Re: [PATCH net-next v3 1/4] net: stmmac: Switch to zero-copy in non-XDP RX path

[Furong Xu <0x1207@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 963 bytes --]

Hi Jon,

On Thu, 23 Jan 2025 14:06:42 +0000, Jon Hunter wrote: 
> We have noticed a boot regression on -next when booting with NFS.
> Bisect is pointing to this commit and reverting this on top of -next
> does fix the problem.
> 
> I only see this on Tegra234 which uses the 
> drivers/net/ethernet/stmicro/stmmac/dwmac-tegra.c driver. Tegra194
> which uses the
> drivers/net/ethernet/stmicro/stmmac/dwmac-dwc-qos-eth.c driver
> appears to be fine.

What is the MTU of Tegra234 and NFS server? Are they both 1500?

Could you please try attached patch to confirm if this regression is
fixed?

If the attached patch fixes this regression, and so it seems to be a
cache coherence issue specific to Tegra234, since this patch avoid
memcpy and the page buffers may be modified by upper network stack of
course, then cache lines of page buffers may become dirty. But by
reverting this patch, cache lines of page buffers never become dirty,
this is the core difference.

[-- Attachment #2: force-disable-rx-checksum.diff --]
[-- Type: text/x-patch, Size: 744 bytes --]

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index edbf8994455d..f00bcfc65dd0 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -5442,7 +5442,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
        struct stmmac_rx_queue *rx_q = &priv->dma_conf.rx_queue[queue];
        struct stmmac_channel *ch = &priv->channel[queue];
        unsigned int count = 0, error = 0, len = 0;
-       int status = 0, coe = priv->hw->rx_csum;
+       int status = 0, coe = 0;
        unsigned int next_entry = rx_q->cur_rx;
        enum dma_data_direction dma_dir;
        unsigned int desc_size;