| Message ID | 1593746858-6548-1-git-send-email-moshe@mellanox.com |
|---|---|
| Headers | show |
| Series | Add devlink-health support for devlink ports | expand |
On Fri, 3 Jul 2020 06:27:31 +0300 Moshe Shemesh wrote: > Implement support for devlink health reporters on per-port basis. First > part in the series prepares common functions parts for health reporter > implementation. Second introduces required API to devlink-health and > mlx5e ones demonstrate its usage and effectively implement the feature > for mlx5 driver. > The per-port reporter functionality is achieved by adding a list of > devlink_health_reporters to devlink_port struct in a manner similar to > existing device infrastructure. This is the only major difference and > it makes possible to fully reuse device reporters operations. > The effect will be seen in conjunction with iproute2 additions and > will affect all devlink health commands. User can distinguish between > device and port reporters by looking at a devlink handle. Port reporters > have a port index at the end of the address and such addresses can be > provided as a parameter in every place where devlink-health accepted it. > These can be obtained from devlink port show command. > For example: > $ devlink health show > pci/0000:00:0a.0: > reporter fw > state healthy error 0 recover 0 auto_dump true > pci/0000:00:0a.0/1: > reporter tx > state healthy error 0 recover 0 grace_period 500 auto_recover true auto_dump true > $ devlink health set pci/0000:00:0a.0/1 reporter tx grace_period 1000 \ > auto_recover false auto_dump false > $ devlink health show pci/0000:00:0a.0/1 reporter tx > pci/0000:00:0a.0/1: > reporter tx > state healthy error 0 recover 0 grace_period 1000 auto_recover flase auto_dump false What's the motivation, though? This patch series achieves nothing that couldn't be previously achieved. Is there no concern of uAPI breakage with moving the existing health reporters in patch 7?
Sat, Jul 04, 2020 at 01:44:39AM CEST, kuba@kernel.org wrote: >On Fri, 3 Jul 2020 06:27:31 +0300 Moshe Shemesh wrote: >> Implement support for devlink health reporters on per-port basis. First >> part in the series prepares common functions parts for health reporter >> implementation. Second introduces required API to devlink-health and >> mlx5e ones demonstrate its usage and effectively implement the feature >> for mlx5 driver. >> The per-port reporter functionality is achieved by adding a list of >> devlink_health_reporters to devlink_port struct in a manner similar to >> existing device infrastructure. This is the only major difference and >> it makes possible to fully reuse device reporters operations. >> The effect will be seen in conjunction with iproute2 additions and >> will affect all devlink health commands. User can distinguish between >> device and port reporters by looking at a devlink handle. Port reporters >> have a port index at the end of the address and such addresses can be >> provided as a parameter in every place where devlink-health accepted it. >> These can be obtained from devlink port show command. >> For example: >> $ devlink health show >> pci/0000:00:0a.0: >> reporter fw >> state healthy error 0 recover 0 auto_dump true >> pci/0000:00:0a.0/1: >> reporter tx >> state healthy error 0 recover 0 grace_period 500 auto_recover true auto_dump true >> $ devlink health set pci/0000:00:0a.0/1 reporter tx grace_period 1000 \ >> auto_recover false auto_dump false >> $ devlink health show pci/0000:00:0a.0/1 reporter tx >> pci/0000:00:0a.0/1: >> reporter tx >> state healthy error 0 recover 0 grace_period 1000 auto_recover flase auto_dump false > >What's the motivation, though? > >This patch series achieves nothing that couldn't be previously achieved. Well, not really. If you have 2 ports, you have 2 set's of tx/rx health reporters. Cannot achieve that w/o per-port health reporters. > >Is there no concern of uAPI breakage with moving the existing health >reporters in patch 7? No. This is bug by design that we are fixing now. No other way around :/ This is mlx5 only.
On Sat, 4 Jul 2020 16:16:42 +0200 Jiri Pirko wrote: > Sat, Jul 04, 2020 at 01:44:39AM CEST, kuba@kernel.org wrote: > >On Fri, 3 Jul 2020 06:27:31 +0300 Moshe Shemesh wrote: > >> Implement support for devlink health reporters on per-port basis. First > >> part in the series prepares common functions parts for health reporter > >> implementation. Second introduces required API to devlink-health and > >> mlx5e ones demonstrate its usage and effectively implement the feature > >> for mlx5 driver. > >> The per-port reporter functionality is achieved by adding a list of > >> devlink_health_reporters to devlink_port struct in a manner similar to > >> existing device infrastructure. This is the only major difference and > >> it makes possible to fully reuse device reporters operations. > >> The effect will be seen in conjunction with iproute2 additions and > >> will affect all devlink health commands. User can distinguish between > >> device and port reporters by looking at a devlink handle. Port reporters > >> have a port index at the end of the address and such addresses can be > >> provided as a parameter in every place where devlink-health accepted it. > >> These can be obtained from devlink port show command. > >> For example: > >> $ devlink health show > >> pci/0000:00:0a.0: > >> reporter fw > >> state healthy error 0 recover 0 auto_dump true > >> pci/0000:00:0a.0/1: > >> reporter tx > >> state healthy error 0 recover 0 grace_period 500 auto_recover true auto_dump true > >> $ devlink health set pci/0000:00:0a.0/1 reporter tx grace_period 1000 \ > >> auto_recover false auto_dump false > >> $ devlink health show pci/0000:00:0a.0/1 reporter tx > >> pci/0000:00:0a.0/1: > >> reporter tx > >> state healthy error 0 recover 0 grace_period 1000 auto_recover flase auto_dump false > > > >What's the motivation, though? > > > >This patch series achieves nothing that couldn't be previously achieved. > > Well, not really. If you have 2 ports, you have 2 set's of tx/rx health > reporters. Cannot achieve that w/o per-port health reporters. Which mlx5 doesn't. Each port has its own instance of devlink today. > >Is there no concern of uAPI breakage with moving the existing health > >reporters in patch 7? > > No. This is bug by design that we are fixing now. No other way around :/ > This is mlx5 only. Please repost including in the cover letter a proper an explanation of why the change is necessary, what benefits it will bring us, what are next steps, and why it doesn't matter much that the health reporters move for mlx5. The cover letter describes code not reasoning, which is IMHO unacceptable for patches that "change" uAPI.
On 7/5/2020 7:48 PM, Jakub Kicinski wrote: > On Sat, 4 Jul 2020 16:16:42 +0200 Jiri Pirko wrote: >> Sat, Jul 04, 2020 at 01:44:39AM CEST, kuba@kernel.org wrote: >>> On Fri, 3 Jul 2020 06:27:31 +0300 Moshe Shemesh wrote: >>>> Implement support for devlink health reporters on per-port basis. First >>>> part in the series prepares common functions parts for health reporter >>>> implementation. Second introduces required API to devlink-health and >>>> mlx5e ones demonstrate its usage and effectively implement the feature >>>> for mlx5 driver. >>>> The per-port reporter functionality is achieved by adding a list of >>>> devlink_health_reporters to devlink_port struct in a manner similar to >>>> existing device infrastructure. This is the only major difference and >>>> it makes possible to fully reuse device reporters operations. >>>> The effect will be seen in conjunction with iproute2 additions and >>>> will affect all devlink health commands. User can distinguish between >>>> device and port reporters by looking at a devlink handle. Port reporters >>>> have a port index at the end of the address and such addresses can be >>>> provided as a parameter in every place where devlink-health accepted it. >>>> These can be obtained from devlink port show command. >>>> For example: >>>> $ devlink health show >>>> pci/0000:00:0a.0: >>>> reporter fw >>>> state healthy error 0 recover 0 auto_dump true >>>> pci/0000:00:0a.0/1: >>>> reporter tx >>>> state healthy error 0 recover 0 grace_period 500 auto_recover true auto_dump true >>>> $ devlink health set pci/0000:00:0a.0/1 reporter tx grace_period 1000 \ >>>> auto_recover false auto_dump false >>>> $ devlink health show pci/0000:00:0a.0/1 reporter tx >>>> pci/0000:00:0a.0/1: >>>> reporter tx >>>> state healthy error 0 recover 0 grace_period 1000 auto_recover flase auto_dump false >>> What's the motivation, though? >>> >>> This patch series achieves nothing that couldn't be previously achieved. >> Well, not really. If you have 2 ports, you have 2 set's of tx/rx health >> reporters. Cannot achieve that w/o per-port health reporters. > Which mlx5 doesn't. Each port has its own instance of devlink today. That's right for mlx5, but in the general case Tx and Rx should be per port and not per device. Better to fix the API before more drivers use it for such reporters. >>> Is there no concern of uAPI breakage with moving the existing health >>> reporters in patch 7? >> No. This is bug by design that we are fixing now. No other way around :/ >> This is mlx5 only. > Please repost including in the cover letter a proper an explanation of > why the change is necessary, what benefits it will bring us, what are > next steps, and why it doesn't matter much that the health reporters > move for mlx5. > > The cover letter describes code not reasoning, which is IMHO > unacceptable for patches that "change" uAPI. Sure, will fix and resend.
Implement support for devlink health reporters on per-port basis. First part in the series prepares common functions parts for health reporter implementation. Second introduces required API to devlink-health and mlx5e ones demonstrate its usage and effectively implement the feature for mlx5 driver. The per-port reporter functionality is achieved by adding a list of devlink_health_reporters to devlink_port struct in a manner similar to existing device infrastructure. This is the only major difference and it makes possible to fully reuse device reporters operations. The effect will be seen in conjunction with iproute2 additions and will affect all devlink health commands. User can distinguish between device and port reporters by looking at a devlink handle. Port reporters have a port index at the end of the address and such addresses can be provided as a parameter in every place where devlink-health accepted it. These can be obtained from devlink port show command. For example: $ devlink health show pci/0000:00:0a.0: reporter fw state healthy error 0 recover 0 auto_dump true pci/0000:00:0a.0/1: reporter tx state healthy error 0 recover 0 grace_period 500 auto_recover true auto_dump true $ devlink health set pci/0000:00:0a.0/1 reporter tx grace_period 1000 \ auto_recover false auto_dump false $ devlink health show pci/0000:00:0a.0/1 reporter tx pci/0000:00:0a.0/1: reporter tx state healthy error 0 recover 0 grace_period 1000 auto_recover flase auto_dump false Changes v1 -> v2: Fixed functions comment to match parameters list. Vladyslav Tarasiuk (7): devlink: Refactor devlink health reporter constructor devlink: Rework devlink health reporter destructor devlink: Create generic devlink health reporter search function devlink: Implement devlink health reporters on per-port basis devlink: Add devlink health port reporters API net/mlx5e: Move devlink port register and unregister calls net/mlx5e: Move devlink-health rx and tx reporters to devlink port .../ethernet/mellanox/mlx5/core/en/reporter_rx.c | 9 +- .../ethernet/mellanox/mlx5/core/en/reporter_tx.c | 13 +- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 15 +- include/net/devlink.h | 11 + net/core/devlink.c | 244 ++++++++++++++++----- 5 files changed, 216 insertions(+), 76 deletions(-)