All of lore.kernel.org
 help / color / mirror / Atom feed
From: Serge Semin <Sergey.Semin@baikalelectronics.ru>
To: Mark Brown <broonie@kernel.org>
Cc: Serge Semin <Sergey.Semin@baikalelectronics.ru>,
	Serge Semin <fancer.lancer@gmail.com>,
	Alexey Malahov <Alexey.Malahov@baikalelectronics.ru>,
	Ramil Zaripov <Ramil.Zaripov@baikalelectronics.ru>,
	Pavel Parkhomenko <Pavel.Parkhomenko@baikalelectronics.ru>,
	Andy Shevchenko <andy.shevchenko@gmail.com>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
	Lars Povlsen <lars.povlsen@microchip.com>,
	"wuxu . wu" <wuxu.wu@huawei.com>, Feng Tang <feng.tang@intel.com>,
	Rob Herring <robh+dt@kernel.org>, <linux-spi@vger.kernel.org>,
	<devicetree@vger.kernel.org>, <linux-kernel@vger.kernel.org>
Subject: [PATCH 04/30] Revert: spi: spi-dw: Add lock protect dw_spi rx/tx to prevent concurrent calls
Date: Sun, 20 Sep 2020 14:28:48 +0300	[thread overview]
Message-ID: <20200920112914.26501-5-Sergey.Semin@baikalelectronics.ru> (raw)
In-Reply-To: <20200920112914.26501-1-Sergey.Semin@baikalelectronics.ru>

There is no point in having the commit 19b61392c5a8 ("spi: spi-dw: Add
lock protect dw_spi rx/tx to prevent concurrent calls") applied. The
commit author made an assumption that the problem with the rx data
mismatch was due to the lack of the data protection. While most likely it
was caused by the lack of the memory barrier. So having the
commit bfda044533b2 ("spi: dw: use "smp_mb()" to avoid sending spi data
error") applied would be enough to fix the problem.

Indeed the spin unlock operation makes sure each memory operation issued
before the release will be completed before it's completed. In other words
it works as an implicit one way memory barrier. So having both smp_mb()
and the spin_unlock_irqrestore() here is just redundant. One of them would
be enough. It's better to leave the smp_mb() since the Tx/Rx buffers
consistency is provided by the data transfer algorithm implementation:
first we initialize the buffers pointers, then make sure the assignments
are visible by the other CPUs by calling the smp_mb(), only after that
enable the interrupt, which handler uses the buffers.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>

---

Folks. I have also a doubt whether the SMP memory barrier is required there
because the normal IO-methods like readl/writel imply a full memory barrier. So
any memory operation performed before them are supposed to be seen by devices
and another CPUs [1]. So most likely there could have been a problem with those
IOs implementation on the subject platform or a spurious interrupt could
have been raised during the data initialization. What do you think?
Am I missing something?

[1] "LINUX KERNEL MEMORY BARRIERS", Documentation/memory-barriers.txt,
    Section "KERNEL I/O BARRIER EFFECTS"
---
 drivers/spi/spi-dw-core.c | 14 ++------------
 drivers/spi/spi-dw.h      |  1 -
 2 files changed, 2 insertions(+), 13 deletions(-)

diff --git a/drivers/spi/spi-dw-core.c b/drivers/spi/spi-dw-core.c
index 1af74362461d..18411f5b9954 100644
--- a/drivers/spi/spi-dw-core.c
+++ b/drivers/spi/spi-dw-core.c
@@ -142,11 +142,9 @@ static inline u32 rx_max(struct dw_spi *dws)
 
 static void dw_writer(struct dw_spi *dws)
 {
-	u32 max;
+	u32 max = tx_max(dws);
 	u16 txw = 0;
 
-	spin_lock(&dws->buf_lock);
-	max = tx_max(dws);
 	while (max--) {
 		/* Set the tx word if the transfer's original "tx" is not null */
 		if (dws->tx_end - dws->len) {
@@ -158,16 +156,13 @@ static void dw_writer(struct dw_spi *dws)
 		dw_write_io_reg(dws, DW_SPI_DR, txw);
 		dws->tx += dws->n_bytes;
 	}
-	spin_unlock(&dws->buf_lock);
 }
 
 static void dw_reader(struct dw_spi *dws)
 {
-	u32 max;
+	u32 max = rx_max(dws);
 	u16 rxw;
 
-	spin_lock(&dws->buf_lock);
-	max = rx_max(dws);
 	while (max--) {
 		rxw = dw_read_io_reg(dws, DW_SPI_DR);
 		/* Care rx only if the transfer's original "rx" is not null */
@@ -179,7 +174,6 @@ static void dw_reader(struct dw_spi *dws)
 		}
 		dws->rx += dws->n_bytes;
 	}
-	spin_unlock(&dws->buf_lock);
 }
 
 static void int_error_stop(struct dw_spi *dws, const char *msg)
@@ -291,21 +285,18 @@ static int dw_spi_transfer_one(struct spi_controller *master,
 {
 	struct dw_spi *dws = spi_controller_get_devdata(master);
 	struct chip_data *chip = spi_get_ctldata(spi);
-	unsigned long flags;
 	u8 imask = 0;
 	u16 txlevel = 0;
 	u32 cr0;
 	int ret;
 
 	dws->dma_mapped = 0;
-	spin_lock_irqsave(&dws->buf_lock, flags);
 	dws->n_bytes = DIV_ROUND_UP(transfer->bits_per_word, BITS_PER_BYTE);
 	dws->tx = (void *)transfer->tx_buf;
 	dws->tx_end = dws->tx + transfer->len;
 	dws->rx = transfer->rx_buf;
 	dws->rx_end = dws->rx + transfer->len;
 	dws->len = transfer->len;
-	spin_unlock_irqrestore(&dws->buf_lock, flags);
 
 	/* Ensure dw->rx and dw->rx_end are visible */
 	smp_mb();
@@ -464,7 +455,6 @@ int dw_spi_add_host(struct device *dev, struct dw_spi *dws)
 	dws->master = master;
 	dws->type = SSI_MOTO_SPI;
 	dws->dma_addr = (dma_addr_t)(dws->paddr + DW_SPI_DR);
-	spin_lock_init(&dws->buf_lock);
 
 	spi_controller_set_devdata(master, dws);
 
diff --git a/drivers/spi/spi-dw.h b/drivers/spi/spi-dw.h
index 51bab30b9f85..1ab704d1ebd8 100644
--- a/drivers/spi/spi-dw.h
+++ b/drivers/spi/spi-dw.h
@@ -131,7 +131,6 @@ struct dw_spi {
 	size_t			len;
 	void			*tx;
 	void			*tx_end;
-	spinlock_t		buf_lock;
 	void			*rx;
 	void			*rx_end;
 	int			dma_mapped;
-- 
2.27.0


  parent reply	other threads:[~2020-09-20 11:31 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-20 11:28 [PATCH 00/30] spi: dw: Add full Baikal-T1 SPI Controllers support Serge Semin
2020-09-20 11:28 ` [PATCH 01/30] spi: dw: Discard IRQ threshold macro Serge Semin
2020-09-20 11:28 ` [PATCH 02/30] spi: dw: Use ternary op to init set_cs callback Serge Semin
2020-09-29 13:11   ` Mark Brown
2020-09-29 21:55     ` Serge Semin
2020-09-30 14:57       ` Serge Semin
2020-09-30 15:01         ` Mark Brown
2020-09-30 15:07           ` Serge Semin
2020-09-20 11:28 ` [PATCH 03/30] spi: dw: Initialize n_bytes before the memory barrier Serge Semin
2020-09-29 13:12   ` Mark Brown
2020-09-29 22:05     ` Serge Semin
2020-09-20 11:28 ` Serge Semin [this message]
2020-09-29 13:28   ` [PATCH 04/30] Revert: spi: spi-dw: Add lock protect dw_spi rx/tx to prevent concurrent calls Mark Brown
2020-09-29 22:08     ` Serge Semin
2020-09-20 11:28 ` [PATCH 05/30] spi: dw: Clear IRQ status on DW SPI controller reset Serge Semin
2020-09-20 11:28 ` [PATCH 06/30] spi: dw: Disable all IRQs when controller is unused Serge Semin
2020-09-20 11:28 ` [PATCH 07/30] spi: dw: Use relaxed IO-methods to access FIFOs Serge Semin
2020-09-20 11:28 ` [PATCH 08/30] spi: dw: Discard DW SSI chip type storages Serge Semin
2020-09-20 11:28 ` [PATCH 09/30] spi: dw: Convert CS-override to DW SPI capabilities Serge Semin
2020-09-20 11:28 ` [PATCH 10/30] spi: dw: Add KeemBay Master capability Serge Semin
2020-09-20 11:28 ` [PATCH 11/30] spi: dw: Add DWC SSI capability Serge Semin
2020-09-29 13:52   ` Mark Brown
2020-09-29 22:17     ` Serge Semin
2020-09-30 15:03       ` Serge Semin
2020-09-30 15:41         ` Mark Brown
2020-09-30 15:53           ` Serge Semin
2020-09-20 11:28 ` [PATCH 12/30] spi: dw: Detach SPI device specific CR0 config method Serge Semin
2020-09-20 11:28 ` [PATCH 13/30] spi: dw: Update SPI bus speed in a config function Serge Semin
2020-09-20 11:28 ` [PATCH 14/30] spi: dw: Simplify the SPI bus speed config procedure Serge Semin
2020-09-20 11:28 ` [PATCH 15/30] spi: dw: Update Rx sample delay in the config function Serge Semin
2020-09-20 11:29 ` [PATCH 16/30] spi: dw: Add DW SPI controller config structure Serge Semin
2020-09-20 11:29 ` [PATCH 17/30] spi: dw: Refactor data IO procedure Serge Semin
2020-09-20 11:29 ` [PATCH 18/30] spi: dw: Refactor IRQ-based SPI transfer procedure Serge Semin
2020-09-20 11:29 ` [PATCH 19/30] spi: dw: Perform IRQ setup in a dedicated function Serge Semin
2020-09-20 11:29 ` [PATCH 20/30] spi: dw: Unmask IRQs after enabling the chip Serge Semin
2020-09-20 11:29 ` [PATCH 21/30] spi: dw: Discard chip enabling on DMA setup error Serge Semin
2020-09-20 11:29 ` [PATCH 22/30] spi: dw: De-assert chip-select on reset Serge Semin
2020-09-20 11:29 ` [PATCH 23/30] spi: dw: Explicitly de-assert CS on SPI transfer completion Serge Semin
2020-09-20 11:29 ` [PATCH 24/30] spi: dw: Move num-of retries parameter to the header file Serge Semin
2020-09-20 11:29 ` [PATCH 25/30] spi: dw: Add generic DW SSI status-check method Serge Semin
2020-09-20 11:29 ` [PATCH 26/30] spi: dw: Add memory operations support Serge Semin
2020-09-20 11:29 ` [PATCH 27/30] spi: dw: Introduce max mem-ops SPI bus frequency setting Serge Semin
2020-09-20 11:29 ` [PATCH 28/30] spi: dw: Add poll-based SPI transfers support Serge Semin
2020-09-20 11:29 ` [PATCH 29/30] dt-bindings: spi: dw: Add Baikal-T1 SPI Controllers Serge Semin
2020-09-29 16:58   ` Rob Herring
2020-09-20 11:41 ` [PATCH 30/30] spi: dw: Add Baikal-T1 SPI Controller glue driver Serge Semin
2020-09-29 14:43 ` [PATCH 00/30] spi: dw: Add full Baikal-T1 SPI Controllers support Mark Brown
2020-09-29 22:43   ` Serge Semin
2020-09-30 11:04     ` Mark Brown
2020-09-30 13:11       ` Serge Semin
2020-09-30 14:43         ` Mark Brown
2020-09-29 16:23 ` Mark Brown

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200920112914.26501-5-Sergey.Semin@baikalelectronics.ru \
    --to=sergey.semin@baikalelectronics.ru \
    --cc=Alexey.Malahov@baikalelectronics.ru \
    --cc=Pavel.Parkhomenko@baikalelectronics.ru \
    --cc=Ramil.Zaripov@baikalelectronics.ru \
    --cc=andriy.shevchenko@linux.intel.com \
    --cc=andy.shevchenko@gmail.com \
    --cc=broonie@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=fancer.lancer@gmail.com \
    --cc=feng.tang@intel.com \
    --cc=lars.povlsen@microchip.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-spi@vger.kernel.org \
    --cc=robh+dt@kernel.org \
    --cc=wuxu.wu@huawei.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.