On Mon, May 05, 2014 at 10:54:06AM -0600, Bjorn Helgaas wrote:
> On Mon, May 5, 2014 at 7:46 AM, Jason Cooper <jason@xxxxxxxxxxxxxx> wrote:
> > Greg,
> >
> > + Bjorn Helgaas
> >
> > On Wed, Apr 30, 2014 at 05:41:43PM -0700, gregkh@xxxxxxxxxxxxxxxxxxx wrote:
> >>
> >> The patch below does not apply to the 3.14-stable tree.
> >> If someone wants it applied there, or to any other stable or longterm
> >> tree, then please email the backport, including the original git commit
> >> id to <stable@xxxxxxxxxxxxxxx>.
> >
> > Please disregard for v3.14.y. It looks as though Bjorn Helgaas
> > committed it as:
> >
> > 322a8e91844f PCI: mvebu: Use Device ID and revision from underlying endpoint
> >
> > which has been in since v3.14-rc4. The commit you found, below
> > (a760d2fb2c70) has only been in since v3.15-rc1.
> >
> > That's why it didn't apply.
> >
> > I'm not sure what I f'd up here. There's no way it could've been
> > applied twice without throwing the error Greg saw and I saw.
> >
> > Bjorn, did I send you something wrongly?
>
> I don't think so, but I might have messed it up by the way I applied
> it. Here's what I think I did:
>
> - applied this change to my pci/host-mvebu branch
> - cherry-picked it to my for-linus branch
ahh, this is where the wheels came off the cart. This created the same
change with a different commit ID.
> - asked Linus to pull it as pci-v3.14-fixes-1 after the v3.14 merge window
> - added other changes to pci/host-mvebu
> - merged pci/host-mvebu to my "next" branch
> - asked Linus to pull "next" as pci-v3.15-changes during v3.15 merge window
>
> So the same commit was in both pull requests (pci-v3.14-fixes-1 and
> pci-v3.15-changes).
>
> Maybe my git workflow is broken? I'm open to suggestions for improvements.
Here's what I do:
- apply a fix to mvebu/fixes
- merge mvebu/fixes into mvebu/for-next (let it cook for a few days)
- send pull request for mvebu/fixes to arm-soc.
- If I find that a commit in mvebu/fixes is needed in another branch,
say mvebu/drivers, I merge mvebu/fixes into mvebu/drivers.
- I also give arm-soc a heads up on the dependency
- ideally, the merge commit is *before* the commits that need it
- send the pull request to arm-soc for mvebu/drivers.
Once I send off a pull request (we only use signed-tags), I consider
that branch up to the tag to be immutable. Those are the commit IDs
that are going to be in mainline.
Does that help?
I'm really curious how both commits got into mainline without a conflict
or raising any eyebrows...
thx,
Jason.
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html