linux-kernel - Re: [PATCH] Staging: comedi: drivers: comedi

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: <20170221163141.GA22866@linux-Precision-WorkStation-T5500>
Date:   Wed, 22 Feb 2017 00:31:42 +0800
From:   Cheah Kok Cheong <thrust73@...il.com>
To:     Valentin Rothberg <valentinrothberg@...il.com>
Cc:     Ian Abbott <abbotti@....co.uk>, hsweeten@...ionengravers.com,
        gregkh@...uxfoundation.org, devel@...verdev.osuosl.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] Staging: comedi: drivers: comedi_test: Avoid multiple
 line dereference

On Tue, Feb 21, 2017 at 11:20:08AM +0100, Valentin Rothberg wrote:
> On Feb 21 '17 10:12, Ian Abbott wrote:
> > On 21/02/2017 09:33, Cheah Kok Cheong wrote:
> > > On Mon, Feb 20, 2017 at 05:36:52PM +0000, Ian Abbott wrote:
> > > > On 20/02/17 16:02, Cheah Kok Cheong wrote:
> > > > > On Mon, Feb 20, 2017 at 10:03:39AM +0000, Ian Abbott wrote:
> > > > > > On 20/02/17 08:28, Cheah Kok Cheong wrote:
> > > > > > > Fix checkpatch warning "Avoid multiple line dereference"
> > > > > > > using a local variable to avoid line wrap.
> > > > > > > 
> > > > > > > Signed-off-by: Cheah Kok Cheong <thrust73@...il.com>
> > > > > > > ---
> > > > > > > drivers/staging/comedi/drivers/comedi_test.c | 6 ++----
> > > > > > > 1 file changed, 2 insertions(+), 4 deletions(-)
> > > > > > > 
> > > > > > > diff --git a/drivers/staging/comedi/drivers/comedi_test.c b/drivers/staging/comedi/drivers/comedi_test.c
> > > > > > > index 2a063f0..fde83e0 100644
> > > > > > > --- a/drivers/staging/comedi/drivers/comedi_test.c
> > > > > > > +++ b/drivers/staging/comedi/drivers/comedi_test.c
> > > > > > > @@ -480,11 +480,9 @@ static void waveform_ao_timer(unsigned long arg)
> > > > > > > 			/* output the last scan */
> > > > > > > 			for (i = 0; i < cmd->scan_end_arg; i++) {
> > > > > > > 				unsigned int chan = CR_CHAN(cmd->chanlist[i]);
> > > > > > > +				unsigned short d = devpriv->ao_loopbacks[chan];
> > > > > > > 
> > > > > > > -				if (comedi_buf_read_samples(s,
> > > > > > > -							    &devpriv->
> > > > > > > -							     ao_loopbacks[chan],
> > > > > > > -							    1) == 0) {
> > > > > > > +				if (!comedi_buf_read_samples(s, &d, 1)) {
> > > > > > > 					/* unexpected underrun! (cancelled?) */
> > > > > > > 					async->events |= COMEDI_CB_OVERFLOW;
> > > > > > > 					goto underrun;
> > > > > > > 
> > > > > > 
> > > > > > NAK.  This leaves devpriv->ao_loopbacks[chan] unchanged.
> > > > > > 
> > > > > 
> > > > > Thanks for pointing this out. In that case will assigning the variable to
> > > > > devpriv->ao_loopbacks[chan] be acceptable? Please review below snippet.
> > > > > 
> > > > > Otherwise I'll just drop the variable and adjust the lines to avoid
> > > > > checkpatch warning.
> > > > > 
> > > > > Sorry for the inconvenience caused.
> > > > > 
> > > > > [ Snip ]
> > > > > 
> > > > > 			/* output the last scan */
> > > > > 			for (i = 0; i < cmd->scan_end_arg; i++) {
> > > > > 				unsigned int chan = CR_CHAN(cmd->chanlist[i]);
> > > > > 				unsigned short data;
> > > > > 
> > > > > 				if (!comedi_buf_read_samples(s, &data, 1)) {
> > > > > 					/* unexpected underrun! (cancelled?) */
> > > > > 					async->events |= COMEDI_CB_OVERFLOW;
> > > > > 					goto underrun;
> > > > > 				}
> > > > > 
> > > > > 				devpriv->ao_loopbacks[chan] = data;
> > > > > 			}
> > > > > 			/* advance time of last scan */
> > > > > 
> > > > > [ Snip ]
> > > > 
> > > > It will work, but you could just use a pointer variable set to
> > > > &devpriv->ao_loopbacks[chan] and pass that to comedi_buf_read_samples().
> > > > 
> > > 
> > > Thanks for the suggestion. I tried below snippet 1 with the shortest pointer
> > > name but 80 characters is exceeded. The declaration and initialisation
> > > will have to be splitted. Will this be acceptable or am I doing it wrong
> > > again?
> > > 
> > > Sorry for the trouble.
> > > 
> > > Snippet 1:
> > > [ Snip ]
> > > 
> > > 			/* output the last scan */
> > > 			for (i = 0; i < cmd->scan_end_arg; i++) {
> > > 				unsigned int chan = CR_CHAN(cmd->chanlist[i]);
> > > 				unsigned short *p = &devpriv->ao_loopbacks[chan];
> > > 
> > > 				if (!comedi_buf_read_samples(s, p, 1)) {
> > > 					/* unexpected underrun! (cancelled?) */
> > > 					async->events |= COMEDI_CB_OVERFLOW;
> > > 					goto underrun;
> > > 				}
> > > 			}
> > > 			/* advance time of last scan */
> > > 
> > > [ Snip ]
> > > 
> > > Snippet 2:
> > > [ Snip ]
> > > 
> > > 			/* output the last scan */
> > > 			for (i = 0; i < cmd->scan_end_arg; i++) {
> > > 				unsigned int chan = CR_CHAN(cmd->chanlist[i]);
> > > 				unsigned short *pd;
> > > 
> > > 				pd = &devpriv->ao_loopbacks[chan];
> > > 
> > > 				if (!comedi_buf_read_samples(s, pd, 1)) {
> > > 					/* unexpected underrun! (cancelled?) */
> > > 					async->events |= COMEDI_CB_OVERFLOW;
> > > 					goto underrun;
> > > 				}
> > > 			}
> > > 
> > > [ Snip ]
> > 
> > Snippet 2 looks fine.  Alternatives are to modify Snippet 1 to split the
> > initialization of the pointer variable after the '=', or to shorten the the
> > name of the 'chan' variable.

I'm tempted to shorten the 'chan' variable but this will break consistency
since it's also use in static int waveform_ai_insn_read() and
static int waveform_ao_insn_write(). I'll send Snippet 2 as V2.

> 
> Another option could be using the typedefs from include/linux/types.h,
> e.g. ushort.  However, this might require changing other declarations as
> well to keep consistency.

Thanks for the idea. I counted seven instances of 'unsigned short'
in this file so it's not viable for our situation.

Thks.
Brgds,
CheahKC