[Qemu-devel] Re: [PATCH 1/4] Add a scatter-gather list type and accessors

[Avi Kivity]

Anthony Liguori wrote:
> Avi Kivity wrote:
> > Scatter-gather lists are used extensively in dma-capable devices; a
> > single data structure allows more code reuse later on.
> >
> > diff --git a/dma-helpers.c b/dma-helpers.c
> > new file mode 100644
> > index 0000000..315834e
> > --- /dev/null
> > +++ b/dma-helpers.c
> > @@ -0,0 +1,29 @@
> >   
>
> Needs copyright/license.

Sure.

Is it possible to have a blanket license for files which don't have 
explicit terms?  I don't much like boilerplate.

> > +#include "dma.h"
> > +
> > +
> > +void qemu_sglist_init(QEMUSGList *qsg, int alloc_hint)
> > +{
> > +    qsg->sg = qemu_malloc(alloc_hint * sizeof(ScatterGatherEntry));
> >   
>
> Would be nice to check for malloc failures and fail gracefully at least.

Do you mean an exit(1)?  If so we could just put it in qemu_malloc().

Propagation is usually not possible since hardware is not expected to 
run out of memory.  It will also be a never-tested code path.

> > +void qemu_sglist_add(QEMUSGList *qsg, target_phys_addr_t base,
> > +                     target_phys_addr_t len)
> > +{
> > +    if (qsg->nsg == qsg->nalloc) {
> > +        qsg->nalloc = 2 * qsg->nalloc + 1;
> >   
>
> Do you really want exponential growth verses linear growth?  The sg 
> lists should be relatively small so linear growth should be fine.

I expect this to trigger rarely since the allocation hint should suffice 
nearly 100% of the time.  But in case we miss, it's better to reallocate 
as little as possible.

(what I really want is std::vector<>)

--
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.