| Gregory Bean | 70816e2 | 2010-08-28 10:05:45 -0700 | [diff] [blame] | 1 | This document provides an overview of the msm_gpiomux interface, which | 
 | 2 | is used to provide gpio pin multiplexing and configuration on mach-msm | 
 | 3 | targets. | 
 | 4 |  | 
 | 5 | History | 
 | 6 | ======= | 
 | 7 |  | 
 | 8 | The first-generation API for gpio configuration & multiplexing on msm | 
 | 9 | is the function gpio_tlmm_config().  This function has a few notable | 
 | 10 | shortcomings, which led to its deprecation and replacement by gpiomux: | 
 | 11 |  | 
 | 12 | The 'disable' parameter:  Setting the second parameter to | 
 | 13 | gpio_tlmm_config to GPIO_CFG_DISABLE tells the peripheral | 
 | 14 | processor in charge of the subsystem to perform a look-up into a | 
 | 15 | low-power table and apply the low-power/sleep setting for the pin. | 
 | 16 | As the msm family evolved this became problematic. Not all pins | 
 | 17 | have sleep settings, not all peripheral processors will accept requests | 
 | 18 | to apply said sleep settings, and not all msm targets have their gpio | 
 | 19 | subsystems managed by a peripheral processor. In order to get consistent | 
 | 20 | behavior on all targets, drivers are forced to ignore this parameter, | 
 | 21 | rendering it useless. | 
 | 22 |  | 
 | 23 | The 'direction' flag: for all mux-settings other than raw-gpio (0), | 
 | 24 | the output-enable bit of a gpio is hard-wired to a known | 
 | 25 | input (usually VDD or ground).  For those settings, the direction flag | 
 | 26 | is meaningless at best, and deceptive at worst.  In addition, using the | 
 | 27 | direction flag to change output-enable (OE) directly can cause trouble in | 
 | 28 | gpiolib, which has no visibility into gpio direction changes made | 
 | 29 | in this way.  Direction control in gpio mode should be made through gpiolib. | 
 | 30 |  | 
 | 31 | Key Features of gpiomux | 
 | 32 | ======================= | 
 | 33 |  | 
 | 34 | - A consistent interface across all generations of msm.  Drivers can expect | 
 | 35 | the same results on every target. | 
 | 36 | - gpiomux plays nicely with gpiolib.  Functions that should belong to gpiolib | 
 | 37 | are left to gpiolib and not duplicated here.  gpiomux is written with the | 
 | 38 | intent that gpio_chips will call gpiomux reference-counting methods | 
 | 39 | from their request() and free() hooks, providing full integration. | 
 | 40 | - Tabular configuration.  Instead of having to call gpio_tlmm_config | 
 | 41 | hundreds of times, gpio configuration is placed in a single table. | 
 | 42 | - Per-gpio sleep.  Each gpio is individually reference counted, allowing only | 
 | 43 | those lines which are in use to be put in high-power states. | 
 | 44 | - 0 means 'do nothing': all flags are designed so that the default memset-zero | 
 | 45 | equates to a sensible default of 'no configuration', preventing users | 
 | 46 | from having to provide hundreds of 'no-op' configs for unused or | 
 | 47 | unwanted lines. | 
 | 48 |  | 
 | 49 | Usage | 
 | 50 | ===== | 
 | 51 |  | 
 | 52 | To use gpiomux, provide configuration information for relevant gpio lines | 
 | 53 | in the msm_gpiomux_configs table.  Since a 0 equates to "unconfigured", | 
 | 54 | only those lines to be managed by gpiomux need to be specified.  Here | 
 | 55 | is a completely fictional example: | 
 | 56 |  | 
 | 57 | struct msm_gpiomux_config msm_gpiomux_configs[GPIOMUX_NGPIOS] = { | 
 | 58 | 	[12] = { | 
 | 59 | 		.active = GPIOMUX_VALID | GPIOMUX_DRV_8MA | GPIOMUX_FUNC_1, | 
 | 60 | 		.suspended = GPIOMUX_VALID | GPIOMUX_PULL_DOWN, | 
 | 61 | 	}, | 
 | 62 | 	[34] = { | 
 | 63 | 		.suspended = GPIOMUX_VALID | GPIOMUX_PULL_DOWN, | 
 | 64 | 	}, | 
 | 65 | }; | 
 | 66 |  | 
 | 67 | To indicate that a gpio is in use, call msm_gpiomux_get() to increase | 
 | 68 | its reference count.  To decrease the reference count, call msm_gpiomux_put(). | 
 | 69 |  | 
 | 70 | The effect of this configuration is as follows: | 
 | 71 |  | 
 | 72 | When the system boots, gpios 12 and 34 will be initialized with their | 
 | 73 | 'suspended' configurations.  All other gpios, which were left unconfigured, | 
 | 74 | will not be touched. | 
 | 75 |  | 
 | 76 | When msm_gpiomux_get() is called on gpio 12 to raise its reference count | 
 | 77 | above 0, its active configuration will be applied.  Since no other gpio | 
 | 78 | line has a valid active configuration, msm_gpiomux_get() will have no | 
 | 79 | effect on any other line. | 
 | 80 |  | 
 | 81 | When msm_gpiomux_put() is called on gpio 12 or 34 to drop their reference | 
 | 82 | count to 0, their suspended configurations will be applied. | 
 | 83 | Since no other gpio line has a valid suspended configuration, no other | 
 | 84 | gpio line will be effected by msm_gpiomux_put().  Since gpio 34 has no valid | 
 | 85 | active configuration, this is effectively a no-op for gpio 34 as well, | 
 | 86 | with one small caveat, see the section "About Output-Enable Settings". | 
 | 87 |  | 
 | 88 | All of the GPIOMUX_VALID flags may seem like unnecessary overhead, but | 
 | 89 | they address some important issues.  As unused entries (all those | 
 | 90 | except 12 and 34) are zero-filled, gpiomux needs a way to distinguish | 
 | 91 | the used fields from the unused.  In addition, the all-zero pattern | 
 | 92 | is a valid configuration!  Therefore, gpiomux defines an additional bit | 
 | 93 | which is used to indicate when a field is used.  This has the pleasant | 
 | 94 | side-effect of allowing calls to msm_gpiomux_write to use '0' to indicate | 
 | 95 | that a value should not be changed: | 
 | 96 |  | 
 | 97 |   msm_gpiomux_write(0, GPIOMUX_VALID, 0); | 
 | 98 |  | 
 | 99 | replaces the active configuration of gpio 0 with an all-zero configuration, | 
 | 100 | but leaves the suspended configuration as it was. | 
 | 101 |  | 
 | 102 | Static Configurations | 
 | 103 | ===================== | 
 | 104 |  | 
 | 105 | To install a static configuration, which is applied at boot and does | 
 | 106 | not change after that, install a configuration with a suspended component | 
 | 107 | but no active component, as in the previous example: | 
 | 108 |  | 
 | 109 | 	[34] = { | 
 | 110 | 		.suspended = GPIOMUX_VALID | GPIOMUX_PULL_DOWN, | 
 | 111 | 	}, | 
 | 112 |  | 
 | 113 | The suspended setting is applied during boot, and the lack of any valid | 
 | 114 | active setting prevents any other setting from being applied at runtime. | 
 | 115 | If other subsystems attempting to access the line is a concern, one could | 
 | 116 | *really* anchor the configuration down by calling msm_gpiomux_get on the | 
 | 117 | line at initialization to move the line into active mode.  With the line | 
 | 118 | held, it will never be re-suspended, and with no valid active configuration, | 
 | 119 | no new configurations will be applied. | 
 | 120 |  | 
 | 121 | But then, if having other subsystems grabbing for the line is truly a concern, | 
 | 122 | it should be reserved with gpio_request instead, which carries an implicit | 
 | 123 | msm_gpiomux_get. | 
 | 124 |  | 
 | 125 | gpiomux and gpiolib | 
 | 126 | =================== | 
 | 127 |  | 
 | 128 | It is expected that msm gpio_chips will call msm_gpiomux_get() and | 
 | 129 | msm_gpiomux_put() from their request and free hooks, like this fictional | 
 | 130 | example: | 
 | 131 |  | 
 | 132 | static int request(struct gpio_chip *chip, unsigned offset) | 
 | 133 | { | 
 | 134 |         return msm_gpiomux_get(chip->base + offset); | 
 | 135 | } | 
 | 136 |  | 
 | 137 | static void free(struct gpio_chip *chip, unsigned offset) | 
 | 138 | { | 
 | 139 |         msm_gpiomux_put(chip->base + offset); | 
 | 140 | } | 
 | 141 |  | 
 | 142 | 	...somewhere in a gpio_chip declaration... | 
 | 143 | 	.request = request, | 
 | 144 | 	.free    = free, | 
 | 145 |  | 
 | 146 | This provides important functionality: | 
 | 147 | - It guarantees that a gpio line will have its 'active' config applied | 
 | 148 |   when the line is requested, and will not be suspended while the line | 
 | 149 |   remains requested; and | 
 | 150 | - It guarantees that gpio-direction settings from gpiolib behave sensibly. | 
 | 151 |   See "About Output-Enable Settings." | 
 | 152 |  | 
 | 153 | This mechanism allows for "auto-request" of gpiomux lines via gpiolib | 
 | 154 | when it is suitable.  Drivers wishing more exact control are, of course, | 
 | 155 | free to also use msm_gpiomux_set and msm_gpiomux_get. | 
 | 156 |  | 
 | 157 | About Output-Enable Settings | 
 | 158 | ============================ | 
 | 159 |  | 
 | 160 | Some msm targets do not have the ability to query the current gpio | 
 | 161 | configuration setting.  This means that changes made to the output-enable | 
 | 162 | (OE) bit by gpiolib cannot be consistently detected and preserved by gpiomux. | 
 | 163 | Therefore, when gpiomux applies a configuration setting, any direction | 
 | 164 | settings which may have been applied by gpiolib are lost and the default | 
 | 165 | input settings are re-applied. | 
 | 166 |  | 
 | 167 | For this reason, drivers should not assume that gpio direction settings | 
 | 168 | continue to hold if they free and then re-request a gpio.  This seems like | 
 | 169 | common sense - after all, anybody could have obtained the line in the | 
 | 170 | meantime - but it needs saying. | 
 | 171 |  | 
 | 172 | This also means that calls to msm_gpiomux_write will reset the OE bit, | 
 | 173 | which means that if the gpio line is held by a client of gpiolib and | 
 | 174 | msm_gpiomux_write is called, the direction setting has been lost and | 
 | 175 | gpiolib's internal state has been broken. | 
 | 176 | Release gpio lines before reconfiguring them. |