[Patch 2/2] update generic_defconfig to support sn2.

[Patch 2/2] update generic_defconfig to support sn2.

[Robin Holt]

This patch changes the generic_defconfig so it works on all sn2
platforms I have access to.  There is only one support configuration
which was not tested and that configuration is only a combination of two
tested configurations.  With this patchset applied, a generic kernel can
be booted on either a RHEL 5.2, RHEL5.3, or SLES10 SP1 root and operate.
All features needed by SGI's ProPack are also working.  I have not
tested all features of RHEL or SLES, but they do at least boot.

Signed-off-by: Robin Holt <holt@sgi.com>


Index: contention_unroll/arch/ia64/configs/generic_defconfig
=================================--- contention_unroll.orig/arch/ia64/configs/generic_defconfig	2008-08-02 13:16:56.135972118 -0500
+++ contention_unroll/arch/ia64/configs/generic_defconfig	2008-08-02 13:24:13.322305570 -0500
@@ -1,7 +1,7 @@
 #
 # Automatically generated make config: don't edit
 # Linux kernel version: 2.6.27-rc1
-# Fri Aug  1 19:33:14 2008
+# Sat Aug  2 13:24:08 2008
 #
 CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
 
@@ -23,10 +23,17 @@ CONFIG_POSIX_MQUEUE=y
 CONFIG_IKCONFIG=y
 CONFIG_IKCONFIG_PROC=y
 CONFIG_LOG_BUF_SHIFT 
-# CONFIG_CGROUPS is not set
+CONFIG_CGROUPS=y
+# CONFIG_CGROUP_DEBUG is not set
+# CONFIG_CGROUP_NS is not set
+# CONFIG_CGROUP_DEVICE is not set
+CONFIG_CPUSETS=y
 # CONFIG_GROUP_SCHED is not set
+# CONFIG_CGROUP_CPUACCT is not set
+# CONFIG_RESOURCE_COUNTERS is not set
 CONFIG_SYSFS_DEPRECATED=y
 CONFIG_SYSFS_DEPRECATED_V2=y
+CONFIG_PROC_PID_CPUSET=y
 # CONFIG_RELAY is not set
 CONFIG_NAMESPACES=y
 # CONFIG_UTS_NS is not set
@@ -130,6 +137,7 @@ CONFIG_DMI=y
 CONFIG_EFI=y
 CONFIG_GENERIC_IOMAP=y
 CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER=y
+CONFIG_IA64_UNCACHED_ALLOCATOR=y
 CONFIG_AUDIT_ARCH=y
 CONFIG_IA64_GENERIC=y
 # CONFIG_IA64_DIG is not set
@@ -158,7 +166,7 @@ CONFIG_IOSAPIC=y
 CONFIG_FORCE_MAX_ZONEORDER\x17
 # CONFIG_VIRT_CPU_ACCOUNTING is not set
 CONFIG_SMP=y
-CONFIG_NR_CPUSQ2
+CONFIG_NR_CPUS@96
 CONFIG_HOTPLUG_CPU=y
 CONFIG_ARCH_ENABLE_MEMORY_HOTPLUG=y
 CONFIG_ARCH_ENABLE_MEMORY_HOTREMOVE=y
@@ -369,7 +377,8 @@ CONFIG_EXTRA_FIRMWARE=""
 # CONFIG_DEBUG_DRIVER is not set
 # CONFIG_DEBUG_DEVRES is not set
 # CONFIG_SYS_HYPERVISOR is not set
-# CONFIG_CONNECTOR is not set
+CONFIG_CONNECTOR=y
+CONFIG_PROC_EVENTS=y
 # CONFIG_MTD is not set
 # CONFIG_PARPORT is not set
 CONFIG_PNP=y
@@ -403,7 +412,7 @@ CONFIG_MISC_DEVICES=y
 CONFIG_SGI_IOC4=y
 # CONFIG_TIFM_CORE is not set
 # CONFIG_ENCLOSURE_SERVICES is not set
-# CONFIG_SGI_XP is not set
+CONFIG_SGI_XP=m
 # CONFIG_HP_ILO is not set
 CONFIG_SGI_GRU=m
 # CONFIG_SGI_GRU_DEBUG is not set
@@ -1172,7 +1181,7 @@ CONFIG_INFINIBAND_IPOIB_DEBUG=y
 # CONFIG_RTC_CLASS is not set
 # CONFIG_DMADEVICES is not set
 # CONFIG_UIO is not set
-# CONFIG_MSPEC is not set
+CONFIG_MSPEC=m
 
 #
 # File systems
@@ -1375,6 +1384,7 @@ CONFIG_CRC_ITU_T=m
 CONFIG_CRC32=y
 # CONFIG_CRC7 is not set
 # CONFIG_LIBCRC32C is not set
+CONFIG_GENERIC_ALLOCATOR=y
 CONFIG_PLIST=y
 CONFIG_HAS_IOMEM=y
 CONFIG_HAS_IOPORT=y

RE: [Patch 2/2] update generic_defconfig to support sn2.

[Luck, Tony]

I've locally applied these config updates, but I ran into
a regression while running my usual set of test builds.

I build a few uniprocessor configs (by taking files from
arch/ia64/configs/ and turning off CONFIG_SMP).  The
resulting uniprocessor generic_defconfig fails to
build with:

drivers/misc/sgi-gru/grutables.h:308: error: field `ms_notifier' has incomplete type

I tried this:
diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
index 82af385..6e2dcd8 100644
--- a/drivers/misc/Kconfig
+++ b/drivers/misc/Kconfig
@@ -430,7 +430,7 @@ config SGI_XP
        depends on IA64_GENERIC || IA64_SGI_SN2 || IA64_SGI_UV || (X86_64 && SMP)
        select IA64_UNCACHED_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
        select GENERIC_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
-       select SGI_GRU if IA64_GENERIC || IA64_SGI_UV || (X86_64 && SMP)
+       select SGI_GRU if (IA64_GENERIC || IA64_SGI_UV || X86_64) && SMP
        ---help---
          An SGI machine can be divided into multiple Single System
          Images which act independently of each other and have

As an obvious fix (if SMP is required on X86, then surely it is
also required for other arches too).  But that just let the build
get a bit further until here:

  Building modules, stage 2.
  MODPOST 163 modules
ERROR: "gru_free_message" [drivers/misc/sgi-xp/xpc.ko] undefined!
ERROR: "gru_get_next_message" [drivers/misc/sgi-xp/xpc.ko] undefined!
ERROR: "gru_send_message_gpa" [drivers/misc/sgi-xp/xpc.ko] undefined!
ERROR: "gru_create_message_queue" [drivers/misc/sgi-xp/xpc.ko] undefined!
ERROR: "gru_copy_gpa" [drivers/misc/sgi-xp/xp.ko] undefined!

So there is obviously something else needed.

-Tony

Re: [Patch 2/2] update generic_defconfig to support sn2.

[Robin Holt]

On Mon, Aug 04, 2008 at 12:27:25PM -0700, Luck, Tony wrote:
> I've locally applied these config updates, but I ran into
> a regression while running my usual set of test builds.
> 
> I build a few uniprocessor configs (by taking files from
> arch/ia64/configs/ and turning off CONFIG_SMP).  The
> resulting uniprocessor generic_defconfig fails to
> build with:
> 
> drivers/misc/sgi-gru/grutables.h:308: error: field `ms_notifier' has incomplete type
> 
> I tried this:
> diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
> index 82af385..6e2dcd8 100644
> --- a/drivers/misc/Kconfig
> +++ b/drivers/misc/Kconfig
> @@ -430,7 +430,7 @@ config SGI_XP
>         depends on IA64_GENERIC || IA64_SGI_SN2 || IA64_SGI_UV || (X86_64 && SMP)
>         select IA64_UNCACHED_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
>         select GENERIC_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
> -       select SGI_GRU if IA64_GENERIC || IA64_SGI_UV || (X86_64 && SMP)
> +       select SGI_GRU if (IA64_GENERIC || IA64_SGI_UV || X86_64) && SMP

Wouldn't this allow a !SMP build to not select GRU?  That is not
correct.  I think this is more a depends on thing, but Dean/Jack are
more appropriate for that question.

Robin

Re: [Patch 2/2] update generic_defconfig to support sn2.

[Robin Holt]

On Mon, Aug 04, 2008 at 03:08:15PM -0500, Robin Holt wrote:
> On Mon, Aug 04, 2008 at 12:27:25PM -0700, Luck, Tony wrote:
> > I've locally applied these config updates, but I ran into
> > a regression while running my usual set of test builds.
> > 
> > I build a few uniprocessor configs (by taking files from
> > arch/ia64/configs/ and turning off CONFIG_SMP).  The
> > resulting uniprocessor generic_defconfig fails to
> > build with:
> > 
> > drivers/misc/sgi-gru/grutables.h:308: error: field `ms_notifier' has incomplete type
> > 
> > I tried this:
> > diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
> > index 82af385..6e2dcd8 100644
> > --- a/drivers/misc/Kconfig
> > +++ b/drivers/misc/Kconfig
> > @@ -430,7 +430,7 @@ config SGI_XP
> >         depends on IA64_GENERIC || IA64_SGI_SN2 || IA64_SGI_UV || (X86_64 && SMP)
> >         select IA64_UNCACHED_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
> >         select GENERIC_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
> > -       select SGI_GRU if IA64_GENERIC || IA64_SGI_UV || (X86_64 && SMP)
> > +       select SGI_GRU if (IA64_GENERIC || IA64_SGI_UV || X86_64) && SMP
> 
> Wouldn't this allow a !SMP build to not select GRU?  That is not
> correct.  I think this is more a depends on thing, but Dean/Jack are
> more appropriate for that question.

Dean and I just talked.  We think it should really be:

         depends on (IA64_GENERIC || IA64_SGI_SN2 || IA64_SGI_UV || X86_64) && SMP
         select IA64_UNCACHED_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
         select GENERIC_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
         select SGI_GRU if (IA64_GENERIC || IA64_SGI_UV || X86_64) && SMP

Could you give it a try?  If not, Dean or I can try it tomorrow.

Thanks,
Robin

RE: [Patch 2/2] update generic_defconfig to support sn2.

[Luck, Tony]

> Dean and I just talked.  We think it should really be:
>
>         depends on (IA64_GENERIC || IA64_SGI_SN2 || IA64_SGI_UV || X86_64) && SMP
>         select IA64_UNCACHED_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
>         select GENERIC_ALLOCATOR if IA64_GENERIC || IA64_SGI_SN2
>         select SGI_GRU if (IA64_GENERIC || IA64_SGI_UV || X86_64) && SMP
>
> Could you give it a try?  If not, Dean or I can try it tomorrow.

Yes.  That works (not sure why I didn't manage to spot what is
almost the same messed up expression just three lines above the
one that I did see).

I'll fix it like that. Thanks.

-Tony