Re: mmotm 2008-10-16-00-52 uploaded

Re: mmotm 2008-10-16-00-52 uploaded

[Randy.Dunlap]

On Thu, 16 Oct 2008, akpm@linux-foundation.org wrote:

> The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> 
>    http://userweb.kernel.org/~akpm/mmotm/
> 
> It contains the following patches against 2.6.27:


build-r9169.out:ksysfs.c:(.text+0x1a773): undefined reference to `create_proc_profile'

when CONFIG_PROC_FS=n and CONFIG_SYSFS=n

-- 
~Randy

Re: mmotm 2008-10-16-00-52 uploaded

[Alexey Dobriyan]

On Thu, Oct 16, 2008 at 01:38:13PM -0700, Randy.Dunlap wrote:
> On Thu, 16 Oct 2008, akpm@linux-foundation.org wrote:
> 
> > The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> > 
> >    http://userweb.kernel.org/~akpm/mmotm/
> > 
> > It contains the following patches against 2.6.27:
> 
> 
> build-r9169.out:ksysfs.c:(.text+0x1a773): undefined reference to `create_proc_profile'
> 
> when CONFIG_PROC_FS=n and CONFIG_SYSFS=n

ksysfs.c is compiled if SYSFS=n?

what's up?

Re: mmotm 2008-10-16-00-52 uploaded

[Randy.Dunlap]

On Fri, 17 Oct 2008, Alexey Dobriyan wrote:

> On Thu, Oct 16, 2008 at 01:38:13PM -0700, Randy.Dunlap wrote:
> > On Thu, 16 Oct 2008, akpm@linux-foundation.org wrote:
> > 
> > > The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> > > 
> > >    http://userweb.kernel.org/~akpm/mmotm/
> > > 
> > > It contains the following patches against 2.6.27:
> > 
> > 
> > build-r9169.out:ksysfs.c:(.text+0x1a773): undefined reference to `create_proc_profile'
> > 
> > when CONFIG_PROC_FS=n and CONFIG_SYSFS=n
> 
> ksysfs.c is compiled if SYSFS=n?
> 
> what's up?

Yes, that's what Makefile says...  odd.

-- 
~Randy

Re: mmotm 2008-10-16-00-52 uploaded (cgroup + mm)

[Randy.Dunlap]

On Thu, 16 Oct 2008, akpm@linux-foundation.org wrote:

> The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> 
>    http://userweb.kernel.org/~akpm/mmotm/
> 
> It contains the following patches against 2.6.27:


build-r9168.out:(.text+0x261e6): undefined reference to `lookup_page_cgroup'
build-r9168.out:memcontrol.c:(.text+0x2629f): undefined reference to `lookup_page_cgroup'
build-r9168.out:memcontrol.c:(.text+0x2671a): undefined reference to `lookup_page_cgroup'
build-r9168.out:(.text+0x268f9): undefined reference to `lookup_page_cgroup'
build-r9168.out:memcontrol.c:(.text+0x26e52): undefined reference to `page_cgroup_init'
build-r9168.out:(.text+0x26f44): undefined reference to `lookup_page_cgroup'
build-r9168.out:(.init.text+0xe42): undefined reference to `pgdat_page_cgroup_init'


.config is at http://oss.oracle.com/~rdunlap/kerneltest/configs/config-r9168

-- 
~Randy

Re: mmotm 2008-10-16-00-52 uploaded (HID)

[Randy.Dunlap]

On Thu, 16 Oct 2008, akpm@linux-foundation.org wrote:

> The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> 
>    http://userweb.kernel.org/~akpm/mmotm/
> 
> It contains the following patches against 2.6.27:

mmotm has several various build failures when
HID=y and USB=n
or
HID=y and USB_HID=m.

These are (mostly) caused by drivers/hid/Kconfig doing lots of "select"s
when EMBEDDED=n, such as

config HID
	tristate "Generic HID support"
	depends on INPUT
	default y
        select HID_A4TECH if !EMBEDDED
        select HID_APPLE if !EMBEDDED
        select HID_BELKIN if !EMBEDDED
        select HID_BRIGHT if !EMBEDDED

and HID_BRIGHT (just to pick one of them) says:

config HID_BRIGHT
	tristate "Bright"
	default m
	depends on USB_HID

but that "depends on USB_HID" isn't being "honored" because "select"
doesn't honor (or care) about dependencies.

I think that you already know about this, but drivers/hid/Kconfig
does need fixing.


Build failure examples:

9168:
HID=y, USB=n
hid-bright.c:(.text+0x1013f0): undefined reference to `usbhid_set_leds'
hid-dell.c:(.text+0x1019bc): undefined reference to `usbhid_set_leds'
hid-lg.c:(.text+0x102310): undefined reference to `usbhid_set_leds'
hid-lgff.c:(.text+0x102509): undefined reference to `usbhid_submit_report'
hid-lgff.c:(.text+0x102566): undefined reference to `usbhid_submit_report'
hid-pl.c:(.text+0x102bbb): undefined reference to `usbhid_submit_report'
hid-pl.c:(.text+0x102ddf): undefined reference to `usbhid_submit_report'
hid-sony.c:(.text+0x103339): undefined reference to `usb_control_msg'

9170:
HID=y, USB_HID=m
hid-bright.c:(.text+0xa9d2e): undefined reference to `usbhid_set_leds'
hid-dell.c:(.text+0xaa332): undefined reference to `usbhid_set_leds'
hid-lg.c:(.text+0xaa7e8): undefined reference to `usbhid_set_leds'
hid-pl.c:(.text+0xab669): undefined reference to `usbhid_submit_report'
hid-pl.c:(.text+0xab88d): undefined reference to `usbhid_submit_report'

9174:
HID=y, USB=n
hid-bright.c:(.text+0x1b4cf2): undefined reference to `usbhid_set_leds'
hid-dell.c:(.text+0x1b52e2): undefined reference to `usbhid_set_leds'
hid-lg.c:(.text+0x1b5790): undefined reference to `usbhid_set_leds'
hid-lg2ff.c:(.text+0x1b6077): undefined reference to `usbhid_submit_report'
(.text+0x1b61d2): undefined reference to `usbhid_submit_report'
hid-sony.c:(.text+0x1b6db7): undefined reference to `usb_control_msg'

-- 
~Randy

Re: mmotm 2008-10-16-00-52 uploaded (HID)

[Jiri Kosina]

On Thu, 16 Oct 2008, Randy.Dunlap wrote:

> > The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> >    http://userweb.kernel.org/~akpm/mmotm/
> > It contains the following patches against 2.6.27:
> mmotm has several various build failures when
> HID=y and USB=n
> or
> HID=y and USB_HID=m.
> These are (mostly) caused by drivers/hid/Kconfig doing lots of "select"s
> when EMBEDDED=n, such as
> config HID
> 	tristate "Generic HID support"
> 	depends on INPUT
> 	default y
>         select HID_A4TECH if !EMBEDDED
[ ... ]
> Build failure examples:
> 9168:
> HID=y, USB=n
> hid-bright.c:(.text+0x1013f0): undefined reference to `usbhid_set_leds'
> hid-dell.c:(.text+0x1019bc): undefined reference to `usbhid_set_leds'
> hid-lg.c:(.text+0x102310): undefined reference to `usbhid_set_leds'
> hid-lgff.c:(.text+0x102509): undefined reference to `usbhid_submit_report'
> hid-lgff.c:(.text+0x102566): undefined reference to `usbhid_submit_report'
> hid-pl.c:(.text+0x102bbb): undefined reference to `usbhid_submit_report'
> hid-pl.c:(.text+0x102ddf): undefined reference to `usbhid_submit_report'
> hid-sony.c:(.text+0x103339): undefined reference to `usb_control_msg'

Hi Randy,

yes, that's my bug, and it is already fixed in Linus' tree (commit 
b36299bcc0f).

Thanks,

-- 
Jiri Kosina
SUSE Labs

Re: mmotm 2008-10-16-00-52 uploaded (cgroup + mm)

[KAMEZAWA Hiroyuki]

On Thu, 16 Oct 2008 14:01:48 -0700 (PDT)
"Randy.Dunlap" <rdunlap@xenotime.net> wrote:

> On Thu, 16 Oct 2008, akpm@linux-foundation.org wrote:
> 
> > The mm-of-the-moment snapshot 2008-10-16-00-52 has been uploaded to
> > 
> >    http://userweb.kernel.org/~akpm/mmotm/
> > 
> > It contains the following patches against 2.6.27:
> 
> 
> build-r9168.out:(.text+0x261e6): undefined reference to `lookup_page_cgroup'
> build-r9168.out:memcontrol.c:(.text+0x2629f): undefined reference to `lookup_page_cgroup'
> build-r9168.out:memcontrol.c:(.text+0x2671a): undefined reference to `lookup_page_cgroup'
> build-r9168.out:(.text+0x268f9): undefined reference to `lookup_page_cgroup'
> build-r9168.out:memcontrol.c:(.text+0x26e52): undefined reference to `page_cgroup_init'
> build-r9168.out:(.text+0x26f44): undefined reference to `lookup_page_cgroup'
> build-r9168.out:(.init.text+0xe42): undefined reference to `pgdat_page_cgroup_init'
> 
> 
> .config is at http://oss.oracle.com/~rdunlap/kerneltest/configs/config-r9168
> 
Ouch...

Hmm....it seems

memcg-allocate-all-page_cgroup-at-boot.patch doesn't includes changes to Makefile...

Thank you for report. I'll send a fix soon.

Regards,
-Kame

[mmotm][PATCH] memcg-allocate-all-page_cgroup-at-boot-fix.patch

[KAMEZAWA Hiroyuki]

On Fri, 17 Oct 2008 09:30:46 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> Hmm....it seems
> 
> memcg-allocate-all-page_cgroup-at-boot.patch doesn't includes changes to Makefile...
> 
> Thank you for report. I'll send a fix soon.
> 
This is a fix. for this.

Confirmed vmlinux can be compiled with the config.
(but need to turn off CONFIG_HID_SUPPORT..and found small troube in /samples
 directory's Makefile.

-Kame
==

compile fix to memcg-allocate-all-page_cgroup-at-boot.patch


Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

 mm/Makefile |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6.27/mm/Makefile
===================================================================
--- linux-2.6.27.orig/mm/Makefile
+++ linux-2.6.27/mm/Makefile
@@ -33,5 +33,5 @@ obj-$(CONFIG_FS_XIP) += filemap_xip.o
 obj-$(CONFIG_MIGRATION) += migrate.o
 obj-$(CONFIG_SMP) += allocpercpu.o
 obj-$(CONFIG_QUICKLIST) += quicklist.o
-obj-$(CONFIG_CGROUP_MEM_RES_CTLR) += memcontrol.o
+obj-$(CONFIG_CGROUP_MEM_RES_CTLR) += memcontrol.o page_cgroup.o
 obj-$(CONFIG_CGROUP_MEMRLIMIT_CTLR) += memrlimitcgroup.o