On 12 February 2014 03:36, Peter Crosthwaite <peter.crosthwa...@xilinx.com> wrote: > By just ignoring them and trying again later. This handles the > EGAIN case properly (the previous implementation was only dealing > with short returns and not errors). > > Signed-off-by: Peter Crosthwaite <peter.crosthwa...@xilinx.com> > --- > > hw/char/cadence_uart.c | 7 +++++-- > 1 file changed, 5 insertions(+), 2 deletions(-) > > diff --git a/hw/char/cadence_uart.c b/hw/char/cadence_uart.c > index 1012f1a..1985047 100644 > --- a/hw/char/cadence_uart.c > +++ b/hw/char/cadence_uart.c > @@ -302,8 +302,11 @@ static gboolean cadence_uart_xmit(GIOChannel *chan, > GIOCondition cond, > } > > ret = qemu_chr_fe_write(s->chr, s->tx_fifo, s->tx_count); > - s->tx_count -= ret; > - memmove(s->tx_fifo, s->tx_fifo + ret, s->tx_count); > + > + if (ret >= 0) { > + s->tx_count -= ret; > + memmove(s->tx_fifo, s->tx_fifo + ret, s->tx_count); > + } > > if (s->tx_count) { > int r = qemu_chr_fe_add_watch(s->chr, G_IO_OUT, cadence_uart_xmit, > s);
If we got EAGAIN do we really need to re-add the watch and recompute the UART status, or could we just return early? Incidentally, this looks like the third or fourth patch fixing use of qemu_chr_fe_write() in a UART model; perhaps we should audit the others... thanks -- PMM