[PATCH net-next v5 0/3] ethtool: Dynamic RSS context indirection table resizing

["Björn Töpel" <bjorn@kernel.org>]

Hi!

[ Michael and Pavan, I removed your Tested-by:/Reviewed-by: although
the changes were pretty small. Please re-add if you're still OK with
it. ]

Some NICs (e.g. bnxt) change their RSS indirection table size based on
the queue count, because the hardware table is a shared resource. The
ethtool core locks ctx->indir_size at context creation, so drivers
have to reject channel changes when RSS contexts exist.

This series adds resize helpers and wires them up in bnxt.

Patch 1 adds core helpers:
  ethtool_rxfh_can_resize() - read-only validation
  ethtool_rxfh_resize() - fold/unfold a raw table in place
  ethtool_rxfh_ctxs_can_resize() - validate all non-default contexts
  ethtool_rxfh_ctxs_resize() - resize all non-default contexts,
    with locking and RSS_NTF notifications

Both ethtool_rxfh_can_resize() and ethtool_rxfh_resize() now take a
user_size parameter: the number of indirection table entries the user
originally provided. When shrinking, the table will not fold below
this floor, preserving user intent.

The user_size is tracked in ctx->indir_user_size for non-default RSS
contexts and in dev->ethtool->rss_indir_user_size for context 0. It
is set when the indirection table is configured via netlink or ioctl,
and cleared to zero on reset-to-default.

Patch 2 uses them in bnxt_set_channels(). Validation runs before
bnxt_close_nic(); actual resize is deferred until after. RSS table
size only changes on P5 chips with older firmware.

Patch 3 adds HW tests in rss_drv.py (devices without dynamic table
sizing are skipped):
  resize_periodic - fold/unfold with a non-default [3,2,1,0]
    sub-table (user_size=4), verifying exact content preservation
    (main + ctx)
  resize_below_user_size_reject - periodic sub-table with user_size
    between big and small device table sizes; verifies that shrinking
    below user_size is rejected even when the table is periodic
    (main + ctx)
  resize_nonperiodic_reject - non-periodic table blocks channel
    reduction, with an extra periodic context to exercise
    multi-context validation (main + ctx)
  resize_nonperiodic_no_corruption - failed resize leaves table
    contents and channel count unchanged (main + ctx)

Running the tests:

  # On real hardware
  sudo NETIF=eth0 ./rss_drv.py

Changes v4 -> v5:

 - Track user-provided indirection table size (user_size) as a resize
   floor. Added indir_user_size to ethtool_rxfh_context and
   rss_indir_user_size to ethtool_netdev_state. ethtool_rxfh_can_resize()
   and ethtool_rxfh_resize() now take a user_size parameter and reject
   shrinking below it. (Jakub)
 - Propagated user_size out of rss_set_prep_indir() and stored it on
   successful set in both netlink and ioctl paths.
 - resize_periodic test now sends a 4-entry sub-table (user_size=4)
   instead of replicating to full device table size
 - Added resize_below_user_size_reject test to verify user_size floor.
 - Removed "Open items" section - user_size tracking is now implemented.

Changes v3 -> v4:

 - Rebased onto net-next
 - Added Reviewed-by: from Michael
 - Added missing Cc: to make the pwbots happier

Changes v2 -> v3:

 - Changed ethtool_rxfh_can_resize() to return bool instead of int;
   true means resize is possible, false means it is not. Inverted
   callers accordingly. (Jakub)
 - Added Tested-by from Pavan

Changes v1 -> v2:

 - Dropped netdevsim support and netdevsim selftest (Jakub)
 - Split ethtool_rxfh_contexts_resize_all() into separate validate
   (ethtool_rxfh_ctxs_can_resize) and apply (ethtool_rxfh_ctxs_resize)
   so drivers can validate before closing the device (Jakub)
 - Shortened helper names (Jakub)
 - Replaced scoped_guard(mutex) with explicit mutex_lock/unlock
   (Jakub)
 - Removed defensive zero-size check, bare expressions instead of != 0
   comparisons, ! instead of == 0 (Jakub)
 - In bnxt, moved bnxt_check_rings() before RSS validation and
   deferred actual resize to after bnxt_close_nic() (Jakub, Michael)
 - Added comment that RSS table size only changes on P5 chips with
   older firmware (Michael)
 - Use non-default [3,2,1,0]xN pattern set via netlink to distinguish
   correct fold from driver resetting to defaults (Jakub)
 - Check exact indirection table pattern, not just set(indir) (Jakub)
 - Use ksft_raises() instead of try/except/else (Jakub)
 - Removed queue_count=8 from NetDrvEnv (Jakub)
 - Added ksft_variants to resize_nonperiodic_reject for ctx coverage
 - Added extra periodic context in reject test for multi-context
   validation coverage
 - Added resize_nonperiodic_no_corruption test


Björn Töpel (3):
  ethtool: Add RSS indirection table resize helpers
  bnxt_en: Resize RSS contexts on channel count change
  selftests: rss_drv: Add RSS indirection table resize tests

 .../net/ethernet/broadcom/bnxt/bnxt_ethtool.c |  38 ++-
 include/linux/ethtool.h                       |  11 +
 net/ethtool/common.c                          | 150 +++++++++++
 net/ethtool/ioctl.c                           |  15 +-
 net/ethtool/rss.c                             |  24 +-
 .../selftests/drivers/net/hw/rss_drv.py       | 233 +++++++++++++++++-
 6 files changed, 450 insertions(+), 21 deletions(-)


base-commit: 348baefbb635cbb448e154f38c93657d4cf23936
-- 
2.53.0