| Message ID | 1392573811.28866.22.camel@x220 (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
???????????, 16 ??????? 2014, 19:03 +01:00 ?? Paul Bolle <pebolle@tiscali.nl>: > The Kconfig symbol MACH_MXLADS was added in v2.6.29. It has never been > used. Setting it has no effect. There are no calls for > machine_is_mxlads(). This symbol can safely be removed. > > Signed-off-by: Paul Bolle <pebolle@tiscali.nl> This is used by mach-mx1ads.c ---
On Sun, 2014-02-16 at 22:16 +0400, Alexander Shiyan wrote: > ???????????, 16 ??????? 2014, 19:03 +01:00 ?? Paul Bolle <pebolle@tiscali.nl>: > > The Kconfig symbol MACH_MXLADS was added in v2.6.29. It has never been > > used. Setting it has no effect. There are no calls for > > machine_is_mxlads(). This symbol can safely be removed. > > > > Signed-off-by: Paul Bolle <pebolle@tiscali.nl> > > This is used by mach-mx1ads.c How is it used? Paul Bolle
???????????, 16 ??????? 2014, 19:43 +01:00 ?? Paul Bolle <pebolle@tiscali.nl>: > On Sun, 2014-02-16 at 22:16 +0400, Alexander Shiyan wrote: > > ???????????, 16 ??????? 2014, 19:03 +01:00 ?? Paul Bolle > <pebolle@tiscali.nl>: > > > The Kconfig symbol MACH_MXLADS was added in v2.6.29. It has never been > > > used. Setting it has no effect. There are no calls for > > > machine_is_mxlads(). This symbol can safely be removed. > > > > > > Signed-off-by: Paul Bolle <pebolle@tiscali.nl> > > > > This is used by mach-mx1ads.c > > How is it used? Kconfig selects MACH_MXLADS, this enables machine_arch_type from include/generated/mach-types.h which used for machine_is_xx() macro, so if multiple boards is defined in the kernel, this cause to incorrect matching. Russel, Arnd, fixme. ---
On Sun, 2014-02-16 at 23:06 +0400, Alexander Shiyan wrote: > Kconfig selects MACH_MXLADS, this enables machine_arch_type > from include/generated/mach-types.h which used for machine_is_xx() macro, > so if multiple boards is defined in the kernel, this cause to incorrect matching. But, as I stated in the commit explanation, there are no users of machine_is_mxlads(). > Russel, Arnd, fixme. What does that mean? Paul Bolle
???????????, 16 ??????? 2014, 20:12 +01:00 ?? Paul Bolle <pebolle@tiscali.nl>: > On Sun, 2014-02-16 at 23:06 +0400, Alexander Shiyan wrote: > > Kconfig selects MACH_MXLADS, this enables machine_arch_type > > from include/generated/mach-types.h which used for machine_is_xx() macro, > > so if multiple boards is defined in the kernel, this cause to incorrect > matching. > > But, as I stated in the commit explanation, there are no users of > machine_is_mxlads(). If you have MXLADS and FOO boards support in the kernel (for example) and you want to start MXLADS version, your machine_is_FOO() will be positive, since machine_arch_type will be constant value FOO in this case. ---
On Sun, 2014-02-16 at 23:21 +0400, Alexander Shiyan wrote: > ???????????, 16 ??????? 2014, 20:12 +01:00 ?? Paul Bolle <pebolle@tiscali.nl>: > > On Sun, 2014-02-16 at 23:06 +0400, Alexander Shiyan wrote: > > > Kconfig selects MACH_MXLADS, this enables machine_arch_type > > > from include/generated/mach-types.h which used for machine_is_xx() macro, > > > so if multiple boards is defined in the kernel, this cause to incorrect > > matching. > > > > But, as I stated in the commit explanation, there are no users of > > machine_is_mxlads(). > > If you have MXLADS and FOO boards support in the kernel (for example) > and you want to start MXLADS version, your machine_is_FOO() will be positive, > since machine_arch_type will be constant value FOO in this case. I think it might be helpful - for people trying to understand this thread - to add a quote from https://lkml.org/lkml/2013/3/8/473 : So, the result is that: - de-selected platforms have their if (machine_is_xxx()) { } optimised out of the kernel. - for a kernel built targetting one platform, all the if (machine_is_xxx()) tests are optimised away, leaving only the relevant code behind. - otherwise, we get the _appropriate_ conditional code for the configuration generated. Back to the issue at hand: without my patch ARCH_MX1ADS selects MACH_MXLADS. So it seems in this case people will have to select ARCH_MX1ADS besides selecting FOO. Wouldn't that solve this problem? Note that the MACH_MXLADS is now basically meant to be selected by ARCH_MX1ADS and not set directly anyway. Paul Bolle
???????????, 16 ??????? 2014, 21:35 +01:00 ?? Paul Bolle <pebolle@tiscali.nl>: > On Sun, 2014-02-16 at 23:21 +0400, Alexander Shiyan wrote: > > ???????????, 16 ??????? 2014, 20:12 +01:00 ?? Paul Bolle > <pebolle@tiscali.nl>: > > > On Sun, 2014-02-16 at 23:06 +0400, Alexander Shiyan wrote: > > > > Kconfig selects MACH_MXLADS, this enables machine_arch_type > > > > from include/generated/mach-types.h which used for machine_is_xx() > macro, > > > > so if multiple boards is defined in the kernel, this cause to incorrect > > > matching. > > > > > > But, as I stated in the commit explanation, there are no users of > > > machine_is_mxlads(). > > > > If you have MXLADS and FOO boards support in the kernel (for example) > > and you want to start MXLADS version, your machine_is_FOO() will be > positive, > > since machine_arch_type will be constant value FOO in this case. > > I think it might be helpful - for people trying to understand this > thread - to add a quote from https://lkml.org/lkml/2013/3/8/473 : > So, the result is that: > - de-selected platforms have their if (machine_is_xxx()) { } optimised > out of the kernel. > - for a kernel built targetting one platform, all the > if (machine_is_xxx()) tests are optimised away, leaving only the > relevant code behind. > - otherwise, we get the _appropriate_ conditional code for the > configuration generated. > > Back to the issue at hand: without my patch ARCH_MX1ADS selects > MACH_MXLADS. So it seems in this case people will have to select > ARCH_MX1ADS besides selecting FOO. Wouldn't that solve this problem? > Note that the MACH_MXLADS is now basically meant to be selected by > ARCH_MX1ADS and not set directly anyway. It seems that even in the current state, Kconfig for i.MX contains an error. As a start, we should rename ARCH_MX1ADS to MACH_MX1ADS first. ---
On Mon, 2014-02-17 at 08:43 +0400, Alexander Shiyan wrote: > It seems that even in the current state, Kconfig for i.MX contains an error. > As a start, we should rename ARCH_MX1ADS to MACH_MX1ADS first. Last year I admitted that I didn't entirely grasped the differences between "arch", "platform", and "machine" in the arch/arm universe[1]. Perhaps someone could help me understand why it's better to use MACH_MX1ADS here. Paul Bolle [1] http://driverdev.linuxdriverproject.org/pipermail/driverdev-devel/2013-May/038204.html
On Mon, Feb 17, 2014 at 10:02:48AM +0100, Paul Bolle wrote: > On Mon, 2014-02-17 at 08:43 +0400, Alexander Shiyan wrote: > > It seems that even in the current state, Kconfig for i.MX contains an error. > > As a start, we should rename ARCH_MX1ADS to MACH_MX1ADS first. No. Have a look at arch/arm/tools/mach-types: mxlads MACH_MXLADS MXLADS 1851 mx1ads ARCH_MX1ADS MX1ADS 160 And what gen-mach-types makes from it: #ifdef CONFIG_MACH_MXLADS # ifdef machine_arch_type # undef machine_arch_type # define machine_arch_type __machine_arch_type # else # define machine_arch_type MACH_TYPE_MXLADS # endif # define machine_is_mxlads() (machine_arch_type == MACH_TYPE_MXLADS) #else # define machine_is_mxlads() (0) #endif #ifdef CONFIG_ARCH_MX1ADS # ifdef machine_arch_type # undef machine_arch_type # define machine_arch_type __machine_arch_type # else # define machine_arch_type MACH_TYPE_MX1ADS # endif # define machine_is_mx1ads() (machine_arch_type == MACH_TYPE_MX1ADS) #else # define machine_is_mx1ads() (0) #endif So machine_is_mx1ads() can only ever evaluate to true when CONFIG_ARCH_MX1ADS is set and machine_is_mxlads() can only evaluate to true when CONFIG_MACH_MXLADS is set. Whether to use ARCH or MACH simply depends on what Russell has in the machine registry database for a given board. Sascha
diff --git a/arch/arm/mach-imx/Kconfig b/arch/arm/mach-imx/Kconfig index 33567aa..f564be8 100644 --- a/arch/arm/mach-imx/Kconfig +++ b/arch/arm/mach-imx/Kconfig @@ -144,14 +144,10 @@ config SOC_IMX51 if ARCH_MULTI_V4T comment "MX1 platforms:" -config MACH_MXLADS - bool - config ARCH_MX1ADS bool "MX1ADS platform" select IMX_HAVE_PLATFORM_IMX_I2C select IMX_HAVE_PLATFORM_IMX_UART - select MACH_MXLADS select SOC_IMX1 help Say Y here if you are using Motorola MX1ADS/MXLADS boards
The Kconfig symbol MACH_MXLADS was added in v2.6.29. It has never been used. Setting it has no effect. There are no calls for machine_is_mxlads(). This symbol can safely be removed. Signed-off-by: Paul Bolle <pebolle@tiscali.nl> --- Tested with "git grep". arch/arm/mach-imx/Kconfig | 4 ---- 1 file changed, 4 deletions(-)