linux-kernel - Re: [PATCH v2 4/4] spi: amd: Don't wait for a write-only transfer to finish

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 10 Sep 2021 12:42:23 +0000
From:   Charles Keepax <ckeepax@...nsource.cirrus.com>
To:     Lucas Tanure <tanureal@...nsource.cirrus.com>
CC:     Mark Brown <broonie@...nel.org>,
        Sanjay R Mehta <sanju.mehta@....com>,
        Nehal Bakulchandra Shah <Nehal-Bakulchandra.shah@....com>,
        <linux-kernel@...r.kernel.org>, <linux-spi@...r.kernel.org>,
        <patches@...nsource.cirrus.com>
Subject: Re: [PATCH v2 4/4] spi: amd: Don't wait for a write-only transfer to
 finish

On Fri, Sep 10, 2021 at 12:15:29PM +0100, Lucas Tanure wrote:
> Return from a write-only transfer without waiting for
> it to finish
> But wait before a new transfer as the previous may
> still happening and also wait before reading the data
> from the FIFO
> 
> Signed-off-by: Lucas Tanure <tanureal@...nsource.cirrus.com>
> ---
> -static void amd_spi_execute_opcode(struct amd_spi *amd_spi)
> +static int amd_spi_execute_opcode(struct amd_spi *amd_spi)
>  {
> +	int ret;
> +
> +	ret = amd_spi_busy_wait(amd_spi);
> +	if (ret)
> +		return ret;
> +
>  	/* Set ExecuteOpCode bit in the CTRL0 register */
>  	amd_spi_setclear_reg32(amd_spi, AMD_SPI_CTRL0_REG, AMD_SPI_EXEC_CMD, AMD_SPI_EXEC_CMD);
> -	amd_spi_busy_wait(amd_spi);
> +
> +	return 0;
>  }
>  
>  static int amd_spi_master_setup(struct spi_device *spi)
> @@ -178,6 +185,7 @@ static inline int amd_spi_fifo_xfer(struct amd_spi *amd_spi,
>  			amd_spi_clear_fifo_ptr(amd_spi);
>  			/* Execute command */
>  			amd_spi_execute_opcode(amd_spi);
> +			amd_spi_busy_wait(amd_spi);

Surely the previous transfer can't still be happening if this if
unconditional? Should this not be gated on rx_len?

Thanks,
Charles

>  			/* Read data from FIFO to receive buffer  */
>  			for (i = 0; i < rx_len; i++)
>  				buf[i] = amd_spi_readreg8(amd_spi, AMD_SPI_FIFO_BASE + tx_len + i);
> -- 
> 2.33.0
>