swiotlb-xen tree + xen + linux-next

swiotlb-xen tree + xen + linux-next

[Konrad Rzeszutek Wilk]

Hey Stephen,

Jeremy and I fixed the mega failures you saw and last night Jeremy's
got pulled in while mine was silent. I am enabling my tree today, and while
the previous failure is gone when I do a merge of my tree against linux-next,
I keep on getting one that I am having trouble solving.

The issues is that my patchset touches drivers/pci/Makefile which
had been changed a bit by a): 350a55e9ff6005032407d3234af800f413b03af5
(PCI: use new ccflags variable in Makefile) which does this:
@@ -65,6 +65,4 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
 
 obj-$(CONFIG_PCI_STUB) += pci-stub.o
 
-ifeq ($(CONFIG_PCI_DEBUG),y)
-EXTRA_CFLAGS += -DDEBUG
-endif
+ccflags-$(CONFIG_PCI_DEBUG) := -DDEBUG

And then b) my patch (git commit 956a9202cd1220397933a07beda9f96b3df1fa24
xen-pcifront: Xen PCI frontend driver.) goes in and tries to add:

@@ -65,6 +65,8 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
 
 obj-$(CONFIG_PCI_STUB) += pci-stub.o
 
+obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
+
 ifeq ($(CONFIG_PCI_DEBUG),y)
 EXTRA_CFLAGS += -DDEBUG
 endif

I've tried to put a) patch in the branch and rebase my tree on top of that
and this way, the "ifeq ($(CONFIG_PCI_DEBUG_...)" is gone, but merge still
complains! The error is:

diff --cc drivers/pci/Makefile
index dcd7ace,39c7327..0000000
--- a/drivers/pci/Makefile
+++ b/drivers/pci/Makefile
@@@ -65,4 -65,7 +65,10 @@@ obj-$(CONFIG_PCI_SYSCALL) += syscall.
  
  obj-$(CONFIG_PCI_STUB) += pci-stub.o
  
++<<<<<<< HEAD
++=======
+ # Xen PCI frontend driver
+ obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
+ 
++>>>>>>> stable/xen-pcifront-0.8.3
  ccflags-$(CONFIG_PCI_DEBUG) := -DDEBUG

[That branch with the a) patch is stable/xen-pcifront-0.8.3, which differs
from #linux-next (which is linked to stable/xen-pcifront-0.8.2) by just
having the patch a) in it. I am not going to use that for #linux-next as
it does not seem to make merging automatic].

Any ideas why it is doing this?

In the internim and so that the merge tonight won't be broken, I've
attached a merge conflict patch instead:

diff --git a/drivers/pci/Makefile b/drivers/pci/Makefile
index dcd7ace..f01e344 100644
--- a/drivers/pci/Makefile
+++ b/drivers/pci/Makefile
@@ -65,4 +65,6 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
 
 obj-$(CONFIG_PCI_STUB) += pci-stub.o
 
+obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
+
 ccflags-$(CONFIG_PCI_DEBUG) := -DDEBUG

Re: swiotlb-xen tree + xen + linux-next

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 3329 bytes --]

Hi Konrad,

On Tue, 19 Oct 2010 10:25:39 -0400 Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> wrote:
>
> Jeremy and I fixed the mega failures you saw and last night Jeremy's
> got pulled in while mine was silent. I am enabling my tree today, and while
> the previous failure is gone when I do a merge of my tree against linux-next,
> I keep on getting one that I am having trouble solving.

Thanks for fixing this all up.

> The issues is that my patchset touches drivers/pci/Makefile which
> had been changed a bit by a): 350a55e9ff6005032407d3234af800f413b03af5
> (PCI: use new ccflags variable in Makefile) which does this:
> @@ -65,6 +65,4 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
>  
>  obj-$(CONFIG_PCI_STUB) += pci-stub.o
>  
> -ifeq ($(CONFIG_PCI_DEBUG),y)
> -EXTRA_CFLAGS += -DDEBUG
> -endif
> +ccflags-$(CONFIG_PCI_DEBUG) := -DDEBUG
> 
> And then b) my patch (git commit 956a9202cd1220397933a07beda9f96b3df1fa24
> xen-pcifront: Xen PCI frontend driver.) goes in and tries to add:
> 
> @@ -65,6 +65,8 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
>  
>  obj-$(CONFIG_PCI_STUB) += pci-stub.o
>  
> +obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
> +
>  ifeq ($(CONFIG_PCI_DEBUG),y)
>  EXTRA_CFLAGS += -DDEBUG
>  endif
> 
> I've tried to put a) patch in the branch and rebase my tree on top of that
> and this way, the "ifeq ($(CONFIG_PCI_DEBUG_...)" is gone, but merge still
> complains! The error is:
> 
> diff --cc drivers/pci/Makefile
> index dcd7ace,39c7327..0000000
> --- a/drivers/pci/Makefile
> +++ b/drivers/pci/Makefile
> @@@ -65,4 -65,7 +65,10 @@@ obj-$(CONFIG_PCI_SYSCALL) += syscall.
>   
>   obj-$(CONFIG_PCI_STUB) += pci-stub.o
>   
> ++<<<<<<< HEAD
> ++=======
> + # Xen PCI frontend driver
> + obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
> + 
> ++>>>>>>> stable/xen-pcifront-0.8.3
>   ccflags-$(CONFIG_PCI_DEBUG) := -DDEBUG
> 
> [That branch with the a) patch is stable/xen-pcifront-0.8.3, which differs
> from #linux-next (which is linked to stable/xen-pcifront-0.8.2) by just
> having the patch a) in it. I am not going to use that for #linux-next as
> it does not seem to make merging automatic].
> 
> Any ideas why it is doing this?

I think that if git sees this as the "ccflags" change on one side of the
merge and "ccflags" change plus your change on the other and decides that
it conflicts.  It considers the two "ccflags" changes to be different
because they are not the same commit (even if they have the same patch).

> In the internim and so that the merge tonight won't be broken, I've
> attached a merge conflict patch instead:
> 
> diff --git a/drivers/pci/Makefile b/drivers/pci/Makefile
> index dcd7ace..f01e344 100644
> --- a/drivers/pci/Makefile
> +++ b/drivers/pci/Makefile
> @@ -65,4 +65,6 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
>  
>  obj-$(CONFIG_PCI_STUB) += pci-stub.o
>  
> +obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
> +
>  ccflags-$(CONFIG_PCI_DEBUG) := -DDEBUG
> 

Thanks.  You don't need to worry about simple conflicts like this, Linus
and I are both quite happy with fixing them up - though I will still
report them, just so people know what is going on.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]