mbox series

[00/19,SRU,J/gcp] Update gve/gvnic code

Message ID 20221011055704.642271-1-khalid.elmously@canonical.com
Headers show
Series Update gve/gvnic code | expand

Message

Khalid Elmously Oct. 11, 2022, 5:56 a.m. UTC 
BugLink: https://bugs.launchpad.net/bugs/1953575

Update gvnic driver code to match upstream v5.16

These changes were part of a larger patchset that was backported to the 5.4 (and 5.13) kernel already. Some patches should have been applied to 5.15 as well. The noticable problem is lack of jumbo-frame support on the 5.15 kernels.

GVE team at GCP has tested these changes and have +1'd them.


Ameer Hamza (1):
  gve: fix for null pointer dereference.

Catherine Sullivan (6):
  gve: Add netif_set_xps_queue call
  gve: Add rx buffer pagecnt bias
  gve: Implement suspend/resume/shutdown
  gve: Move the irq db indexes out of the ntfy block struct
  gve: Update gve_free_queue_page_list signature
  gve: remove memory barrier around seqno

Dan Carpenter (1):
  gve: fix unmatched u64_stats_update_end()

David Awogbemila (2):
  gve: Add RX context.
  gve: Implement packet continuation for RX.

Gustavo A. R. Silva (1):
  gve: Use kvcalloc() instead of kvzalloc()

Jeroen de Borst (1):
  gve: Correct order of processing device options

Jordan Kim (2):
  gve: Allow pageflips on larger pages
  gve: Add consumed counts to ethtool stats

Shailend Chand (1):
  gve: Add a jumbo-frame device option.

Tao Liu (2):
  gve: Add tx|rx-coalesce-usec for DQO
  gve: Do lazy cleanup in TX path

Willem de Bruijn (1):
  gve: Add optional metadata descriptor type GVE_TXD_MTD

Yangchun Fu (1):
  gve: Switch to use napi_complete_done

 drivers/net/ethernet/google/gve/gve.h         |  56 ++-
 drivers/net/ethernet/google/gve/gve_adminq.c  |  69 ++-
 drivers/net/ethernet/google/gve/gve_adminq.h  |  14 +
 drivers/net/ethernet/google/gve/gve_desc.h    |  33 +-
 drivers/net/ethernet/google/gve/gve_dqo.h     |  24 +-
 drivers/net/ethernet/google/gve/gve_ethtool.c |  89 +++-
 drivers/net/ethernet/google/gve/gve_main.c    | 172 +++++---
 drivers/net/ethernet/google/gve/gve_rx.c      | 409 +++++++++++++-----
 drivers/net/ethernet/google/gve/gve_rx_dqo.c  |  68 +--
 drivers/net/ethernet/google/gve/gve_tx.c      | 167 ++++---
 drivers/net/ethernet/google/gve/gve_utils.c   |  40 +-
 drivers/net/ethernet/google/gve/gve_utils.h   |   2 +-
 12 files changed, 820 insertions(+), 323 deletions(-)

Comments

Tim Gardner Oct. 11, 2022, 12:45 p.m. UTC | #1 
On 10/10/22 11:56 PM, Khalid Elmously wrote:
> BugLink: https://bugs.launchpad.net/bugs/1953575
> 
> Update gvnic driver code to match upstream v5.16
> 
> These changes were part of a larger patchset that was backported to the 5.4 (and 5.13) kernel already. Some patches should have been applied to 5.15 as well. The noticable problem is lack of jumbo-frame support on the 5.15 kernels.
> 
> GVE team at GCP has tested these changes and have +1'd them.
> 
> 
> Ameer Hamza (1):
>    gve: fix for null pointer dereference.
> 
> Catherine Sullivan (6):
>    gve: Add netif_set_xps_queue call
>    gve: Add rx buffer pagecnt bias
>    gve: Implement suspend/resume/shutdown
>    gve: Move the irq db indexes out of the ntfy block struct
>    gve: Update gve_free_queue_page_list signature
>    gve: remove memory barrier around seqno
> 
> Dan Carpenter (1):
>    gve: fix unmatched u64_stats_update_end()
> 
> David Awogbemila (2):
>    gve: Add RX context.
>    gve: Implement packet continuation for RX.
> 
> Gustavo A. R. Silva (1):
>    gve: Use kvcalloc() instead of kvzalloc()
> 
> Jeroen de Borst (1):
>    gve: Correct order of processing device options
> 
> Jordan Kim (2):
>    gve: Allow pageflips on larger pages
>    gve: Add consumed counts to ethtool stats
> 
> Shailend Chand (1):
>    gve: Add a jumbo-frame device option.
> 
> Tao Liu (2):
>    gve: Add tx|rx-coalesce-usec for DQO
>    gve: Do lazy cleanup in TX path
> 
> Willem de Bruijn (1):
>    gve: Add optional metadata descriptor type GVE_TXD_MTD
> 
> Yangchun Fu (1):
>    gve: Switch to use napi_complete_done
> 
>   drivers/net/ethernet/google/gve/gve.h         |  56 ++-
>   drivers/net/ethernet/google/gve/gve_adminq.c  |  69 ++-
>   drivers/net/ethernet/google/gve/gve_adminq.h  |  14 +
>   drivers/net/ethernet/google/gve/gve_desc.h    |  33 +-
>   drivers/net/ethernet/google/gve/gve_dqo.h     |  24 +-
>   drivers/net/ethernet/google/gve/gve_ethtool.c |  89 +++-
>   drivers/net/ethernet/google/gve/gve_main.c    | 172 +++++---
>   drivers/net/ethernet/google/gve/gve_rx.c      | 409 +++++++++++++-----
>   drivers/net/ethernet/google/gve/gve_rx_dqo.c  |  68 +--
>   drivers/net/ethernet/google/gve/gve_tx.c      | 167 ++++---
>   drivers/net/ethernet/google/gve/gve_utils.c   |  40 +-
>   drivers/net/ethernet/google/gve/gve_utils.h   |   2 +-
>   12 files changed, 820 insertions(+), 323 deletions(-)
> 

I extracted the cherry pick SHAs from each patch and attempted to apply 
them cleanly. '[PATCH 16/19] gve: Implement packet continuation for RX.' 
does not appear to be a clean cherry pick. I didn't look beyond that.

P.S. I'm of the opinion that anything over about 8 patches should be a 
pull request.

rtg