Upstreaming pending -net patches from the export branch, with conflicts

Upstreaming pending -net patches from the export branch, with conflicts

[Mat Martineau]

Hi Paolo -

There are 7 pending -net patches in the export branch 
(export/20220217T054646):

a92e1a369e45 selftests: mptcp: be more conservative with cookie MPJ limits
0d1f61719931 selftests: mptcp: more robust signal race test
f9ce14971cd2 mptcp: add mibs counter for ignored incoming options
474d31e3d81f mptcp: fix race in incoming ADD_ADDR option processing
cb3236ecdf7f mptcp: fix race in overlapping signal events
ec621fe172ae selftests: mptcp: improve 'fair usage on close' stability
3dbd5c179fa0 selftests: mptcp: fix diag instability

I was doing a test run to see how these applied to net/master and ran in 
to two issues:

"selftests: mptcp: improve 'fair usage on close' stability" creates a 
net/net-next merge conflict due to the renaming of "wait_for_tw" to 
"wait_attempt_fail". I propose to drop this rename for the -net patch and 
queue a new patch for mptcp-next with just the rename.


"selftests: mptcp: more robust signal race test" creates a net/net-next 
conflict that's due to changes in the patch context (switching from 
direct calls to "ip netns exec..." to "pm_nl_add_endpoint"). I think the 
conflict resolution is super obvious for the netdev maintainers if we 
remove the blank line added here:

  	# signal addresses race test
  	reset
+
  	pm_nl_set_limits $ns1 4 4
  	pm_nl_set_limits $ns2 4 4

So, I propose to remove that whitespace change and mention the expected 
merge conflict when posting to netdev.

Sound good?

--
Mat Martineau
Intel

Re: Upstreaming pending -net patches from the export branch, with conflicts

[Paolo Abeni]

On Thu, 2022-02-17 at 14:43 -0800, Mat Martineau wrote:
> There are 7 pending -net patches in the export branch 
> (export/20220217T054646):
> 
> a92e1a369e45 selftests: mptcp: be more conservative with cookie MPJ limits
> 0d1f61719931 selftests: mptcp: more robust signal race test
> f9ce14971cd2 mptcp: add mibs counter for ignored incoming options
> 474d31e3d81f mptcp: fix race in incoming ADD_ADDR option processing
> cb3236ecdf7f mptcp: fix race in overlapping signal events
> ec621fe172ae selftests: mptcp: improve 'fair usage on close' stability
> 3dbd5c179fa0 selftests: mptcp: fix diag instability
> 
> I was doing a test run to see how these applied to net/master and ran in 
> to two issues:
> 
> "selftests: mptcp: improve 'fair usage on close' stability" creates a 
> net/net-next merge conflict due to the renaming of "wait_for_tw" to 
> "wait_attempt_fail". I propose to drop this rename for the -net patch and 
> queue a new patch for mptcp-next with just the rename.
> 
> 
> "selftests: mptcp: more robust signal race test" creates a net/net-next 
> conflict that's due to changes in the patch context (switching from 
> direct calls to "ip netns exec..." to "pm_nl_add_endpoint"). I think the 
> conflict resolution is super obvious for the netdev maintainers if we 
> remove the blank line added here:
> 
>   	# signal addresses race test
>   	reset
> +
>   	pm_nl_set_limits $ns1 4 4
>   	pm_nl_set_limits $ns2 4 4
> 
> So, I propose to remove that whitespace change and mention the expected 
> merge conflict when posting to netdev.
> 
> Sound good?

LGTM! Thanks for accurate checks!

Paolo