + bl sdram_in_selfrefresh @ put SDRAM in self refresh, idle SDRC + bl configure_core_dpll @ change the DPLL3 M2 divider + bl enable_sdrc @ take SDRC out of idle + cmp r4, #0x1 @ wait for DLL status to change bleq wait_dll_unlock blne wait_dll_lock - cmp r3, #0x1 - beq return_to_sdram - bl configure_sdrc - mov r12, r5 @ if slowing, wait for SDRC to stabilize - bl wait_clk_stable + cmp r3, #0x1 @ if increasing SDRC clk rate, + beq return_to_sdram @ return to SDRAM code, otherwise, + bl configure_sdrc @ reprogram SDRC regs now + mov r12, r5 + bl wait_clk_stable @ wait for SDRC to stabilize return_to_sdram: isb @ prevent speculative exec past here mov r0, #0 @ return value @@ -93,6 +95,7 @@ sdram_in_selfrefresh: bic r12, r12, #0x4 @ clear PWDENA str r12, [r11] @ write back to SDRC_POWER register ldr r12, [r11] @ posted-write barrier for SDRC +idle_sdrc: ldr r11, omap3_cm_iclken1_core @ read the CM_ICLKEN1_CORE reg ldr r12, [r11] bic r12, r12, #0x2 @ disable iclk bit for SDRC -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 16/23] OMAP3 SRAM: convert SRAM code to use macros rather than magic numbers 2008-12-11 15:52 ` [PATCH 15/23] OMAP3 SRAM: add more comments on the SRAM code Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 17/23] OMAP3: PM: Fixed VDD2 control to work from both sysfs and SRF API Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap; +Cc: Paul Walmsley From: Paul Walmsley <paul@pwsan.com> Convert omap3_sram_configure_core_dpll() to use macros rather than magic numbers. Signed-off-by: Paul Walmsley <paul@pwsan.com> --- arch/arm/mach-omap2/sram34xx.S | 53 ++++++++++++++++++++++++++++----------- 1 files changed, 38 insertions(+), 15 deletions(-) diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 37a1e1f..16eb4ef 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -36,6 +36,29 @@ .text +/* r4 parameters */ +#define SDRC_NO_UNLOCK_DLL 0x0 +#define SDRC_UNLOCK_DLL 0x1 + +/* SDRC_DLLA_CTRL bit settings */ +#define DLLIDLE_MASK 0x4 + +/* SDRC_DLLA_STATUS bit settings */ +#define LOCKSTATUS_MASK 0x4 + +/* SDRC_POWER bit settings */ +#define SRFRONIDLEREQ_MASK 0x40 +#define PWDENA_MASK 0x4 + +/* CM_IDLEST1_CORE bit settings */ +#define ST_SDRC_MASK 0x2 + +/* CM_ICLKEN1_CORE bit settings */ +#define EN_SDRC_MASK 0x2 + +/* CM_CLKSEL1_PLL bit settings */ +#define CORE_DPLL_CLKOUT_DIV_SHIFT 0x1b + /* * omap3_sram_configure_core_dpll - change DPLL3 M2 divider * r0 = new SDRC_RFR_CTRL register contents @@ -57,13 +80,13 @@ ENTRY(omap3_sram_configure_core_dpll) dsb @ flush buffered writes to interconnect cmp r3, #0x2 @ if increasing SDRC clk rate, blne configure_sdrc @ program the SDRC regs early (for RFR) - cmp r4, #0x1 @ set the intended DLL state + cmp r4, #SDRC_UNLOCK_DLL @ set the intended DLL state bleq unlock_dll blne lock_dll bl sdram_in_selfrefresh @ put SDRAM in self refresh, idle SDRC bl configure_core_dpll @ change the DPLL3 M2 divider bl enable_sdrc @ take SDRC out of idle - cmp r4, #0x1 @ wait for DLL status to change + cmp r4, #SDRC_UNLOCK_DLL @ wait for DLL status to change bleq wait_dll_unlock blne wait_dll_lock cmp r3, #0x1 @ if increasing SDRC clk rate, @@ -78,33 +101,33 @@ return_to_sdram: unlock_dll: ldr r11, omap3_sdrc_dlla_ctrl ldr r12, [r11] - orr r12, r12, #0x4 + orr r12, r12, #DLLIDLE_MASK str r12, [r11] @ (no OCP barrier needed) bx lr lock_dll: ldr r11, omap3_sdrc_dlla_ctrl ldr r12, [r11] - bic r12, r12, #0x4 + bic r12, r12, #DLLIDLE_MASK str r12, [r11] @ (no OCP barrier needed) bx lr sdram_in_selfrefresh: ldr r11, omap3_sdrc_power @ read the SDRC_POWER register ldr r12, [r11] @ read the contents of SDRC_POWER mov r9, r12 @ keep a copy of SDRC_POWER bits - orr r12, r12, #0x40 @ enable self refresh on idle req - bic r12, r12, #0x4 @ clear PWDENA + orr r12, r12, #SRFRONIDLEREQ_MASK @ enable self refresh on idle + bic r12, r12, #PWDENA_MASK @ clear PWDENA str r12, [r11] @ write back to SDRC_POWER register ldr r12, [r11] @ posted-write barrier for SDRC idle_sdrc: ldr r11, omap3_cm_iclken1_core @ read the CM_ICLKEN1_CORE reg ldr r12, [r11] - bic r12, r12, #0x2 @ disable iclk bit for SDRC + bic r12, r12, #EN_SDRC_MASK @ disable iclk bit for SDRC str r12, [r11] wait_sdrc_idle: ldr r11, omap3_cm_idlest1_core ldr r12, [r11] - and r12, r12, #0x2 @ check for SDRC idle - cmp r12, #2 + and r12, r12, #ST_SDRC_MASK @ check for SDRC idle + cmp r12, #ST_SDRC_MASK bne wait_sdrc_idle bx lr configure_core_dpll: @@ -112,7 +135,7 @@ configure_core_dpll: ldr r12, [r11] ldr r10, core_m2_mask_val @ modify m2 for core dpll and r12, r12, r10 - orr r12, r12, r3, lsl #0x1B @ r3 contains the M2 val + orr r12, r12, r3, lsl #CORE_DPLL_CLKOUT_DIV_SHIFT str r12, [r11] ldr r12, [r11] @ posted-write barrier for CM bx lr @@ -123,12 +146,12 @@ wait_clk_stable: enable_sdrc: ldr r11, omap3_cm_iclken1_core ldr r12, [r11] - orr r12, r12, #0x2 @ enable iclk bit for SDRC + orr r12, r12, #EN_SDRC_MASK @ enable iclk bit for SDRC str r12, [r11] wait_sdrc_idle1: ldr r11, omap3_cm_idlest1_core ldr r12, [r11] - and r12, r12, #0x2 + and r12, r12, #ST_SDRC_MASK cmp r12, #0 bne wait_sdrc_idle1 restore_sdrc_power_val: @@ -138,14 +161,14 @@ restore_sdrc_power_val: wait_dll_lock: ldr r11, omap3_sdrc_dlla_status ldr r12, [r11] - and r12, r12, #0x4 - cmp r12, #0x4 + and r12, r12, #LOCKSTATUS_MASK + cmp r12, #LOCKSTATUS_MASK bne wait_dll_lock bx lr wait_dll_unlock: ldr r11, omap3_sdrc_dlla_status ldr r12, [r11] - and r12, r12, #0x4 + and r12, r12, #LOCKSTATUS_MASK cmp r12, #0x0 bne wait_dll_unlock bx lr -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 17/23] OMAP3: PM: Fixed VDD2 control to work from both sysfs and SRF API 2008-12-11 15:52 ` [PATCH 16/23] OMAP3 SRAM: convert SRAM code to use macros rather than magic numbers Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 18/23] OMAP3: PM: Added DVFS OPP locking interface for VDD1 and VDD2 Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap SRF API for setting throughput requirement attempted to use throughput values as parameter to set_opp() and sysfs was attempting to use OPP levels (1..3.) Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/pm.c | 5 +- arch/arm/mach-omap2/pm.h | 1 + arch/arm/mach-omap2/resource34xx.c | 62 ++++++++++++++++++---------- arch/arm/plat-omap/include/mach/omap34xx.h | 4 ++ 4 files changed, 47 insertions(+), 25 deletions(-) diff --git a/arch/arm/mach-omap2/pm.c b/arch/arm/mach-omap2/pm.c index 285f596..2dd1341 100644 --- a/arch/arm/mach-omap2/pm.c +++ b/arch/arm/mach-omap2/pm.c @@ -61,7 +61,6 @@ static struct kobj_attribute voltage_off_while_idle_attr = __ATTR(voltage_off_while_idle, 0644, idle_show, idle_store); #ifdef CONFIG_OMAP_PM_SRF -static struct device dummy_sysfs_dev; static ssize_t vdd_opp_show(struct kobject *, struct kobj_attribute *, char *); static ssize_t vdd_opp_store(struct kobject *k, struct kobj_attribute *, const char *buf, size_t n); @@ -145,13 +144,13 @@ static ssize_t vdd_opp_store(struct kobject *kobj, struct kobj_attribute *attr, printk(KERN_ERR "vdd_opp_store: Invalid value\n"); return -EINVAL; } - resource_request("vdd1_opp", &dummy_sysfs_dev, value); + set_opp_level(VDD1_OPP, value); } else if (attr == &vdd2_opp_attr) { if (value < 2 || value > 3) { printk(KERN_ERR "vdd_opp_store: Invalid value\n"); return -EINVAL; } - resource_request("vdd2_opp", &dummy_sysfs_dev, value); + set_opp_level(VDD2_OPP, value); } else { return -EINVAL; } diff --git a/arch/arm/mach-omap2/pm.h b/arch/arm/mach-omap2/pm.h index a209819..4c0052f 100644 --- a/arch/arm/mach-omap2/pm.h +++ b/arch/arm/mach-omap2/pm.h @@ -42,6 +42,7 @@ extern int omap3_pm_set_suspend_state(struct powerdomain *pwrdm, int state); #define omap3_pm_set_suspend_state(pwrdm,state) do {} while (0); #endif extern int set_pwrdm_state(struct powerdomain *pwrdm, u32 state); +extern int set_opp_level(int res, u32 target_level); #ifdef CONFIG_PM_DEBUG extern void omap2_pm_dump(int mode, int resume, unsigned int us); diff --git a/arch/arm/mach-omap2/resource34xx.c b/arch/arm/mach-omap2/resource34xx.c index 034318b..d2c4e1f 100644 --- a/arch/arm/mach-omap2/resource34xx.c +++ b/arch/arm/mach-omap2/resource34xx.c @@ -135,6 +135,8 @@ int set_pd_latency(struct shared_resource *resp, u32 latency) static struct clk *vdd1_clk; static struct clk *vdd2_clk; +static struct shared_resource *vdd1_resp; +static struct shared_resource *vdd2_resp; static struct device dummy_mpu_dev; static struct device dummy_dsp_dev; @@ -154,19 +156,27 @@ void init_opp(struct shared_resource *resp) if (strcmp(resp->name, "vdd1_opp") == 0) { resp->curr_level = curr_vdd1_prcm_set->opp_id; vdd1_clk = clk_get(NULL, "virt_vdd1_prcm_set"); + vdd1_resp = resp; } else if (strcmp(resp->name, "vdd2_opp") == 0) { resp->curr_level = curr_vdd2_prcm_set->opp_id; vdd2_clk = clk_get(NULL, "virt_vdd2_prcm_set"); + vdd2_resp = resp; } return; } -int set_opp(struct shared_resource *resp, u32 target_level) +int set_opp_level(int res, u32 target_level) { - unsigned long mpu_freq, mpu_old_freq, l3_freq, tput, t_opp; - int ind; - struct bus_throughput_db *tput_db; + unsigned long mpu_freq, mpu_old_freq, l3_freq, t_opp; struct cpufreq_freqs freqs_notify; + struct shared_resource *resp; + + if (res == VDD1_OPP) + resp = vdd1_resp; + else if (res == VDD2_OPP) + resp = vdd2_resp; + else + return 0; if (resp->curr_level == target_level) return 0; @@ -174,7 +184,7 @@ int set_opp(struct shared_resource *resp, u32 target_level) if (!mpu_opps || !dsp_opps || !l3_opps) return 0; - if (strcmp(resp->name, "vdd1_opp") == 0) { + if (res == VDD1_OPP) { mpu_old_freq = get_freq(mpu_opps + MAX_VDD1_OPP, curr_vdd1_prcm_set->opp_id); mpu_freq = get_freq(mpu_opps + MAX_VDD1_OPP, @@ -208,23 +218,7 @@ int set_opp(struct shared_resource *resp, u32 target_level) /* Send a post notification to CPUFreq */ cpufreq_notify_transition(&freqs_notify, CPUFREQ_POSTCHANGE); #endif - } else if (strcmp(resp->name, "vdd2_opp") == 0) { - tput_db = resp->resource_data; - tput = target_level; - /* using the throughput db map to the appropriate L3 Freq */ - for (ind = 1; ind < MAX_VDD2_OPP; ind++) - if (tput_db->throughput[ind] > tput) - target_level = ind; - - /* Set the highest OPP possible */ - if (ind == MAX_VDD2_OPP) - target_level = ind-1; - - if (resp->curr_level == target_level) - return 0; - - resp->curr_level = target_level; - + } else { l3_freq = get_freq(l3_opps + MAX_VDD2_OPP, target_level); t_opp = ID_VDD(PRCM_VDD2) | @@ -249,6 +243,30 @@ int set_opp(struct shared_resource *resp, u32 target_level) return 0; } +int set_opp(struct shared_resource *resp, u32 target_level) +{ + unsigned long tput; + int ind; + struct bus_throughput_db *tput_db; + + if (resp == vdd1_resp) { + set_opp_level(VDD1_OPP, target_level); + } else if (resp == vdd2_resp) { + tput_db = resp->resource_data; + tput = target_level; + /* using the throughput db map to the appropriate L3 Freq */ + for (ind = 1; ind < MAX_VDD2_OPP; ind++) + if (tput_db->throughput[ind] > tput) + target_level = ind; + + /* Set the highest OPP possible */ + if (ind == MAX_VDD2_OPP) + target_level = ind-1; + set_opp_level(VDD2_OPP, target_level); + } + return 0; +} + /** * validate_opp - Validates if valid VDD1 OPP's are passed as the * target_level. diff --git a/arch/arm/plat-omap/include/mach/omap34xx.h b/arch/arm/plat-omap/include/mach/omap34xx.h index b64375e..6db9a6c 100644 --- a/arch/arm/plat-omap/include/mach/omap34xx.h +++ b/arch/arm/plat-omap/include/mach/omap34xx.h @@ -72,6 +72,10 @@ #define OMAP34XX_DSP_IPI_BASE (OMAP34XX_DSP_BASE + 0x1000000) #define OMAP34XX_DSP_MMU_BASE (OMAP34XX_DSP_BASE + 0x2000000) +/* VDD OPP identifiers */ +#define VDD1_OPP 0x1 +#define VDD2_OPP 0x2 + /* VDD1 OPPS */ #define VDD1_OPP1 0x1 #define VDD1_OPP2 0x2 -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 18/23] OMAP3: PM: Added DVFS OPP locking interface for VDD1 and VDD2 2008-12-11 15:52 ` [PATCH 17/23] OMAP3: PM: Fixed VDD2 control to work from both sysfs and SRF API Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 19/23] OMAP3: Fix rate calculation bug in omap3_select_table_rate Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap Added two new files under /sys/power for controlling OPP locks. vdd1_lock and vdd2_lock. You can write to these to select desired OPP level and it will be locked at that level. Lock can be cleared by writing 0 to the same file. Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/pm.c | 55 ++++++++++++++++++++++++++++++++++- arch/arm/mach-omap2/pm.h | 8 ++++- arch/arm/mach-omap2/resource34xx.c | 24 +++++++++++++-- 3 files changed, 81 insertions(+), 6 deletions(-) diff --git a/arch/arm/mach-omap2/pm.c b/arch/arm/mach-omap2/pm.c index 2dd1341..155bd8d 100644 --- a/arch/arm/mach-omap2/pm.c +++ b/arch/arm/mach-omap2/pm.c @@ -43,6 +43,8 @@ unsigned short clocks_off_while_idle; unsigned short enable_off_mode; unsigned short voltage_off_while_idle; atomic_t sleep_block = ATOMIC_INIT(0); +static int vdd1_locked; +static int vdd2_locked; static ssize_t idle_show(struct kobject *, struct kobj_attribute *, char *); static ssize_t idle_store(struct kobject *k, struct kobj_attribute *, @@ -69,6 +71,11 @@ static struct kobj_attribute vdd1_opp_attr = static struct kobj_attribute vdd2_opp_attr = __ATTR(vdd2_opp, 0644, vdd_opp_show, vdd_opp_store); +static struct kobj_attribute vdd1_lock_attr = + __ATTR(vdd1_lock, 0644, vdd_opp_show, vdd_opp_store); +static struct kobj_attribute vdd2_lock_attr = + __ATTR(vdd2_lock, 0644, vdd_opp_show, vdd_opp_store); + #endif static ssize_t idle_show(struct kobject *kobj, struct kobj_attribute *attr, @@ -127,6 +134,10 @@ static ssize_t vdd_opp_show(struct kobject *kobj, struct kobj_attribute *attr, return sprintf(buf, "%hu\n", resource_get_level("vdd1_opp")); else if (attr == &vdd2_opp_attr) return sprintf(buf, "%hu\n", resource_get_level("vdd2_opp")); + else if (attr == &vdd1_lock_attr) + return sprintf(buf, "%hu\n", resource_get_opp_lock(VDD1_OPP)); + else if (attr == &vdd2_lock_attr) + return sprintf(buf, "%hu\n", resource_get_opp_lock(VDD2_OPP)); else return -EINVAL; } @@ -135,22 +146,50 @@ static ssize_t vdd_opp_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t n) { unsigned short value; + int flags = 0; if (sscanf(buf, "%hu", &value) != 1) return -EINVAL; + /* Check locks */ + if (attr == &vdd1_lock_attr) { + flags = OPP_IGNORE_LOCK; + attr = &vdd1_opp_attr; + if (vdd1_locked && value == 0) { + resource_unlock_opp(VDD1_OPP); + vdd1_locked = 0; + return n; + } + if (vdd1_locked == 0 && value != 0) { + resource_lock_opp(VDD1_OPP); + vdd1_locked = 1; + } + } else if (attr == &vdd2_lock_attr) { + flags = OPP_IGNORE_LOCK; + attr = &vdd2_opp_attr; + if (vdd2_locked && value == 0) { + resource_unlock_opp(VDD2_OPP); + vdd2_locked = 0; + return n; + } + if (vdd2_locked == 0 && value != 0) { + resource_lock_opp(VDD2_OPP); + vdd2_locked = 1; + } + } + if (attr == &vdd1_opp_attr) { if (value < 1 || value > 5) { printk(KERN_ERR "vdd_opp_store: Invalid value\n"); return -EINVAL; } - set_opp_level(VDD1_OPP, value); + resource_set_opp_level(VDD1_OPP, value, flags); } else if (attr == &vdd2_opp_attr) { if (value < 2 || value > 3) { printk(KERN_ERR "vdd_opp_store: Invalid value\n"); return -EINVAL; } - set_opp_level(VDD2_OPP, value); + resource_set_opp_level(VDD2_OPP, value, flags); } else { return -EINVAL; } @@ -227,6 +266,18 @@ static int __init omap_pm_init(void) printk(KERN_ERR "sysfs_create_file failed: %d\n", error); return error; } + + error = sysfs_create_file(power_kobj, &vdd1_lock_attr.attr); + if (error) { + printk(KERN_ERR "sysfs_create_file failed: %d\n", error); + return error; + } + + error = sysfs_create_file(power_kobj, &vdd2_lock_attr.attr); + if (error) { + printk(KERN_ERR "sysfs_create_file failed: %d\n", error); + return error; + } #endif voltage_off_while_idle = 0; /* Going to 0V on anything under ES2.1 will eventually cause a crash */ diff --git a/arch/arm/mach-omap2/pm.h b/arch/arm/mach-omap2/pm.h index 4c0052f..4f98c86 100644 --- a/arch/arm/mach-omap2/pm.h +++ b/arch/arm/mach-omap2/pm.h @@ -42,7 +42,13 @@ extern int omap3_pm_set_suspend_state(struct powerdomain *pwrdm, int state); #define omap3_pm_set_suspend_state(pwrdm,state) do {} while (0); #endif extern int set_pwrdm_state(struct powerdomain *pwrdm, u32 state); -extern int set_opp_level(int res, u32 target_level); +extern int resource_set_opp_level(int res, u32 target_level, int flags); +extern int resource_access_opp_lock(int res, int delta); +#define resource_lock_opp(res) resource_access_opp_lock(res, 1) +#define resource_unlock_opp(res) resource_access_opp_lock(res, -1) +#define resource_get_opp_lock(res) resource_access_opp_lock(res, 0) + +#define OPP_IGNORE_LOCK 0x1 #ifdef CONFIG_PM_DEBUG extern void omap2_pm_dump(int mode, int resume, unsigned int us); diff --git a/arch/arm/mach-omap2/resource34xx.c b/arch/arm/mach-omap2/resource34xx.c index d2c4e1f..e52260d 100644 --- a/arch/arm/mach-omap2/resource34xx.c +++ b/arch/arm/mach-omap2/resource34xx.c @@ -139,6 +139,8 @@ static struct shared_resource *vdd1_resp; static struct shared_resource *vdd2_resp; static struct device dummy_mpu_dev; static struct device dummy_dsp_dev; +static int vdd1_lock; +static int vdd2_lock; /** * init_opp - Initialize the OPP resource @@ -165,7 +167,19 @@ void init_opp(struct shared_resource *resp) return; } -int set_opp_level(int res, u32 target_level) +int resource_access_opp_lock(int res, int delta) +{ + if (res == VDD1_OPP) { + vdd1_lock += delta; + return vdd1_lock; + } else if (res == VDD2_OPP) { + vdd2_lock += delta; + return vdd2_lock; + } + return -EINVAL; +} + +int resource_set_opp_level(int res, u32 target_level, int flags) { unsigned long mpu_freq, mpu_old_freq, l3_freq, t_opp; struct cpufreq_freqs freqs_notify; @@ -185,6 +199,8 @@ int set_opp_level(int res, u32 target_level) return 0; if (res == VDD1_OPP) { + if (flags != OPP_IGNORE_LOCK && vdd1_lock) + return 0; mpu_old_freq = get_freq(mpu_opps + MAX_VDD1_OPP, curr_vdd1_prcm_set->opp_id); mpu_freq = get_freq(mpu_opps + MAX_VDD1_OPP, @@ -219,6 +235,8 @@ int set_opp_level(int res, u32 target_level) cpufreq_notify_transition(&freqs_notify, CPUFREQ_POSTCHANGE); #endif } else { + if (flags != OPP_IGNORE_LOCK && vdd2_lock) + return 0; l3_freq = get_freq(l3_opps + MAX_VDD2_OPP, target_level); t_opp = ID_VDD(PRCM_VDD2) | @@ -250,7 +268,7 @@ int set_opp(struct shared_resource *resp, u32 target_level) struct bus_throughput_db *tput_db; if (resp == vdd1_resp) { - set_opp_level(VDD1_OPP, target_level); + resource_set_opp_level(VDD1_OPP, target_level, 0); } else if (resp == vdd2_resp) { tput_db = resp->resource_data; tput = target_level; @@ -262,7 +280,7 @@ int set_opp(struct shared_resource *resp, u32 target_level) /* Set the highest OPP possible */ if (ind == MAX_VDD2_OPP) target_level = ind-1; - set_opp_level(VDD2_OPP, target_level); + resource_set_opp_level(VDD2_OPP, target_level, 0); } return 0; } -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 19/23] OMAP3: Fix rate calculation bug in omap3_select_table_rate 2008-12-11 15:52 ` [PATCH 18/23] OMAP3: PM: Added DVFS OPP locking interface for VDD1 and VDD2 Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap L3 divisor is not taken into account when trying to set DPLL3 clock. Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/clock34xx.c | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index 47a8465..cfd7f0f 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -939,6 +939,7 @@ static int omap3_select_table_rate(struct clk *clk, unsigned long rate) struct omap_opp *prcm_vdd; unsigned long found_speed = 0, curr_mpu_speed; int index; + int l3_div; if ((clk != &virt_vdd1_prcm_set) && (clk != &virt_vdd2_prcm_set)) return -EINVAL; @@ -984,7 +985,9 @@ static int omap3_select_table_rate(struct clk *clk, unsigned long rate) curr_mpu_speed/1000, found_speed/1000); #endif } else { - clk_set_rate(dpll3_clk, prcm_vdd->rate); + l3_div = cm_read_mod_reg(CORE_MOD, CM_CLKSEL) & + OMAP3430_CLKSEL_L3_MASK; + clk_set_rate(dpll3_clk, prcm_vdd->rate * l3_div); curr_vdd2_prcm_set = prcm_vdd; omap2_clksel_recalc(&core_ck); propagate_rate(&core_ck); -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 2008-12-11 15:52 ` [PATCH 19/23] OMAP3: Fix rate calculation bug in omap3_select_table_rate Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 21/23] OMAP3: PM: Prevented DVFS state switches when enabling off-mode Tero Kristo 2008-12-11 16:06 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Paul Walmsley 0 siblings, 2 replies; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap Previously only 1 and 2 was supported. This is needed for DVFS VDD2 control. Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/clock34xx.c | 9 +++------ arch/arm/mach-omap2/sram34xx.S | 7 ++++--- arch/arm/plat-omap/include/mach/sram.h | 7 +++++-- arch/arm/plat-omap/sram.c | 6 +++--- 4 files changed, 15 insertions(+), 14 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index cfd7f0f..55b1425 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -516,9 +516,9 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) sdrcrate = sdrc_ick.rate; if (rate > clk->rate) - sdrcrate <<= ((rate / clk->rate) - 1); + sdrcrate <<= ((rate / clk->rate) >> 1); else - sdrcrate >>= ((clk->rate / rate) - 1); + sdrcrate >>= ((clk->rate / rate) >> 1); sp = omap2_sdrc_get_params(sdrcrate); if (!sp) @@ -545,13 +545,10 @@ static int omap3_core_dpll_m2_set_rate(struct clk *clk, unsigned long rate) pr_debug("clock: SDRC timing params used: %08x %08x %08x\n", sp->rfr_ctrl, sp->actim_ctrla, sp->actim_ctrlb); - /* REVISIT: SRAM code doesn't support other M2 divisors yet */ - WARN_ON(new_div != 1 && new_div != 2); - local_irq_disable(); omap3_configure_core_dpll(sp->rfr_ctrl, sp->actim_ctrla, sp->actim_ctrlb, new_div, unlock_dll, c, - sp->mr); + sp->mr, rate > clk->rate); local_irq_enable(); omap2_clksel_recalc(clk); diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S index 16eb4ef..832cd76 100644 --- a/arch/arm/mach-omap2/sram34xx.S +++ b/arch/arm/mach-omap2/sram34xx.S @@ -70,6 +70,7 @@ * r5 = number of MPU cycles to wait for SDRC to stabilize after * reprogramming the SDRC when switching to a slower MPU speed * r6 = new SDRC_MR_0 register value + * r7 = increasing SDRC rate? (1 = yes, 0 = no) * */ ENTRY(omap3_sram_configure_core_dpll) @@ -78,8 +79,8 @@ ENTRY(omap3_sram_configure_core_dpll) ldr r5, [sp, #56] @ load extra args from the stack ldr r6, [sp, #60] @ load extra args from the stack dsb @ flush buffered writes to interconnect - cmp r3, #0x2 @ if increasing SDRC clk rate, - blne configure_sdrc @ program the SDRC regs early (for RFR) + cmp r7, #1 @ if increasing SDRC clk rate, + bleq configure_sdrc @ program the SDRC regs early (for RFR) cmp r4, #SDRC_UNLOCK_DLL @ set the intended DLL state bleq unlock_dll blne lock_dll @@ -89,7 +90,7 @@ ENTRY(omap3_sram_configure_core_dpll) cmp r4, #SDRC_UNLOCK_DLL @ wait for DLL status to change bleq wait_dll_unlock blne wait_dll_lock - cmp r3, #0x1 @ if increasing SDRC clk rate, + cmp r7, #1 @ if increasing SDRC clk rate, beq return_to_sdram @ return to SDRAM code, otherwise, bl configure_sdrc @ reprogram SDRC regs now mov r12, r5 diff --git a/arch/arm/plat-omap/include/mach/sram.h b/arch/arm/plat-omap/include/mach/sram.h index d07da3b..ad0a600 100644 --- a/arch/arm/plat-omap/include/mach/sram.h +++ b/arch/arm/plat-omap/include/mach/sram.h @@ -24,7 +24,8 @@ extern u32 omap2_set_prcm(u32 dpll_ctrl_val, u32 sdrc_rfr_val, int bypass); extern u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, - u32 unlock_dll, u32 f, u32 sdrc_mr); + u32 unlock_dll, u32 f, u32 sdrc_mr, + u32 inc); extern void omap3_sram_restore_context(void); /* Do not use these */ @@ -62,7 +63,9 @@ extern unsigned long omap243x_sram_reprogram_sdrc_sz; extern u32 omap3_sram_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, - u32 sdrc_actim_ctrlb, u32 m2); + u32 sdrc_actim_ctrlb, u32 m2, + u32 unlock_dll, u32 f, u32 sdrc_mr, + u32 inc); extern unsigned long omap3_sram_configure_core_dpll_sz; #ifdef CONFIG_PM diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c index 68a1f69..9f6232f 100644 --- a/arch/arm/plat-omap/sram.c +++ b/arch/arm/plat-omap/sram.c @@ -372,10 +372,10 @@ static u32 (*_omap3_sram_configure_core_dpll)(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll, - u32 f, u32 sdrc_mr); + u32 f, u32 sdrc_mr, u32 inc); u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll, - u32 f, u32 sdrc_mr) + u32 f, u32 sdrc_mr, u32 inc) { if (!_omap3_sram_configure_core_dpll) omap_sram_error(); @@ -383,7 +383,7 @@ u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, return _omap3_sram_configure_core_dpll(sdrc_rfr_ctrl, sdrc_actim_ctrla, sdrc_actim_ctrlb, m2, - unlock_dll, f, sdrc_mr); + unlock_dll, f, sdrc_mr, inc); } #ifdef CONFIG_PM -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 21/23] OMAP3: PM: Prevented DVFS state switches when enabling off-mode 2008-12-11 15:52 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 22/23] OMAP3: PM: Enable VDD2 OPP1 Tero Kristo 2008-12-11 16:06 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Paul Walmsley 1 sibling, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap resource_refresh() call tried to refresh DVFS OPP levels also. This should eventually be fixed in a way that DVFS VDD2 level is actually governed by something. Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/pm34xx.c | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/arch/arm/mach-omap2/pm34xx.c b/arch/arm/mach-omap2/pm34xx.c index e0c952e..7367f68 100644 --- a/arch/arm/mach-omap2/pm34xx.c +++ b/arch/arm/mach-omap2/pm34xx.c @@ -783,8 +783,12 @@ void omap3_pm_off_mode_enable(int enable) state = PWRDM_POWER_RET; #ifdef CONFIG_OMAP_PM_SRF + resource_lock_opp(VDD1_OPP); + resource_lock_opp(VDD2_OPP); if (resource_refresh()) printk(KERN_ERR "Error: could not refresh resources\n"); + resource_unlock_opp(VDD1_OPP); + resource_unlock_opp(VDD2_OPP); #endif list_for_each_entry(pwrst, &pwrst_list, node) { pwrst->next_state = state; -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 22/23] OMAP3: PM: Enable VDD2 OPP1 2008-12-11 15:52 ` [PATCH 21/23] OMAP3: PM: Prevented DVFS state switches when enabling off-mode Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 2008-12-11 15:52 ` [PATCH 23/23] OMAP3: PM: Fix linker error without CONFIG_PM option Tero Kristo 0 siblings, 1 reply; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap VDD2 OPP1 works at least with minimal kernel setup, there are some corner cases where it fails though, like when you use USB. Board setup should make sure and enable only working OPPs. Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/pm.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/arch/arm/mach-omap2/pm.c b/arch/arm/mach-omap2/pm.c index 155bd8d..1c6c93a 100644 --- a/arch/arm/mach-omap2/pm.c +++ b/arch/arm/mach-omap2/pm.c @@ -185,7 +185,7 @@ static ssize_t vdd_opp_store(struct kobject *kobj, struct kobj_attribute *attr, } resource_set_opp_level(VDD1_OPP, value, flags); } else if (attr == &vdd2_opp_attr) { - if (value < 2 || value > 3) { + if (value < 1 || value > 3) { printk(KERN_ERR "vdd_opp_store: Invalid value\n"); return -EINVAL; } -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* [PATCH 23/23] OMAP3: PM: Fix linker error without CONFIG_PM option 2008-12-11 15:52 ` [PATCH 22/23] OMAP3: PM: Enable VDD2 OPP1 Tero Kristo @ 2008-12-11 15:52 ` Tero Kristo 0 siblings, 0 replies; 32+ messages in thread From: Tero Kristo @ 2008-12-11 15:52 UTC (permalink / raw) To: linux-omap omap3_save_scratchpad_contents() is not available without CONFIG_PM. Signed-off-by: Tero Kristo <tero.kristo@nokia.com> --- arch/arm/mach-omap2/clock34xx.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/arch/arm/mach-omap2/clock34xx.c b/arch/arm/mach-omap2/clock34xx.c index 55b1425..b38cb30 100644 --- a/arch/arm/mach-omap2/clock34xx.c +++ b/arch/arm/mach-omap2/clock34xx.c @@ -990,7 +990,9 @@ static int omap3_select_table_rate(struct clk *clk, unsigned long rate) propagate_rate(&core_ck); } +#ifdef CONFIG_PM omap3_save_scratchpad_contents(); +#endif return 0; } -- 1.5.4.3 ^ permalink raw reply related [flat|nested] 32+ messages in thread
* Re: [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 2008-12-11 15:52 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Tero Kristo 2008-12-11 15:52 ` [PATCH 21/23] OMAP3: PM: Prevented DVFS state switches when enabling off-mode Tero Kristo @ 2008-12-11 16:06 ` Paul Walmsley 2008-12-11 16:38 ` Tero.Kristo 1 sibling, 1 reply; 32+ messages in thread From: Paul Walmsley @ 2008-12-11 16:06 UTC (permalink / raw) To: Tero Kristo; +Cc: linux-omap Hi Tero one comment on this patch ... On Thu, 11 Dec 2008, Tero Kristo wrote: > Previously only 1 and 2 was supported. This is needed for DVFS VDD2 control. > diff --git a/arch/arm/mach-omap2/sram34xx.S b/arch/arm/mach-omap2/sram34xx.S > index 16eb4ef..832cd76 100644 > --- a/arch/arm/mach-omap2/sram34xx.S > +++ b/arch/arm/mach-omap2/sram34xx.S > @@ -70,6 +70,7 @@ > * r5 = number of MPU cycles to wait for SDRC to stabilize after > * reprogramming the SDRC when switching to a slower MPU speed > * r6 = new SDRC_MR_0 register value > + * r7 = increasing SDRC rate? (1 = yes, 0 = no) > * > */ > ENTRY(omap3_sram_configure_core_dpll) > @@ -78,8 +79,8 @@ ENTRY(omap3_sram_configure_core_dpll) > ldr r5, [sp, #56] @ load extra args from the stack > ldr r6, [sp, #60] @ load extra args from the stack you will also need a ldr r7, [sp, #64] @ load extra args from the stack here > dsb @ flush buffered writes to interconnect > - cmp r3, #0x2 @ if increasing SDRC clk rate, > - blne configure_sdrc @ program the SDRC regs early (for RFR) > + cmp r7, #1 @ if increasing SDRC clk rate, > + bleq configure_sdrc @ program the SDRC regs early (for RFR) > cmp r4, #SDRC_UNLOCK_DLL @ set the intended DLL state > bleq unlock_dll > blne lock_dll > @@ -89,7 +90,7 @@ ENTRY(omap3_sram_configure_core_dpll) > cmp r4, #SDRC_UNLOCK_DLL @ wait for DLL status to change > bleq wait_dll_unlock > blne wait_dll_lock > - cmp r3, #0x1 @ if increasing SDRC clk rate, > + cmp r7, #1 @ if increasing SDRC clk rate, > beq return_to_sdram @ return to SDRAM code, otherwise, > bl configure_sdrc @ reprogram SDRC regs now > mov r12, r5 > diff --git a/arch/arm/plat-omap/include/mach/sram.h b/arch/arm/plat-omap/include/mach/sram.h > index d07da3b..ad0a600 100644 > --- a/arch/arm/plat-omap/include/mach/sram.h > +++ b/arch/arm/plat-omap/include/mach/sram.h > @@ -24,7 +24,8 @@ extern u32 omap2_set_prcm(u32 dpll_ctrl_val, u32 sdrc_rfr_val, int bypass); > extern u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, > u32 sdrc_actim_ctrla, > u32 sdrc_actim_ctrlb, u32 m2, > - u32 unlock_dll, u32 f, u32 sdrc_mr); > + u32 unlock_dll, u32 f, u32 sdrc_mr, > + u32 inc); > extern void omap3_sram_restore_context(void); > > /* Do not use these */ > @@ -62,7 +63,9 @@ extern unsigned long omap243x_sram_reprogram_sdrc_sz; > > extern u32 omap3_sram_configure_core_dpll(u32 sdrc_rfr_ctrl, > u32 sdrc_actim_ctrla, > - u32 sdrc_actim_ctrlb, u32 m2); > + u32 sdrc_actim_ctrlb, u32 m2, > + u32 unlock_dll, u32 f, u32 sdrc_mr, > + u32 inc); > extern unsigned long omap3_sram_configure_core_dpll_sz; > > #ifdef CONFIG_PM > diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c > index 68a1f69..9f6232f 100644 > --- a/arch/arm/plat-omap/sram.c > +++ b/arch/arm/plat-omap/sram.c > @@ -372,10 +372,10 @@ static u32 (*_omap3_sram_configure_core_dpll)(u32 sdrc_rfr_ctrl, > u32 sdrc_actim_ctrla, > u32 sdrc_actim_ctrlb, > u32 m2, u32 unlock_dll, > - u32 f, u32 sdrc_mr); > + u32 f, u32 sdrc_mr, u32 inc); > u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, > u32 sdrc_actim_ctrlb, u32 m2, u32 unlock_dll, > - u32 f, u32 sdrc_mr) > + u32 f, u32 sdrc_mr, u32 inc) > { > if (!_omap3_sram_configure_core_dpll) > omap_sram_error(); > @@ -383,7 +383,7 @@ u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, > return _omap3_sram_configure_core_dpll(sdrc_rfr_ctrl, > sdrc_actim_ctrla, > sdrc_actim_ctrlb, m2, > - unlock_dll, f, sdrc_mr); > + unlock_dll, f, sdrc_mr, inc); > } > > #ifdef CONFIG_PM > -- > 1.5.4.3 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-omap" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > - Paul ^ permalink raw reply [flat|nested] 32+ messages in thread
* RE: [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 2008-12-11 16:06 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Paul Walmsley @ 2008-12-11 16:38 ` Tero.Kristo 0 siblings, 0 replies; 32+ messages in thread From: Tero.Kristo @ 2008-12-11 16:38 UTC (permalink / raw) To: paul; +Cc: linux-omap Hi, >Hi Tero > >one comment on this patch ... > >On Thu, 11 Dec 2008, Tero Kristo wrote: > >> Previously only 1 and 2 was supported. This is needed for >DVFS VDD2 control. > >> diff --git a/arch/arm/mach-omap2/sram34xx.S >> b/arch/arm/mach-omap2/sram34xx.S index 16eb4ef..832cd76 100644 >> --- a/arch/arm/mach-omap2/sram34xx.S >> +++ b/arch/arm/mach-omap2/sram34xx.S >> @@ -70,6 +70,7 @@ >> * r5 = number of MPU cycles to wait for SDRC to stabilize after >> * reprogramming the SDRC when switching to a slower MPU speed >> * r6 = new SDRC_MR_0 register value >> + * r7 = increasing SDRC rate? (1 = yes, 0 = no) >> * >> */ >> ENTRY(omap3_sram_configure_core_dpll) >> @@ -78,8 +79,8 @@ ENTRY(omap3_sram_configure_core_dpll) >> ldr r5, [sp, #56] @ load extra args from the stack >> ldr r6, [sp, #60] @ load extra args from the stack > >you will also need a > > ldr r7, [sp, #64] @ load extra args from the stack > >here Woops. Resending in a bit. Funny though that this patch does not seem to affect system stability too much at initial testing, it did not cause any crash when I tried this out. Might cause problems in longer run / stress testing. >> dsb @ flush buffered writes >to interconnect >> - cmp r3, #0x2 @ if increasing SDRC clk rate, >> - blne configure_sdrc @ program the SDRC regs >early (for RFR) >> + cmp r7, #1 @ if increasing SDRC clk rate, >> + bleq configure_sdrc @ program the SDRC regs >early (for RFR) >> cmp r4, #SDRC_UNLOCK_DLL @ set the intended DLL state >> bleq unlock_dll >> blne lock_dll >> @@ -89,7 +90,7 @@ ENTRY(omap3_sram_configure_core_dpll) >> cmp r4, #SDRC_UNLOCK_DLL @ wait for DLL status to change >> bleq wait_dll_unlock >> blne wait_dll_lock >> - cmp r3, #0x1 @ if increasing SDRC clk rate, >> + cmp r7, #1 @ if increasing SDRC clk rate, >> beq return_to_sdram @ return to SDRAM code, >otherwise, >> bl configure_sdrc @ reprogram SDRC regs now >> mov r12, r5 >> diff --git a/arch/arm/plat-omap/include/mach/sram.h >> b/arch/arm/plat-omap/include/mach/sram.h >> index d07da3b..ad0a600 100644 >> --- a/arch/arm/plat-omap/include/mach/sram.h >> +++ b/arch/arm/plat-omap/include/mach/sram.h >> @@ -24,7 +24,8 @@ extern u32 omap2_set_prcm(u32 dpll_ctrl_val, u32 >> sdrc_rfr_val, int bypass); extern u32 >omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, >> u32 sdrc_actim_ctrla, >> u32 sdrc_actim_ctrlb, u32 m2, >> - u32 unlock_dll, u32 f, u32 >sdrc_mr); >> + u32 unlock_dll, u32 f, u32 sdrc_mr, >> + u32 inc); >> extern void omap3_sram_restore_context(void); >> >> /* Do not use these */ >> @@ -62,7 +63,9 @@ extern unsigned long >> omap243x_sram_reprogram_sdrc_sz; >> >> extern u32 omap3_sram_configure_core_dpll(u32 sdrc_rfr_ctrl, >> u32 sdrc_actim_ctrla, >> - u32 sdrc_actim_ctrlb, u32 m2); >> + u32 sdrc_actim_ctrlb, u32 m2, >> + u32 unlock_dll, u32 >f, u32 sdrc_mr, >> + u32 inc); >> extern unsigned long omap3_sram_configure_core_dpll_sz; >> >> #ifdef CONFIG_PM >> diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c >> index 68a1f69..9f6232f 100644 >> --- a/arch/arm/plat-omap/sram.c >> +++ b/arch/arm/plat-omap/sram.c >> @@ -372,10 +372,10 @@ static u32 >(*_omap3_sram_configure_core_dpll)(u32 sdrc_rfr_ctrl, >> u32 sdrc_actim_ctrla, >> u32 sdrc_actim_ctrlb, >> u32 m2, u32 unlock_dll, >> - u32 f, u32 sdrc_mr); >> + u32 f, u32 >sdrc_mr, u32 inc); >> u32 omap3_configure_core_dpll(u32 sdrc_rfr_ctrl, u32 >sdrc_actim_ctrla, >> u32 sdrc_actim_ctrlb, u32 m2, u32 >unlock_dll, >> - u32 f, u32 sdrc_mr) >> + u32 f, u32 sdrc_mr, u32 inc) >> { >> if (!_omap3_sram_configure_core_dpll) >> omap_sram_error(); >> @@ -383,7 +383,7 @@ u32 omap3_configure_core_dpll(u32 >sdrc_rfr_ctrl, u32 sdrc_actim_ctrla, >> return _omap3_sram_configure_core_dpll(sdrc_rfr_ctrl, >> sdrc_actim_ctrla, >> sdrc_actim_ctrlb, m2, >> - unlock_dll, f, sdrc_mr); >> + unlock_dll, f, >sdrc_mr, inc); >> } >> >> #ifdef CONFIG_PM >> -- >> 1.5.4.3 >> >> -- >> To unsubscribe from this list: send the line "unsubscribe >linux-omap" >> in the body of a message to majordomo@vger.kernel.org More majordomo >> info at http://vger.kernel.org/majordomo-info.html >> > > >- Paul > ^ permalink raw reply [flat|nested] 32+ messages in thread
* Re: [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory 2008-12-11 15:52 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tero Kristo 2008-12-11 15:52 ` [PATCH 03/23] OMAP3 SRAM: add ARM barriers to omap3_sram_configure_core_dpll Tero Kristo @ 2008-12-15 19:07 ` Tony Lindgren 1 sibling, 0 replies; 32+ messages in thread From: Tony Lindgren @ 2008-12-15 19:07 UTC (permalink / raw) To: Tero Kristo; +Cc: linux-omap, Paul Walmsley * Tero Kristo <tero.kristo@nokia.com> [081211 07:58]: > From: Paul Walmsley <paul@pwsan.com> > > Mark the SRAM (aka OCM RAM) as Non-cacheable Normal memory[1]. This > is to prevent the ARM from evicting existing cache lines to SDRAM > while code is executing from the SRAM. Necessary since one of the > primary uses for the SRAM is to hold the code and data for the CORE > DPLL M2 divider reprogramming code, which must execute while the SDRC > is idled. If the ARM attempts to write cache lines back to the while > the SRAM code is running, the ARM will stall[2]. > > TI deals with this problem in the CDP kernel by marking the SRAM as > Strongly-ordered memory. > > Tero Kristo <tero.kristo@nokia.com> caught a bug in an earlier version of > this patch - thanks Tero. As I talked with Paul, we might want to set up two sections in SRAM. One section that is cached and another that is not. Tony > > ... > > 1. ARMv7 ARM (DDI 0406A) pp. A3-30, A3-31, B3-32. > > 2. Private communication with Richard Woodruff <r-woodruff2@ti.com> > > Signed-off-by: Paul Walmsley <paul@pwsan.com> > Cc: Tero Kristo <tero.kristo@nokia.com> > Cc: Richard Woodruff <r-woodruff2@ti.com> > --- > arch/arm/plat-omap/sram.c | 9 +++++++++ > 1 files changed, 9 insertions(+), 0 deletions(-) > mode change 100755 => 100644 arch/arm/plat-omap/sram.c > > diff --git a/arch/arm/plat-omap/sram.c b/arch/arm/plat-omap/sram.c > old mode 100755 > new mode 100644 > index abcc05b..04214e1 > --- a/arch/arm/plat-omap/sram.c > +++ b/arch/arm/plat-omap/sram.c > @@ -207,6 +207,15 @@ void __init omap_map_sram(void) > base = OMAP3_SRAM_PA; > base = ROUND_DOWN(base, PAGE_SIZE); > omap_sram_io_desc[0].pfn = __phys_to_pfn(base); > + > + /* > + * SRAM must be marked as non-cached on OMAP3 since the > + * CORE DPLL M2 divider change code (in SRAM) runs with the > + * SDRAM controller disabled, and if it is marked cached, > + * the ARM may attempt to write cache lines back to SDRAM > + * which will cause the system to hang. > + */ > + omap_sram_io_desc[0].type = MT_MEMORY_NONCACHED; > } > > omap_sram_io_desc[0].length = 1024 * 1024; /* Use section desc */ > -- > 1.5.4.3 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-omap" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html ^ permalink raw reply [flat|nested] 32+ messages in thread
* Re: [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type 2008-12-11 15:52 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tero Kristo 2008-12-11 15:52 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tero Kristo @ 2008-12-15 19:06 ` Tony Lindgren 2008-12-15 19:58 ` Paul Walmsley 1 sibling, 1 reply; 32+ messages in thread From: Tony Lindgren @ 2008-12-15 19:06 UTC (permalink / raw) To: Tero Kristo; +Cc: linux-omap, Paul Walmsley * Tero Kristo <tero.kristo@nokia.com> [081211 07:59]: > From: Paul Walmsley <paul@pwsan.com> > > Add a Non-cacheable Normal ARM executable memory type, > MT_MEMORY_NONCACHED. This is needed for the OMAP3 SDRAM clock change > code, which must run from SRAM. The SRAM must be marked as > non-cacheable memory to avoid dirty cache line writebacks to SDRAM > while the SDRAM controller is paused. This should be discussed on LAKML. Tony > Signed-off-by: Paul Walmsley <paul@pwsan.com> > --- > arch/arm/include/asm/mach/map.h | 1 + > arch/arm/mm/mmu.c | 23 +++++++++++++++++++++++ > 2 files changed, 24 insertions(+), 0 deletions(-) > > diff --git a/arch/arm/include/asm/mach/map.h b/arch/arm/include/asm/mach/map.h > index 39d949b..58cf91f 100644 > --- a/arch/arm/include/asm/mach/map.h > +++ b/arch/arm/include/asm/mach/map.h > @@ -26,6 +26,7 @@ struct map_desc { > #define MT_HIGH_VECTORS 8 > #define MT_MEMORY 9 > #define MT_ROM 10 > +#define MT_MEMORY_NONCACHED 11 > > #ifdef CONFIG_MMU > extern void iotable_init(struct map_desc *, int); > diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c > index 7f36c82..9ad6413 100644 > --- a/arch/arm/mm/mmu.c > +++ b/arch/arm/mm/mmu.c > @@ -242,6 +242,10 @@ static struct mem_type mem_types[] = { > .prot_sect = PMD_TYPE_SECT, > .domain = DOMAIN_KERNEL, > }, > + [MT_MEMORY_NONCACHED] = { > + .prot_sect = PMD_TYPE_SECT | PMD_SECT_AP_WRITE, > + .domain = DOMAIN_KERNEL, > + }, > }; > > const struct mem_type *get_mem_type(unsigned int type) > @@ -405,9 +409,28 @@ static void __init build_mem_type_table(void) > kern_pgprot |= L_PTE_SHARED; > vecs_pgprot |= L_PTE_SHARED; > mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S; > + mem_types[MT_MEMORY_NONCACHED].prot_sect |= PMD_SECT_S; > #endif > } > > + /* > + * Non-cacheable Normal - intended for memory areas that must > + * not cause cache line evictions when used > + */ > + if (cpu_arch >= CPU_ARCH_ARMv6) { > + if (cpu_arch >= CPU_ARCH_ARMv7 && (cr & CR_TRE)) { > + /* Non-cacheable Normal is XCB = 001 */ > + mem_types[MT_MEMORY_NONCACHED].prot_sect |= > + PMD_SECT_BUFFERED; > + } else { > + /* For both ARMv6 and non-TEX-remapping ARMv7 */ > + mem_types[MT_MEMORY_NONCACHED].prot_sect |= > + PMD_SECT_TEX(1); > + } > + } else { > + mem_types[MT_MEMORY_NONCACHED].prot_sect |= PMD_SECT_BUFFERABLE; > + } > + > for (i = 0; i < 16; i++) { > unsigned long v = pgprot_val(protection_map[i]); > protection_map[i] = __pgprot(v | user_pgprot); > -- > 1.5.4.3 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-omap" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html ^ permalink raw reply [flat|nested] 32+ messages in thread
* Re: [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type 2008-12-15 19:06 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tony Lindgren @ 2008-12-15 19:58 ` Paul Walmsley 0 siblings, 0 replies; 32+ messages in thread From: Paul Walmsley @ 2008-12-15 19:58 UTC (permalink / raw) To: Tony Lindgren; +Cc: Tero Kristo, linux-omap On Mon, 15 Dec 2008, Tony Lindgren wrote: > * Tero Kristo <tero.kristo@nokia.com> [081211 07:59]: > > From: Paul Walmsley <paul@pwsan.com> > > > > Add a Non-cacheable Normal ARM executable memory type, > > MT_MEMORY_NONCACHED. This is needed for the OMAP3 SDRAM clock change > > code, which must run from SRAM. The SRAM must be marked as > > non-cacheable memory to avoid dirty cache line writebacks to SDRAM > > while the SDRAM controller is paused. > > This should be discussed on LAKML. Yeah, it's my understanding that these patches are targeted for the PM branch. - Paul ^ permalink raw reply [flat|nested] 32+ messages in thread
* Re: OMAP3: DVFS core patch set 2008-12-11 15:52 OMAP3: DVFS core patch set Tero Kristo 2008-12-11 15:52 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tero Kristo @ 2008-12-12 1:59 ` Paul Walmsley 2008-12-18 8:22 ` Paul Walmsley 2008-12-15 21:48 ` Kevin Hilman 2 siblings, 1 reply; 32+ messages in thread From: Paul Walmsley @ 2008-12-12 1:59 UTC (permalink / raw) To: linux-omap; +Cc: Tero Kristo, peter.de-schrijver Hello, For anyone interested, here are some worst case measurements for CORE DVFS transition times with patches 1-16. The test board was a 3430SDP ES2.0 GP. Tests were run both with and without changing the VDD2 voltage; and at two different MPU frequencies. No voltage change: MPU @ 381MHz: 13 ticks (0.40 ms) worst-case MPU @ 83MHz: 38 ticks (1.19 ms) worst-case With voltage change: MPU @ 381MHz: 34 ticks (1.04 ms) worst-case MPU @ 83MHz: 130 ticks (3.97 ms) worst-case Some important notes: * Mean CORE DVFS times will be quite a bit lower, based on eyeballing the individual test run durations. Variability was quite high; the above numbers would probably be considered outliers. * The TI "September 11" u-boot on this board programs the CORE DPLL for a slightly derated frequency. This causes the SDRC DLL to unlock on transitions to a lower frequency, and causes it to relock on transitions to a higher frequency. Since a fixed bootloader would use 83000000 Hz and 166000000 Hz, the DLL would lock at both frequencies, which may affect the above figures. * Environmental parameters weren't tracked since the board is remote. * Since the 32KiHz sync timer was used, there is a max roundoff variability of ((1/32768) * 2) seconds, or about 61 microseconds. * The test script and additional test patches are available upon request by E-mailing me. * I did not test at the highest MPU rate, for no real reason aside from 381MHz is what the bootloader set the MPU to * The tests were relatively short-duration; perhaps a few hundred transitions in each case. Commentary: * The variability caused by MPU rate was quite surprising. I thought that the bulk of the time in this process would be MPU rate-invariant. - Paul ^ permalink raw reply [flat|nested] 32+ messages in thread
* Re: OMAP3: DVFS core patch set 2008-12-12 1:59 ` OMAP3: DVFS core patch set Paul Walmsley @ 2008-12-18 8:22 ` Paul Walmsley 0 siblings, 0 replies; 32+ messages in thread From: Paul Walmsley @ 2008-12-18 8:22 UTC (permalink / raw) To: linux-omap; +Cc: Tero Kristo, peter.de-schrijver a quick followup on this: On Thu, 11 Dec 2008, Paul Walmsley wrote: > With voltage change: > MPU @ 83MHz: 130 ticks (3.97 ms) worst-case It recently occurred to me that part of this 130 tick time is due to the udelay()s in the SmartReflex code. My test code did not update loops_per_jiffy when it did the MPU rate, so the udelay()s there would loop for longer than they should. So that needs to be fixed and re-measured. Presumably further improvements would be gained by converting sr_voltagescale_vcbypass() to be interrupt-driven, rather than polling for the voltage change to complete. That should reduce the CORE DVFS time further. - Paul ^ permalink raw reply [flat|nested] 32+ messages in thread
* Re: OMAP3: DVFS core patch set 2008-12-11 15:52 OMAP3: DVFS core patch set Tero Kristo 2008-12-11 15:52 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tero Kristo 2008-12-12 1:59 ` OMAP3: DVFS core patch set Paul Walmsley @ 2008-12-15 21:48 ` Kevin Hilman 2 siblings, 0 replies; 32+ messages in thread From: Kevin Hilman @ 2008-12-15 21:48 UTC (permalink / raw) To: Tero Kristo; +Cc: linux-omap Tero Kristo <tero.kristo@nokia.com> writes: > This patch set provides DVFS support for OMAP3. First 16 patches contain > clock framework modifications from Paul Walmsley, which are needed for VDD2 > OPP clock control to work properly. Rest of the patches are misc fixes for > DVFS control, providing locking interface for OPPs and providing support > for DPLL3 M2 rates higher than 2. Tero, Paul, It seems most of these can be applied dirctly to l-o, correct? Can you separate this out into two series? One which goes straight to l-o for all the SRAM, SDRC and clock stuff. And a separate one which applies to the PM branch for the DVFS fixes, updates etc. Thanks, Kevin ^ permalink raw reply [flat|nested] 32+ messages in thread
end of thread, other threads:[~2008-12-18 8:22 UTC | newest] Thread overview: 32+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2008-12-11 15:52 OMAP3: DVFS core patch set Tero Kristo 2008-12-11 15:52 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tero Kristo 2008-12-11 15:52 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tero Kristo 2008-12-11 15:52 ` [PATCH 03/23] OMAP3 SRAM: add ARM barriers to omap3_sram_configure_core_dpll Tero Kristo 2008-12-11 15:52 ` [PATCH 04/23] OMAP3 clock: add interconnect barriers to CORE DPLL M2 change Tero Kristo 2008-12-11 15:52 ` [PATCH 05/23] OMAP3 SRAM: clear the SDRC PWRENA bit during SDRC frequency change Tero Kristo 2008-12-11 15:52 ` [PATCH 06/23] OMAP3 SDRC: Add 166MHz, 83MHz SDRC settings for the BeagleBoard Tero Kristo 2008-12-11 15:52 ` [PATCH 07/23] OMAP3 SDRC: initialize SDRC_POWER at boot Tero Kristo 2008-12-11 15:52 ` [PATCH 08/23] OMAP3 SRAM: renumber registers to make space for argument passing Tero Kristo 2008-12-11 15:52 ` [PATCH 09/23] OMAP3 clock: only unlock SDRC DLL if SDRC clk < 83MHz Tero Kristo 2008-12-11 15:52 ` [PATCH 10/23] OMAP3 clock: use pr_debug() rather than pr_info() in some clock change code Tero Kristo 2008-12-11 15:52 ` [PATCH 11/23] OMAP3 clock: remove wait for DPLL3 M2 clock to stabilize Tero Kristo 2008-12-11 15:52 ` [PATCH 12/23] OMAP3 clock: initialize SDRC timings at kernel start Tero Kristo 2008-12-11 15:52 ` [PATCH 13/23] OMAP3 clock: add a short delay when lowering CORE clk rate Tero Kristo 2008-12-11 15:52 ` [PATCH 14/23] OMAP3 clock/SDRC: program SDRC_MR register during SDRC clock change Tero Kristo 2008-12-11 15:52 ` [PATCH 15/23] OMAP3 SRAM: add more comments on the SRAM code Tero Kristo 2008-12-11 15:52 ` [PATCH 16/23] OMAP3 SRAM: convert SRAM code to use macros rather than magic numbers Tero Kristo 2008-12-11 15:52 ` [PATCH 17/23] OMAP3: PM: Fixed VDD2 control to work from both sysfs and SRF API Tero Kristo 2008-12-11 15:52 ` [PATCH 18/23] OMAP3: PM: Added DVFS OPP locking interface for VDD1 and VDD2 Tero Kristo 2008-12-11 15:52 ` [PATCH 19/23] OMAP3: Fix rate calculation bug in omap3_select_table_rate Tero Kristo 2008-12-11 15:52 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Tero Kristo 2008-12-11 15:52 ` [PATCH 21/23] OMAP3: PM: Prevented DVFS state switches when enabling off-mode Tero Kristo 2008-12-11 15:52 ` [PATCH 22/23] OMAP3: PM: Enable VDD2 OPP1 Tero Kristo 2008-12-11 15:52 ` [PATCH 23/23] OMAP3: PM: Fix linker error without CONFIG_PM option Tero Kristo 2008-12-11 16:06 ` [PATCH 20/23] OMAP3: Add support for DPLL3 divisor values higher than 2 Paul Walmsley 2008-12-11 16:38 ` Tero.Kristo 2008-12-15 19:07 ` [PATCH 02/23] OMAP3 SRAM: mark OCM RAM as Non-cacheable Normal memory Tony Lindgren 2008-12-15 19:06 ` [PATCH 01/23] ARM: MMU: add a Non-cacheable Normal executable memory type Tony Lindgren 2008-12-15 19:58 ` Paul Walmsley 2008-12-12 1:59 ` OMAP3: DVFS core patch set Paul Walmsley 2008-12-18 8:22 ` Paul Walmsley 2008-12-15 21:48 ` Kevin Hilman
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox