Messages in this thread | | | From | Ratheesh Kannoth <> | Subject | RE: [PATCH net-next v18 02/13] rtase: Implement the .ndo_open function | Date | Thu, 9 May 2024 09:09:00 +0000 |
| |
> From: Justin Lai <justinlai0215@realtek.com> > Sent: Thursday, May 9, 2024 2:29 PM > > > + > > > + /* rx and tx descriptors needs 256 bytes alignment. > > > + * dma_alloc_coherent provides more. > > > + */ > > > + for (i = 0; i < tp->func_tx_queue_num; i++) { > > > + tp->tx_ring[i].desc = > > > + dma_alloc_coherent(&pdev->dev, > > > + > > RTASE_TX_RING_DESC_SIZE, > > > + > > &tp->tx_ring[i].phy_addr, > > > + GFP_KERNEL); > > > + if (!tp->tx_ring[i].desc) > > You have handled errors gracefully very where else. why not here ? > > I would like to ask you, are you referring to other places where there are error > description messages, but not here? other functions, you are freeing allocated resources in case of failure, but here, you are returning error directly.
> > Did you mark the skb for recycle ? Hmm ... did i miss to find the code ? > > > We have done this part when using the skb and before finally releasing the skb > resource. Do you think it would be better to do this part of the process when > allocating the skb? i think, you added skb_for_recycle() in the following patch. Sorry I missed it . ignore my comment.
> > > > + > > > +err_free_all_allocated_irq: > > You are allocating from i = 1, but freeing from j = 0; > > Hi Ratheesh, > I have done request_irq() once before the for loop, so there should be no > problem starting free from j=0 here. Thanks for pointing out.
| |