On Tue, Feb 21, 2012 at 10:09:01AM +0100, Kevin Wolf wrote:
> Am 21.02.2012 09:35, schrieb Paolo Bonzini:
> > On 02/20/2012 11:50 AM, Alexander Graf wrote:
> >>>> DMAAIOCB *dbs = qemu_aio_get(&dma_aio_pool, bs, cb, opaque);
> >>>>
> >>>> - trace_dma_bdrv_io(dbs, bs, sector_num, to_dev);
> >>>> + trace_dma_bdrv_io(dbs, bs, sector_num, dir);
> >> Was the trace wrong before or is it now? I don't see its definition
> >> changed anywhere.
> >
> > Not sure what you mean. :)
>
> trace-events:
>
> dma_bdrv_io(void *dbs, void *bs, int64_t sector_num, bool to_dev)
> "dbs=%p bs=%p sector_num=%" PRId64 " to_dev=%d"
Ah, damnit. Didn't find that file. I'll resubmit with trace-events
updated too.
> to_dev is declared bool here, and it should also be renamed to dir (the
> unfortunate thing about DMADirection is that it swaps 0 and 1 compared
> to bool to_dev... We need to check carefully that all occurrences have
> been caught.)
Yeah. And even more unfortunate that afaict there's no way to make
gcc warn on enum<->bool conversions. Still there are only about 4
callers of dma_bdrv_io() - nearly everything uses the
dma_bdrv_{read,write}() wrappers - so I'm confident I got them all.
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
