[PATCH v3 26/26] media: dvb-frontends: tda10048: Make the range of z explicit.

Michael Ira Krufky mkrufky at linuxtv.org
Mon May 13 07:25:46 PDT 2024


On Mon, May 13, 2024 at 9:38 AM Ricardo Ribalda <ribalda at chromium.org> wrote:
>
> On Fri, 3 May 2024 at 16:08, Dragan Simic <dsimic at manjaro.org> wrote:
> >
> > Hello Ricardo,
> >
> > On 2024-05-03 13:56, Ricardo Ribalda wrote:
> > > I am trying to get the DS, but
> > > https://www.nxp.com/acrobat_download/literature/9397/75015931.pdf is a
> > > dead links now.
> > >
> > > Anyone have access to the datasheet?
> >
> > It's kind of available on the link below, but for some strange reason
> > the download fails after downloading the first 128 KB or so.
> >
> > https://web.archive.org/web/20080907185532/https://www.nxp.com/acrobat_download/literature/9397/75015931.pdf\
>
> Mike, by any chance do you have a copy of the DS?
>
>
> >
> >
> > > On Fri, 3 May 2024 at 13:55, Dan Carpenter <dan.carpenter at linaro.org>
> > > wrote:
> > >>
> > >> On Fri, May 03, 2024 at 11:27:58AM +0100, Mauro Carvalho Chehab wrote:
> > >> > Em Mon, 29 Apr 2024 15:05:05 +0000
> > >> > Ricardo Ribalda <ribalda at chromium.org> escreveu:
> > >> >
> > >> > > We do not expect the sample_freq to be over 613MHz.
> > >> > >
> > >> > > Found by cocci:
> > >> > > drivers/media/dvb-frontends/tda10048.c:345:1-7: WARNING: do_div() does a 64-by-32 division, please consider using div64_u64 instead.
> > >> > >
> > >> > > Signed-off-by: Ricardo Ribalda <ribalda at chromium.org>
> > >> > > ---
> > >> > >  drivers/media/dvb-frontends/tda10048.c | 4 +++-
> > >> > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > >> > >
> > >> > > diff --git a/drivers/media/dvb-frontends/tda10048.c b/drivers/media/dvb-frontends/tda10048.c
> > >> > > index 3e725cdcc66b..1886f733dbbf 100644
> > >> > > --- a/drivers/media/dvb-frontends/tda10048.c
> > >> > > +++ b/drivers/media/dvb-frontends/tda10048.c
> > >> > > @@ -328,7 +328,8 @@ static int tda10048_set_wref(struct dvb_frontend *fe, u32 sample_freq_hz,
> > >> > >                          u32 bw)
> > >> > >  {
> > >> > >     struct tda10048_state *state = fe->demodulator_priv;
> > >> > > -   u64 t, z;
> > >> > > +   u32 z;
> > >> > > +   u64 t;
> > >> > >
> > >> > >     dprintk(1, "%s()\n", __func__);
> > >> > >
> > >> > > @@ -341,6 +342,7 @@ static int tda10048_set_wref(struct dvb_frontend *fe, u32 sample_freq_hz,
> > >> > >     /* t *= 2147483648 on 32bit platforms */
> > >> > >     t *= (2048 * 1024);
> > >> > >     t *= 1024;
> > >> > > +   /* Sample frequency is under 613MHz */
> > >> >
> > >> > Are you sure about that? Some DVB devices have very high frequency
> > >> > clocks, specially if they're also used for satellite, so I can't
> > >> > be sure by just looking at the driver's code.
> > >> >
> > >> > Also, we had already a bunch of regressions with "fixes" like this
> > >> > that actually broke frontend drivers.
> > >>
> > >> This patch preserves the existing behavior. The sample_freq_hz
> > >> variable
> > >> is a u32 so, in the original code, z couldn't have been more than
> > >> U32_MAX even though it was declared as a u64.
>
>
> I agree with Dan, we keep the existing behaviour. So it wont hurt to
> merge the code...
>
> All  that said, if someone has access to the DS, I do not mind reviewing it.
>
>
> > >>
> > >> It's possible that the original code was wrong.  We have seen that in
> > >> other places in this patchset.  Adding a note about the datasheet is
> > >> also a good idea.
> > >>
> > >> regards,
> > >> dan carpenter
> > >>
>
>
>
> --
> Ricardo Ribalda
>

Nice to hear from you!  :-)

I believe that I may have a copy of it on an old "spinny" hard drive
somewhere in one of the ancient desktop computers I have lining my
basement walls, lol.  It will take me some time to locate it.  I hope
this isn't urgent o:-)

...It so happens that the dev box I used when I worked on that driver
is up right now, but the datasheet isn't in my home directory.  There
are two other drives in the chassis but not connected / powered - I'll
give these a look and let you know if I find anything.

Best,
Michael Krufky



More information about the Linux-mediatek mailing list