net: wan: fsl_qmc_hdlc: Discard received CRC
Herve Codina
maintainerPATCH 1/1 net: wan: fsl_qmc_hdlc: Discard received CRC
Received frame from QMC contains the CRC.
Upper layers don't need this CRC and tcpdump mentioned trailing junk data due to this CRC presence.
As some other HDLC driver, simply discard this CRC.
--- drivers/net/wan/fsl_qmc_hdlc.c | 24 ++++++++++++++++++------ 1 file changed, 18 insertions(+), 6 deletions(-)
drivers/net/wan/fsl_qmc_hdlc.c +18 -6
@@ -250,6 +250,7 @@ static void qmc_hcld_recv_complete(void *context, size_t length, unsigned int flstruct qmc_hdlc_desc *desc = context;struct net_device *netdev;struct qmc_hdlc *qmc_hdlc; +size_t crc_size;int ret; netdev = desc->netdev;
@@ -268,15 +269,26 @@ static void qmc_hcld_recv_complete(void *context, size_t length, unsigned int flif (flags &QMC_RX_FLAG_HDLC_CRC )/* CRC error */ netdev->stats.rx_crc_errors++;kfree_skb (desc->skb); - }else { - netdev->stats.rx_packets++; - netdev->stats.rx_bytes += length; + goto re_queue; + } -skb_put (desc->skb, length); - desc->skb->protocol =hdlc_type_trans (desc->skb, netdev); -netif_rx (desc->skb); +/* Discard the CRC */ + crc_size = qmc_hdlc->is_crc32 ?4 :2 ; +if (length < crc_size) { + netdev->stats.rx_length_errors++; +kfree_skb (desc->skb); + goto re_queue; } + length -= crc_size; + + netdev->stats.rx_packets++; + netdev->stats.rx_bytes += length; + +skb_put (desc->skb, length); + desc->skb->protocol =hdlc_type_trans (desc->skb, netdev); +netif_rx (desc->skb); +re_queue:/* Re-queue a transfer using the same descriptor */ ret =qmc_hdlc_recv_queue (qmc_hdlc, desc, desc->dma_size);if (ret) {
Signature · 1 line
2.45.0
Quoted 5 lines · Herve Codina
Received frame from QMC contains the CRC.
Upper layers don't need this CRC and tcpdump mentioned trailing junk data due to this CRC presence.As some other HDLC driver, simply discard this CRC.
It might be nice to specifically site an example.
But yes, I see this pattern in hdlc_rx_done().
Quoted 7 lines
Fixes d0f2258e79fd ("net: wan: Add support for QMC HDLC")Signed-off-by Herve Codina <[email protected]>--- drivers/net/wan/fsl_qmc_hdlc.c | 24 ++++++++++++++++++------ 1 file changed, 18 insertions(+), 6 deletions(-)
The above notwithstanding, this looks good to me.
From: Simon Horman
Quoted 11 lines
Sent: 31 July 2024 09:45
Quoted 5 lines · Herve Codina
Received frame from QMC contains the CRC.
Upper layers don't need this CRC and tcpdump mentioned trailing junk data due to this CRC presence.As some other HDLC driver, simply discard this CRC.
It might be nice to specifically site an example.
But yes, I see this pattern inhdlc_rx_done().
Pretty much the only reason you'd want the received CRC is to do error recovery assuming a single short error burst.
Not that I've ever seen a driver contain the required code. 'Back of envelope' calculation: 32bit crc, 2kbyte frame, so 18bits of crc per data bit, sqrt for 'birthday paradox' - I bet you can recover 8-bit error bursts.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
This patch was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:
Hello:
This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:
Quoted 11 lines · On Tue, 30 Jul 2024 08:31:33 +0200 you wrote:
Received frame from QMC contains the CRC.
Upper layers don't need this CRC and tcpdump mentioned trailing junk data due to this CRC presence.As some other HDLC driver, simply discard this CRC.
Fixes d0f2258e79fd ("net: wan: Add support for QMC HDLC")Signed-off-by Herve Codina <[email protected]>[...]
Here is the summary with links:
- [net,v1] net: wan: fsl_qmc_hdlc: Discard received CRC
https://git.kernel.org/netdev/net/c/e549360069b4
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html