6.18-stable review patch. If anyone has any objections, please let me know.
------------------
From: Wei Fang wei.fang@nxp.com
[ Upstream commit a48e232210009be50591fdea8ba7c07b0f566a13 ]
There is a crash issue when running zero copy XDP_TX action, the crash log is shown below.
[ 216.122464] Unable to handle kernel paging request at virtual address fffeffff80000000 [ 216.187524] Internal error: Oops: 0000000096000144 [#1] SMP [ 216.301694] Call trace: [ 216.304130] dcache_clean_poc+0x20/0x38 (P) [ 216.308308] __dma_sync_single_for_device+0x1bc/0x1e0 [ 216.313351] stmmac_xdp_xmit_xdpf+0x354/0x400 [ 216.317701] __stmmac_xdp_run_prog+0x164/0x368 [ 216.322139] stmmac_napi_poll_rxtx+0xba8/0xf00 [ 216.326576] __napi_poll+0x40/0x218 [ 216.408054] Kernel panic - not syncing: Oops: Fatal exception in interrupt
For XDP_TX action, the xdp_buff is converted to xdp_frame by xdp_convert_buff_to_frame(). The memory type of the resulting xdp_frame depends on the memory type of the xdp_buff. For page pool based xdp_buff it produces xdp_frame with memory type MEM_TYPE_PAGE_POOL. For zero copy XSK pool based xdp_buff it produces xdp_frame with memory type MEM_TYPE_PAGE_ORDER0. However, stmmac_xdp_xmit_back() does not check the memory type and always uses the page pool type, this leads to invalid mappings and causes the crash. Therefore, check the xdp_buff memory type in stmmac_xdp_xmit_back() to fix this issue.
Fixes: bba2556efad6 ("net: stmmac: Enable RX via AF_XDP zero-copy") Signed-off-by: Wei Fang wei.fang@nxp.com Reviewed-by: Hariprasad Kelam hkelam@marvell.com Link: https://patch.msgid.link/20251204071332.1907111-1-wei.fang@nxp.com Signed-off-by: Paolo Abeni pabeni@redhat.com Signed-off-by: Sasha Levin sashal@kernel.org --- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 86e912471dea..0dd17179c85d 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -88,6 +88,7 @@ MODULE_PARM_DESC(phyaddr, "Physical device address"); #define STMMAC_XDP_CONSUMED BIT(0) #define STMMAC_XDP_TX BIT(1) #define STMMAC_XDP_REDIRECT BIT(2) +#define STMMAC_XSK_CONSUMED BIT(3)
static int flow_ctrl = 0xdead; module_param(flow_ctrl, int, 0644); @@ -4988,6 +4989,7 @@ static int stmmac_xdp_get_tx_queue(struct stmmac_priv *priv, static int stmmac_xdp_xmit_back(struct stmmac_priv *priv, struct xdp_buff *xdp) { + bool zc = !!(xdp->rxq->mem.type == MEM_TYPE_XSK_BUFF_POOL); struct xdp_frame *xdpf = xdp_convert_buff_to_frame(xdp); int cpu = smp_processor_id(); struct netdev_queue *nq; @@ -5004,9 +5006,18 @@ static int stmmac_xdp_xmit_back(struct stmmac_priv *priv, /* Avoids TX time-out as we are sharing with slow path */ txq_trans_cond_update(nq);
- res = stmmac_xdp_xmit_xdpf(priv, queue, xdpf, false); - if (res == STMMAC_XDP_TX) + /* For zero copy XDP_TX action, dma_map is true */ + res = stmmac_xdp_xmit_xdpf(priv, queue, xdpf, zc); + if (res == STMMAC_XDP_TX) { stmmac_flush_tx_descriptors(priv, queue); + } else if (res == STMMAC_XDP_CONSUMED && zc) { + /* xdp has been freed by xdp_convert_buff_to_frame(), + * no need to call xsk_buff_free() again, so return + * STMMAC_XSK_CONSUMED. + */ + res = STMMAC_XSK_CONSUMED; + xdp_return_frame(xdpf); + }
__netif_tx_unlock(nq);
@@ -5356,6 +5367,8 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue) break; case STMMAC_XDP_CONSUMED: xsk_buff_free(buf->xdp); + fallthrough; + case STMMAC_XSK_CONSUMED: rx_dropped++; break; case STMMAC_XDP_TX: