[PATCH 2/6] spi/spi-pl022: Use GFP_NOWAIT for allocation from tasklet

[Jassi Brar]

On Tue, Aug 9, 2011 at 4:35 PM, Viresh Kumar <viresh.kumar at st.com> wrote:
> tasklets don't allow invocation to sleeping routines. In configure_dma()
> routine, sg_alloc_table() was called with GFP_KERNEL flag and so this causes
> crash when called from tasklet.
>
> Replace GFP_KERNEL with GFP_NOWAIT to get this fixed.
>
> Signed-off-by: Viresh Kumar <viresh.kumar at st.com>
> ---
>  drivers/spi/spi-pl022.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/spi/spi-pl022.c b/drivers/spi/spi-pl022.c
> index f600d00..cbd9afe 100644
> --- a/drivers/spi/spi-pl022.c
> +++ b/drivers/spi/spi-pl022.c
> @@ -1019,11 +1019,11 @@ static int configure_dma(struct pl022 *pl022)
>        pages = (pl022->cur_transfer->len >> PAGE_SHIFT) + 1;
>        dev_dbg(&pl022->adev->dev, "using %d pages for transfer\n", pages);
>
> -       ret = sg_alloc_table(&pl022->sgt_rx, pages, GFP_KERNEL);
> +       ret = sg_alloc_table(&pl022->sgt_rx, pages, GFP_NOWAIT);
>        if (ret)
>                goto err_alloc_rx_sg;
>
> -       ret = sg_alloc_table(&pl022->sgt_tx, pages, GFP_KERNEL);
> +       ret = sg_alloc_table(&pl022->sgt_tx, pages, GFP_NOWAIT);
>        if (ret)
>                goto err_alloc_tx_sg;

Since this could be called from when we actually need the transfer start,
maybe we could try harder using GFP_ATOMIC instead ?