Re: mmotm 2008-09-22-01-36 uploaded (kvm)

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Randy Dunlap]

On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:

> The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> 
>    http://userweb.kernel.org/~akpm/mmotm/
> 
> It contains the following patches against 2.6.27-rc7:



ERROR: "intel_iommu_found" [arch/x86/kvm/kvm.ko] undefined!

when CONFIG_DMAR=n.

---
~Randy

Re: mmotm 2008-09-22-01-36 uploaded (profiling)

[Randy Dunlap]

On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:

> The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> 
>    http://userweb.kernel.org/~akpm/mmotm/
> 
> It contains the following patches against 2.6.27-rc7:

kernel/built-in.o: In function `profiling_store':
ksysfs.c:(.text+0x1b9b1): undefined reference to `create_proc_profile'
make[1]: *** [.tmp_vmlinux1] Error 1


when CONFIG_PROC_FS=n.


---
~Randy

Re: mmotm 2008-09-22-01-36 uploaded (memory_hotplug)

[Randy Dunlap]

On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:

> The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> 
>    http://userweb.kernel.org/~akpm/mmotm/
> 
> It contains the following patches against 2.6.27-rc7:


mm/built-in.o: In function `remove_memory':
(.text+0x497ed): multiple definition of `remove_memory'
arch/x86/mm/built-in.o:(.text+0x3d8): first defined here
make[1]: *** [vmlinux.o] Error 1


so it is defined in both mm/memory_hotplug.c and
arch/x86/mm/init_64.c when CONFIG_MEMORY_HOTREMOVE=y.

---
~Randy

Re: mmotm 2008-09-22-01-36 uploaded (profiling)

[Andrew Morton]

On Mon, 22 Sep 2008 11:19:56 -0700
Randy Dunlap <randy.dunlap@oracle.com> wrote:

> On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> 
> > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > 
> >    http://userweb.kernel.org/~akpm/mmotm/
> > 
> > It contains the following patches against 2.6.27-rc7:
> 
> kernel/built-in.o: In function `profiling_store':
> ksysfs.c:(.text+0x1b9b1): undefined reference to `create_proc_profile'
> make[1]: *** [.tmp_vmlinux1] Error 1
> 
> 
> when CONFIG_PROC_FS=n.
> 

That'll be Dave's profiling-dynamically-enable-readprofile-at-runtime.patch

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Andrew Morton]

On Mon, 22 Sep 2008 11:19:47 -0700
Randy Dunlap <randy.dunlap@oracle.com> wrote:

> On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> 
> > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > 
> >    http://userweb.kernel.org/~akpm/mmotm/
> > 
> > It contains the following patches against 2.6.27-rc7:
> 
> 
> 
> ERROR: "intel_iommu_found" [arch/x86/kvm/kvm.ko] undefined!
> 
> when CONFIG_DMAR=n.
> 

Yes, I got that too.  i386 allmodconfig fails similarly.  It's still
unclear which tree in linux-next broke it.  PCI or async_tx, perhaps.

Re: mmotm 2008-09-22-01-36 uploaded (memory_hotplug)

[Andrew Morton]

On Mon, 22 Sep 2008 11:29:18 -0700
Randy Dunlap <randy.dunlap@oracle.com> wrote:

> On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> 
> > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > 
> >    http://userweb.kernel.org/~akpm/mmotm/
> > 
> > It contains the following patches against 2.6.27-rc7:
> 
> 
> mm/built-in.o: In function `remove_memory':
> (.text+0x497ed): multiple definition of `remove_memory'
> arch/x86/mm/built-in.o:(.text+0x3d8): first defined here
> make[1]: *** [vmlinux.o] Error 1
> 
> 
> so it is defined in both mm/memory_hotplug.c and
> arch/x86/mm/init_64.c when CONFIG_MEMORY_HOTREMOVE=y.
> 

OK, thanks, it looks like people are changing things under our feet.

Badari, can you please check this fix against
mm-cleanup-to-make-remove_memory-arch-neutral.patch?



diff -puN arch/sh/mm/init.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix arch/sh/mm/init.c
--- a/arch/sh/mm/init.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix
+++ a/arch/sh/mm/init.c
@@ -293,20 +293,4 @@ int memory_add_physaddr_to_nid(u64 addr)
 EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid);
 #endif
 
-#ifdef CONFIG_MEMORY_HOTREMOVE
-int remove_memory(u64 start, u64 size)
-{
-	unsigned long start_pfn = start >> PAGE_SHIFT;
-	unsigned long end_pfn = start_pfn + (size >> PAGE_SHIFT);
-	int ret;
-
-	ret = offline_pages(start_pfn, end_pfn, 120 * HZ);
-	if (unlikely(ret))
-		printk("%s: Failed, offline_pages() == %d\n", __func__, ret);
-
-	return ret;
-}
-EXPORT_SYMBOL_GPL(remove_memory);
-#endif
-
 #endif /* CONFIG_MEMORY_HOTPLUG */
diff -puN arch/x86/mm/init_64.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix arch/x86/mm/init_64.c
--- a/arch/x86/mm/init_64.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix
+++ a/arch/x86/mm/init_64.c
@@ -796,24 +796,6 @@ int memory_add_physaddr_to_nid(u64 start
 EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid);
 #endif
 
-#ifdef CONFIG_MEMORY_HOTREMOVE
-int remove_memory(u64 start, u64 size)
-{
-	unsigned long start_pfn, end_pfn;
-	unsigned long timeout = 120 * HZ;
-	int ret;
-	start_pfn = start >> PAGE_SHIFT;
-	end_pfn = start_pfn + (size >> PAGE_SHIFT);
-	ret = offline_pages(start_pfn, end_pfn, timeout);
-	if (ret)
-		goto out;
-	/* Arch-specific calls go here */
-out:
-	return ret;
-}
-EXPORT_SYMBOL_GPL(remove_memory);
-#endif /* CONFIG_MEMORY_HOTREMOVE */
-
 #endif /* CONFIG_MEMORY_HOTPLUG */
 
 /*
_

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Jesse Barnes]

On Monday, September 22, 2008 12:52 pm Andrew Morton wrote:
> On Mon, 22 Sep 2008 11:19:47 -0700
>
> Randy Dunlap <randy.dunlap@oracle.com> wrote:
> > On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> > > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > >
> > >    http://userweb.kernel.org/~akpm/mmotm/
> > >
> > > It contains the following patches against 2.6.27-rc7:
> >
> > ERROR: "intel_iommu_found" [arch/x86/kvm/kvm.ko] undefined!
> >
> > when CONFIG_DMAR=n.
>
> Yes, I got that too.  i386 allmodconfig fails similarly.  It's still
> unclear which tree in linux-next broke it.  PCI or async_tx, perhaps.

I don't have that symbol in my tree (yet).  It's part of one of the IOMMU KVM 
patch sets that's currently in-flight.  I'll add it to my linux-next tree 
today or tomorrow, assuming I get an ack from David about it (I'll bounce it 
over to him now).

Thanks,
Jesse

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Jesse Barnes]

On Monday, September 22, 2008 2:58 pm Jesse Barnes wrote:
> On Monday, September 22, 2008 12:52 pm Andrew Morton wrote:
> > On Mon, 22 Sep 2008 11:19:47 -0700
> >
> > Randy Dunlap <randy.dunlap@oracle.com> wrote:
> > > On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> > > > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > > >
> > > >    http://userweb.kernel.org/~akpm/mmotm/
> > > >
> > > > It contains the following patches against 2.6.27-rc7:
> > >
> > > ERROR: "intel_iommu_found" [arch/x86/kvm/kvm.ko] undefined!
> > >
> > > when CONFIG_DMAR=n.
> >
> > Yes, I got that too.  i386 allmodconfig fails similarly.  It's still
> > unclear which tree in linux-next broke it.  PCI or async_tx, perhaps.
>
> I don't have that symbol in my tree (yet).  It's part of one of the IOMMU
> KVM patch sets that's currently in-flight.  I'll add it to my linux-next
> tree today or tomorrow, assuming I get an ack from David about it (I'll
> bounce it over to him now).

Oh nm, I already have an ack from Mark, I'll go ahead and push it to my 
linux-next branch.

Jesse

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Andrew Morton]

On Mon, 22 Sep 2008 14:58:52 -0700
Jesse Barnes <jbarnes@virtuousgeek.org> wrote:

> On Monday, September 22, 2008 12:52 pm Andrew Morton wrote:
> > On Mon, 22 Sep 2008 11:19:47 -0700
> >
> > Randy Dunlap <randy.dunlap@oracle.com> wrote:
> > > On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> > > > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > > >
> > > >    http://userweb.kernel.org/~akpm/mmotm/
> > > >
> > > > It contains the following patches against 2.6.27-rc7:
> > >
> > > ERROR: "intel_iommu_found" [arch/x86/kvm/kvm.ko] undefined!
> > >
> > > when CONFIG_DMAR=n.
> >
> > Yes, I got that too.  i386 allmodconfig fails similarly.  It's still
> > unclear which tree in linux-next broke it.  PCI or async_tx, perhaps.
> 
> I don't have that symbol in my tree (yet).  It's part of one of the IOMMU KVM 
> patch sets that's currently in-flight.  I'll add it to my linux-next tree 
> today or tomorrow, assuming I get an ack from David about it (I'll bounce it 
> over to him now).
> 

It's already there, and exported in linux-next.  But it just doesn't get
compiled in due to some Kconfig snafu.

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Amit Shah]

* On Tuesday 23 Sep 2008 03:40:26 Andrew Morton wrote:
> On Mon, 22 Sep 2008 14:58:52 -0700
>
> Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> > On Monday, September 22, 2008 12:52 pm Andrew Morton wrote:
> > > On Mon, 22 Sep 2008 11:19:47 -0700
> > >
> > > Randy Dunlap <randy.dunlap@oracle.com> wrote:
> > > > On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> > > > > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > > > >
> > > > >    http://userweb.kernel.org/~akpm/mmotm/
> > > > >
> > > > > It contains the following patches against 2.6.27-rc7:
> > > >
> > > > ERROR: "intel_iommu_found" [arch/x86/kvm/kvm.ko] undefined!
> > > >
> > > > when CONFIG_DMAR=n.
> > >
> > > Yes, I got that too.  i386 allmodconfig fails similarly.  It's still
> > > unclear which tree in linux-next broke it.  PCI or async_tx, perhaps.
> >
> > I don't have that symbol in my tree (yet).  It's part of one of the IOMMU
> > KVM patch sets that's currently in-flight.  I'll add it to my linux-next
> > tree today or tomorrow, assuming I get an ack from David about it (I'll
> > bounce it over to him now).
>
> It's already there, and exported in linux-next.  But it just doesn't get
> compiled in due to some Kconfig snafu.

This was introduced in the KVM tree.

Avi, please pick this up:

http://article.gmane.org/gmane.comp.emulators.kvm.devel/21770

or Mike Day's patches that fix the kernel as well as userspace build:

1/2: http://article.gmane.org/gmane.comp.emulators.kvm.devel/21813

2/2: http://article.gmane.org/gmane.comp.emulators.kvm.devel/21814

Re: mmotm 2008-09-22-01-36 uploaded (kvm)

[Avi Kivity]

Amit Shah wrote:
> Avi, please pick this up:
>
> http://article.gmane.org/gmane.comp.emulators.kvm.devel/21770
>
>   

I applied this, thanks.

-- 
error compiling committee.c: too many arguments to function

Re: mmotm 2008-09-22-01-36 uploaded (memory_hotplug)

[Badari Pulavarty]

On Mon, 2008-09-22 at 12:57 -0700, Andrew Morton wrote:
> On Mon, 22 Sep 2008 11:29:18 -0700
> Randy Dunlap <randy.dunlap@oracle.com> wrote:
> 
> > On Mon, 22 Sep 2008 01:38:58 -0700 akpm@linux-foundation.org wrote:
> > 
> > > The mm-of-the-moment snapshot 2008-09-22-01-36 has been uploaded to
> > > 
> > >    http://userweb.kernel.org/~akpm/mmotm/
> > > 
> > > It contains the following patches against 2.6.27-rc7:
> > 
> > 
> > mm/built-in.o: In function `remove_memory':
> > (.text+0x497ed): multiple definition of `remove_memory'
> > arch/x86/mm/built-in.o:(.text+0x3d8): first defined here
> > make[1]: *** [vmlinux.o] Error 1
> > 
> > 
> > so it is defined in both mm/memory_hotplug.c and
> > arch/x86/mm/init_64.c when CONFIG_MEMORY_HOTREMOVE=y.
> > 
> 
> OK, thanks, it looks like people are changing things under our feet.
> 
> Badari, can you please check this fix against
> mm-cleanup-to-make-remove_memory-arch-neutral.patch?
> 


When I made the patch, only ppc64, ia64 and s390 had
memory_remove() support in mainline. I sent a patch against
x86 to add hotplug memory remove support. I guess you
merged Gary's patch and sh-arch patch.

I noticed that you cleaned up all these and added to -mm.
Do you want me to merge all these into a single patch and 
resend it (against -mm) ? (It will look exactly the same
anyway).


Thanks,
Badari

> 
> diff -puN arch/sh/mm/init.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix arch/sh/mm/init.c
> --- a/arch/sh/mm/init.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix
> +++ a/arch/sh/mm/init.c
> @@ -293,20 +293,4 @@ int memory_add_physaddr_to_nid(u64 addr)
>  EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid);
>  #endif
> 
> -#ifdef CONFIG_MEMORY_HOTREMOVE
> -int remove_memory(u64 start, u64 size)
> -{
> -	unsigned long start_pfn = start >> PAGE_SHIFT;
> -	unsigned long end_pfn = start_pfn + (size >> PAGE_SHIFT);
> -	int ret;
> -
> -	ret = offline_pages(start_pfn, end_pfn, 120 * HZ);
> -	if (unlikely(ret))
> -		printk("%s: Failed, offline_pages() == %d\n", __func__, ret);
> -
> -	return ret;
> -}
> -EXPORT_SYMBOL_GPL(remove_memory);
> -#endif
> -
>  #endif /* CONFIG_MEMORY_HOTPLUG */
> diff -puN arch/x86/mm/init_64.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix arch/x86/mm/init_64.c
> --- a/arch/x86/mm/init_64.c~mm-cleanup-to-make-remove_memory-arch-neutral-fix-fix
> +++ a/arch/x86/mm/init_64.c
> @@ -796,24 +796,6 @@ int memory_add_physaddr_to_nid(u64 start
>  EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid);
>  #endif
> 
> -#ifdef CONFIG_MEMORY_HOTREMOVE
> -int remove_memory(u64 start, u64 size)
> -{
> -	unsigned long start_pfn, end_pfn;
> -	unsigned long timeout = 120 * HZ;
> -	int ret;
> -	start_pfn = start >> PAGE_SHIFT;
> -	end_pfn = start_pfn + (size >> PAGE_SHIFT);
> -	ret = offline_pages(start_pfn, end_pfn, timeout);
> -	if (ret)
> -		goto out;
> -	/* Arch-specific calls go here */
> -out:
> -	return ret;
> -}
> -EXPORT_SYMBOL_GPL(remove_memory);
> -#endif /* CONFIG_MEMORY_HOTREMOVE */
> -
>  #endif /* CONFIG_MEMORY_HOTPLUG */
> 
>  /*
> _
> 

Re: mmotm 2008-09-22-01-36 uploaded (memory_hotplug)

[Andrew Morton]

On Tue, 23 Sep 2008 15:08:22 -0700
Badari Pulavarty <pbadari@us.ibm.com> wrote:

> > > so it is defined in both mm/memory_hotplug.c and
> > > arch/x86/mm/init_64.c when CONFIG_MEMORY_HOTREMOVE=y.
> > > 
> > 
> > OK, thanks, it looks like people are changing things under our feet.
> > 
> > Badari, can you please check this fix against
> > mm-cleanup-to-make-remove_memory-arch-neutral.patch?
> > 
> 
> 
> When I made the patch, only ppc64, ia64 and s390 had
> memory_remove() support in mainline. I sent a patch against
> x86 to add hotplug memory remove support. I guess you
> merged Gary's patch and sh-arch patch.
> 
> I noticed that you cleaned up all these and added to -mm.
> Do you want me to merge all these into a single patch and 
> resend it (against -mm) ?

No, that's OK.

> (It will look exactly the same
> anyway).

Does that mean you reviewed all the fixes I added?  We're sure that all
the per-arch implementations of remove_memory() can be replaced by the
generic one?

Re: mmotm 2008-09-22-01-36 uploaded (memory_hotplug)

[Badari Pulavarty]

On Tue, 2008-09-23 at 15:17 -0700, Andrew Morton wrote:
> On Tue, 23 Sep 2008 15:08:22 -0700
> Badari Pulavarty <pbadari@us.ibm.com> wrote:
> 
> > > > so it is defined in both mm/memory_hotplug.c and
> > > > arch/x86/mm/init_64.c when CONFIG_MEMORY_HOTREMOVE=y.
> > > > 
> > > 
> > > OK, thanks, it looks like people are changing things under our feet.
> > > 
> > > Badari, can you please check this fix against
> > > mm-cleanup-to-make-remove_memory-arch-neutral.patch?
> > > 
> > 
> > 
> > When I made the patch, only ppc64, ia64 and s390 had
> > memory_remove() support in mainline. I sent a patch against
> > x86 to add hotplug memory remove support. I guess you
> > merged Gary's patch and sh-arch patch.
> > 
> > I noticed that you cleaned up all these and added to -mm.
> > Do you want me to merge all these into a single patch and 
> > resend it (against -mm) ?
> 
> No, that's OK.
> 
> > (It will look exactly the same
> > anyway).
> 
> Does that mean you reviewed all the fixes I added?  We're sure that all
> the per-arch implementations of remove_memory() can be replaced by the
> generic one?

Yes. I just reviewed all these patches in mmtom 2008-09-23. 
They looked fine. There is no need for arch-specific ones.

Thanks,
Badari