[PATCH v2] driver core: Add CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT

[PATCH v2] driver core: Add CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT

[Saravana Kannan]

Add a build time equivalent of fw_devlink.sync_state=timeout so that
board specific kernels could enable it and not have to deal with setting
or cluttering the kernel commandline.

Cc: Doug Anderson <dianders@chromium.org>
Signed-off-by: Saravana Kannan <saravanak@google.com>
---
 drivers/base/Kconfig | 12 ++++++++++++
 drivers/base/core.c  |  5 +++++
 2 files changed, 17 insertions(+)

diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
index 6f04b831a5c0..aac247512d69 100644
--- a/drivers/base/Kconfig
+++ b/drivers/base/Kconfig
@@ -230,4 +230,16 @@ config GENERIC_ARCH_NUMA
 	  Enable support for generic NUMA implementation. Currently, RISC-V
 	  and ARM64 use it.
 
+config FW_DEVLINK_SYNC_STATE_TIMEOUT
+	bool
+	help
+	  This is build time equivalent of adding kernel command line parameter
+	  "fw_devlink.sync_state=timeout". Give up waiting on consumers and
+	  call sync_state() on any devices that haven't yet received their
+	  sync_state() calls after deferred_probe_timeout has expired or by
+	  late_initcall() if !CONFIG_MODULES. You should almost always want to
+	  select N here unless you have already successfully tested with the
+	  command line option on every system/board your kernel is expected to
+	  work on.
+
 endmenu
diff --git a/drivers/base/core.c b/drivers/base/core.c
index fe74a786e2c3..adc81871829f 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -1672,7 +1672,12 @@ early_param("fw_devlink.strict", fw_devlink_strict_setup);
 #define FW_DEVLINK_SYNC_STATE_STRICT	0
 #define FW_DEVLINK_SYNC_STATE_TIMEOUT	1
 
+#ifndef CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT
 static int fw_devlink_sync_state;
+#else
+static int fw_devlink_sync_state = FW_DEVLINK_SYNC_STATE_TIMEOUT;
+#endif
+
 static int __init fw_devlink_sync_state_setup(char *arg)
 {
 	if (!arg)
-- 
2.40.0.rc1.284.g88254d51c5-goog

Re: [PATCH v2] driver core: Add CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT

[Greg Kroah-Hartman]

On Tue, Mar 14, 2023 at 06:07:31PM -0700, Saravana Kannan wrote:
> Add a build time equivalent of fw_devlink.sync_state=timeout so that
> board specific kernels could enable it and not have to deal with setting
> or cluttering the kernel commandline.
> 
> Cc: Doug Anderson <dianders@chromium.org>
> Signed-off-by: Saravana Kannan <saravanak@google.com>
> ---
>  drivers/base/Kconfig | 12 ++++++++++++
>  drivers/base/core.c  |  5 +++++
>  2 files changed, 17 insertions(+)
> 
> diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
> index 6f04b831a5c0..aac247512d69 100644
> --- a/drivers/base/Kconfig
> +++ b/drivers/base/Kconfig
> @@ -230,4 +230,16 @@ config GENERIC_ARCH_NUMA
>  	  Enable support for generic NUMA implementation. Currently, RISC-V
>  	  and ARM64 use it.
>  
> +config FW_DEVLINK_SYNC_STATE_TIMEOUT
> +	bool
> +	help
> +	  This is build time equivalent of adding kernel command line parameter
> +	  "fw_devlink.sync_state=timeout". Give up waiting on consumers and
> +	  call sync_state() on any devices that haven't yet received their
> +	  sync_state() calls after deferred_probe_timeout has expired or by
> +	  late_initcall() if !CONFIG_MODULES. You should almost always want to
> +	  select N here unless you have already successfully tested with the
> +	  command line option on every system/board your kernel is expected to
> +	  work on.

As nothing can actually select this, it doesn't make sense to add this
now, right?  We need a user, otherwise the automated tools will come
along and remove this option when they figure out that it can't be ever
used.

thanks,

greg k-h

Re: [PATCH v2] driver core: Add CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT

[Doug Anderson]

Hi,

On Fri, Mar 17, 2023 at 7:12 AM Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
>
> On Tue, Mar 14, 2023 at 06:07:31PM -0700, Saravana Kannan wrote:
> > Add a build time equivalent of fw_devlink.sync_state=timeout so that
> > board specific kernels could enable it and not have to deal with setting
> > or cluttering the kernel commandline.
> >
> > Cc: Doug Anderson <dianders@chromium.org>
> > Signed-off-by: Saravana Kannan <saravanak@google.com>
> > ---
> >  drivers/base/Kconfig | 12 ++++++++++++
> >  drivers/base/core.c  |  5 +++++
> >  2 files changed, 17 insertions(+)
> >
> > diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
> > index 6f04b831a5c0..aac247512d69 100644
> > --- a/drivers/base/Kconfig
> > +++ b/drivers/base/Kconfig
> > @@ -230,4 +230,16 @@ config GENERIC_ARCH_NUMA
> >         Enable support for generic NUMA implementation. Currently, RISC-V
> >         and ARM64 use it.
> >
> > +config FW_DEVLINK_SYNC_STATE_TIMEOUT
> > +     bool
> > +     help
> > +       This is build time equivalent of adding kernel command line parameter
> > +       "fw_devlink.sync_state=timeout". Give up waiting on consumers and
> > +       call sync_state() on any devices that haven't yet received their
> > +       sync_state() calls after deferred_probe_timeout has expired or by
> > +       late_initcall() if !CONFIG_MODULES. You should almost always want to
> > +       select N here unless you have already successfully tested with the
> > +       command line option on every system/board your kernel is expected to
> > +       work on.
>
> As nothing can actually select this, it doesn't make sense to add this
> now, right?  We need a user, otherwise the automated tools will come
> along and remove this option when they figure out that it can't be ever
> used.

I think the hope was that it could be setup as a config that would
show up as user selectable (like in menuconfig). Saravana: the problem
is your bare "bool" above. That should be changed to a description
shown to the user. Like maybe you'd want to change that to:

  bool "sync_state behavior defaults to timeout instead of strict"

-Doug

Re: [PATCH v2] driver core: Add CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT

[Saravana Kannan]

On Fri, Mar 17, 2023 at 9:57 AM Doug Anderson <dianders@chromium.org> wrote:
>
> Hi,
>
> On Fri, Mar 17, 2023 at 7:12 AM Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
> >
> > On Tue, Mar 14, 2023 at 06:07:31PM -0700, Saravana Kannan wrote:
> > > Add a build time equivalent of fw_devlink.sync_state=timeout so that
> > > board specific kernels could enable it and not have to deal with setting
> > > or cluttering the kernel commandline.
> > >
> > > Cc: Doug Anderson <dianders@chromium.org>
> > > Signed-off-by: Saravana Kannan <saravanak@google.com>
> > > ---
> > >  drivers/base/Kconfig | 12 ++++++++++++
> > >  drivers/base/core.c  |  5 +++++
> > >  2 files changed, 17 insertions(+)
> > >
> > > diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
> > > index 6f04b831a5c0..aac247512d69 100644
> > > --- a/drivers/base/Kconfig
> > > +++ b/drivers/base/Kconfig
> > > @@ -230,4 +230,16 @@ config GENERIC_ARCH_NUMA
> > >         Enable support for generic NUMA implementation. Currently, RISC-V
> > >         and ARM64 use it.
> > >
> > > +config FW_DEVLINK_SYNC_STATE_TIMEOUT
> > > +     bool
> > > +     help
> > > +       This is build time equivalent of adding kernel command line parameter
> > > +       "fw_devlink.sync_state=timeout". Give up waiting on consumers and
> > > +       call sync_state() on any devices that haven't yet received their
> > > +       sync_state() calls after deferred_probe_timeout has expired or by
> > > +       late_initcall() if !CONFIG_MODULES. You should almost always want to
> > > +       select N here unless you have already successfully tested with the
> > > +       command line option on every system/board your kernel is expected to
> > > +       work on.
> >
> > As nothing can actually select this, it doesn't make sense to add this
> > now, right?  We need a user, otherwise the automated tools will come
> > along and remove this option when they figure out that it can't be ever
> > used.
>
> I think the hope was that it could be setup as a config that would
> show up as user selectable (like in menuconfig). Saravana: the problem
> is your bare "bool" above. That should be changed to a description
> shown to the user. Like maybe you'd want to change that to:
>
>   bool "sync_state behavior defaults to timeout instead of strict"

Ah, thanks. Will do this.

-Saravana