On Thu, Oct 18, 2012 at 4:59 PM, Felipe Balbi <[email protected]> wrote: > On Thu, Oct 18, 2012 at 02:09:43PM +0300, Andy Shevchenko wrote: >> On Thu, 2012-10-18 at 13:34 +0300, Felipe Balbi wrote: >> > On Thu, Oct 18, 2012 at 11:15:31AM +0300, Andy Shevchenko wrote: >> > > On Wed, 2012-10-17 at 16:53 +0300, Felipe Balbi wrote: >> > > > On Wed, Oct 17, 2012 at 04:36:58PM +0300, Andy Shevchenko wrote: >> > > > > On Wed, 2012-10-17 at 16:09 +0300, Felipe Balbi wrote: >> >> > > > > > > diff --git a/drivers/dma/dw_dmac.c b/drivers/dma/dw_dmac.c >> >> > > > > > > @@ -492,10 +491,8 @@ static void dwc_handle_error(struct dw_dma >> > > > > > > *dw, struct dw_dma_chan *dwc) >> > > > > > > * controller flagged an error instead of scribbling over >> > > > > > > * random memory locations. >> > > > > > > */ >> > > > > > > - dev_printk(KERN_CRIT, chan2dev(&dwc->chan), >> > > > > > > - "Bad descriptor submitted for DMA!\n"); >> > > > > > > - dev_printk(KERN_CRIT, chan2dev(&dwc->chan), >> > > > > > > - " cookie: %d\n", bad_desc->txd.cookie); >> > > > > > > + dev_crit(chan2dev(&dwc->chan), "Bad descriptor submitted for >> > > > > > > DMA!\n"); >> > > > > > > + dev_crit(chan2dev(&dwc->chan), " cookie: %d\n", >> > > > > > > bad_desc->txd.cookie); >> > > > > > >> > > > > > now this is critical, indeed. I would suggest using >> > > > > > dev_WARN_ONCE() so >> > > > > > that it's noisy enough to catch the failing user. >> > > > > To this and upper comment, there is an explanation why it's >> > > > > critical. I >> > > > > guess the WARN_ONCE is not good enough, for example if we have more >> > > > > than >> > > > > one user making such noise. >> > > > >> > > > then use dev_WARN() >> > > I can't see how dev_WARN could be more useful here than the dev_crit. In >> > > current message we have channel and cookie to link back to the user. >> > > What does WARN add meaningful? >> > >> > a dump_stack() >> >> How could it be useful? The dwc_handle_error is called from a tasklet >> that is called from scheduler asynchronously. The tasklet is queued in >> interrupt handler. > > even if it's not useful, it's a lot more verbose and more likely to get > user's attention. If someone's passing broken DMA descriptor, it should > be a really big fat warning so we can get user reports early enoough.
I will do separate patch for this. If Viresh and/or Vinoud is okay with it, user will see fat message. -- With Best Regards, Andy Shevchenko -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
