DPDK patches and discussions
 help / color / mirror / Atom feed
From: Bruce Richardson <bruce.richardson@intel.com>
To: Kevin Laatz <kevin.laatz@intel.com>
Cc: dev@dpdk.org
Subject: Re: [PATCH v2] dma/idxd: add generic option for queue config
Date: Fri, 1 Apr 2022 10:08:44 +0100	[thread overview]
Message-ID: <YkbBHHq7dVAEVSQm@bricha3-MOBL.ger.corp.intel.com> (raw)
In-Reply-To: <20220331172112.575972-1-kevin.laatz@intel.com>

On Thu, Mar 31, 2022 at 06:21:12PM +0100, Kevin Laatz wrote:
> The device config script currently uses some defaults to configure
> devices in a generic way.
> 
> With the addition of this option, users have more control over how
> queues are configured.
> 
> Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
> ---
>  drivers/dma/idxd/dpdk_idxd_cfg.py | 39 ++++++++++++++++++++++++-------
>  1 file changed, 30 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/dma/idxd/dpdk_idxd_cfg.py b/drivers/dma/idxd/dpdk_idxd_cfg.py
> index 3f5d5ee752..704cf775ed 100755
> --- a/drivers/dma/idxd/dpdk_idxd_cfg.py
> +++ b/drivers/dma/idxd/dpdk_idxd_cfg.py
> @@ -63,8 +63,24 @@ def get_dsa_id(pci):
>      sys.exit(f"Could not get device ID for device {pci}")
>  
>  
> -def configure_dsa(dsa_id, queues, prefix):
> +def parse_wq_opts(wq_opts):
> +    "Parse user-specified queue configuration, creating a dict of options"
> +    opt_dict = {}
> +    for wq_opt in wq_opts:
> +        try:
> +            opt, val = wq_opt.split("=")
> +        except ValueError:
> +            sys.exit("Invalid --wq-option format, use format 'option=value'")
> +        opt_dict[opt] = val
> +    return opt_dict
> +

Can I suggest a shorter version of this:

def parse_wq_opts(wq_opts):
    "Parse ... "
    try:
        return {o.split('=')[0]: o.split('=')[1] for o in wq_opts}
    except:
        ...


> +
> +def configure_dsa(dsa_id, args):
>      "Configure the DSA instance with appropriate number of queues"
> +    queues = args.q
> +    prefix = args.prefix
> +    wq_opts = args.wq_option
> +

Do we really save much using all these variables? The "queues" one might be
useful and clearer, but I wonder how much we save by avoiding the "args."
part of the rest? For example, wq_opts is only used once.

>      dsa_dir = SysfsDir(f"/sys/bus/dsa/devices/dsa{dsa_id}")
>  
>      max_groups = dsa_dir.read_int("max_groups")
> @@ -83,14 +99,17 @@ def configure_dsa(dsa_id, queues, prefix):
>  
>      # configure each queue
>      for q in range(nb_queues):
> +        default_dict = {"group_id": q % nb_groups,
> +                        "type": "user",
> +                        "mode": "dedicated",
> +                        "name": f"{prefix}_wq{dsa_id}.{q}",
> +                        "priority": 1,
> +                        "max_batch_size": 1024,
> +                        "size": int(max_work_queues_size / nb_queues)}
> +        config_dict = parse_wq_opts(wq_opts)
> +        write_dict = {**default_dict, **config_dict}

These two lines could be replaced by:
	default_dict.update(parse_wq_opts(...))
though in this case, you probably want to rename default_dict to
write_dict - or maybe even to "qcfg" or something more meaningful.

>          wq_dir = SysfsDir(os.path.join(dsa_dir.path, f"wq{dsa_id}.{q}"))
> -        wq_dir.write_values({"group_id": q % nb_groups,
> -                             "type": "user",
> -                             "mode": "dedicated",
> -                             "name": f"{prefix}_wq{dsa_id}.{q}",
> -                             "priority": 1,
> -                             "max_batch_size": 1024,
> -                             "size": int(max_work_queues_size / nb_queues)})
> +        wq_dir.write_values(write_dict)
>  
>      # enable device and then queues
>      idxd_dir = SysfsDir(get_drv_dir("idxd"))
> @@ -112,6 +131,8 @@ def main(args):
>      arg_p.add_argument('--name-prefix', metavar='prefix', dest='prefix',
>                         default="dpdk",
>                         help="Prefix for workqueue name to mark for DPDK use [default: 'dpdk']")
> +    arg_p.add_argument('--wq-option', action='append',
> +                       help="Provide additional config option for queues (format 'x=y')")
>      arg_p.add_argument('--reset', action='store_true',
>                         help="Reset DSA device and its queues")
>      parsed_args = arg_p.parse_args(args[1:])
> @@ -121,7 +142,7 @@ def main(args):
>      if parsed_args.reset:
>          reset_device(dsa_id)
>      else:
> -        configure_dsa(dsa_id, parsed_args.q, parsed_args.prefix)
> +        configure_dsa(dsa_id, parsed_args)
>  
>  
>  if __name__ == "__main__":
> -- 
> 2.31.1
> 

  reply	other threads:[~2022-04-01  9:08 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-03-30 15:07 [PATCH] " Kevin Laatz
2022-03-31 14:57 ` Bruce Richardson
2022-03-31 15:47   ` Kevin Laatz
2022-03-31 17:21 ` [PATCH v2] " Kevin Laatz
2022-04-01  9:08   ` Bruce Richardson [this message]
2022-04-01 10:35   ` [PATCH v3] " Kevin Laatz
2022-04-01 10:51     ` Bruce Richardson
2022-06-07 10:47       ` Thomas Monjalon
2022-04-04  9:58     ` Pai G, Sunil

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=YkbBHHq7dVAEVSQm@bricha3-MOBL.ger.corp.intel.com \
    --to=bruce.richardson@intel.com \
    --cc=dev@dpdk.org \
    --cc=kevin.laatz@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).