Date: Mon, 7 Oct 2013 23:20:42 +0200 From: Zbigniew Bodek <zbb@freebsd.org> To: Ian Lepore <ian@freebsd.org> Cc: "freebsd-arm@freebsd.org" <freebsd-arm@freebsd.org>, freebsd-current <freebsd-current@freebsd.org>, "freebsd-embedded@freebsd.org" <freebsd-embedded@freebsd.org> Subject: Re: Changes to UART ns8250 Message-ID: <CALF_Txkt35eSLgc4d52JyCWLRUW%2BsVDq3h=hr_4nWJ53k4PQDg@mail.gmail.com> In-Reply-To: <1381179800.1130.21.camel@revolution.hippie.lan> References: <CALF_Tx=AwVnr0d75-K-yu97iVgmTJC7aaABoix73zHD%2B5eKJnQ@mail.gmail.com> <CAJ-VmokZHvLpUvvD9s8ES0PT%2BE1oVfHWhdbiXF77osBOK2_Yqg@mail.gmail.com> <CALF_TxmfGGeTzgMNVdz-HUn6G=mdgZjiTo8%2BLC8tcrn%2BVvShHg@mail.gmail.com> <1381179385.1130.18.camel@revolution.hippie.lan> <1381179800.1130.21.camel@revolution.hippie.lan>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello Ian.
Well you can't really perform wake event basing on busy interrupt because
we are busy waiting/sleeping i.a. to avoid that interrupt to occur. I tried
to use pause() that should switch to another task for the declared amount
of time and I got:
"panic: mi_switch: switch in a critical section"
Best regards
Zbigniew Bodek
2013/10/7 Ian Lepore <ian@freebsd.org>
> On Mon, 2013-10-07 at 14:56 -0600, Ian Lepore wrote:
> > On Mon, 2013-10-07 at 22:36 +0200, Zbigniew Bodek wrote:
> > > Hello Adrian,
> > >
> > > Thank you for your remarks.
> > > Please check my answers in-line.
> > >
> > > Best regards
> > > Zbigniew Bodek
> > >
> > >
> > > 2013/10/7 Adrian Chadd <adrian@freebsd.org>
> > >
> > > > Hi,
> > > >
> > > > You should add:
> > > >
> > > > * a loop counter, to break out after a while;
> > > >
> > > [ZBB] In general as long as UART is busy we cannot proceed but if the
> > > timeout occurs we could return an error. Do you agree?
> > >
> > > > * a DELAY(1) or something.
> > > >
> > > [ZBB] DELAY is also a busy wait after all. The reasonable solution
> might be
> > > to use ns8250_delay() to get the transmission time and use it for
> timeout
> > > from the first point. I would DELAY(1) in each loop and decrement value
> > > acquired from ns8250_delay(). The loop should break during that time
> or we
> > > return an error. What do you think?
> > >
> >
> > Is it possible to not busy-wait at all? Something like the attached?
> >
> > -- Ian
> >
> > differences between files attachment (temp.diff)
> > Index: sys/dev/uart/uart_dev_ns8250.c
> > ===================================================================
> > --- sys/dev/uart/uart_dev_ns8250.c (revision 255916)
> > +++ sys/dev/uart/uart_dev_ns8250.c (working copy)
> > @@ -614,6 +614,7 @@ ns8250_bus_ipend(struct uart_softc *sc)
> >
> > if (ns8250->busy_detect && (iir & IIR_BUSY) == IIR_BUSY) {
> > (void)uart_getreg(bas, DW_REG_USR);
> > + wakeup(&ns8250->busy_detect);
> > uart_unlock(sc->sc_hwmtx);
> > return (0);
> > }
> > @@ -652,6 +653,16 @@ ns8250_bus_param(struct uart_softc *sc, int baudra
> >
> > bas = &sc->sc_bas;
> > uart_lock(sc->sc_hwmtx);
> > + /*
> > + * When using DW UART with BUSY detection it is necessary to wait
> > + * until all serial transfers are finished before manipulating the
> > + * line control. LCR will not be affected when UART is busy.
> > + */
> > + while (ns8250->busy_detect &&
> > + (uart_getreg(bas, DW_REG_USR) & USR_BUSY) == 0) {
> > + msleep(&ns8250->busy_detect, sc->sc_hwmtx, "dwbusy", 10);
> > + }
> > +
> > error = ns8250_param(bas, baudrate, databits, stopbits, parity);
> > uart_unlock(sc->sc_hwmtx);
> > return (error);
>
> Ooops, that should be != 0 in that USR_BUSY test. Also, I haven't
> tested this at all because I don't I have any boards that use DW uarts.
>
> -- Ian
>
>
>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CALF_Txkt35eSLgc4d52JyCWLRUW%2BsVDq3h=hr_4nWJ53k4PQDg>