On Fri, Apr 6, 2018 at 7:44 PM, Mauro Carvalho Chehab
<mchehab@xxxxxxxxxx> wrote:
>
> - if (output->dma)
> + if (has_dma)
> spin_lock_irq(&output->dma->lock);
>
> ddbwritel(dev, 0, TS_CONTROL(output));
>
> - if (output->dma) {
> + if (has_dma) {
> ddbwritel(dev, 0, DMA_BUFFER_CONTROL(output->dma));
> output->dma->running = 0;
> spin_unlock_irq(&output->dma->lock);
>
> Even in this case, where has_dma is const, sparse still generates the
> warning. Just in case, I also tried changing has_dma to just
> "bool" and just "int".
Sparse will always warn for this, for the simple reason that the basic
block there in the middle that does that
ddbwritel(dev, 0, TS_CONTROL(output));
has two different locking contexts.
So the only way to get sparse to not warn is to have something like
if (output->dma) {
spin_lock_irq(&output->dma->lock);
ddbwritel(dev, 0, TS_CONTROL(output));
ddbwritel(dev, 0, DMA_BUFFER_CONTROL(output->dma));
output->dma->running = 0;
spin_unlock_irq(&output->dma->lock);
} else {
ddbwritel(dev, 0, TS_CONTROL(output));
}
which probably generates better code too when that "shared" region is
smaller than the rest of the code.
Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
