linux-kernel - Re: [PATCH v2] staging: wfx: refactor to avoid duplication at hif

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <2040746.q8W4dvF0dS@pc-42>
Date:   Mon, 10 Aug 2020 11:38:33 +0200
From:   Jérôme Pouiller <jerome.pouiller@...abs.com>
To:     Tomer Samara <tomersamara98@...il.com>
Cc:     gregkh@...uxfoundation.org, devel@...verdev.osuosl.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] staging: wfx: refactor to avoid duplication at hif_tx.c

Hello Tomer,

On Wednesday 5 August 2020 14:14:42 CEST Tomer Samara wrote:
> 
> Add functions wfx_full_send(), wfx_full_send_no_reply_async(),
> wfx_full_send_no_reply() and wfx_full_send_no_reply_free()
> which works as follow:
> wfx_full_send() - simple wrapper for both wfx_fill_header()
>                   and wfx_cmd_send().
> wfx_full_send_no_reply_async() - wrapper for both but with
>                                  NULL as reply and size zero.
> wfx_full_send_no_reply() - same as wfx_full_send_no_reply_async()
>                            but with false async value
> wfx_full_send_no_reply_free() - same as wfx_full_send_no_reply()
>                                 but also free the struct hif_msg.
> 
> Signed-off-by: Tomer Samara <tomersamara98@...il.com>
> ---
> Changes in v2:
>  - Changed these functions to static
> 
> drivers/staging/wfx/hif_tx.c | 180 ++++++++++++++++-------------------
>  1 file changed, 80 insertions(+), 100 deletions(-)
> 
> diff --git a/drivers/staging/wfx/hif_tx.c b/drivers/staging/wfx/hif_tx.c
> index 5110f9b93762..17f668fa40a0 100644
> --- a/drivers/staging/wfx/hif_tx.c
> +++ b/drivers/staging/wfx/hif_tx.c
> @@ -40,7 +40,7 @@ static void wfx_fill_header(struct hif_msg *hif, int if_id,
> 
>  static void *wfx_alloc_hif(size_t body_len, struct hif_msg **hif)
>  {
> -       *hif = kzalloc(sizeof(struct hif_msg) + body_len, GFP_KERNEL);
> +       *hif = kzalloc(sizeof(*hif) + body_len, GFP_KERNEL);

This change is not related to the rest of the patch. It should probably be
split out.

>         if (*hif)
>                 return (*hif)->body;
>         else
> @@ -123,9 +123,39 @@ int wfx_cmd_send(struct wfx_dev *wdev, struct hif_msg *request,
>         return ret;
>  }
> 
> +static int wfx_full_send(struct wfx_dev *wdev, struct hif_msg *hif, void *reply,
> +                        size_t reply_len, bool async, int if_id, unsigned int cmd,
> +                        int size)
> +{
> +       wfx_fill_header(hif, if_id, cmd, size);
> +       return wfx_cmd_send(wdev, hif, reply, reply_len, async);
> +}

This function takes 8 parameters. Are you sure it simplifies the code?

In add, it does two actions: modify hif and send it. I would keep these
two actions separated.

> +
> +static int wfx_full_send_no_reply_async(struct wfx_dev *wdev, struct hif_msg *hif, int if_id,
> +                                       unsigned int cmd, int size, bool async)
> +{
> +       return wfx_full_send(wdev, hif, NULL, 0, async, if_id, cmd, size);
> +}

Does it make sense to have a parameter 'async' to
wfx_full_send_no_reply_async()? It is weird to call this function with
async=false, no?

> +
> +static int wfx_full_send_no_reply(struct wfx_dev *wdev, struct hif_msg *hif, int if_id,
> +                                 unsigned int cmd, int size)
> +{
> +       return wfx_full_send_no_reply_async(wdev, hif, if_id, cmd, size, false);
> +}
> +
> +static int wfx_full_send_no_reply_free(struct wfx_dev *wdev, struct hif_msg *hif, int if_id,
> +                                      unsigned int cmd, int size)
> +{
> +       int ret;
> +
> +       ret = wfx_full_send_no_reply(wdev, hif, if_id, cmd, size);
> +       kfree(hif);
> +       return ret;
> +}

One more time, sending the data and releasing are unrelated actions.
Indeed, it saves a few lines of code, but is it really an improvement?

> +
>  // This function is special. After HIF_REQ_ID_SHUT_DOWN, chip won't reply to any
>  // request anymore. We need to slightly hack struct wfx_hif_cmd for that job. Be
> -// carefull to only call this funcion during device unregister.
> +// careful to only call this function during device unregister.

Not related to the rest of the patch.

[...]

As it stands, I think this change does not improve the code. Obviously, it
is a subjective opinion. What the other developers think about it?

-- 
Jérôme Pouiller