On Wed, Jan 24, 2024 at 04:27:51PM +0900, Tomasz Figa wrote:
> On Wed, Jan 24, 2024 at 4:18 PM Sergey Senozhatsky
> <senozhatsky@xxxxxxxxxxxx> wrote:
> >
> > On (24/01/24 11:31), Chen, Jason Z wrote:
> > [..]
> > > /* Write registers up to 4 at a time */
> > > static int ov08x40_write_reg(struct ov08x40 *ov08x,
> > > u16 reg, u32 len, u32 __val)
> > > @@ -2936,6 +1827,19 @@ static int ov08x40_start_streaming(struct ov08x40 *ov08x)
> > > return ret;
> > > }
> > >
> > > + /* Use i2c burst to write register on full size registers */
> > > + if (ov08x->cur_mode->exposure_shift == 1) {
> > > + ret = ov08x40_burst_fill_regs(ov08x, OV08X40_REG_XTALK_FIRST_A,
> > > + OV08X40_REG_XTALK_LAST_A, 0x75);
> > > + ret = ov08x40_burst_fill_regs(ov08x, OV08X40_REG_XTALK_FIRST_B,
> > > + OV08X40_REG_XTALK_LAST_B, 0x75);
> >
> > If we have to be pedantic then second ov08x40_burst_fill_regs()
> > overwrites whatever first ov08x40_burst_fill_regs() returns.
> >
> > But it's unlikely to be a real problem.
>
> Hmm, good point. I guess in theory we could have a situation when the
> first write fails and the second succeeds and no error is returned
> ultimately... I guess I'll leave the final call on this one to Sakari,
> as the maintainer. :)
It's a bug, isn't it?
--
Sakari Ailus
