current BK compilation failure on ppc32

current BK compilation failure on ppc32

[Christoph Hellwig]

kernel/power/smp.c: In function `smp_pause':
kernel/power/smp.c:24: error: storage size of `ctxt' isn't known
kernel/power/smp.c:24: warning: unused variable `ctxt'

kernel/power/smp.c seems to be inherently swsusp-specific but is
compiled for CONFIG_PM. (Same seems to be true for amny other files
in kernel/power/, but as they compile it only causes bloat..)


--- 1.10/kernel/power/Makefile	2004-07-02 07:23:47 +02:00
+++ edited/kernel/power/Makefile	2004-07-03 22:07:29 +02:00
@@ -1,5 +1,7 @@
 obj-y				:= main.o process.o console.o pm.o
+ifeq ($(CONFIG_SOFTWARE_SUSPEND), y)
 obj-$(CONFIG_SMP)		+= smp.o
+endif
 obj-$(CONFIG_SOFTWARE_SUSPEND)	+= swsusp.o
 obj-$(CONFIG_PM_DISK)		+= disk.o pmdisk.o
 

Re: current BK compilation failure on ppc32

[Pavel Machek]

Hi!

> kernel/power/smp.c: In function `smp_pause':
> kernel/power/smp.c:24: error: storage size of `ctxt' isn't known
> kernel/power/smp.c:24: warning: unused variable `ctxt'
> 
> kernel/power/smp.c seems to be inherently swsusp-specific but is
> compiled for CONFIG_PM. (Same seems to be true for amny other files
> in kernel/power/, but as they compile it only causes bloat..)

Patch is good, thanks and sorry for breakage.
								Pavel

> --- 1.10/kernel/power/Makefile	2004-07-02 07:23:47 +02:00
> +++ edited/kernel/power/Makefile	2004-07-03 22:07:29 +02:00
> @@ -1,5 +1,7 @@
>  obj-y				:= main.o process.o console.o pm.o
> +ifeq ($(CONFIG_SOFTWARE_SUSPEND), y)
>  obj-$(CONFIG_SMP)		+= smp.o
> +endif
>  obj-$(CONFIG_SOFTWARE_SUSPEND)	+= swsusp.o
>  obj-$(CONFIG_PM_DISK)		+= disk.o pmdisk.o
>  

-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!

Re: current BK compilation failure on ppc32

[Linus Torvalds]

On Sat, 3 Jul 2004, Christoph Hellwig wrote:
> 
> kernel/power/smp.c seems to be inherently swsusp-specific but is
> compiled for CONFIG_PM. (Same seems to be true for amny other files
> in kernel/power/, but as they compile it only causes bloat..)
> 
> 
> --- 1.10/kernel/power/Makefile	2004-07-02 07:23:47 +02:00
> +++ edited/kernel/power/Makefile	2004-07-03 22:07:29 +02:00
> @@ -1,5 +1,7 @@
>  obj-y				:= main.o process.o console.o pm.o
> +ifeq ($(CONFIG_SOFTWARE_SUSPEND), y)
>  obj-$(CONFIG_SMP)		+= smp.o
> +endif

Don't do it like that.

Instead, do something like

	smp-power-$(CONFIG_SMP)	+= smp.o
	obj-$(CONFIG_SOFTWARE_SUSPEND) += $(smp-power-y)

which not only is shorter, but gets a _lot_ more readable after a while.

It's also extremely useful for constructs like "include this file X is
either 'y' or 'm'". From fs/Makefile:

	..

	nfsd-$(CONFIG_NFSD)             := nfsctl.o
	obj-y                           += $(nfsd-y) $(nfsd-m)

	..

which just means that "nfsctl.o" will be compiled in if nfsd is 
compiled-in or a module.

You can make pretty complex decision trees this way - much more readably 
than by explicit comparisons.

		Linus

Re: current BK compilation failure on ppc32

[Christoph Hellwig]

> Don't do it like that.
> 
> Instead, do something like
> 
> 	smp-power-$(CONFIG_SMP)	+= smp.o
> 	obj-$(CONFIG_SOFTWARE_SUSPEND) += $(smp-power-y)
> 
> which not only is shorter, but gets a _lot_ more readable after a while.

New patch below:


--- 1.10/kernel/power/Makefile	2004-07-02 07:23:47 +02:00
+++ edited/kernel/power/Makefile	2004-07-04 12:21:57 +02:00
@@ -1,6 +1,8 @@
+
+swsusp-smp-$(CONFIG_SMP)	+= smp.o
+
 obj-y				:= main.o process.o console.o pm.o
-obj-$(CONFIG_SMP)		+= smp.o
-obj-$(CONFIG_SOFTWARE_SUSPEND)	+= swsusp.o
+obj-$(CONFIG_SOFTWARE_SUSPEND)	+= swsusp.o $(swsusp-smp-y)
 obj-$(CONFIG_PM_DISK)		+= disk.o pmdisk.o
 
 obj-$(CONFIG_MAGIC_SYSRQ)	+= poweroff.o