| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Mon, 9 Sep 2019 11:35:26 +0530
From: Vignesh Raghavendra <vigneshr@...com>
To: "Ramuthevar,Vadivel MuruganX"
<vadivel.muruganx.ramuthevar@...ux.intel.com>,
<linux-mtd@...ts.infradead.org>
CC: <linux-kernel@...r.kernel.org>, <devicetree@...r.kernel.org>,
<dwmw2@...radead.org>, <computersforpeace@...il.com>,
<richard@....at>, <jwboyer@...il.com>,
<boris.brezillon@...e-electrons.com>, <cyrille.pitchen@...el.com>,
<david.oberhollenzer@...ma-star.at>, <miquel.raynal@...tlin.com>,
<tudor.ambarus@...il.com>, <andriy.shevchenko@...el.com>,
<cheol.yong.kim@...el.com>, <qi-ming.wu@...el.com>
Subject: Re: [PATCH v2 2/3] mtd: spi-nor: cadence-quadspi: disable DMA and DAC
for Intel LGM
Hi,
On 27/08/19 9:28 AM, Ramuthevar,Vadivel MuruganX wrote:
> From: Ramuthevar Vadivel Murugan <vadivel.muruganx.ramuthevar@...ux.intel.com>
>
> on Intel's Lightning Mountain(LGM) SoCs QSPI controller do not use
> Direct Memory Access(DMA) and Direct Access Controller(DAC).
>
> This patch introduces to properly disable the DMA and DAC
> for data transfer instead it uses indirect data transfer.
>
> Signed-off-by: Ramuthevar Vadivel Murugan <vadivel.muruganx.ramuthevar@...ux.intel.com>
> ---
> drivers/mtd/spi-nor/Kconfig | 2 +-
> drivers/mtd/spi-nor/cadence-quadspi.c | 21 ++++++++++++++++++---
> 2 files changed, 19 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
> index 6de83277ce8b..ba2e372ae514 100644
> --- a/drivers/mtd/spi-nor/Kconfig
> +++ b/drivers/mtd/spi-nor/Kconfig
> @@ -34,7 +34,7 @@ config SPI_ASPEED_SMC
>
> config SPI_CADENCE_QUADSPI
> tristate "Cadence Quad SPI controller"
> - depends on OF && (ARM || ARM64 || COMPILE_TEST)
> + depends on OF && (ARM || ARM64 || COMPILE_TEST || X86)
> help
> Enable support for the Cadence Quad SPI Flash controller.
>
> diff --git a/drivers/mtd/spi-nor/cadence-quadspi.c b/drivers/mtd/spi-nor/cadence-quadspi.c
> index 67f15a1f16fd..69fa13e95110 100644
> --- a/drivers/mtd/spi-nor/cadence-quadspi.c
> +++ b/drivers/mtd/spi-nor/cadence-quadspi.c
> @@ -517,12 +517,16 @@ static int cqspi_indirect_read_execute(struct spi_nor *nor, u8 *rxbuf,
> struct cqspi_st *cqspi = f_pdata->cqspi;
> void __iomem *reg_base = cqspi->iobase;
> void __iomem *ahb_base = cqspi->ahb_base;
> + u32 trigger_address = cqspi->trigger_address;
> unsigned int remaining = n_rx;
> unsigned int mod_bytes = n_rx % 4;
> unsigned int bytes_to_read = 0;
> u8 *rxbuf_end = rxbuf + n_rx;
> int ret = 0;
>
> + if (!f_pdata->use_direct_mode)
> + writel(trigger_address, reg_base + CQSPI_REG_INDIRECTTRIGGER);
> +
Again, as I pointed out previously, this should not be needed.
cqspi_controller_init() already does above configuration and no need to
touch this register on every read.
> writel(from_addr, reg_base + CQSPI_REG_INDIRECTRDSTARTADDR);
> writel(remaining, reg_base + CQSPI_REG_INDIRECTRDBYTES);
>
> @@ -609,6 +613,14 @@ static int cqspi_write_setup(struct spi_nor *nor)
> struct cqspi_st *cqspi = f_pdata->cqspi;
> void __iomem *reg_base = cqspi->iobase;
>
> + /* Disable the DMA and direct access controller */
> + if (!f_pdata->use_direct_mode) {
> + reg = readl(reg_base + CQSPI_REG_CONFIG);
> + reg &= ~CQSPI_REG_CONFIG_ENB_DIR_ACC_CTRL;
> + reg &= ~CQSPI_REG_CONFIG_DMA_MASK;
> + writel(reg, reg_base + CQSPI_REG_CONFIG);
> + }
> +
You did not respond to my previous comment. Why would you need to clear
CQSPI_REG_CONFIG_DMA_MASK field, if reset default is 0 anyway?
> /* Set opcode. */
> reg = nor->program_opcode << CQSPI_REG_WR_INSTR_OPCODE_LSB;
> writel(reg, reg_base + CQSPI_REG_WR_INSTR);
> @@ -1171,7 +1183,8 @@ static int cqspi_of_get_pdata(struct platform_device *pdev)
> return -ENXIO;
> }
>
> - cqspi->rclk_en = of_property_read_bool(np, "cdns,rclk-en");
> + if (!of_device_is_compatible(np, "intel,lgm-qspi"))
> + cqspi->rclk_en = of_property_read_bool(np, "cdns,rclk-en");
>
If you don't want to use this property, then just drop it from your DT.
Don't override it in the driver based on compatible.
> return 0;
> }
> @@ -1301,7 +1314,8 @@ static int cqspi_setup_flash(struct cqspi_st *cqspi, struct device_node *np)
> f_pdata->registered = true;
>
> if (mtd->size <= cqspi->ahb_size) {
> - f_pdata->use_direct_mode = true;
> + f_pdata->use_direct_mode =
> + !(of_device_is_compatible(np, "intel,lgm-qspi"));
Looks like, you haven't followed any of my advice. Please add a quirk
flag to disable DAC mode. Something like:
#define CQSPI_DISABLE_DAC_MODE BIT(1)
static const struct cqspi_driver_platdata intel_lgm_qspi = {
.hwcaps_mask = CQSPI_BASE_HWCAPS_MASK,
.quirks = CQSPI_DISABLE_DAC_MODE,
};
static const struct of_device_id cqspi_dt_ids[] = {
...
{
.compatible = "intel,lgm-qspi",
.data = &intel_lgm_qspi,
},
...
};
Then in cqspi_setup_flash(),
if (mtd->size <= cqspi->ahb_size &&
!ddata->quirks & CQSPI_DISABLE_DAC_MODE) {
f_pdata->use_direct_mode = true;
...
}
> dev_dbg(nor->dev, "using direct mode for %s\n",
> mtd->name);
>
> @@ -1347,7 +1361,7 @@ static int cqspi_probe(struct platform_device *pdev)
> }
>
> /* Obtain QSPI clock. */
> - cqspi->clk = devm_clk_get(dev, NULL);
> + cqspi->clk = devm_clk_get(dev, "qspi");
This will break DT backward compatibility. Existing DTs don't have
clock-names = "qspi". Hence, this code will error out.
What I meant in the previous mail was: if device does not have multiple
clk inputs then there is no need for clock-names and there is no need to
touch this part of code.
cqspi->clk = devm_clk_get(dev, NULL);
This should just work fine even on your board. So drop this hunk.
> if (IS_ERR(cqspi->clk)) {
> dev_err(dev, "Cannot claim QSPI clock.\n");
> return PTR_ERR(cqspi->clk);
> @@ -1369,6 +1383,7 @@ static int cqspi_probe(struct platform_device *pdev)
> return PTR_ERR(cqspi->ahb_base);
> }
> cqspi->mmap_phys_base = (dma_addr_t)res_ahb->start;
> + cqspi->trigger_address = res_ahb->start;
Nope, this is not needed. See:
https://elixir.bootlin.com/linux/v5.3-rc6/source/drivers/mtd/spi-nor/cadence-quadspi.c#L1168
Populate the trigger-address using cdns,trigger-address property in DT
> cqspi->ahb_size = resource_size(res_ahb);
>
> init_completion(&cqspi->transfer_complete);
>
--
Regards
Vignesh
Powered by blists - more mailing lists