Hi
On Thu, Nov 14, 2013 at 7:43 PM, Hardik Patel hardik.patel@volansystech.com wrote:
On Thu, Nov 14, 2013 at 10:02 PM, Tom Rini trini@ti.com wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 11/14/2013 11:28 AM, Michael Trimarchi wrote:
Hi
On Thu, Nov 14, 2013 at 5:19 PM, Tom Rini trini@ti.com wrote: On 11/14/2013 05:06 AM, Hardik wrote:
From: Hardik Patel hardik.patel@volansystech.com
Signed-off-by: Hardik Patel hardik.patel@volansystech.com
arch/arm/cpu/armv7/omap4/sdram_elpida.c | 2 +- board/ti/panda/panda.c | 24 ++++++++++++++++++++++++ 2 files changed, 25 insertions(+), 1 deletion(-)
diff --git a/arch/arm/cpu/armv7/omap4/sdram_elpida.c b/arch/arm/cpu/armv7/omap4/sdram_elpida.c index e4c8316..9fbdeea 100644 --- a/arch/arm/cpu/armv7/omap4/sdram_elpida.c +++ b/arch/arm/cpu/armv7/omap4/sdram_elpida.c @@ -123,7 +123,7 @@ static void emif_get_reg_dump_sdp(u32 emif_nr, const struct emif_regs **regs) *regs = &emif_regs_elpida_200_mhz_2cs; else if (omap4_rev == OMAP4430_ES2_3) *regs = &emif_regs_elpida_400_mhz_1cs;
else if (omap4_rev < OMAP4470_ES1_0)
else if (omap4_rev < OMAP4470_ES1_0 && !is_panda_es_rev_b3()) *regs = &emif_regs_elpida_400_mhz_2cs; else *regs = &emif_regs_elpida_400_mhz_1cs;diff --git a/board/ti/panda/panda.c b/board/ti/panda/panda.c index c104024..5a7f80a 100644 --- a/board/ti/panda/panda.c +++ b/board/ti/panda/panda.c @@ -122,6 +122,30 @@ int get_board_revision(void) return board_id; }
+/* +* Routine: is_panda_es_rev_b3 +* Description: Detect if we are running on B3 version of ES panda board, +* This can be done by reading the level of GPIO 171 +* and checking the processor revisions. +* GPIO171: 1 => Panda ES Rev B3 +*/ +u8 is_panda_es_rev_b3(void) +{
int processor_rev = omap_revision();int ret = 0;if ((processor_rev >= OMAP4460_ES1_0 &&processor_rev <= OMAP4460_ES1_1)) {/* Setup the mux for the common board ID pins (gpio
- */
writew((IEN | M3), (*ctrl)->control_padconf_core_base- UNIPRO_TX0);
/* if processor_rev is panda ES and GPIO171 is 1, itis rev b3 */
ret = gpio_get_value(PANDA_BOARD_ID_2_GPIO);}return ret;+}
/**
- @brief misc_init_r - Configure Panda board specific configurations
- such as power configurations, ethernet initialization as phase2 of
This isn't right. What I was saying is that board/ti/panda/panda.c needs to provide its own emif_get_reg_dump function, which will (as designed) override the current one from arch/arm/cpu/armv7/omap4/sdram_elpida.c
Any problem to have visibility of emif_regs array of lpddr?
Things that need exposing so that we can do what was intended before should be exposed, in <asm/arch-omap4/sys_proto.h> I think.
Tom -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
iQIcBAEBAgAGBQJShPszAAoJENk4IS6UOR1W2TMP/RzK77vL75h5HYP23RKP/CWx i7A/xbStKYT8ozh2s1IyHsKohVmnFmIh6lIa4cTYMzrqQBIJxnw9NB1p73qRFPFS M8Bwx0QS3IU9Wqybk2w1x26Ymm65NWcAQ9SvPh4EBCpdnu1JnJoaceekw1fp55Pb B+H/44Xr5D57F4R0DAuLqT2Kc6uoMf69X6I/DU7CuxhcceoMAMaFrZq+WrLAqIeY M5qsCTsyOIQdzV4pheR9gH7F6dbzlA9uee2lshIP0yN5JJmWkZc1CjMR0SsStSpy RKFJuNOgv32v7CGwvyA/HAk6KdKy5VSWUhHT/+1B3fMzIZhxypOlL90Vbz2Zlh5N J5E5RNWsQaBJnVaEpo/4wWOTWmq/L5+lQLyJVPUkZFGocHgRBT8gqkcREe/+mSKx DaxX4mLNlCvooMrNSakPPZHtPHXnVieMAvhdKDDdhLRKHAST+1hZD/ztBD63bwcJ eSvGBe6SDv/K/mScU9eeNtJzVYs6O1VZm9sqgJBPhgUTN/Mfv7NbZVF3m4rMpkeB PbQ/eSzxERBxeiInrPh2Ns62oDxCXTodOw5qy+jUQ72WPI/3zMNRHaBbyXTNn9nZ NAj4XSex3tiQUpJYw21wbshEuCz4/Lo5LBWXeksdg0bfSxWX9jG/mpkOE6NACSyE 7TeL/tTeum0zVByyfhjG =ZmUi -----END PGP SIGNATURE-----
Hi Tom, Michel,
Thanks for your feedback on this.
We had implemented override of "emif_get_reg_dump" function in panda.c. However, we thought checking detection of pandaboard in sdram_elpida.c be smaller change and it was also under omap4. However, now I realized after discussing that this may break in case of non panda.c OMAP4 boards.
Based on your feedback,we are going to correct it like below. I am trying to explain it in brief:
- We are going to extern emif_regs in <asm/arch-omap4/sys_proto.h>, so that
we can access them in panda.c #ifdef CONFIG_SYS_EMIF_PRECALCULATED_TIMING_REGS extern const struct emif_regs emif_regs_elpida_200_mhz_2cs; extern const struct emif_regs emif_regs_elpida_380_mhz_1cs; extern const struct emif_regs emif_regs_elpida_400_mhz_1cs; extern const struct emif_regs emif_regs_elpida_400_mhz_2cs; #endif
I prefer having them in type name and have a function that give back the pointer of the regs?
- "emif_regs_elpida_200_mhz_2cs" & "emif_regs_elpida_380_mhz_1cs" are
defined as "static const" in sdram_elpida.c. So, we need to remove "static" out of them as we are accessing it outside. This will be change in sdram_elpida.c.
- In Panda.c, we will write "emif_get_reg_dump" strong function as below to
override one of sdram_elpida.c. We would still need "is_panda_es_rev_b3()" function to identify rev b3. However we will use it properly by NOT using NOT OPERATOR( ! is_panda_es_rev_b3() ) .
#ifdef CONFIG_SYS_EMIF_PRECALCULATED_TIMING_REGS static void emif_get_reg_dump(u32 emif_nr, const struct emif_regs **regs) { u32 omap4_rev = omap_revision();
/* Same devices and geometry on both EMIFs */ if (omap4_rev == OMAP4430_ES1_0) *regs = &emif_regs_elpida_380_mhz_1cs; else if (omap4_rev == OMAP4430_ES2_0) *regs = &emif_regs_elpida_200_mhz_2cs; else if (omap4_rev == OMAP4430_ES2_3) *regs = &emif_regs_elpida_400_mhz_1cs; else if (omap4_rev < OMAP4470_ES1_0) { if( is_panda_es_rev_b3() ) *regs = &emif_regs_elpida_400_mhz_1cs; else *regs = &emif_regs_elpida_400_mhz_2cs; } else *regs = &emif_regs_elpida_400_mhz_1cs;}
Is it possible to arrange it in a switch?
Michael
#endif
All changes will be protected by "#ifdef CONFIG_SYS_EMIF_PRECALCULATED_TIMING_REGS" macro.
Please let me know if above changes makes sense.
Thanks, Hardik