| Message ID | 20210220001322.1311139-2-dje@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Add support for ipv6 host forwarding | expand |
Hi Doug,
On 2/20/21 1:13 AM, Doug Evans via wrote:
When updating submodules, the commit description is a good
good place to include the output of:
$ git shortlog 8f43a99..26ae658
See for example QEMU commit f350d78f102 ("Update SLOF").
Anyhow up to the maintainer merging your patch.
> Signed-off-by: Doug Evans <dje@google.com>
> ---
>
> Changes from v4:
> NOTE TO REVIEWERS: I need some hand-holding to know what The Right
> way to submit this particular patch is.
>
> - no change
>
> Changes from v3:
> - pick up latest libslirp patch to reject ipv6 addr-any for guest address
> - libslirp currently only provides a stateless DHCPv6 server, which means
> it can't know in advance what the guest's IP address is, and thus
> cannot do the "addr-any -> guest ip address" translation that is done
> for ipv4
>
> Changes from v2:
> - this patch is new in v3, split out from v2
>
> slirp | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/slirp b/slirp
> index 8f43a99191..26ae658a83 160000
> --- a/slirp
> +++ b/slirp
> @@ -1 +1 @@
> -Subproject commit 8f43a99191afb47ca3f3c6972f6306209f367ece
> +Subproject commit 26ae658a83eeca16780cf5615c8247cbb151c3fa
>
On Fri, Feb 19, 2021 at 4:20 PM Philippe Mathieu-Daudé <philmd@redhat.com> wrote: > Hi Doug, > > On 2/20/21 1:13 AM, Doug Evans via wrote: > > When updating submodules, the commit description is a good > good place to include the output of: > > $ git shortlog 8f43a99..26ae658 > > See for example QEMU commit f350d78f102 ("Update SLOF"). > > Anyhow up to the maintainer merging your patch. > Thanks, that helps a little. The issue here is that qemu is using branch 4.2 of libslirp, whereas the patch is currently just on libslirp's master branch. Part of the problem is the commit description, which you've helped with. But what about the functional part of the patch? I can only get git format-patch to include a commit id, and the only commit id available is on the libslirp master branch. Is there an additional step I need to do like submit libslirp changes in three parts? Step 1: Submit patch to libslirp's repo (not qemu's copy, but the libslirp master repo) Step 2: Submit a patch to libslirp's repo to add the patch to its 4.2 branch Step 2b: Wait for qemu's copy of libslirp's 4.2 branch to appear in qemu's libslirp repo Step 3: Submit patch to advance qemu's libslirp submodule I've done steps 1,3, it's just effecting the equivalent of step2 that I'm fuzzy on. > > Signed-off-by: Doug Evans <dje@google.com> > > --- > > > > Changes from v4: > > NOTE TO REVIEWERS: I need some hand-holding to know what The Right > > way to submit this particular patch is. > > > > - no change > > > > Changes from v3: > > - pick up latest libslirp patch to reject ipv6 addr-any for guest address > > - libslirp currently only provides a stateless DHCPv6 server, which > means > > it can't know in advance what the guest's IP address is, and thus > > cannot do the "addr-any -> guest ip address" translation that is done > > for ipv4 > > > > Changes from v2: > > - this patch is new in v3, split out from v2 > > > > slirp | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/slirp b/slirp > > index 8f43a99191..26ae658a83 160000 > > --- a/slirp > > +++ b/slirp > > @@ -1 +1 @@ > > -Subproject commit 8f43a99191afb47ca3f3c6972f6306209f367ece > > +Subproject commit 26ae658a83eeca16780cf5615c8247cbb151c3fa > > > >
Hi On Wed, Feb 24, 2021 at 12:20 AM Doug Evans <dje@google.com> wrote: > On Fri, Feb 19, 2021 at 4:20 PM Philippe Mathieu-Daudé <philmd@redhat.com> > wrote: > >> Hi Doug, >> >> On 2/20/21 1:13 AM, Doug Evans via wrote: >> >> When updating submodules, the commit description is a good >> good place to include the output of: >> >> $ git shortlog 8f43a99..26ae658 >> >> See for example QEMU commit f350d78f102 ("Update SLOF"). >> >> Anyhow up to the maintainer merging your patch. >> > > > Thanks, that helps a little. > The issue here is that qemu is using branch 4.2 of libslirp, whereas the > patch is currently just on libslirp's master branch. > Part of the problem is the commit description, which you've helped with. > But what about the functional part of the patch? > I can only get git format-patch to include a commit id, and the only > commit id available is on the libslirp master branch. > > Is there an additional step I need to do like submit libslirp changes in > three parts? > Step 1: Submit patch to libslirp's repo (not qemu's copy, but the > libslirp master repo) > Step 2: Submit a patch to libslirp's repo to add the patch to its 4.2 > branch > Step 2b: Wait for qemu's copy of libslirp's 4.2 branch to appear in > qemu's libslirp repo > Step 3: Submit patch to advance qemu's libslirp submodule > > I've done steps 1,3, it's just effecting the equivalent of step2 that I'm > fuzzy on. > Only bug fixes should go in libslirp 4.2 branch. You can submit MR against the stable-4.2 branch in gitlab. We may update qemu to libslirp master while not in freeze (as long as feature versioning is handled correctly, which may be error prone). I tried to update QEMU to libslirp master some time ago, unfortunately it failed to pass the merge because of the next commit changing the build system ( https://patchew.org/QEMU/20210125073427.3970606-1-marcandre.lureau@redhat.com/20210125073427.3970606-2-marcandre.lureau@redhat.com/). But just updating libslirp to master should be fine. You can see in the commit message that I use a git tool "cherry-diff" to highlight the differences between the branches. > >> > Signed-off-by: Doug Evans <dje@google.com> >> > --- >> > >> > Changes from v4: >> > NOTE TO REVIEWERS: I need some hand-holding to know what The Right >> > way to submit this particular patch is. >> > >> > - no change >> > >> > Changes from v3: >> > - pick up latest libslirp patch to reject ipv6 addr-any for guest >> address >> > - libslirp currently only provides a stateless DHCPv6 server, which >> means >> > it can't know in advance what the guest's IP address is, and thus >> > cannot do the "addr-any -> guest ip address" translation that is >> done >> > for ipv4 >> > >> > Changes from v2: >> > - this patch is new in v3, split out from v2 >> > >> > slirp | 2 +- >> > 1 file changed, 1 insertion(+), 1 deletion(-) >> > >> > diff --git a/slirp b/slirp >> > index 8f43a99191..26ae658a83 160000 >> > --- a/slirp >> > +++ b/slirp >> > @@ -1 +1 @@ >> > -Subproject commit 8f43a99191afb47ca3f3c6972f6306209f367ece >> > +Subproject commit 26ae658a83eeca16780cf5615c8247cbb151c3fa >> > >> >>
diff --git a/slirp b/slirp index 8f43a99191..26ae658a83 160000 --- a/slirp +++ b/slirp @@ -1 +1 @@ -Subproject commit 8f43a99191afb47ca3f3c6972f6306209f367ece +Subproject commit 26ae658a83eeca16780cf5615c8247cbb151c3fa
Signed-off-by: Doug Evans <dje@google.com> --- Changes from v4: NOTE TO REVIEWERS: I need some hand-holding to know what The Right way to submit this particular patch is. - no change Changes from v3: - pick up latest libslirp patch to reject ipv6 addr-any for guest address - libslirp currently only provides a stateless DHCPv6 server, which means it can't know in advance what the guest's IP address is, and thus cannot do the "addr-any -> guest ip address" translation that is done for ipv4 Changes from v2: - this patch is new in v3, split out from v2 slirp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)