Re: [Qemu-devel] [PATCH v2] xilinx_spips: Correct usage of an uninitialized local variable

[francisco iglesias]

On Monday, 22 January 2018, Peter Maydell <address@hidden> wrote:

> On 17 January 2018 at 21:34, Francisco Iglesias
> <address@hidden> wrote:
> > Coverity found that the variable tx_rx in the function
> > xilinx_spips_flush_txfifo was being used uninitialized (CID 1383841).
> This
> > patch corrects this by always initializing tx_rx to zeros.
> >
> > Signed-off-by: Francisco Iglesias <address@hidden>
> >
> > ---
> > v2. Add a sanity check on the num_busses property when realizing the
> >     devices.
> > ---
> >  hw/ssi/xilinx_spips.c | 13 ++++++++++++-
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> >
> > diff --git a/hw/ssi/xilinx_spips.c b/hw/ssi/xilinx_spips.c
> > index 85c5d0c..12f1de9 100644
> > --- a/hw/ssi/xilinx_spips.c
> > +++ b/hw/ssi/xilinx_spips.c
> > @@ -210,6 +210,9 @@
> >  #define SNOOP_NONE 0xEE
> >  #define SNOOP_STRIPING 0
> >
> > +#define MIN_NUM_BUSSES 1
> > +#define MAX_NUM_BUSSES 2
> > +
> >  static inline int num_effective_busses(XilinxSPIPS *s)
> >  {
> >      return (s->regs[R_LQSPI_CFG] & LQSPI_CFG_SEP_BUS &&
> > @@ -573,7 +576,7 @@ static void xilinx_spips_flush_txfifo(XilinxSPIPS
> *s)
> >      for (;;) {
> >          int i;
> >          uint8_t tx = 0;
> > -        uint8_t tx_rx[num_effective_busses(s)];
> > +        uint8_t tx_rx[MAX_NUM_BUSSES] = { 0 };
> >          uint8_t dummy_cycles = 0;
> >          uint8_t addr_length;
> >
> > @@ -1221,6 +1224,14 @@ static void xilinx_spips_realize(DeviceState
> *dev, Error **errp)
> >
> >      DB_PRINT_L(0, "realized spips\n");
> >
> > +    if (s->num_busses < MIN_NUM_BUSSES || s->num_busses >
> MAX_NUM_BUSSES) {
> > +        qemu_log_mask(LOG_GUEST_ERROR,
> > +                      "%s: Invalid number of spi busses requested, "
> > +                      "will default to use one bus\n",
> > +                      __func__);
> > +        s->num_busses = 1;
>
> Realize functions should report errors in their parameters via the Error**
> parameter (which will cause the realize operation to fail, which the caller
> can deal with appropriately). Here's an example from the realize function
> for the GICv3:
>
>     if (s->num_irq > GICV3_MAXIRQ) {
>         error_setg(errp,
>                    "requested %u interrupt lines exceeds GIC maximum %d",
>                    s->num_irq, GICV3_MAXIRQ);
>         return;
>     }
>
> (Note no trailing \n required for error_setg strings, unlike
> qemu_log_mask.)
>
> thanks
> -- PMM
>


Hi Peter,

Thank you very much for reviewing the patch again! I'll correct this in the
next version (3) of the patch.

Best regards,
Francisco Iglesias