Message ID | 20201012202511.21000-1-gvrose8192@gmail.com |
---|---|
Headers | show |
Series | Add support for Linux kernels up to 5.8.x | expand |
On 10/12/20 10:24 PM, Greg Rose wrote: > This patch set will add support for Linux kernels up to 5.8. In > addition there are quite a few patches for openvswitch on the Linux > kernel mailing list that have not been backported - here is a first > batch attempting to catch up on some of that technical debt. There > will be a follow up batch of patches to this one but I didn't want > the patch bomb to get too large. > > V4 passes Travis here: > https://travis-ci.org/github/gvrose8192/ovs-experimental/builds/735114390 Thanks! I'm wondering if there was changes in patches beside #5 and #11? If not, I guess, I could add Acks from Yi-Hung from v3 before applying the series. Yi-Hung, could you, please, take a look at patches #5 and #11 and check if comments was addressed? Best regards, Ilya Maximets. > > --- > V2 - V2 of this patch set changes the NEWS as suggested by Ilya > - Moves the acinclude patch for 5.8 support to the end of the > patch series > - Reduces targeted Linux kernel support to 5.8 since 5.9 is > still not baked > - Updates the travis kernel test list > - Adds tags from authors from the first patch series. > V3 - Add tag from Eelco Chaldron > - Mix mistaken warning message in acinclude.m4 > - Remember to add link to passing Travis build in cover letter > V4 - Add portion change in ovs_dp_cmd_set missed in previous patch set > - Simplify the WRITE_ONCE macro implemention for compat. > > Eelco Chaudron (1): > datapath: return an error instead of doing BUG_ON() > > Enrico Weigelt (1): > datapath: drop unneeded likely() call around IS_ERR() > > Greg Rose (3): > acinclude: Enable builds up to Linux 5.8 > travis: Update kernel list as of 5.8 > Documentation: Update faq and NEWS for kernel 5.8 > > Guillaume Nault (1): > datapath: fix GFP flags in rtnl_net_notifyid() > > Jason A. Donenfeld (1): > datapath: use skb_list_walk_safe helper for gso segments > > Kees Cook (1): > datapath: Distribute switch variables for initialization > > Paolo Abeni (3): > datapath: fix flow command message size > datapath: drop unneeded BUG_ON() in ovs_flow_cmd_build_info() > datapath: remove another BUG_ON() > > Paul Blakey (1): > datapath: Set OvS recirc_id from tc chain index > > Taehee Yoo (1): > datapath: do not update max_headroom if new headroom is equal to old > headroom > > Tonghao Zhang (9): > datapath: don't unlock mutex when changing the user_features fails > datapath: optimize flow-mask looking up > datapath: simplify the flow_hash > datapath: add likely in flow_lookup > datapath: fix possible memleak on destroy flow-table > datapath: simplify the ovs_dp_cmd_new > datapath: select vport upcall portid directly > datapath: don't call pad_packet if not necessary > datapath: use hlist_for_each_entry_rcu instead of hlist_for_each_entry > > Yifeng Sun (1): > datapath: Print error when ovs_execute_actions() fails > > aaron conole (1): > datapath: support asymmetric conntrack > > .travis.yml | 4 +- > Documentation/faq/releases.rst | 1 + > NEWS | 2 + > acinclude.m4 | 7 +- > datapath/conntrack.c | 11 + > datapath/datapath.c | 224 ++++++++++++------ > datapath/datapath.h | 2 + > datapath/flow.c | 13 + > datapath/flow_netlink.c | 18 +- > datapath/flow_table.c | 214 +++++++++-------- > .../linux/compat/include/linux/compiler.h | 8 + > datapath/linux/compat/include/linux/skbuff.h | 7 + > .../linux/compat/include/linux/static_key.h | 7 + > datapath/vport.c | 5 +- > 14 files changed, 330 insertions(+), 193 deletions(-) >
On 10/14/2020 8:41 AM, Ilya Maximets wrote: > On 10/12/20 10:24 PM, Greg Rose wrote: >> This patch set will add support for Linux kernels up to 5.8. In >> addition there are quite a few patches for openvswitch on the Linux >> kernel mailing list that have not been backported - here is a first >> batch attempting to catch up on some of that technical debt. There >> will be a follow up batch of patches to this one but I didn't want >> the patch bomb to get too large. >> >> V4 passes Travis here: >> https://travis-ci.org/github/gvrose8192/ovs-experimental/builds/735114390 > > Thanks! > > I'm wondering if there was changes in patches beside #5 and #11? > If not, I guess, I could add Acks from Yi-Hung from v3 before applying > the series. > > Yi-Hung, could you, please, take a look at patches #5 and #11 and check > if comments was addressed? > > Best regards, Ilya Maximets. Hi Ilya, you are correct, the only changes in V4 of this series are in patches #5 and #11. Thanks, - Greg > >> >> --- >> V2 - V2 of this patch set changes the NEWS as suggested by Ilya >> - Moves the acinclude patch for 5.8 support to the end of the >> patch series >> - Reduces targeted Linux kernel support to 5.8 since 5.9 is >> still not baked >> - Updates the travis kernel test list >> - Adds tags from authors from the first patch series. >> V3 - Add tag from Eelco Chaldron >> - Mix mistaken warning message in acinclude.m4 >> - Remember to add link to passing Travis build in cover letter >> V4 - Add portion change in ovs_dp_cmd_set missed in previous patch set >> - Simplify the WRITE_ONCE macro implemention for compat. >> >> Eelco Chaudron (1): >> datapath: return an error instead of doing BUG_ON() >> >> Enrico Weigelt (1): >> datapath: drop unneeded likely() call around IS_ERR() >> >> Greg Rose (3): >> acinclude: Enable builds up to Linux 5.8 >> travis: Update kernel list as of 5.8 >> Documentation: Update faq and NEWS for kernel 5.8 >> >> Guillaume Nault (1): >> datapath: fix GFP flags in rtnl_net_notifyid() >> >> Jason A. Donenfeld (1): >> datapath: use skb_list_walk_safe helper for gso segments >> >> Kees Cook (1): >> datapath: Distribute switch variables for initialization >> >> Paolo Abeni (3): >> datapath: fix flow command message size >> datapath: drop unneeded BUG_ON() in ovs_flow_cmd_build_info() >> datapath: remove another BUG_ON() >> >> Paul Blakey (1): >> datapath: Set OvS recirc_id from tc chain index >> >> Taehee Yoo (1): >> datapath: do not update max_headroom if new headroom is equal to old >> headroom >> >> Tonghao Zhang (9): >> datapath: don't unlock mutex when changing the user_features fails >> datapath: optimize flow-mask looking up >> datapath: simplify the flow_hash >> datapath: add likely in flow_lookup >> datapath: fix possible memleak on destroy flow-table >> datapath: simplify the ovs_dp_cmd_new >> datapath: select vport upcall portid directly >> datapath: don't call pad_packet if not necessary >> datapath: use hlist_for_each_entry_rcu instead of hlist_for_each_entry >> >> Yifeng Sun (1): >> datapath: Print error when ovs_execute_actions() fails >> >> aaron conole (1): >> datapath: support asymmetric conntrack >> >> .travis.yml | 4 +- >> Documentation/faq/releases.rst | 1 + >> NEWS | 2 + >> acinclude.m4 | 7 +- >> datapath/conntrack.c | 11 + >> datapath/datapath.c | 224 ++++++++++++------ >> datapath/datapath.h | 2 + >> datapath/flow.c | 13 + >> datapath/flow_netlink.c | 18 +- >> datapath/flow_table.c | 214 +++++++++-------- >> .../linux/compat/include/linux/compiler.h | 8 + >> datapath/linux/compat/include/linux/skbuff.h | 7 + >> .../linux/compat/include/linux/static_key.h | 7 + >> datapath/vport.c | 5 +- >> 14 files changed, 330 insertions(+), 193 deletions(-) >> >
On Wed, Oct 14, 2020 at 8:41 AM Ilya Maximets <i.maximets@ovn.org> wrote: > > On 10/12/20 10:24 PM, Greg Rose wrote: > > This patch set will add support for Linux kernels up to 5.8. In > > addition there are quite a few patches for openvswitch on the Linux > > kernel mailing list that have not been backported - here is a first > > batch attempting to catch up on some of that technical debt. There > > will be a follow up batch of patches to this one but I didn't want > > the patch bomb to get too large. > > > > V4 passes Travis here: > > https://travis-ci.org/github/gvrose8192/ovs-experimental/builds/735114390 > > Thanks! > > I'm wondering if there was changes in patches beside #5 and #11? > If not, I guess, I could add Acks from Yi-Hung from v3 before applying > the series. > > Yi-Hung, could you, please, take a look at patches #5 and #11 and check > if comments was addressed? > > Best regards, Ilya Maximets. Hi IIya, Thanks for reminding me. I just acked both patch 5 and 11. Thanks, -Yi-Hung
On 10/14/20 7:47 PM, Yi-Hung Wei wrote: > On Wed, Oct 14, 2020 at 8:41 AM Ilya Maximets <i.maximets@ovn.org> wrote: >> >> On 10/12/20 10:24 PM, Greg Rose wrote: >>> This patch set will add support for Linux kernels up to 5.8. In >>> addition there are quite a few patches for openvswitch on the Linux >>> kernel mailing list that have not been backported - here is a first >>> batch attempting to catch up on some of that technical debt. There >>> will be a follow up batch of patches to this one but I didn't want >>> the patch bomb to get too large. >>> >>> V4 passes Travis here: >>> https://travis-ci.org/github/gvrose8192/ovs-experimental/builds/735114390 >> >> Thanks! >> >> I'm wondering if there was changes in patches beside #5 and #11? >> If not, I guess, I could add Acks from Yi-Hung from v3 before applying >> the series. >> >> Yi-Hung, could you, please, take a look at patches #5 and #11 and check >> if comments was addressed? >> >> Best regards, Ilya Maximets. > > Hi IIya, > > Thanks for reminding me. I just acked both patch 5 and 11. Thanks for review! Best regards, Ilya Maximets.
On 10/14/20 7:46 PM, Gregory Rose wrote: > > > On 10/14/2020 8:41 AM, Ilya Maximets wrote: >> On 10/12/20 10:24 PM, Greg Rose wrote: >>> This patch set will add support for Linux kernels up to 5.8. In >>> addition there are quite a few patches for openvswitch on the Linux >>> kernel mailing list that have not been backported - here is a first >>> batch attempting to catch up on some of that technical debt. There >>> will be a follow up batch of patches to this one but I didn't want >>> the patch bomb to get too large. >>> >>> V4 passes Travis here: >>> https://travis-ci.org/github/gvrose8192/ovs-experimental/builds/735114390 >> >> Thanks! >> >> I'm wondering if there was changes in patches beside #5 and #11? >> If not, I guess, I could add Acks from Yi-Hung from v3 before applying >> the series. >> >> Yi-Hung, could you, please, take a look at patches #5 and #11 and check >> if comments was addressed? >> >> Best regards, Ilya Maximets. > > Hi Ilya, > > you are correct, the only changes in V4 of this series are in > patches #5 and #11. Thanks! I added ACKs from v3 and applied series to master. Best regards, Ilya Maximets. > > Thanks, > > - Greg > > >> >>> >>> --- >>> V2 - V2 of this patch set changes the NEWS as suggested by Ilya >>> - Moves the acinclude patch for 5.8 support to the end of the >>> patch series >>> - Reduces targeted Linux kernel support to 5.8 since 5.9 is >>> still not baked >>> - Updates the travis kernel test list >>> - Adds tags from authors from the first patch series. >>> V3 - Add tag from Eelco Chaldron >>> - Mix mistaken warning message in acinclude.m4 >>> - Remember to add link to passing Travis build in cover letter >>> V4 - Add portion change in ovs_dp_cmd_set missed in previous patch set >>> - Simplify the WRITE_ONCE macro implemention for compat. >>> >>> Eelco Chaudron (1): >>> datapath: return an error instead of doing BUG_ON() >>> >>> Enrico Weigelt (1): >>> datapath: drop unneeded likely() call around IS_ERR() >>> >>> Greg Rose (3): >>> acinclude: Enable builds up to Linux 5.8 >>> travis: Update kernel list as of 5.8 >>> Documentation: Update faq and NEWS for kernel 5.8 >>> >>> Guillaume Nault (1): >>> datapath: fix GFP flags in rtnl_net_notifyid() >>> >>> Jason A. Donenfeld (1): >>> datapath: use skb_list_walk_safe helper for gso segments >>> >>> Kees Cook (1): >>> datapath: Distribute switch variables for initialization >>> >>> Paolo Abeni (3): >>> datapath: fix flow command message size >>> datapath: drop unneeded BUG_ON() in ovs_flow_cmd_build_info() >>> datapath: remove another BUG_ON() >>> >>> Paul Blakey (1): >>> datapath: Set OvS recirc_id from tc chain index >>> >>> Taehee Yoo (1): >>> datapath: do not update max_headroom if new headroom is equal to old >>> headroom >>> >>> Tonghao Zhang (9): >>> datapath: don't unlock mutex when changing the user_features fails >>> datapath: optimize flow-mask looking up >>> datapath: simplify the flow_hash >>> datapath: add likely in flow_lookup >>> datapath: fix possible memleak on destroy flow-table >>> datapath: simplify the ovs_dp_cmd_new >>> datapath: select vport upcall portid directly >>> datapath: don't call pad_packet if not necessary >>> datapath: use hlist_for_each_entry_rcu instead of hlist_for_each_entry >>> >>> Yifeng Sun (1): >>> datapath: Print error when ovs_execute_actions() fails >>> >>> aaron conole (1): >>> datapath: support asymmetric conntrack >>> >>> .travis.yml | 4 +- >>> Documentation/faq/releases.rst | 1 + >>> NEWS | 2 + >>> acinclude.m4 | 7 +- >>> datapath/conntrack.c | 11 + >>> datapath/datapath.c | 224 ++++++++++++------ >>> datapath/datapath.h | 2 + >>> datapath/flow.c | 13 + >>> datapath/flow_netlink.c | 18 +- >>> datapath/flow_table.c | 214 +++++++++-------- >>> .../linux/compat/include/linux/compiler.h | 8 + >>> datapath/linux/compat/include/linux/skbuff.h | 7 + >>> .../linux/compat/include/linux/static_key.h | 7 + >>> datapath/vport.c | 5 +- >>> 14 files changed, 330 insertions(+), 193 deletions(-) >>> >>
On 10/19/2020 6:25 AM, Ilya Maximets wrote: > On 10/14/20 7:46 PM, Gregory Rose wrote: >> >> >> On 10/14/2020 8:41 AM, Ilya Maximets wrote: >>> On 10/12/20 10:24 PM, Greg Rose wrote: >>>> This patch set will add support for Linux kernels up to 5.8. In >>>> addition there are quite a few patches for openvswitch on the Linux >>>> kernel mailing list that have not been backported - here is a first >>>> batch attempting to catch up on some of that technical debt. There >>>> will be a follow up batch of patches to this one but I didn't want >>>> the patch bomb to get too large. >>>> >>>> V4 passes Travis here: >>>> https://travis-ci.org/github/gvrose8192/ovs-experimental/builds/735114390 >>> >>> Thanks! >>> >>> I'm wondering if there was changes in patches beside #5 and #11? >>> If not, I guess, I could add Acks from Yi-Hung from v3 before applying >>> the series. >>> >>> Yi-Hung, could you, please, take a look at patches #5 and #11 and check >>> if comments was addressed? >>> >>> Best regards, Ilya Maximets. >> >> Hi Ilya, >> >> you are correct, the only changes in V4 of this series are in >> patches #5 and #11. > > Thanks! > > I added ACKs from v3 and applied series to master. > > Best regards, Ilya Maximets. Awesome! Thanks Ilya. - Greg > >> >> Thanks, >> >> - Greg >> >> >>> >>>> >>>> --- >>>> V2 - V2 of this patch set changes the NEWS as suggested by Ilya >>>> - Moves the acinclude patch for 5.8 support to the end of the >>>> patch series >>>> - Reduces targeted Linux kernel support to 5.8 since 5.9 is >>>> still not baked >>>> - Updates the travis kernel test list >>>> - Adds tags from authors from the first patch series. >>>> V3 - Add tag from Eelco Chaldron >>>> - Mix mistaken warning message in acinclude.m4 >>>> - Remember to add link to passing Travis build in cover letter >>>> V4 - Add portion change in ovs_dp_cmd_set missed in previous patch set >>>> - Simplify the WRITE_ONCE macro implemention for compat. >>>> >>>> Eelco Chaudron (1): >>>> datapath: return an error instead of doing BUG_ON() >>>> >>>> Enrico Weigelt (1): >>>> datapath: drop unneeded likely() call around IS_ERR() >>>> >>>> Greg Rose (3): >>>> acinclude: Enable builds up to Linux 5.8 >>>> travis: Update kernel list as of 5.8 >>>> Documentation: Update faq and NEWS for kernel 5.8 >>>> >>>> Guillaume Nault (1): >>>> datapath: fix GFP flags in rtnl_net_notifyid() >>>> >>>> Jason A. Donenfeld (1): >>>> datapath: use skb_list_walk_safe helper for gso segments >>>> >>>> Kees Cook (1): >>>> datapath: Distribute switch variables for initialization >>>> >>>> Paolo Abeni (3): >>>> datapath: fix flow command message size >>>> datapath: drop unneeded BUG_ON() in ovs_flow_cmd_build_info() >>>> datapath: remove another BUG_ON() >>>> >>>> Paul Blakey (1): >>>> datapath: Set OvS recirc_id from tc chain index >>>> >>>> Taehee Yoo (1): >>>> datapath: do not update max_headroom if new headroom is equal to old >>>> headroom >>>> >>>> Tonghao Zhang (9): >>>> datapath: don't unlock mutex when changing the user_features fails >>>> datapath: optimize flow-mask looking up >>>> datapath: simplify the flow_hash >>>> datapath: add likely in flow_lookup >>>> datapath: fix possible memleak on destroy flow-table >>>> datapath: simplify the ovs_dp_cmd_new >>>> datapath: select vport upcall portid directly >>>> datapath: don't call pad_packet if not necessary >>>> datapath: use hlist_for_each_entry_rcu instead of hlist_for_each_entry >>>> >>>> Yifeng Sun (1): >>>> datapath: Print error when ovs_execute_actions() fails >>>> >>>> aaron conole (1): >>>> datapath: support asymmetric conntrack >>>> >>>> .travis.yml | 4 +- >>>> Documentation/faq/releases.rst | 1 + >>>> NEWS | 2 + >>>> acinclude.m4 | 7 +- >>>> datapath/conntrack.c | 11 + >>>> datapath/datapath.c | 224 ++++++++++++------ >>>> datapath/datapath.h | 2 + >>>> datapath/flow.c | 13 + >>>> datapath/flow_netlink.c | 18 +- >>>> datapath/flow_table.c | 214 +++++++++-------- >>>> .../linux/compat/include/linux/compiler.h | 8 + >>>> datapath/linux/compat/include/linux/skbuff.h | 7 + >>>> .../linux/compat/include/linux/static_key.h | 7 + >>>> datapath/vport.c | 5 +- >>>> 14 files changed, 330 insertions(+), 193 deletions(-) >>>> >>> >