Re: [PATCH 3/6] staging: r8188eu: cnt is set but not used

From: Martin Kaiser <martin@kaiser.cx>
To: Pavel Skripkin <paskripkin@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Larry Finger <Larry.Finger@lwfinger.net>,
	Phillip Potter <phil@philpotter.co.uk>,
	Michael Straube <straube.linux@gmail.com>,
	linux-staging@lists.linux.dev, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 3/6] staging: r8188eu: cnt is set but not used
Date: Sat, 26 Feb 2022 19:07:06 +0100	[thread overview]
Message-ID: <20220226180706.wierfb2ancom32zn@viti.kaiser.cx> (raw)
In-Reply-To: <184f741c-60ed-fe32-5705-08f9769e926a@gmail.com>

Thus wrote Pavel Skripkin (paskripkin@gmail.com):

> On 2/26/22 19:57, Pavel Skripkin wrote:
> > Hi Martin,

> > On 2/26/22 17:48, Martin Kaiser wrote:
> > > In function recv_func, the cnt variable is set but not used.
> > > It can be removed.

> > > Signed-off-by: Martin Kaiser <martin@kaiser.cx>
> > > ---
> > >   drivers/staging/r8188eu/core/rtw_recv.c | 2 --
> > >   1 file changed, 2 deletions(-)

> > > diff --git a/drivers/staging/r8188eu/core/rtw_recv.c b/drivers/staging/r8188eu/core/rtw_recv.c
> > > index 0144c4642911..9a2e2bc2e294 100644
> > > --- a/drivers/staging/r8188eu/core/rtw_recv.c
> > > +++ b/drivers/staging/r8188eu/core/rtw_recv.c
> > > @@ -1798,11 +1798,9 @@ static int recv_func(struct adapter *padapter, struct recv_frame *rframe)
> > >   	if (check_fwstate(mlmepriv, WIFI_STATION_STATE) &&
> > >   	    psecuritypriv->busetkipkey) {
> > >   		struct recv_frame *pending_frame;
> > > -		int cnt = 0;
> > >   		pending_frame = rtw_alloc_recvframe(&padapter->recvpriv.uc_swdec_pending_queue);
> > >   		while (pending_frame) {

> > Just out of curiosity: is this thing infinity loop?

> Hm.

> This function is called only inside a tasklet. IIRC it's not ok to have this
> kind of loops in softirq context

What exactly is this doing that's not allowed in a tasklet? Does it call
anything that could potentially block? 

If I pull rtw_alloc_recvframe into the loop, that function uses spin_lock_bh +
spin_unlock_bh, I guess this is ok. As for recv_func_posthandle, I don't
see anything where we could be stuck...

Thanks,
Martin