Re: [PATCH iwl-next 06/14] libeth: add bookkeeping support for control queue messages

[Simon Horman <horms@kernel.org>]

On Tue, Apr 08, 2025 at 02:47:52PM +0200, Larysa Zaremba wrote:
> From: Phani R Burra <phani.r.burra@intel.com>
> 
> All send control queue messages are allocated/freed in libeth itself
> and tracked with the unique transaction (Xn) ids until they receive
> response or time out. Responses can be received out of order, therefore
> transactions are stored in an array and tracked though a bitmap.
> 
> Pre-allocated DMA memory is used where possible. It reduces the driver
> overhead in handling memory allocation/free and message timeouts.
> 
> Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
> Signed-off-by: Phani R Burra <phani.r.burra@intel.com>
> Co-developed-by: Victor Raj <victor.raj@intel.com>
> Signed-off-by: Victor Raj <victor.raj@intel.com>
> Co-developed-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
> Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
> Co-developed-by: Larysa Zaremba <larysa.zaremba@intel.com>
> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com>
> ---
>  drivers/net/ethernet/intel/libeth/controlq.c | 578 +++++++++++++++++++
>  include/net/libeth/controlq.h                | 169 ++++++
>  2 files changed, 747 insertions(+)
> 
> diff --git a/drivers/net/ethernet/intel/libeth/controlq.c b/drivers/net/ethernet/intel/libeth/controlq.c

...

> +/**
> + * libeth_ctlq_xn_deinit - deallocate and free the transaction manager resources
> + * @xnm: pointer to the transaction manager
> + * @ctx: controlq context structure
> + *
> + * All Rx processing must be stopped beforehand.
> + */
> +void libeth_ctlq_xn_deinit(struct libeth_ctlq_xn_manager *xnm,
> +			   struct libeth_ctlq_ctx *ctx)
> +{
> +	bool must_wait = false;
> +	u32 i;
> +
> +	/* Should be no new clear bits after this */
> +	spin_lock(&xnm->free_xns_bm_lock);
> +		xnm->shutdown = true;

nit: The line above is not correctly indented.

     Flagged by Smatch.

> +
> +	for_each_clear_bit(i, xnm->free_xns_bm, LIBETH_CTLQ_MAX_XN_ENTRIES) {
> +		struct libeth_ctlq_xn *xn = &xnm->ring[i];
> +
> +		spin_lock(&xn->xn_lock);
> +
> +		if (xn->state == LIBETH_CTLQ_XN_WAITING ||
> +		    xn->state == LIBETH_CTLQ_XN_IDLE) {
> +			complete(&xn->cmd_completion_event);
> +			must_wait = true;
> +		} else if (xn->state == LIBETH_CTLQ_XN_ASYNC) {
> +			__libeth_ctlq_xn_push_free(xnm, xn);
> +		}
> +
> +		spin_unlock(&xn->xn_lock);
> +	}
> +
> +	spin_unlock(&xnm->free_xns_bm_lock);
> +
> +	if (must_wait)
> +		wait_for_completion(&xnm->can_destroy);
> +
> +	libeth_ctlq_xn_deinit_dma(&ctx->mmio_info.pdev->dev, xnm,
> +				  LIBETH_CTLQ_MAX_XN_ENTRIES);
> +	kfree(xnm);
> +	libeth_ctlq_deinit(ctx);
> +}
> +EXPORT_SYMBOL_NS_GPL(libeth_ctlq_xn_deinit, "LIBETH_CP");

...