| 1 |
.. SPDX-License-Identifier: GPL-2.0 |
| 2 |
|
| 3 |
======================================== |
| 4 |
GPMC (General Purpose Memory Controller) |
| 5 |
======================================== |
| 6 |
|
| 7 |
GPMC is an unified memory controller dedicated to interfacing external |
| 8 |
memory devices like |
| 9 |
|
| 10 |
* Asynchronous SRAM like memories and application specific integrated |
| 11 |
circuit devices. |
| 12 |
* Asynchronous, synchronous, and page mode burst NOR flash devices |
| 13 |
NAND flash |
| 14 |
* Pseudo-SRAM devices |
| 15 |
|
| 16 |
GPMC is found on Texas Instruments SoC's (OMAP based) |
| 17 |
IP details: https://www.ti.com/lit/pdf/spruh73 section 7.1 |
| 18 |
|
| 19 |
|
| 20 |
GPMC generic timing calculation: |
| 21 |
================================ |
| 22 |
|
| 23 |
GPMC has certain timings that has to be programmed for proper |
| 24 |
functioning of the peripheral, while peripheral has another set of |
| 25 |
timings. To have peripheral work with gpmc, peripheral timings has to |
| 26 |
be translated to the form gpmc can understand. The way it has to be |
| 27 |
translated depends on the connected peripheral. Also there is a |
| 28 |
dependency for certain gpmc timings on gpmc clock frequency. Hence a |
| 29 |
generic timing routine was developed to achieve above requirements. |
| 30 |
|
| 31 |
Generic routine provides a generic method to calculate gpmc timings |
| 32 |
from gpmc peripheral timings. struct gpmc_device_timings fields has to |
| 33 |
be updated with timings from the datasheet of the peripheral that is |
| 34 |
connected to gpmc. A few of the peripheral timings can be fed either |
| 35 |
in time or in cycles, provision to handle this scenario has been |
| 36 |
provided (refer struct gpmc_device_timings definition). It may so |
| 37 |
happen that timing as specified by peripheral datasheet is not present |
| 38 |
in timing structure, in this scenario, try to correlate peripheral |
| 39 |
timing to the one available. If that doesn't work, try to add a new |
| 40 |
field as required by peripheral, educate generic timing routine to |
| 41 |
handle it, make sure that it does not break any of the existing. |
| 42 |
Then there may be cases where peripheral datasheet doesn't mention |
| 43 |
certain fields of struct gpmc_device_timings, zero those entries. |
| 44 |
|
| 45 |
Generic timing routine has been verified to work properly on |
| 46 |
multiple onenand's and tusb6010 peripherals. |
| 47 |
|
| 48 |
A word of caution: generic timing routine has been developed based |
| 49 |
on understanding of gpmc timings, peripheral timings, available |
| 50 |
custom timing routines, a kind of reverse engineering without |
| 51 |
most of the datasheets & hardware (to be exact none of those supported |
| 52 |
in mainline having custom timing routine) and by simulation. |
| 53 |
|
| 54 |
gpmc timing dependency on peripheral timings: |
| 55 |
|
| 56 |
[<gpmc_timing>: <peripheral timing1>, <peripheral timing2> ...] |
| 57 |
|
| 58 |
1. common |
| 59 |
|
| 60 |
cs_on: |
| 61 |
t_ceasu |
| 62 |
adv_on: |
| 63 |
t_avdasu, t_ceavd |
| 64 |
|
| 65 |
2. sync common |
| 66 |
|
| 67 |
sync_clk: |
| 68 |
clk |
| 69 |
page_burst_access: |
| 70 |
t_bacc |
| 71 |
clk_activation: |
| 72 |
t_ces, t_avds |
| 73 |
|
| 74 |
3. read async muxed |
| 75 |
|
| 76 |
adv_rd_off: |
| 77 |
t_avdp_r |
| 78 |
oe_on: |
| 79 |
t_oeasu, t_aavdh |
| 80 |
access: |
| 81 |
t_iaa, t_oe, t_ce, t_aa |
| 82 |
rd_cycle: |
| 83 |
t_rd_cycle, t_cez_r, t_oez |
| 84 |
|
| 85 |
4. read async non-muxed |
| 86 |
|
| 87 |
adv_rd_off: |
| 88 |
t_avdp_r |
| 89 |
oe_on: |
| 90 |
t_oeasu |
| 91 |
access: |
| 92 |
t_iaa, t_oe, t_ce, t_aa |
| 93 |
rd_cycle: |
| 94 |
t_rd_cycle, t_cez_r, t_oez |
| 95 |
|
| 96 |
5. read sync muxed |
| 97 |
|
| 98 |
adv_rd_off: |
| 99 |
t_avdp_r, t_avdh |
| 100 |
oe_on: |
| 101 |
t_oeasu, t_ach, cyc_aavdh_oe |
| 102 |
access: |
| 103 |
t_iaa, cyc_iaa, cyc_oe |
| 104 |
rd_cycle: |
| 105 |
t_cez_r, t_oez, t_ce_rdyz |
| 106 |
|
| 107 |
6. read sync non-muxed |
| 108 |
|
| 109 |
adv_rd_off: |
| 110 |
t_avdp_r |
| 111 |
oe_on: |
| 112 |
t_oeasu |
| 113 |
access: |
| 114 |
t_iaa, cyc_iaa, cyc_oe |
| 115 |
rd_cycle: |
| 116 |
t_cez_r, t_oez, t_ce_rdyz |
| 117 |
|
| 118 |
7. write async muxed |
| 119 |
|
| 120 |
adv_wr_off: |
| 121 |
t_avdp_w |
| 122 |
we_on, wr_data_mux_bus: |
| 123 |
t_weasu, t_aavdh, cyc_aavhd_we |
| 124 |
we_off: |
| 125 |
t_wpl |
| 126 |
cs_wr_off: |
| 127 |
t_wph |
| 128 |
wr_cycle: |
| 129 |
t_cez_w, t_wr_cycle |
| 130 |
|
| 131 |
8. write async non-muxed |
| 132 |
|
| 133 |
adv_wr_off: |
| 134 |
t_avdp_w |
| 135 |
we_on, wr_data_mux_bus: |
| 136 |
t_weasu |
| 137 |
we_off: |
| 138 |
t_wpl |
| 139 |
cs_wr_off: |
| 140 |
t_wph |
| 141 |
wr_cycle: |
| 142 |
t_cez_w, t_wr_cycle |
| 143 |
|
| 144 |
9. write sync muxed |
| 145 |
|
| 146 |
adv_wr_off: |
| 147 |
t_avdp_w, t_avdh |
| 148 |
we_on, wr_data_mux_bus: |
| 149 |
t_weasu, t_rdyo, t_aavdh, cyc_aavhd_we |
| 150 |
we_off: |
| 151 |
t_wpl, cyc_wpl |
| 152 |
cs_wr_off: |
| 153 |
t_wph |
| 154 |
wr_cycle: |
| 155 |
t_cez_w, t_ce_rdyz |
| 156 |
|
| 157 |
10. write sync non-muxed |
| 158 |
|
| 159 |
adv_wr_off: |
| 160 |
t_avdp_w |
| 161 |
we_on, wr_data_mux_bus: |
| 162 |
t_weasu, t_rdyo |
| 163 |
we_off: |
| 164 |
t_wpl, cyc_wpl |
| 165 |
cs_wr_off: |
| 166 |
t_wph |
| 167 |
wr_cycle: |
| 168 |
t_cez_w, t_ce_rdyz |
| 169 |
|
| 170 |
|
| 171 |
Note: |
| 172 |
Many of gpmc timings are dependent on other gpmc timings (a few |
| 173 |
gpmc timings purely dependent on other gpmc timings, a reason that |
| 174 |
some of the gpmc timings are missing above), and it will result in |
| 175 |
indirect dependency of peripheral timings to gpmc timings other than |
| 176 |
mentioned above, refer timing routine for more details. To know what |
| 177 |
these peripheral timings correspond to, please see explanations in |
| 178 |
struct gpmc_device_timings definition. And for gpmc timings refer |
| 179 |
IP details (link above). |