On 2014/10/28 17:41, Markus Armbruster wrote: > Gonglei <arei.gong...@huawei.com> writes: > >> On 2014/10/28 15:35, Markus Armbruster wrote: >> >>> The next commit will exploit the fact it never fails. This one makes >>> it obvious. >>> >>> Signed-off-by: Markus Armbruster <arm...@redhat.com> >>> --- >>> hw/net/lance.c | 3 ++- >>> hw/net/pcnet-pci.c | 3 ++- >>> hw/net/pcnet.c | 4 +--- >>> hw/net/pcnet.h | 2 +- >>> 4 files changed, 6 insertions(+), 6 deletions(-) >>> >>> diff --git a/hw/net/lance.c b/hw/net/lance.c >>> index a1c49f1..3663340 100644 >>> --- a/hw/net/lance.c >>> +++ b/hw/net/lance.c >>> @@ -134,7 +134,8 @@ static int lance_init(SysBusDevice *sbd) >>> >>> s->phys_mem_read = ledma_memory_read; >>> s->phys_mem_write = ledma_memory_write; >>> - return pcnet_common_init(dev, s, &net_lance_info); >>> + pcnet_common_init(dev, s, &net_lance_info); >>> + return 0; >>> } >>> >>> static void lance_reset(DeviceState *dev) >>> diff --git a/hw/net/pcnet-pci.c b/hw/net/pcnet-pci.c >>> index fb5f5d6..50eb069 100644 >>> --- a/hw/net/pcnet-pci.c >>> +++ b/hw/net/pcnet-pci.c >>> @@ -335,7 +335,8 @@ static int pci_pcnet_init(PCIDevice *pci_dev) >>> s->phys_mem_write = pci_physical_memory_write; >>> s->dma_opaque = pci_dev; >>> >>> - return pcnet_common_init(DEVICE(pci_dev), s, &net_pci_pcnet_info); >>> + pcnet_common_init(DEVICE(pci_dev), s, &net_pci_pcnet_info); >>> + return 0; >>> } >>> >>> static void pci_reset(DeviceState *dev) >>> diff --git a/hw/net/pcnet.c b/hw/net/pcnet.c >>> index d344c15..5a081c4 100644 >>> --- a/hw/net/pcnet.c >>> +++ b/hw/net/pcnet.c >>> @@ -1724,7 +1724,7 @@ void pcnet_common_cleanup(PCNetState *d) >>> d->nic = NULL; >>> } >>> >>> -int pcnet_common_init(DeviceState *dev, PCNetState *s, NetClientInfo *info) >>> +void pcnet_common_init(DeviceState *dev, PCNetState *s, NetClientInfo >>> *info) >> >> Do we need consider to pass an Error **errp argument to it? > > This function can't fail. The point of thimy patch is to make "can't > fail" obvious. If we add an errp parameter, the caller needs to check > it, for robustness.
Yes, it is. > I prefer to keep things simple, and add the error > checking only when it's actually needed. > OK, it's fine :) Reviewed-by: Gonglei <arei.gong...@huawei.com> Best regards, -Gonglei
