From: Jan Weitzel <J.Weitzel@phytec.de>
To: vj <vicencb@gmail.com>
Cc: barebox@lists.infradead.org
Subject: Re: [PATCH 6/6] OMAP4: clean voltage switch
Date: Mon, 22 Oct 2012 11:00:30 +0200 [thread overview]
Message-ID: <1350896430.3205.4.camel@lws-weitzel> (raw)
In-Reply-To: <CAAMcf8Dv6wE+pwNZqW+64t-eKrcLSMb9PE8jN8zgvyYq6dX2Jg@mail.gmail.com>
Am Sonntag, den 21.10.2012, 01:49 +0200 schrieb vj:
> On Fri, Oct 19, 2012 at 12:43 PM, Jan Weitzel <J.Weitzel@phytec.de> wrote:
> > Am Freitag, den 19.10.2012, 00:30 +0200 schrieb Vicente Bergas:
> >> OMAP4 requires that parent domains scale ahead of dependent domains.
> >> This is due to the restrictions in timing closure. To ensure
> >> a consistent behavior across all OMAP4 SoC, ensure that
> >> vdd_core scale first, then vdd_mpu and finally vdd_iva.
> >>
> >> TPS SET0/SET1 register is selected by a GPIO pin on OMAP4460 platforms.
> >> Currently we control this pin with a mux configuration as part of
> >> boot sequence.
> >> Current configuration results in the following voltage waveform:
> >> |---------------| (SET1 default 1.4V)
> >> | --------(programmed voltage)
> >> | <- (This switch happens on mux7,pullup)
> >> vdd_mpu(TPS) -----/ (OPP boot voltage)
> >> --------- (programmed voltage)
> >> vdd_core(TWL6030) -----------------------/ (OPP boot voltage)
> >> Problem 1) |<----- Tx ------>|
> >> timing violation for a duration Tx close to few milliseconds.
> >> Problem 2) voltage of MPU goes beyond spec for even the highest of MPU OPP.
> >>
> >> By using GPIO as recommended as standard procedure by TI, the sequence
> >> changes to:
> >> -------- (programmed voltage)
> >> vdd_mpu(TPS) ------------/ (Opp boot voltage)
> >> --------- (programmed voltage)
> >> vdd_core(TWL6030) -------------/ (OPP boot voltage)
> >>
> >> original patch from:
> >> http://lists.denx.de/pipermail/u-boot/2012-March/119149.html
> >> http://lists.denx.de/pipermail/u-boot/2012-March/119150.html
> >>
> >> Signed-off-by: Vicente Bergas <vicencb@gmail.com>
> >> ---
> >> arch/arm/boards/archosg9/mux.c | 9 +----
> >> arch/arm/mach-omap/include/mach/omap4-clock.h | 2 +-
> >> arch/arm/mach-omap/include/mach/omap4-silicon.h | 2 +-
> >> arch/arm/mach-omap/omap4_clock.c | 37 +++++++++++++++++-
> >> arch/arm/mach-omap/omap4_generic.c | 52 ++++---------------------
> >> 5 files changed, 46 insertions(+), 56 deletions(-)
> >>
> >> diff --git a/arch/arm/boards/archosg9/mux.c b/arch/arm/boards/archosg9/mux.c
> >> index 7ca5057..c8695c0 100644
> >> --- a/arch/arm/boards/archosg9/mux.c
> >> +++ b/arch/arm/boards/archosg9/mux.c
> >> @@ -441,7 +441,7 @@ static const struct pad_conf_entry wkup_padconf_array[] = {
> >> /* gpio_wk30 */
> >> { FREF_CLK3_REQ , M3 },
> >> /* gpio_wk7 */ /* tps62361_vsel0 */
> >> - { FREF_CLK4_REQ , IEN | PTU | M3 },
> >> + { FREF_CLK4_REQ , M3 },
> >> };
> > This should also be fixed for the panda board.
> >> void set_muxconf_regs(void){
> >> @@ -449,11 +449,4 @@ void set_muxconf_regs(void){
> >> core_padconf_array, ARRAY_SIZE(core_padconf_array));
> >> omap4_do_set_mux(OMAP44XX_CONTROL_PADCONF_WKUP,
> >> wkup_padconf_array, ARRAY_SIZE(wkup_padconf_array));
> >> -
> >> - /* gpio_wk7 is used for controlling TPS on 4460 */
> >> - if (omap4_revision() >= OMAP4460_ES1_0) {
> >> - writew(M3, OMAP44XX_CONTROL_PADCONF_WKUP + FREF_CLK4_REQ);
> >> - /* Enable GPIO-1 clocks before TPS initialization */
> >> - omap4_enable_gpio1_wup_clocks();
> >> - }
> >> }
> >> diff --git a/arch/arm/mach-omap/include/mach/omap4-clock.h b/arch/arm/mach-omap/include/mach/omap4-clock.h
> >> index e5302d6..8cbc061 100644
> >> --- a/arch/arm/mach-omap/include/mach/omap4-clock.h
> >> +++ b/arch/arm/mach-omap/include/mach/omap4-clock.h
> >> @@ -335,5 +335,5 @@ void omap4_lock_core_dpll_shadow(const struct dpll_param *param);
> >> void omap4_enable_gpio1_wup_clocks(void);
> >> void omap4_enable_gpio_clocks(void);
> >> void omap4_enable_all_clocks(void);
> >> -void omap4_do_scale_tps62361(u32 reg, u32 volt_mv);
> >> +void omap4_do_scale_tps62361(int vsel0_pin, u32 reg, u32 volt_mv);
> >>
> >> diff --git a/arch/arm/mach-omap/include/mach/omap4-silicon.h b/arch/arm/mach-omap/include/mach/omap4-silicon.h
> >> index 71ffe39..137cc88 100644
> >> --- a/arch/arm/mach-omap/include/mach/omap4-silicon.h
> >> +++ b/arch/arm/mach-omap/include/mach/omap4-silicon.h
> >> @@ -196,6 +196,6 @@ struct dpll_param;
> >> void omap4_ddr_init(const struct ddr_regs *, const struct dpll_param *);
> >> void omap4_power_i2c_send(u32);
> >> unsigned int omap4_revision(void);
> >> -noinline int omap4_scale_vcores(unsigned vsel0_pin);
> >> +noinline int omap4_scale_vcores(int vsel0_pin);
> >>
> >> #endif
> >> diff --git a/arch/arm/mach-omap/omap4_clock.c b/arch/arm/mach-omap/omap4_clock.c
> >> index 1481f16..f7927d2 100644
> >> --- a/arch/arm/mach-omap/omap4_clock.c
> >> +++ b/arch/arm/mach-omap/omap4_clock.c
> >> @@ -394,9 +394,38 @@ void omap4_enable_all_clocks(void)
> >> sr32(CM_L3INIT_USBPHY_CLKCTRL, 0, 32, 0x301);
> >> }
> >>
> >> -void omap4_do_scale_tps62361(u32 reg, u32 volt_mv)
> >> +#define GPIO_MASK 0x1f
> >> +#define GPIO_OE 0x34
> >> +#define GPIO_CLEARDATAOUT 0x90
> >> +#define GPIO_SETDATAOUT 0x94
> >> +
> >> +static void __iomem *omap4_get_gpio_base(unsigned gpio)
> >> +{
> >> + void __iomem *base;
> >> +
> >> + if (gpio < 32)
> >> + base = (void *)0x4a310100;
> >> + else
> >> + base = (void *)(0x48053100 + ((gpio & ~GPIO_MASK) << 8));
> >> +
> >> + return base;
> >> +}
> >> +
> >> +void omap4_do_scale_tps62361(int vsel0_pin, u32 reg, u32 volt_mv)
> >> {
> >> u32 temp, step;
> >> + void __iomem *base;
> >> + u32 val;
> >> +
> >> + /* Pull the GPIO low to select SET0 register, while we program SET1 */
> >> + if (vsel0_pin >= 0) {
> >> + base = omap4_get_gpio_base(vsel0_pin);
> >> + val = 1 << (vsel0_pin & GPIO_MASK);
> >> + writel(val, base + GPIO_CLEARDATAOUT);
> >> + val = readl(base + GPIO_OE);
> >> + val &= ~(1 << (vsel0_pin & GPIO_MASK));
> >> + writel(val, base + GPIO_OE);
> >> + }
> >>
> >> step = volt_mv - TPS62361_BASE_VOLT_MV;
> >> step /= 10;
> >> @@ -411,4 +440,10 @@ void omap4_do_scale_tps62361(u32 reg, u32 volt_mv)
> >> if (!wait_on_value(OMAP44XX_PRM_VC_VAL_BYPASS_VALID_BIT, 0,
> >> OMAP44XX_PRM_VC_VAL_BYPASS, LDELAY))
> >> puts("Scaling voltage failed for vdd_mpu from TPS\n");
> >> +
> >> + /* Pull the GPIO high to select SET1 register */
> >> + if (vsel0_pin >= 0) {
> >> + val = 1 << (vsel0_pin & GPIO_MASK);
> >> + writel(val, base + GPIO_SETDATAOUT);
> >> + }
> >> }
> >> diff --git a/arch/arm/mach-omap/omap4_generic.c b/arch/arm/mach-omap/omap4_generic.c
> >> index 765fcb5..8e692ba 100644
> >> --- a/arch/arm/mach-omap/omap4_generic.c
> >> +++ b/arch/arm/mach-omap/omap4_generic.c
> >> @@ -495,27 +495,11 @@ enum omap_boot_src omap4_bootsrc(void)
> >> return OMAP_BOOTSRC_UNKNOWN;
> >> }
> >>
> >> -#define GPIO_MASK 0x1f
> >> -
> >> -static void __iomem *omap4_get_gpio_base(unsigned gpio)
> >> -{
> >> - void __iomem *base;
> >> -
> >> - if (gpio < 32)
> >> - base = (void *)0x4a310000;
> >> - else
> >> - base = (void *)(0x48053000 + ((gpio & ~GPIO_MASK) << 8));
> >> -
> >> - return base;
> >> -}
> >> -
> >> #define I2C_SLAVE 0x12
> >>
> >> -noinline int omap4_scale_vcores(unsigned vsel0_pin)
> >> +noinline int omap4_scale_vcores(int vsel0_pin)
> >> {
> >> - void __iomem *base;
> >> unsigned int rev = omap4_revision();
> >> - u32 val = 0;
> >>
> >> /* For VC bypass only VCOREx_CGF_FORCE is necessary and
> >> * VCOREx_CFG_VOLTAGE changes can be discarded
> >> @@ -523,33 +507,6 @@ noinline int omap4_scale_vcores(unsigned vsel0_pin)
> >> writel(0, OMAP44XX_PRM_VC_CFG_I2C_MODE);
> >> writel(0x6026, OMAP44XX_PRM_VC_CFG_I2C_CLK);
> >>
> >> - /* TPS - supplies vdd_mpu on 4460 */
> >> - if (rev >= OMAP4460_ES1_0) {
> >> - /*
> >> - * Setup SET1 and SET0 with right values so that kernel
> >> - * can use either of them based on its needs.
> >> - */
> >> - omap4_do_scale_tps62361(TPS62361_REG_ADDR_SET0, 1430);
> >> - omap4_do_scale_tps62361(TPS62361_REG_ADDR_SET1, 1430);
> >> -
> >> - /*
> >> - * Select SET1 in TPS62361:
> >> - * VSEL1 is grounded on board. So the following selects
> >> - * VSEL1 = 0 and VSEL0 = 1
> >> - */
> >> - base = omap4_get_gpio_base(vsel0_pin);
> >> -
> >> - val = 1 << (vsel0_pin & GPIO_MASK);
> >> - writel(val, base + 0x190);
> >> -
> >> - val = readl(base + 0x134);
> >> - val &= (1 << (vsel0_pin & GPIO_MASK));
> >> - writel(val, base + 0x134);
> >> -
> >> - val = 1 << (vsel0_pin & GPIO_MASK);
> >> - writel(val, base + 0x194);
> >> - }
> >> -
> >> /* set VCORE1 force VSEL */
> >> /*
> >> * 4430 : supplies vdd_mpu
> >> @@ -568,8 +525,13 @@ noinline int omap4_scale_vcores(unsigned vsel0_pin)
> >> /* 0x55: i2c addr, 28: ~ 1200 mvolts*/
> >> omap4_power_i2c_send((0x2855 << 8) | I2C_SLAVE);
> >>
> >> + /* TPS - supplies vdd_mpu on 4460 */
> >> + if (rev >= OMAP4460_ES1_0)
> >> + omap4_do_scale_tps62361(vsel0_pin,
> >> + TPS62361_REG_ADDR_SET1, 1203);
> >> +
> > On PCM049 we use 920 MHz OPP_Turbo so this doesn't work. We must make
> > the voltage configurable.
> >
> > Jan
>
> Is there any way to choose the correct VCOREs based only on OMAP revision?
> Or the only solution is to give control of VCOREs to the boards?
> In u-boot the voltages are selected as a function of the OMAP revision only.
> Another possible solution is to choose a lower OPP, if this is possible.
> Afterwards the OS will change it as required.
Going to OPP100 is ok for us. So only the RAM timings and the oscillator are board specific.
Jan
> Also in u-boot the VSEL0 pin is platform-defined, not board-defined. But this
> will break the pcm049 board as it uses a non-standard gpio (182 instead of 7).
>
> I think that as few things are board-dependent the better.
>
> >> /* FIXME: set VCORE2 force VSEL, Check the reset value */
> >> - omap4_power_i2c_send((0x295B << 8) | I2C_SLAVE);
> >> + omap4_power_i2c_send((0x285B << 8) | I2C_SLAVE);
> >>
> >> /* set VCORE3 force VSEL */
> >> switch (rev) {
> >
> >
_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox
next prev parent reply other threads:[~2012-10-22 9:00 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <[PATCH 6/6] OMAP4: clean voltage switch>
2012-10-15 21:42 ` Vicente Bergas
2012-10-18 12:02 ` Teresa Gamez
2012-10-18 22:30 ` Vicente Bergas
2012-10-19 10:43 ` Jan Weitzel
2012-10-20 23:49 ` vj
2012-10-22 9:00 ` Jan Weitzel [this message]
2012-10-15 0:22 [PATCH 0/6] archosg9: improve support for tablet Vicente Bergas
2012-10-15 0:22 ` [PATCH 6/6] OMAP4: clean voltage switch Vicente Bergas
2012-10-15 7:17 ` Sascha Hauer
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1350896430.3205.4.camel@lws-weitzel \
--to=j.weitzel@phytec.de \
--cc=barebox@lists.infradead.org \
--cc=vicencb@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox