| Message ID | 108883742c4b44d338ea0816abeeb2901a2afcd5.1448361028.git.nsekhar@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tuesday 24 November 2015 04:04 PM, Sekhar Nori wrote: > davinci_spi_bufs() uses wait_for_completion_interruptible() > without bothering to handle -ERESTARTSYS. Due to this, > sometime, it returns prematurely when a signal is received. > Since the return value is never checked, userspace eventually > receives a spurious -EIO. > > To fix this, use un-interruptible wait_for_completion_timeout(). > > Signed-off-by: Sekhar Nori <nsekhar@ti.com> > --- > drivers/spi/spi-davinci.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c > index 7d3af3eacf57..38026aa1afd7 100644 > --- a/drivers/spi/spi-davinci.c > +++ b/drivers/spi/spi-davinci.c > @@ -703,7 +703,10 @@ static int davinci_spi_bufs(struct spi_device *spi, struct spi_transfer *t) > > /* Wait for the transfer to complete */ > if (spicfg->io_type != SPI_IO_TYPE_POLL) { > - wait_for_completion_interruptible(&(dspi->done)); > + if (wait_for_completion_timeout(&dspi->done, HZ) == 0) { > + dev_err(&spi->dev, "spi transfer timeout\n"); > + return -ETIMEDOUT; Grygorii pointed out offline that this does not recover correctly in DMA case. I will fix and send a v2. Thanks, Sekhar
diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c index 7d3af3eacf57..38026aa1afd7 100644 --- a/drivers/spi/spi-davinci.c +++ b/drivers/spi/spi-davinci.c @@ -703,7 +703,10 @@ static int davinci_spi_bufs(struct spi_device *spi, struct spi_transfer *t) /* Wait for the transfer to complete */ if (spicfg->io_type != SPI_IO_TYPE_POLL) { - wait_for_completion_interruptible(&(dspi->done)); + if (wait_for_completion_timeout(&dspi->done, HZ) == 0) { + dev_err(&spi->dev, "spi transfer timeout\n"); + return -ETIMEDOUT; + } } else { while (dspi->rcount > 0 || dspi->wcount > 0) { errors = davinci_spi_process_events(dspi);
davinci_spi_bufs() uses wait_for_completion_interruptible() without bothering to handle -ERESTARTSYS. Due to this, sometime, it returns prematurely when a signal is received. Since the return value is never checked, userspace eventually receives a spurious -EIO. To fix this, use un-interruptible wait_for_completion_timeout(). Signed-off-by: Sekhar Nori <nsekhar@ti.com> --- drivers/spi/spi-davinci.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)