On 18.05.22 17:47, Jakub Kicinski wrote:
On Wed, 18 May 2022 11:53:12 +0200 Lorenzo Bianconi wrote:
> > + WRITE_ONCE(desc->txd7, 0);
> > + WRITE_ONCE(desc->txd8, 0);
>
> Why all the WRITE_ONCE()? Don't you just need a barrier between writing
> the descriptor and kicking the HW?
I used this approach just to be aligned with current codebase:
https://github.com/torvalds/linux/blob/master/drivers/net/ethernet/mediatek/mtk_eth_soc.c#L1006
https://github.com/torvalds/linux/blob/master/drivers/net/ethernet/mediatek/mtk_eth_soc.c#L1031
but I guess we can even convert the code to use barrier instead. Agree?
Oh, I didn't realize. No preference on converting the old code
but it looks like a cargo cult to me so in the new code let's
not WRITE_ONCE() all descriptor writes unless there's a reason.
If I remember correctly, the existing places use WRITE_ONCE to prevent
write tearing to uncached memory.
- Felix
