From: Tom Rix <[email protected]>
Clang build fails with
spi-sunplus-sp7021.c:405:2: error: variable 'ret' is used
uninitialized whenever switch default is taken
default:
^~~~~~~
Restore initializing ret.
Fixes: 47e8fe57a66f ("spi: Modify irq request position and modify parameters")
Signed-off-by: Tom Rix <[email protected]>
---
drivers/spi/spi-sunplus-sp7021.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/spi/spi-sunplus-sp7021.c b/drivers/spi/spi-sunplus-sp7021.c
index ba5ed9f7277a3..fc61578ce5c47 100644
--- a/drivers/spi/spi-sunplus-sp7021.c
+++ b/drivers/spi/spi-sunplus-sp7021.c
@@ -375,7 +375,7 @@ static int sp7021_spi_slave_transfer_one(struct spi_controller *ctlr, struct spi
{
struct sp7021_spi_ctlr *pspim = spi_master_get_devdata(ctlr);
struct device *dev = pspim->dev;
- int mode, ret;
+ int mode, ret = 0;
mode = SP7021_SPI_IDLE;
if (xfer->tx_buf && xfer->rx_buf) {
--
2.26.3
On Wed, Feb 09, 2022 at 08:44:21AM -0800, Tom Rix wrote:
> On 2/9/22 8:33 AM, Mark Brown wrote:
> > On Wed, Feb 09, 2022 at 08:30:29AM -0800, [email protected] wrote:
> > > Clang build fails with
> > > spi-sunplus-sp7021.c:405:2: error: variable 'ret' is used
> > > uninitialized whenever switch default is taken
> > > default:
> > > ^~~~~~~
> > > Restore initializing ret.
> > Are you sure this is a good fix and that the compiler hasn't instead
> > identifed that we should be returning an error in the default case?
> If setting mode to idle is not really a valid state, then default should
> return an error.
Yes, that's what I'm asking.
> The old behavior returned 0 and I am treating this as a regression.
The goal here isn't just to shut up warnings, it's to fix any problems
that they identify. Unconditionally initialising return values is a
common way of fixing warnings while leaving real problems in place.
>Yes, that's what I'm asking.
>The original setting was ret=0; I just wanted to remove it when defining
>the ret, but it looks like it's required.
>Yes, it should return error at default
>I will submit new patch for this.
> The old behavior returned 0 and I am treating this as a regression.
The goal here isn't just to shut up warnings, it's to fix any problems
that they identify. Unconditionally initialising return values is a
common way of fixing warnings while leaving real problems in place.
The original setting was ret=0; I just wanted to remove it when defining
the ret, but it looks like it's required.
Yes, it should return error at default
I will submit new patch for this.