brintos

brintos / linux-shallow public Read only

0
0
Text · 18.1 KiB · 301aa30 Raw
415 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3=======================4STM32 DMA-MDMA chaining5=======================6 7 8Introduction9------------10 11  This document describes the STM32 DMA-MDMA chaining feature. But before going12  further, let's introduce the peripherals involved.13 14  To offload data transfers from the CPU, STM32 microprocessors (MPUs) embed15  direct memory access controllers (DMA).16 17  STM32MP1 SoCs embed both STM32 DMA and STM32 MDMA controllers. STM32 DMA18  request routing capabilities are enhanced by a DMA request multiplexer19  (STM32 DMAMUX).20 21  **STM32 DMAMUX**22 23  STM32 DMAMUX routes any DMA request from a given peripheral to any STM32 DMA24  controller (STM32MP1 counts two STM32 DMA controllers) channels.25 26  **STM32 DMA**27 28  STM32 DMA is mainly used to implement central data buffer storage (usually in29  the system SRAM) for different peripheral. It can access external RAMs but30  without the ability to generate convenient burst transfer ensuring the best31  load of the AXI.32 33  **STM32 MDMA**34 35  STM32 MDMA (Master DMA) is mainly used to manage direct data transfers between36  RAM data buffers without CPU intervention. It can also be used in a37  hierarchical structure that uses STM32 DMA as first level data buffer38  interfaces for AHB peripherals, while the STM32 MDMA acts as a second level39  DMA with better performance. As a AXI/AHB master, STM32 MDMA can take control40  of the AXI/AHB bus.41 42 43Principles44----------45 46  STM32 DMA-MDMA chaining feature relies on the strengths of STM32 DMA and47  STM32 MDMA controllers.48 49  STM32 DMA has a circular Double Buffer Mode (DBM). At each end of transaction50  (when DMA data counter - DMA_SxNDTR - reaches 0), the memory pointers51  (configured with DMA_SxSM0AR and DMA_SxM1AR) are swapped and the DMA data52  counter is automatically reloaded. This allows the SW or the STM32 MDMA to53  process one memory area while the second memory area is being filled/used by54  the STM32 DMA transfer.55 56  With STM32 MDMA linked-list mode, a single request initiates the data array57  (collection of nodes) to be transferred until the linked-list pointer for the58  channel is null. The channel transfer complete of the last node is the end of59  transfer, unless first and last nodes are linked to each other, in such a60  case, the linked-list loops on to create a circular MDMA transfer.61 62  STM32 MDMA has direct connections with STM32 DMA. This enables autonomous63  communication and synchronization between peripherals, thus saving CPU64  resources and bus congestion. Transfer Complete signal of STM32 DMA channel65  can triggers STM32 MDMA transfer. STM32 MDMA can clear the request generated66  by the STM32 DMA by writing to its Interrupt Clear register (whose address is67  stored in MDMA_CxMAR, and bit mask in MDMA_CxMDR).68 69  .. table:: STM32 MDMA interconnect table with STM32 DMA70 71    +--------------+----------------+-----------+------------+72    | STM32 DMAMUX | STM32 DMA      | STM32 DMA | STM32 MDMA |73    | channels     | channels       | Transfer  | request    |74    |              |                | complete  |            |75    |              |                | signal    |            |76    +==============+================+===========+============+77    | Channel *0*  | DMA1 channel 0 | dma1_tcf0 | *0x00*     |78    +--------------+----------------+-----------+------------+79    | Channel *1*  | DMA1 channel 1 | dma1_tcf1 | *0x01*     |80    +--------------+----------------+-----------+------------+81    | Channel *2*  | DMA1 channel 2 | dma1_tcf2 | *0x02*     |82    +--------------+----------------+-----------+------------+83    | Channel *3*  | DMA1 channel 3 | dma1_tcf3 | *0x03*     |84    +--------------+----------------+-----------+------------+85    | Channel *4*  | DMA1 channel 4 | dma1_tcf4 | *0x04*     |86    +--------------+----------------+-----------+------------+87    | Channel *5*  | DMA1 channel 5 | dma1_tcf5 | *0x05*     |88    +--------------+----------------+-----------+------------+89    | Channel *6*  | DMA1 channel 6 | dma1_tcf6 | *0x06*     |90    +--------------+----------------+-----------+------------+91    | Channel *7*  | DMA1 channel 7 | dma1_tcf7 | *0x07*     |92    +--------------+----------------+-----------+------------+93    | Channel *8*  | DMA2 channel 0 | dma2_tcf0 | *0x08*     |94    +--------------+----------------+-----------+------------+95    | Channel *9*  | DMA2 channel 1 | dma2_tcf1 | *0x09*     |96    +--------------+----------------+-----------+------------+97    | Channel *10* | DMA2 channel 2 | dma2_tcf2 | *0x0A*     |98    +--------------+----------------+-----------+------------+99    | Channel *11* | DMA2 channel 3 | dma2_tcf3 | *0x0B*     |100    +--------------+----------------+-----------+------------+101    | Channel *12* | DMA2 channel 4 | dma2_tcf4 | *0x0C*     |102    +--------------+----------------+-----------+------------+103    | Channel *13* | DMA2 channel 5 | dma2_tcf5 | *0x0D*     |104    +--------------+----------------+-----------+------------+105    | Channel *14* | DMA2 channel 6 | dma2_tcf6 | *0x0E*     |106    +--------------+----------------+-----------+------------+107    | Channel *15* | DMA2 channel 7 | dma2_tcf7 | *0x0F*     |108    +--------------+----------------+-----------+------------+109 110  STM32 DMA-MDMA chaining feature then uses a SRAM buffer. STM32MP1 SoCs embed111  three fast access static internal RAMs of various size, used for data storage.112  Due to STM32 DMA legacy (within microcontrollers), STM32 DMA performances are113  bad with DDR, while they are optimal with SRAM. Hence the SRAM buffer used114  between STM32 DMA and STM32 MDMA. This buffer is split in two equal periods115  and STM32 DMA uses one period while STM32 MDMA uses the other period116  simultaneously.117  ::118 119                    dma[1:2]-tcf[0:7]120                   .----------------.121     ____________ '    _________     V____________122    | STM32 DMA  |    /  __|>_  \    | STM32 MDMA |123    |------------|   |  /     \  |   |------------|124    | DMA_SxM0AR |<=>| | SRAM  | |<=>| []-[]...[] |125    | DMA_SxM1AR |   |  \_____/  |   |            |126    |____________|    \___<|____/    |____________|127 128  STM32 DMA-MDMA chaining uses (struct dma_slave_config).peripheral_config to129  exchange the parameters needed to configure MDMA. These parameters are130  gathered into a u32 array with three values:131 132  * the STM32 MDMA request (which is actually the DMAMUX channel ID),133  * the address of the STM32 DMA register to clear the Transfer Complete134    interrupt flag,135  * the mask of the Transfer Complete interrupt flag of the STM32 DMA channel.136 137Device Tree updates for STM32 DMA-MDMA chaining support138-------------------------------------------------------139 140  **1. Allocate a SRAM buffer**141 142    SRAM device tree node is defined in SoC device tree. You can refer to it in143    your board device tree to define your SRAM pool.144    ::145 146          &sram {147                  my_foo_device_dma_pool: dma-sram@0 {148                          reg = <0x0 0x1000>;149                  };150          };151 152    Be careful of the start index, in case there are other SRAM consumers.153    Define your pool size strategically: to optimise chaining, the idea is that154    STM32 DMA and STM32 MDMA can work simultaneously, on each buffer of the155    SRAM.156    If the SRAM period is greater than the expected DMA transfer, then STM32 DMA157    and STM32 MDMA will work sequentially instead of simultaneously. It is not a158    functional issue but it is not optimal.159 160    Don't forget to refer to your SRAM pool in your device node. You need to161    define a new property.162    ::163 164          &my_foo_device {165                  ...166                  my_dma_pool = &my_foo_device_dma_pool;167          };168 169    Then get this SRAM pool in your foo driver and allocate your SRAM buffer.170 171  **2. Allocate a STM32 DMA channel and a STM32 MDMA channel**172 173    You need to define an extra channel in your device tree node, in addition to174    the one you should already have for "classic" DMA operation.175 176    This new channel must be taken from STM32 MDMA channels, so, the phandle of177    the DMA controller to use is the MDMA controller's one.178    ::179 180          &my_foo_device {181                  [...]182                  my_dma_pool = &my_foo_device_dma_pool;183                  dmas = <&dmamux1 ...>,                // STM32 DMA channel184                         <&mdma1 0 0x3 0x1200000a 0 0>; // + STM32 MDMA channel185          };186 187    Concerning STM32 MDMA bindings:188 189    1. The request line number : whatever the value here, it will be overwritten190    by MDMA driver with the STM32 DMAMUX channel ID passed through191    (struct dma_slave_config).peripheral_config192 193    2. The priority level : choose Very High (0x3) so that your channel will194    take priority other the other during request arbitration195 196    3. A 32bit mask specifying the DMA channel configuration : source and197    destination address increment, block transfer with 128 bytes per single198    transfer199 200    4. The 32bit value specifying the register to be used to acknowledge the201    request: it will be overwritten by MDMA driver, with the DMA channel202    interrupt flag clear register address passed through203    (struct dma_slave_config).peripheral_config204 205    5. The 32bit mask specifying the value to be written to acknowledge the206    request: it will be overwritten by MDMA driver, with the DMA channel207    Transfer Complete flag passed through208    (struct dma_slave_config).peripheral_config209 210Driver updates for STM32 DMA-MDMA chaining support in foo driver211----------------------------------------------------------------212 213  **0. (optional) Refactor the original sg_table if dmaengine_prep_slave_sg()**214 215    In case of dmaengine_prep_slave_sg(), the original sg_table can't be used as216    is. Two new sg_tables must be created from the original one. One for217    STM32 DMA transfer (where memory address targets now the SRAM buffer instead218    of DDR buffer) and one for STM32 MDMA transfer (where memory address targets219    the DDR buffer).220 221    The new sg_list items must fit SRAM period length. Here is an example for222    DMA_DEV_TO_MEM:223    ::224 225      /*226        * Assuming sgl and nents, respectively the initial scatterlist and its227        * length.228        * Assuming sram_dma_buf and sram_period, respectively the memory229        * allocated from the pool for DMA usage, and the length of the period,230        * which is half of the sram_buf size.231        */232      struct sg_table new_dma_sgt, new_mdma_sgt;233      struct scatterlist *s, *_sgl;234      dma_addr_t ddr_dma_buf;235      u32 new_nents = 0, len;236      int i;237 238      /* Count the number of entries needed */239      for_each_sg(sgl, s, nents, i)240              if (sg_dma_len(s) > sram_period)241                      new_nents += DIV_ROUND_UP(sg_dma_len(s), sram_period);242              else243                      new_nents++;244 245      /* Create sg table for STM32 DMA channel */246      ret = sg_alloc_table(&new_dma_sgt, new_nents, GFP_ATOMIC);247      if (ret)248              dev_err(dev, "DMA sg table alloc failed\n");249 250      for_each_sg(new_dma_sgt.sgl, s, new_dma_sgt.nents, i) {251              _sgl = sgl;252              sg_dma_len(s) = min(sg_dma_len(_sgl), sram_period);253              /* Targets the beginning = first half of the sram_buf */254              s->dma_address = sram_buf;255              /*256                * Targets the second half of the sram_buf257                * for odd indexes of the item of the sg_list258                */259              if (i & 1)260                      s->dma_address += sram_period;261      }262 263      /* Create sg table for STM32 MDMA channel */264      ret = sg_alloc_table(&new_mdma_sgt, new_nents, GFP_ATOMIC);265      if (ret)266              dev_err(dev, "MDMA sg_table alloc failed\n");267 268      _sgl = sgl;269      len = sg_dma_len(sgl);270      ddr_dma_buf = sg_dma_address(sgl);271      for_each_sg(mdma_sgt.sgl, s, mdma_sgt.nents, i) {272              size_t bytes = min_t(size_t, len, sram_period);273 274              sg_dma_len(s) = bytes;275              sg_dma_address(s) = ddr_dma_buf;276              len -= bytes;277 278              if (!len && sg_next(_sgl)) {279                      _sgl = sg_next(_sgl);280                      len = sg_dma_len(_sgl);281                      ddr_dma_buf = sg_dma_address(_sgl);282              } else {283                      ddr_dma_buf += bytes;284              }285      }286 287    Don't forget to release these new sg_tables after getting the descriptors288    with dmaengine_prep_slave_sg().289 290  **1. Set controller specific parameters**291 292    First, use dmaengine_slave_config() with a struct dma_slave_config to293    configure STM32 DMA channel. You just have to take care of DMA addresses,294    the memory address (depending on the transfer direction) must point on your295    SRAM buffer, and set (struct dma_slave_config).peripheral_size != 0.296 297    STM32 DMA driver will check (struct dma_slave_config).peripheral_size to298    determine if chaining is being used or not. If it is used, then STM32 DMA299    driver fills (struct dma_slave_config).peripheral_config with an array of300    three u32 : the first one containing STM32 DMAMUX channel ID, the second one301    the channel interrupt flag clear register address, and the third one the302    channel Transfer Complete flag mask.303 304    Then, use dmaengine_slave_config with another struct dma_slave_config to305    configure STM32 MDMA channel. Take care of DMA addresses, the device address306    (depending on the transfer direction) must point on your SRAM buffer, and307    the memory address must point to the buffer originally used for "classic"308    DMA operation. Use the previous (struct dma_slave_config).peripheral_size309    and .peripheral_config that have been updated by STM32 DMA driver, to set310    (struct dma_slave_config).peripheral_size and .peripheral_config of the311    struct dma_slave_config to configure STM32 MDMA channel.312    ::313 314      struct dma_slave_config dma_conf;315      struct dma_slave_config mdma_conf;316 317      memset(&dma_conf, 0, sizeof(dma_conf));318      [...]319      config.direction = DMA_DEV_TO_MEM;320      config.dst_addr = sram_dma_buf;        // SRAM buffer321      config.peripheral_size = 1;            // peripheral_size != 0 => chaining322 323      dmaengine_slave_config(dma_chan, &dma_config);324 325      memset(&mdma_conf, 0, sizeof(mdma_conf));326      config.direction = DMA_DEV_TO_MEM;327      mdma_conf.src_addr = sram_dma_buf;     // SRAM buffer328      mdma_conf.dst_addr = rx_dma_buf;       // original memory buffer329      mdma_conf.peripheral_size = dma_conf.peripheral_size;       // <- dma_conf330      mdma_conf.peripheral_config = dma_config.peripheral_config; // <- dma_conf331 332      dmaengine_slave_config(mdma_chan, &mdma_conf);333 334  **2. Get a descriptor for STM32 DMA channel transaction**335 336    In the same way you get your descriptor for your "classic" DMA operation,337    you just have to replace the original sg_list (in case of338    dmaengine_prep_slave_sg()) with the new sg_list using SRAM buffer, or to339    replace the original buffer address, length and period (in case of340    dmaengine_prep_dma_cyclic()) with the new SRAM buffer.341 342  **3. Get a descriptor for STM32 MDMA channel transaction**343 344    If you previously get descriptor (for STM32 DMA) with345 346    * dmaengine_prep_slave_sg(), then use dmaengine_prep_slave_sg() for347      STM32 MDMA;348    * dmaengine_prep_dma_cyclic(), then use dmaengine_prep_dma_cyclic() for349      STM32 MDMA.350 351    Use the new sg_list using SRAM buffer (in case of dmaengine_prep_slave_sg())352    or, depending on the transfer direction, either the original DDR buffer (in353    case of DMA_DEV_TO_MEM) or the SRAM buffer (in case of DMA_MEM_TO_DEV), the354    source address being previously set with dmaengine_slave_config().355 356  **4. Submit both transactions**357 358    Before submitting your transactions, you may need to define on which359    descriptor you want a callback to be called at the end of the transfer360    (dmaengine_prep_slave_sg()) or the period (dmaengine_prep_dma_cyclic()).361    Depending on the direction, set the callback on the descriptor that finishes362    the overall transfer:363 364    * DMA_DEV_TO_MEM: set the callback on the "MDMA" descriptor365    * DMA_MEM_TO_DEV: set the callback on the "DMA" descriptor366 367    Then, submit the descriptors whatever the order, with dmaengine_tx_submit().368 369  **5. Issue pending requests (and wait for callback notification)**370 371  As STM32 MDMA channel transfer is triggered by STM32 DMA, you must issue372  STM32 MDMA channel before STM32 DMA channel.373 374  If any, your callback will be called to warn you about the end of the overall375  transfer or the period completion.376 377  Don't forget to terminate both channels. STM32 DMA channel is configured in378  cyclic Double-Buffer mode so it won't be disabled by HW, you need to terminate379  it. STM32 MDMA channel will be stopped by HW in case of sg transfer, but not380  in case of cyclic transfer. You can terminate it whatever the kind of transfer.381 382  **STM32 DMA-MDMA chaining DMA_MEM_TO_DEV special case**383 384  STM32 DMA-MDMA chaining in DMA_MEM_TO_DEV is a special case. Indeed, the385  STM32 MDMA feeds the SRAM buffer with the DDR data, and the STM32 DMA reads386  data from SRAM buffer. So some data (the first period) have to be copied in387  SRAM buffer when the STM32 DMA starts to read.388 389  A trick could be pausing the STM32 DMA channel (that will raise a Transfer390  Complete signal, triggering the STM32 MDMA channel), but the first data read391  by the STM32 DMA could be "wrong". The proper way is to prepare the first SRAM392  period with dmaengine_prep_dma_memcpy(). Then this first period should be393  "removed" from the sg or the cyclic transfer.394 395  Due to this complexity, rather use the STM32 DMA-MDMA chaining for396  DMA_DEV_TO_MEM and keep the "classic" DMA usage for DMA_MEM_TO_DEV, unless397  you're not afraid.398 399Resources400---------401 402  Application note, datasheet and reference manual are available on ST website403  (STM32MP1_).404 405  Dedicated focus on three application notes (AN5224_, AN4031_ & AN5001_)406  dealing with STM32 DMAMUX, STM32 DMA and STM32 MDMA.407 408.. _STM32MP1: https://www.st.com/en/microcontrollers-microprocessors/stm32mp1-series.html409.. _AN5224: https://www.st.com/resource/en/application_note/an5224-stm32-dmamux-the-dma-request-router-stmicroelectronics.pdf410.. _AN4031: https://www.st.com/resource/en/application_note/dm00046011-using-the-stm32f2-stm32f4-and-stm32f7-series-dma-controller-stmicroelectronics.pdf411.. _AN5001: https://www.st.com/resource/en/application_note/an5001-stm32cube-expansion-package-for-stm32h7-series-mdma-stmicroelectronics.pdf412 413:Authors:414 415- Amelie Delaunay <amelie.delaunay@foss.st.com>