If the audio route has SSI only, audio-dma stop timing is not correct.
The correct order is as follows.
  1. Stop SSI
  2. Stop audio-dma

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This patch fixes the problem that DMA doesn't stop.
This patch changes order.
 old : Stop SSI. Then, stop the Audio DMAC-pp.
 new : Stop Audio DMAC-pp. Then, stop the SSI.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This patch fixes definition mistake of SCU_SYSTEM_INT_ENABLE1 address.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

The setting of WS continue bit moves before start of SSI.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This patch changes set timing of SRC_BUSIF_DALIGN.
Furthermore the code is put in order with this correction.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

SRC_ROUTE_MODE0 should be set before start.
The setting of SRC_SRCIR moves other place.
By this change, rsnd_src_start() function is removed.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

In case of not using DVC,
 capture : SSI(SSIU) -> SCU(SRC) -> External Memory.
 1. stop SCU.SRC.start_out
 2. stop audio dma-pp
 3. stop SCU.SRC.start_in

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

Change order to follows.

  Base sequence at play:
   1. start Audio DMAC-pp
   2. start Audio DMAC
   3. start SSI
   4. start SSIU
   5. start SCU(CMD)
   6. start SCU(SRC)

   7. stop  SSI
   8. stop  Audio DMAC-pp
   9. stop  SSIU
  10. stop  SCU(CMD)
  11. stop  SCU(SRC)
  12. stop  Audio DMAC

  Base sequence at capture:
   1. start Audio DMAC-pp
   2. start Audio DMAC
   3. start SCU(CMD)
   4. start SCU(SRC)
   5. start SSIU
   6. start SSI

   7. stop SCU(CMD)
   8. stop Audio DMAC-pp
   9. stop SCU(SRC)
  10. stop SSIU
  11. stop SSI
  12. stop Audio DMAC

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This control depends on the rsnd_soc_dai_trigger() function.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This control depends on the rsnd_soc_dai_trigger() function.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

There is a need to modify the audio sequence. This patch is
preparatory code to control the audio sequence.
Add new operation ".dma_start" and ".dma_stop".

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This patch waits for idle mode status of SSI.

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

Fix the calculation of top location in post cropping
with PI conversion enabled.

Not for upstream merge.

Signed-off-by: Hiroki Negishi <hiroki.negishi.zr@hitachi-solutions.com>
Signed-off-by: Yoshifumi Hosoya <yoshifumi.hosoya.wj@renesas.com>

In PI conversion mode, VSP1 driver repeats a same buffer
if there is a buffer in driver's buffer queue.
The repeats are needed in case of DU output in order to continue to display images.

However, in case of memory output, COMPOSITE mode,
the repeats cause incorrect output.
VSP1 extracts Top/Bottom fields from the same buffer,
so an output image is the same as input.
This patch fixes VSP1 driver so that VSP1 outputs correct images without repeats.

Not for upstream merge.

Signed-off-by: Hiroki Negishi <hiroki.negishi.zr@hitachi-solutions.com>
Signed-off-by: Yoshifumi Hosoya <yoshifumi.hosoya.wj@renesas.com>

When one frame is configured as shown below
first field is an odd(1,3,5..) field
second field is an even(2,4,6..) field

Signed-off-by: Yoshifumi Hosoya <yoshifumi.hosoya.wj@renesas.com>

When Tx DMA is only used, Tx FIFO is still not empty after DMA callback.
This patch waits for sweeping data out of the Tx FIFO.

Signed-off-by: Hiromitsu Yamasaki <hiromitsu.yamasaki.ym@renesas.com>
Signed-off-by: Ryo Kataoka <ryo.kataoka.wt@renesas.com>

  - SRC0, SRC7-9 devices not exist. These clock definitions
    and sound route removed. But SRC0 setting is need, set
    to "unused=1".

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>

This patch fixes the problem which lacked the definition of GPIO0 clock
in r8a7794 SoC.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

This reverts commit 67d867aa

.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

Commit c789778d

 ("mmc: sh_mobile_sdhi: Increase SDHI clock rate to
195MHz") can update clk rate with the assigned value, when
"renesas,clk-rate" property is set to device tree.

However, when DIV6 clocks is being used for sdhi clock control, if sdhi
clock is updated by clk_set_rate() during a clock stop, sdhi clock is
enabled after that and clk_get_rate() is called, it returns a value not
intended.

This patch fixes the problem by enabling sdhi clock, before calling
clk_set_rate().

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

Commit 6371c388

 ("mmc: sh_mmcif: Increase MMCIF clock rate to
97.5MHz") can update clk rate with the assigned value, when
"renesas,clk-rate" property is set to device tree.

However, when DIV6 clocks is being used for mmc clock control, if mmc
clock is updated by clk_set_rate() during a clock stop, mmc clock is
enabled after that and clk_get_rate() is called, it returns a value not
intended.

This patch fixes the problem by enabling mmc clock, before calling
clk_set_rate().

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>
Signed-off-by: Gaku Inami <gaku.inami.xw@bp.renesas.com>

Return type of i2c_smbus_read_byte_data is s32 not u8. This patch adds
an appropriate variable and fixes up the assignment.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

This patch holds a supply voltage VIO33 to avoid a shutdown of the DA9063.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

Return type of i2c_smbus_read_byte_data is s32 not u8. This patch adds
an appropriate variable and fixes up the assignment.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

This patch holds a supply voltage VIO33 to avoid a shutdown of the DA9063.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

Return type of i2c_smbus_read_byte_data is s32 not u8. This patch adds
an appropriate variable and fixes up the assignment.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

This patch holds a supply voltage VIO33 to avoid a shutdown of the DA9063.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

Return type of i2c_smbus_read_byte_data is s32 not u8. This patch adds
an appropriate variable and fixes up the assignment.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

This patch holds a supply voltage VIO33 to avoid a shutdown of the DA9063.

Signed-off-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>

There is a case sci_dma_rx_complete() is called in the work_fn_rx() as follows.

work_fn_rx()
 -> dma_async_is_tx_complete()
 -> device_tx_status() == shdma_tx_status()
 -> shdma_chan_ld_cleanup()
 -> __ld_cleanup()
 -> callback() == sci_dma_rx_complete()

In this case, dma_rx_complete() tries to lock spinlock that is already locked from work_fn_rx().
In the result, deadlock occurs.

Fix deadlock as follows.
 - Divide WorkQueue method into two:
     one is work_fn_rx(), for reception timeout.
     the other is a new method, work_fn_rx_complete(), for DMA transfer completion.
 - In reception timeout, work_fn_rx() keeps own process id until end of spinlock.
 - In DMA transfer completion, sci_dma_rx_complete() does not lock
   if kept process id is same as own process id.

Signed-off-by: Mitsuru Ezo <mitsuru.ezo.pz@ps.hitachi-solutions.com>
Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

sci_start_tx() changes the SCSCR register, but does not lock sci_port.
Therefore, a caller should lock sci_port.

sci_tx_dma_release() does not lock sci_port in calling sci_start_tx().
This may cause conflicts of settings to the SCSCR register.
So, this patch adds lock in sci_tx_dma_release().

Signed-off-by: Mitsuru Ezo <mitsuru.ezo.pz@ps.hitachi-solutions.com>
Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

The SCSCR register access in sci_break_ctl() is not locked.
The unlocked access may cause conflicts of settings to the SCSCR register.
Therefore, this patch adds lock during register access in sci_break_ctl().

Signed-off-by: Mitsuru Ezo <mitsuru.ezo.pz@ps.hitachi-solutions.com>
Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

There is deadlock as follows.

work_fn_rx()
  spin_lock_irqsave()      <- spin locked already here
  dev_err()                <- request serial output
  ...
    serial_console_write()
      spin_lock()          <- try to lock the spin
      <waits forever>

This problem is caused by serial output request while spin is already locked.
Serial output process tries to lock the spin and deadlock occured.

Signed-off-by: Mitsuru Ezo <mitsuru.ezo.pz@ps.hitachi-solutions.com>
Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

Change the processing order of start-up and end of SCIF.

 < start-up(sci_startup()) >
    before: IRQ Start(sci_request_irq()) -> DMA Start(sci_request_dma())
    After : DMA Start(sci_request_dma()) -> IRQ Start(sci_request_irq())

 < end(sci_shutdown()) >
    before: DMA End(sci_free_dma()) -> IRQ End(sci_free_irq())
    After : IRQ End(sci_free_irq()) -> DMA End(sci_free_dma())

Signed-off-by: Ryuichi Ando <ryuichi.ando.jz@ps.hitachi-solutions.com>
Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

If CONFIG_SERIAL_SH_SCI_DMA is enabled, the driver doesn't enable TIE
on SCIF or HSCIF. However, this driver may call sci_tx_interrupt()
in sci_er_interrupt(). After that, the driver cannot care of the
interrupt, and then "irq 109: nobody cared" happens on r8a7791/koelsch
board. This patch fixes the issue.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
[geert] Keep kicking tx when using PIO
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
(cherry picked from commit 440cc3afb74e91da35b296934e1efad5270355c7)
Signed-off-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>

Because the value of addr_c1 variable was wrong, this patch modifies
to the corrected value.

Signed-off-by: Koji Matsuoka <koji.matsuoka.xm@renesas.com>

DMA thread and interrupt process might be executed after TRIGGER_STOP.
In that case, they might access NULL pointer since substream is invalid.
This patch avoids these problem.

This patch was created by referring to following commits.
commit 02299d98 ("ASoC: rsnd: spin lock for interrupt handler")
commit 12927a8f ("ASoC: rsnd: don't use rsnd_mod_to_io() on rsnd_dma_xxx()")
commit 75defee0 (""ASoC: rsnd: don't call snd_pcm_period_elapsed() under spin lock")
commit 9b99e9a7 ("ASoC: rsnd: don't use rsnd_mod_to_io() on rsnd_dma_xxx()")
commit d5bbe7de

 ("ASoC: rsnd: move rsnd_mod_is_working() to rsnd_io_is_working()")

Signed-off-by: Ryo Kataoka <ryo.kataoka.wt@renesas.com>

When R-Car I2C detects stop condition of other master after arbitration
lost interruption, it restarts since ESG bit of ICMCR is being set.
This patch avoids restart by clearing ICMCR at arbitration
lost interruption.

Signed-off-by: Ryo Kataoka <ryo.kataoka.wt@renesas.com>