| Message ID | 1441189213-8199-2-git-send-email-j-keerthy@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Sep 02, 2015 at 03:50:10PM +0530, Keerthy wrote: > Currently everytime soc_is calls are made, firstly device tree nodes > are parsed and then string comparisons are made to determine the > soc version. Optimizing it to be done one time and store the result. but is that really a problem ? Are soc_is* calls made in any critical path ? Please provide some benchmarking data.
* Felipe Balbi <balbi@ti.com> [150902 05:34]: > On Wed, Sep 02, 2015 at 03:50:10PM +0530, Keerthy wrote: > > Currently everytime soc_is calls are made, firstly device tree nodes > > are parsed and then string comparisons are made to determine the > > soc version. Optimizing it to be done one time and store the result. > > but is that really a problem ? Are soc_is* calls made in any critical > path ? Please provide some benchmarking data. Regarding the SoC detection, there are multiple issues to consider here: 1. We want to unify the SoC detection so it behaves the same way for all SoCs 2. We don't want the dts files to explode with boards having multiple SoC revisions. So SoC detection and features detection for SGX and DSP etc should be done after setting the major SoC class based on compatible value 3. We don't want to do pointless string comparison over and over Regards, Tony -- 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
* Keerthy <j-keerthy@ti.com> [150902 03:23]: > Currently everytime soc_is calls are made, firstly device tree nodes > are parsed and then string comparisons are made to determine the > soc version. Optimizing it to be done one time and store the result. > Use the stored value in all the subsequent checks for soc_is calls. ... > --- a/arch/arm/mach-omap2/id.c > +++ b/arch/arm/mach-omap2/id.c > @@ -44,12 +44,29 @@ static char soc_name[OMAP_SOC_MAX_NAME_LENGTH]; > static char soc_rev[OMAP_SOC_MAX_NAME_LENGTH]; > u32 omap_features; > > +static int soc_ids[MAX_ID]; Hmm why would we need to keep an array of the non-booted socs? We know they're all false after booting.. A bitmask is usually used here? Regards, Tony -- 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
On Thursday 03 September 2015 09:45 PM, Tony Lindgren wrote: > * Keerthy <j-keerthy@ti.com> [150902 03:23]: >> Currently everytime soc_is calls are made, firstly device tree nodes >> are parsed and then string comparisons are made to determine the >> soc version. Optimizing it to be done one time and store the result. >> Use the stored value in all the subsequent checks for soc_is calls. > ... > >> --- a/arch/arm/mach-omap2/id.c >> +++ b/arch/arm/mach-omap2/id.c >> @@ -44,12 +44,29 @@ static char soc_name[OMAP_SOC_MAX_NAME_LENGTH]; >> static char soc_rev[OMAP_SOC_MAX_NAME_LENGTH]; >> u32 omap_features; >> >> +static int soc_ids[MAX_ID]; > > Hmm why would we need to keep an array of the non-booted > socs? We know they're all false after booting.. > > A bitmask is usually used here? Yes. This will be more optimal. I will implement using a bit mask. > > Regards, > > Tony > -- > 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 > -- 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
diff --git a/arch/arm/mach-omap2/id.c b/arch/arm/mach-omap2/id.c index e3f713f..19289df 100644 --- a/arch/arm/mach-omap2/id.c +++ b/arch/arm/mach-omap2/id.c @@ -44,12 +44,29 @@ static char soc_name[OMAP_SOC_MAX_NAME_LENGTH]; static char soc_rev[OMAP_SOC_MAX_NAME_LENGTH]; u32 omap_features; +static int soc_ids[MAX_ID]; + unsigned int omap_rev(void) { return omap_revision; } EXPORT_SYMBOL(omap_rev); +void init_dra_soc_id(void) +{ + if (of_machine_is_compatible("ti,dra7")) + soc_ids[DRA7XX] = 1; + if (of_machine_is_compatible("ti,dra74")) + soc_ids[DRA74X] = 1; + if (of_machine_is_compatible("ti,dra72")) + soc_ids[DRA72X] = 1; +} + +int check_dra_soc(int id) +{ + return soc_ids[id]; +} + int omap_type(void) { static u32 val = OMAP2_DEVICETYPE_MASK; @@ -643,6 +660,7 @@ void __init dra7xxx_check_revision(void) u16 hawkeye; u8 rev; + init_dra_soc_id(); idcode = read_tap_reg(OMAP_TAP_IDCODE); hawkeye = (idcode >> 12) & 0xffff; rev = (idcode >> 28) & 0xff; diff --git a/arch/arm/mach-omap2/soc.h b/arch/arm/mach-omap2/soc.h index f97654d..cd10260 100644 --- a/arch/arm/mach-omap2/soc.h +++ b/arch/arm/mach-omap2/soc.h @@ -38,6 +38,15 @@ #include <linux/bitops.h> #include <linux/of.h> +enum soc_id_flags { + DRA7XX = 0, + DRA74X, + DRA72X, + MAX_ID, +}; + +int check_dra_soc(int id); + /* * Test if multicore OMAP support is needed */ @@ -397,9 +406,9 @@ IS_OMAP_TYPE(3430, 0x3430) #undef soc_is_dra7xx #undef soc_is_dra74x #undef soc_is_dra72x -#define soc_is_dra7xx() (of_machine_is_compatible("ti,dra7")) -#define soc_is_dra74x() (of_machine_is_compatible("ti,dra74")) -#define soc_is_dra72x() (of_machine_is_compatible("ti,dra72")) +#define soc_is_dra7xx() check_dra_soc(DRA7XX) +#define soc_is_dra74x() check_dra_soc(DRA74X) +#define soc_is_dra72x() check_dra_soc(DRA72X) #endif /* Various silicon revisions for omap2 */
Currently everytime soc_is calls are made, firstly device tree nodes are parsed and then string comparisons are made to determine the soc version. Optimizing it to be done one time and store the result. Use the stored value in all the subsequent checks for soc_is calls. Signed-off-by: Keerthy <j-keerthy@ti.com> --- arch/arm/mach-omap2/id.c | 18 ++++++++++++++++++ arch/arm/mach-omap2/soc.h | 15 ++++++++++++--- 2 files changed, 30 insertions(+), 3 deletions(-)