Re: [PATCH 3/4] firmware: qcom_scm: Refactor code to support multiple download mode

From: Bjorn Andersson
Date: Thu Feb 23 2023 - 09:33:12 EST


On Tue, Feb 21, 2023 at 08:09:40PM +0530, Mukesh Ojha wrote:
> Currently on Qualcomm SoC, download_mode is enabled if
> CONFIG_QCOM_SCM_DOWNLOAD_MODE_DEFAULT is selected. Refactor
> the code such that it supports multiple download modes and
> drop the config.
>
> Signed-off-by: Mukesh Ojha <quic_mojha@xxxxxxxxxxx>
> ---
> drivers/firmware/Kconfig | 11 -----------
> drivers/firmware/qcom_scm.c | 45 +++++++++++++++++++++++++++++++++++++++------
> include/linux/qcom_scm.h | 4 ++++
> 3 files changed, 43 insertions(+), 17 deletions(-)
>
> diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
> index b59e304..ff7e9f3 100644
> --- a/drivers/firmware/Kconfig
> +++ b/drivers/firmware/Kconfig
> @@ -215,17 +215,6 @@ config MTK_ADSP_IPC
> config QCOM_SCM
> tristate
>
> -config QCOM_SCM_DOWNLOAD_MODE_DEFAULT
> - bool "Qualcomm download mode enabled by default"
> - depends on QCOM_SCM
> - help
> - A device with "download mode" enabled will upon an unexpected
> - warm-restart enter a special debug mode that allows the user to
> - "download" memory content over USB for offline postmortem analysis.
> - The feature can be enabled/disabled on the kernel command line.
> -
> - Say Y here to enable "download mode" by default.
> -
> config SYSFB
> bool
> select BOOT_VESA_SUPPORT
> diff --git a/drivers/firmware/qcom_scm.c b/drivers/firmware/qcom_scm.c
> index c376ba8..4975d3c 100644
> --- a/drivers/firmware/qcom_scm.c
> +++ b/drivers/firmware/qcom_scm.c
> @@ -20,8 +20,43 @@
>
> #include "qcom_scm.h"
>
> -static bool download_mode = IS_ENABLED(CONFIG_QCOM_SCM_DOWNLOAD_MODE_DEFAULT);
> -module_param(download_mode, bool, 0);
> +static unsigned int download_mode;
> +static struct qcom_scm *__scm;
> +static void qcom_scm_set_download_mode(bool enable);
> +static int set_dload_mode(const char *val, const struct kernel_param *kp)
> +{
> + int ret;
> + int old_mode = download_mode;
> +
> + ret = param_set_int(val, kp);
> + if (ret)
> + return ret;
> +
> + switch (download_mode) {
> + case QCOM_DOWNLOAD_FULLDUMP:
> + if (__scm)

If you, as I ask below, pass download_mode to
qcom_scm_set_download_mode(), you could replace this switch statement
with just:

if (__scm)
qcom_scm_set_download_mode(download_mode);

> + qcom_scm_set_download_mode(true);
> + break;
> + case QCOM_DOWNLOAD_NODUMP:
> + if (__scm)
> + qcom_scm_set_download_mode(false);
> + break;
> + default:

Please sanity check val before param_set_int() above instead.

> + pr_err("unknown download mode\n");
> + download_mode = old_mode;
> + return -EINVAL;
> + }
> +
> + return 0;
> +}
> +
> +static const struct kernel_param_ops dload_mode_param_ops = {
> + .set = set_dload_mode,
> + .get = param_get_hexint,
> +};
> +module_param_cb(download_mode, &dload_mode_param_ops, &download_mode, 0644);
> +MODULE_PARM_DESC(download_mode,
> + "Download mode: 0x0=no dump mode (default), 0x10=full dump mode");

The module parameters are an interface for humans, so rather than
forcing the operator to play with magical numbers; how about accepting
the values "off", "full" and "minidump"? (And accepting 0 and 1 for any
existing users).

>
> #define SCM_HAS_CORE_CLK BIT(0)
> #define SCM_HAS_IFACE_CLK BIT(1)
> @@ -70,8 +105,6 @@ static const char * const qcom_scm_convention_names[] = {
> [SMC_CONVENTION_LEGACY] = "smc legacy",
> };
>
> -static struct qcom_scm *__scm;
> -
> static int qcom_scm_clk_enable(void)
> {
> int ret;
> @@ -435,8 +468,8 @@ static void qcom_scm_set_download_mode(bool enable)
> }
>
> ret = qcom_scm_io_writel(__scm->dload_mode_addr, enable ?
> - dload_addr_val | QCOM_SCM_BOOT_SET_DLOAD_MODE :
> - dload_addr_val & ~(QCOM_SCM_BOOT_SET_DLOAD_MODE));
> + ((dload_addr_val & ~QCOM_DOWNLOAD_MODE_MASK) | download_mode) :

You're passing 2 pieces of information from set_dload_mode(), one by
argument and one through a global variable - and currently they carry
the same "information", in the next patch enable will be !!download_mode.

Please pass download_mode (u32 mode) as an argument to the function -
even if it's accessible in the global scope.

> + dload_addr_val & ~QCOM_DOWNLOAD_MODE_MASK);
> } else {
> dev_err(__scm->dev,
> "No available mechanism for setting download mode\n");
> diff --git a/include/linux/qcom_scm.h b/include/linux/qcom_scm.h
> index f833564..dd6aced 100644
> --- a/include/linux/qcom_scm.h
> +++ b/include/linux/qcom_scm.h
> @@ -14,6 +14,10 @@
> #define QCOM_SCM_CPU_PWR_DOWN_L2_OFF 0x1
> #define QCOM_SCM_HDCP_MAX_REQ_CNT 5
>
> +#define QCOM_DOWNLOAD_MODE_MASK 0x30
> +#define QCOM_DOWNLOAD_NODUMP 0x0
> +#define QCOM_DOWNLOAD_FULLDUMP 0x10

I don't see who the intended user of these constants are, please move
them into qcom_scm.c.

Thanks,
Bjorn

> +
> struct qcom_scm_hdcp_req {
> u32 addr;
> u32 val;
> --
> 2.7.4
>