[OpenBIOS] Re: [PATCH] escc: explicitly mark variable containing incoming keyboard character as volatile

[Mark Cave-Ayland]

On 25/10/2019 11:28, Segher Boessenkool wrote:
> On Fri, Oct 25, 2019 at 08:42:13AM +0100, Mark Cave-Ayland wrote:
> > On 20/10/2019 13:22, Mark Cave-Ayland wrote:
> > > On 18/10/2019 17:23, Segher Boessenkool wrote:
> > > unsigned char
> > > keyboard_readdata(void)
> > > {
> > >     volatile unsigned char ch;
> > > while (!keyboard_dataready()) { }
> > >
> > > do {
> > >     ch = kbd_dev[2] & 0xff;
> > >     if (ch == 99)
> > >         shiftstate |= 1;
> > >     else if (ch == 110)
> > >         shiftstate |= 2;
> > >     else if (ch == 227)
> > >         shiftstate &= ~1;
> > >     else if (ch == 238)
> > >         shiftstate &= ~2;
> > >     //printk("getch: %d\n", ch);
> > > } // If release, wait for key press
> > > while ((ch & 0x80) == 0x80 || ch == 238 || ch == 227);
> > > //printk("getch rel: %d\n", ch);
> > > ch &= 0x7f;
> > > if (shiftstate)
> > >     ch = sunkbd_keycode_shifted[ch];
> > > else
> > >     ch = sunkbd_keycode[ch];
> > > //printk("getch xlate: %d\n", ch);
> > >
> > > return ch;
> > >
> > > }
>
> I don't see how this would loop, given that kbd_dev is a pointer to volatile.
> > Any further comments? I'd like to get these last couple of patchsets applied fairly
> > soon since it's QEMU freeze starting next week.
>
> I'd use your workaround then?  But something else is going on, it would be
> good to figure out what.

Okay I've gone with this approach for now, probably one to dig into when I have a bit
more time. Applied to master.

ATB,

Mark.