A quick follow-up: Steven Stallion wrote: > Garrett D'Amore wrote: >> 1) no need to initialize macp->m_pdata or macp->p_data_size. (In >> fact, those fields should probably *not* be touched.) > > Gotcha. nemo-design seemed to indicate these needed to be initialized, > (NULL and 0 respectively IIRC) I'll take this out. I'll double check > mac_register() to see if there are any other member initializations that > should be taken out.
mac_alloc() uses kmem_zalloc() so no need for the extra initializations; I dropped all unneccessary member initializations (namely m_instance, m_pdata, and m_pdata_size). >> 3) Just a question: can tx_interrupt_mask in the dnet_mc_tx() routine >> just be treated as a constant? I'm not sure with the removal of the >> GLD_INTR_WAIT code that there is any point to having a separate >> variable here. But I need to look at the whole function to verify that. >> >> 4) Ditto for the handling in dnet_intr(). > > I'll look into this as well. If it can be removed, I'll make sure to get > rid of it. tx_interrupt_mask can be safely removed (replaced with TX_INTERRUPT_MASK directly). I've removed it from dnet_intr() and dnet_mc_tx(). Steve -- Yet magic and hierarchy arise from the same source, and this source has a null pointer. Reference the NULL within NULL, it is the gateway to all wizardry. _______________________________________________ driver-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/driver-discuss
