* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-16 0:58 [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting James Hilliard
@ 2026-11-16 1:03 ` netdev-bot+sinfo
2026-11-16 2:19 ` James Hilliard
2026-11-17 15:20 ` Maxime Chevallier
` (2 subsequent siblings)
3 siblings, 1 reply; 10+ messages in thread
From: netdev-bot+sinfo @ 2026-11-16 1:03 UTC (permalink / raw)
To: James Hilliard
Cc: netdev, Maxime Chevallier, Andrew Lunn, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, Maxime Coquelin,
Alexandre Torgue, Chen-Yu Tsai, Jernej Skrabec, Samuel Holland,
LABBE Corentin, linux-arm-kernel, linux-sunxi, linux-kernel
Hi!
This is an automated message. This series looks like a fix, but its
commit messages seem to be missing some information:
- How the issue was discovered, e.g. hit in production, hit during
development, syzbot report, manual code inspection, LLM or static
analysis tool scan.
- Whether the issue was actually triggered, or is only theoretical
(e.g. found by LLM and not proven at runtime). If it was triggered
please include the symptoms, like the stack trace or kernel messages.
- What hardware the change was tested on. For driver fixes please
mention the device (and if relevant firmware version) used for
testing, or say that the change was not tested on real hardware.
Please do not repost the series just to address the above. Instead,
reply to this email with the missing information, so that reviewers
can take it into account.
*IMPORTANT*: If you do repost (for other reasons) - please include
the relevant info in the cover letter or commit messages.
--
The evaluation is done by an LLM so it may be wrong, if you think
that is the case please reply and explain.
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-16 1:03 ` netdev-bot+sinfo
@ 2026-11-16 2:19 ` James Hilliard
2026-11-16 2:28 ` James Hilliard
0 siblings, 1 reply; 10+ messages in thread
From: James Hilliard @ 2026-11-16 2:19 UTC (permalink / raw)
To: netdev-bot+sinfo
Cc: netdev, Maxime Chevallier, Andrew Lunn, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, Maxime Coquelin,
Alexandre Torgue, Chen-Yu Tsai, Jernej Skrabec, Samuel Holland,
LABBE Corentin, linux-arm-kernel, linux-sunxi, linux-kernel
On Sun, Nov 15, 2026 at 6:03 PM <netdev-bot+sinfo@kernel.org> wrote:
>
> Hi!
>
> This is an automated message. This series looks like a fix, but its
> commit messages seem to be missing some information:
>
> - How the issue was discovered, e.g. hit in production, hit during
> development, syzbot report, manual code inspection, LLM or static
> analysis tool scan.
This was found while investigating RX checksum offload behavior on
H616 and reviewing the shared setup and feature-filtering paths.
>
> - Whether the issue was actually triggered, or is only theoretical
> (e.g. found by LLM and not proven at runtime). If it was triggered
> please include the symptoms, like the stack trace or kernel messages.
The capability loss was reproduced in a source-extracted test using the
actual GMAC, GMAC4 and XGMAC checksum callbacks with mocked MMIO and the
relevant setup and feature-filtering logic.
With the original code, disabling offload and following the reopen path
clears plat->rx_coe. The feature filter then prevents offload from being
enabled again. With the patch, the saved capability survives that
sequence. A requested enable that fails hardware readback still clears
the capability.
>
> - What hardware the change was tested on. For driver fixes please
> mention the device (and if relevant firmware version) used for
> testing, or say that the change was not tested on real hardware.
Tested on an Allwinner H616 board with an AC200 PHY.
>
> Please do not repost the series just to address the above. Instead,
> reply to this email with the missing information, so that reviewers
> can take it into account.
>
> *IMPORTANT*: If you do repost (for other reasons) - please include
> the relevant info in the cover letter or commit messages.
> --
> The evaluation is done by an LLM so it may be wrong, if you think
> that is the case please reply and explain.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-16 2:19 ` James Hilliard
@ 2026-11-16 2:28 ` James Hilliard
0 siblings, 0 replies; 10+ messages in thread
From: James Hilliard @ 2026-11-16 2:28 UTC (permalink / raw)
To: netdev-bot+sinfo
Cc: netdev, Maxime Chevallier, Andrew Lunn, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, Maxime Coquelin,
Alexandre Torgue, Chen-Yu Tsai, Jernej Skrabec, Samuel Holland,
LABBE Corentin, linux-arm-kernel, linux-sunxi, linux-kernel
On Sun, Nov 15, 2026 at 7:19 PM James Hilliard
<james.hilliard1@gmail.com> wrote:
>
> On Sun, Nov 15, 2026 at 6:03 PM <netdev-bot+sinfo@kernel.org> wrote:
> >
> > Hi!
> >
> > This is an automated message. This series looks like a fix, but its
> > commit messages seem to be missing some information:
> >
> > - How the issue was discovered, e.g. hit in production, hit during
> > development, syzbot report, manual code inspection, LLM or static
> > analysis tool scan.
>
> This was found while investigating RX checksum offload behavior on
> H616 and reviewing the shared setup and feature-filtering paths.
>
> >
> > - Whether the issue was actually triggered, or is only theoretical
> > (e.g. found by LLM and not proven at runtime). If it was triggered
> > please include the symptoms, like the stack trace or kernel messages.
>
> The capability loss was reproduced in a source-extracted test using the
> actual GMAC, GMAC4 and XGMAC checksum callbacks with mocked MMIO and the
> relevant setup and feature-filtering logic.
I mixed up the two threads. That explanation belongs to the
separate checksum-capability fix [0].
For this sun8i patch, the issue was found and reproduced during MAC/PHY
receive testing on an Allwinner H616 board with an AC200 PHY. Even after
ethtool reported RX checksumming disabled, EMAC_RX_DO_CRC remained set
and the hardware kept dropping packets with invalid IP or transport
checksums before they reached an RX descriptor.
With the fix, disabling offload cleared that bit. In steady-state tests,
those packets then reached the host and were rejected by software
checksum validation. Valid traffic continued to work, and Ethernet FCS
rejection remained active.
[0] https://lore.kernel.org/r/20261115-submit-stmmac-rx-checksum-capability-v1-v1-1-ea26fa3689be@gmail.com/
>
> With the original code, disabling offload and following the reopen path
> clears plat->rx_coe. The feature filter then prevents offload from being
> enabled again. With the patch, the saved capability survives that
> sequence. A requested enable that fails hardware readback still clears
> the capability.
>
> >
> > - What hardware the change was tested on. For driver fixes please
> > mention the device (and if relevant firmware version) used for
> > testing, or say that the change was not tested on real hardware.
>
> Tested on an Allwinner H616 board with an AC200 PHY.
>
> >
> > Please do not repost the series just to address the above. Instead,
> > reply to this email with the missing information, so that reviewers
> > can take it into account.
> >
> > *IMPORTANT*: If you do repost (for other reasons) - please include
> > the relevant info in the cover letter or commit messages.
> > --
> > The evaluation is done by an LLM so it may be wrong, if you think
> > that is the case please reply and explain.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-16 0:58 [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting James Hilliard
2026-11-16 1:03 ` netdev-bot+sinfo
@ 2026-11-17 15:20 ` Maxime Chevallier
2026-11-17 15:26 ` Lorenzo Bianconi
2026-11-18 4:00 ` netdev-bot+sashiko
2026-11-19 22:48 ` Jakub Kicinski
3 siblings, 1 reply; 10+ messages in thread
From: Maxime Chevallier @ 2026-11-17 15:20 UTC (permalink / raw)
To: James Hilliard, netdev, Andrew Lunn, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, Maxime Coquelin,
Alexandre Torgue, Chen-Yu Tsai, Jernej Skrabec, Samuel Holland,
LABBE Corentin
Cc: linux-arm-kernel, linux-sunxi, linux-kernel
Hi,
On 11/16/26 01:58, James Hilliard wrote:
> stmmac_set_features() clears hw->rx_csum when RX checksum offload is
> disabled, but sun8i_dwmac_rx_ipc_enable() unconditionally enables the
> hardware checksum engine. The MAC therefore keeps dropping packets with
> invalid IP or transport checksums before they reach the host, even after
> ethtool reports RX checksumming as disabled.
>
> Set or clear EMAC_RX_DO_CRC according to hw->rx_csum so that disabling
> offload leaves checksum validation to the network stack. Preserve the
> callback's existing success return value.
>
> Fixes: 9f93ac8d4085 ("net-next: stmmac: Add dwmac-sun8i")
> Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Nice and simple, I tested that on an orangepi zero, with an allwinner
H2+, and I can indeed see the difference as bad chksum packets are
dropped on that core.
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Maxime
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-17 15:20 ` Maxime Chevallier
@ 2026-11-17 15:26 ` Lorenzo Bianconi
0 siblings, 0 replies; 10+ messages in thread
From: Lorenzo Bianconi @ 2026-11-17 15:26 UTC (permalink / raw)
To: Maxime Chevallier
Cc: James Hilliard, netdev, Andrew Lunn, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, Maxime Coquelin,
Alexandre Torgue, Chen-Yu Tsai, Jernej Skrabec, Samuel Holland,
LABBE Corentin, linux-arm-kernel, linux-sunxi, linux-kernel
[-- Attachment #1: Type: text/plain, Size: 1144 bytes --]
> Hi,
>
> On 11/16/26 01:58, James Hilliard wrote:
> > stmmac_set_features() clears hw->rx_csum when RX checksum offload is
> > disabled, but sun8i_dwmac_rx_ipc_enable() unconditionally enables the
> > hardware checksum engine. The MAC therefore keeps dropping packets with
> > invalid IP or transport checksums before they reach the host, even after
> > ethtool reports RX checksumming as disabled.
> >
> > Set or clear EMAC_RX_DO_CRC according to hw->rx_csum so that disabling
> > offload leaves checksum validation to the network stack. Preserve the
> > callback's existing success return value.
> >
> > Fixes: 9f93ac8d4085 ("net-next: stmmac: Add dwmac-sun8i")
> > Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
>
> Nice and simple, I tested that on an orangepi zero, with an allwinner
> H2+, and I can indeed see the difference as bad chksum packets are
> dropped on that core.
>
> Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
> Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
>
> Maxime
>
>
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-16 0:58 [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting James Hilliard
2026-11-16 1:03 ` netdev-bot+sinfo
2026-11-17 15:20 ` Maxime Chevallier
@ 2026-11-18 4:00 ` netdev-bot+sashiko
2026-11-19 17:18 ` James Hilliard
2026-11-19 22:48 ` Jakub Kicinski
3 siblings, 1 reply; 10+ messages in thread
From: netdev-bot+sashiko @ 2026-11-18 4:00 UTC (permalink / raw)
To: james.hilliard1
Cc: netdev, maxime.chevallier, andrew+netdev, davem, edumazet, kuba,
pabeni, mcoquelin.stm32, alexandre.torgue, wens, jernej.skrabec,
samuel, clabbe.montjoie, linux-arm-kernel, linux-sunxi,
linux-kernel
Thank you for your contribution! Sashiko AI review found 2 potential
issue(s) to consider.
Critical: 0 · High: 0 · Medium: 1 · Low: 1
If you believe none of the issues are legitimate, please reply to each
issue of Medium severity or higher and explain why the AI is wrong.
If any of them are legitimate and you plan to respin, please let patchwork
know by including "pw-bot: cr" as a separate line at the end of your reply
(one such reply per series is enough).
- [Medium] Clearing EMAC_RX_DO_CRC (EMAC_RX_CTL0 bit 27) whenever
NETIF_F_RXCSUM is off may also turn off the MAC's Ethernet FCS…
- [Low] When RX checksum offload is turned back on, frames the sun8i MAC
received while EMAC_RX_DO_CRC was clear can be marked…
--
net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
Before this patch, sun8i_dwmac_rx_ipc_enable() always set
EMAC_RX_DO_CRC. With the patch, it sets or clears the bit based on
hw->rx_csum, so that turning off RX checksum offload through ethtool
also turns off the hardware checksum engine.
> diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
> index 2ece7ba9dd73b..98cc7503bddeb 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
> @@ -705,7 +705,10 @@ static int sun8i_dwmac_rx_ipc_enable(struct mac_device_info *hw)
> u32 v;
>
> v = readl(ioaddr + EMAC_RX_CTL0);
> - v |= EMAC_RX_DO_CRC;
> + if (hw->rx_csum)
> + v |= EMAC_RX_DO_CRC;
> + else
> + v &= ~EMAC_RX_DO_CRC;
[Severity: Medium]
Does clearing bit 27 of EMAC_RX_CTL0 also turn off the MAC's Ethernet
FCS check, and not only the IP/L4 checksum engine?
The bit is named after the CRC:
#define EMAC_RX_DO_CRC BIT(27)
The probe code says the hardware features were "copied from Allwinner
drivers". That code sets this bit with the comment "Enable CRC & IPv4
Header Checksum". The Allwinner manuals describe CHECK_CRC as "Calculate
CRC and check the IPv4 Header Checksum".
On this path, sun8i does not have the hardware strip the FCS.
stmmac_rx() removes it in software without checking it:
/* ACS is disabled; strip manually. */
if (likely(!(status & rx_not_ls)))
len -= ETH_FCS_LEN;
The only FCS check left is the hardware RDES0_CRC_ERROR bit, which
ndesc_get_rx_status() turns into discard_frame.
Suppose bit 27 does control CRC calculation. After "ethtool -K ethX rx
off", the path is stmmac_set_features() -> hw->rx_csum = 0 ->
sun8i_dwmac_rx_ipc_enable(). Frames with a corrupted FCS would then no
longer set RDES0_CRC_ERROR. They would reach the stack as good frames,
with the FCS silently stripped. That behaves like NETIF_F_RXALL rather
than "RX checksum offload off".
Is there a datasheet reference or a bad-FCS test showing that bit 27
gates only the L3/L4 checksum? If the hardware ties the two checks
together, a different approach may be needed. For example, keep bit 27
set and, when rx_csum is 0, stop discarding frames whose only error is
the IP or payload checksum.
[Severity: Low]
Can frames that arrive while EMAC_RX_DO_CRC is clear be marked as
checksum-verified once RX checksum offload is turned back on?
When "ethtool -K ethX rx on" runs, stmmac_set_features() publishes the
software flag before the MMIO write that re-enables the engine:
stmmac_set_features() {
...
if (features & NETIF_F_RXCSUM)
priv->hw->rx_csum = priv->plat->rx_coe;
else
priv->hw->rx_csum = 0;
...
stmmac_rx_ipc(priv, priv->hw);
...
}
NAPI is not disabled, RX DMA is not stopped and the ring is not drained.
Meanwhile, stmmac_rx() may run on another CPU and read the flag
directly:
int coe = priv->hw->rx_csum;
...
if (unlikely(!coe) || !stmmac_has_ip_ethertype(skb) ||
(status & csum_none))
skb_checksum_none_assert(skb);
else
skb->ip_summed = CHECKSUM_UNNECESSARY;
sun8i uses normal descriptors. ndesc_get_rx_status() never returns
csum_none, so nothing in the descriptor says whether the hardware
actually checked a given frame.
Any unchecked frame still in the RX ring would then be passed up as
CHECKSUM_UNNECESSARY. That covers frames received before the writel()
in sun8i_dwmac_rx_ipc_enable() and frames received just after it.
Before this patch, the bit was never cleared, so this state could not
happen on sun8i.
Swapping the two writes would narrow the window, but it would not cover
frames already in the ring. Would RX need to be quiesced or drained
across this transition?
> writel(v, ioaddr + EMAC_RX_CTL0);
>
> return 1;
--
Sashiko AI review · https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20261115-submit-sun8i-emac-rx-checksum-v1-v1-1-f186d127b39f%40gmail.com
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-18 4:00 ` netdev-bot+sashiko
@ 2026-11-19 17:18 ` James Hilliard
0 siblings, 0 replies; 10+ messages in thread
From: James Hilliard @ 2026-11-19 17:18 UTC (permalink / raw)
To: netdev-bot+sashiko
Cc: netdev, maxime.chevallier, andrew+netdev, davem, edumazet, kuba,
pabeni, mcoquelin.stm32, alexandre.torgue, wens, jernej.skrabec,
samuel, clabbe.montjoie, linux-arm-kernel, linux-sunxi,
linux-kernel
On Tue, Nov 17, 2026 at 9:00 PM <netdev-bot+sashiko@kernel.org> wrote:
>
> Thank you for your contribution! Sashiko AI review found 2 potential
> issue(s) to consider.
>
> Critical: 0 · High: 0 · Medium: 1 · Low: 1
>
> If you believe none of the issues are legitimate, please reply to each
> issue of Medium severity or higher and explain why the AI is wrong.
> If any of them are legitimate and you plan to respin, please let patchwork
> know by including "pw-bot: cr" as a separate line at the end of your reply
> (one such reply per series is enough).
>
> - [Medium] Clearing EMAC_RX_DO_CRC (EMAC_RX_CTL0 bit 27) whenever
> NETIF_F_RXCSUM is off may also turn off the MAC's Ethernet FCS…
> - [Low] When RX checksum offload is turned back on, frames the sun8i MAC
> received while EMAC_RX_DO_CRC was clear can be marked…
>
> --
>
> net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
>
> Before this patch, sun8i_dwmac_rx_ipc_enable() always set
> EMAC_RX_DO_CRC. With the patch, it sets or clears the bit based on
> hw->rx_csum, so that turning off RX checksum offload through ethtool
> also turns off the hardware checksum engine.
>
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
> > index 2ece7ba9dd73b..98cc7503bddeb 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
> > @@ -705,7 +705,10 @@ static int sun8i_dwmac_rx_ipc_enable(struct mac_device_info *hw)
> > u32 v;
> >
> > v = readl(ioaddr + EMAC_RX_CTL0);
> > - v |= EMAC_RX_DO_CRC;
> > + if (hw->rx_csum)
> > + v |= EMAC_RX_DO_CRC;
> > + else
> > + v &= ~EMAC_RX_DO_CRC;
>
> [Severity: Medium]
> Does clearing bit 27 of EMAC_RX_CTL0 also turn off the MAC's Ethernet
> FCS check, and not only the IP/L4 checksum engine?
It doesn't on H616. I sent 100 frames with a deliberately incorrect
FCS in each setting and checked that bit 27 changed. With normal
filtering, none produced an RX descriptor. With error forwarding
enabled, all 100 produced descriptors with RDES0_CRC_ERROR set in
both settings, and the driver discarded them. All 100 good-FCS
controls were received in both settings.
>
> The bit is named after the CRC:
>
> #define EMAC_RX_DO_CRC BIT(27)
>
> The probe code says the hardware features were "copied from Allwinner
> drivers". That code sets this bit with the comment "Enable CRC & IPv4
> Header Checksum". The Allwinner manuals describe CHECK_CRC as "Calculate
> CRC and check the IPv4 Header Checksum".
>
> On this path, sun8i does not have the hardware strip the FCS.
> stmmac_rx() removes it in software without checking it:
>
> /* ACS is disabled; strip manually. */
> if (likely(!(status & rx_not_ls)))
> len -= ETH_FCS_LEN;
>
> The only FCS check left is the hardware RDES0_CRC_ERROR bit, which
> ndesc_get_rx_status() turns into discard_frame.
>
> Suppose bit 27 does control CRC calculation. After "ethtool -K ethX rx
> off", the path is stmmac_set_features() -> hw->rx_csum = 0 ->
> sun8i_dwmac_rx_ipc_enable(). Frames with a corrupted FCS would then no
> longer set RDES0_CRC_ERROR. They would reach the stack as good frames,
> with the FCS silently stripped. That behaves like NETIF_F_RXALL rather
> than "RX checksum offload off".
>
> Is there a datasheet reference or a bad-FCS test showing that bit 27
> gates only the L3/L4 checksum? If the hardware ties the two checks
> together, a different approach may be needed. For example, keep bit 27
> set and, when rx_csum is 0, stop discarding frames whose only error is
> the IP or payload checksum.
>
> [Severity: Low]
> Can frames that arrive while EMAC_RX_DO_CRC is clear be marked as
> checksum-verified once RX checksum offload is turned back on?
Yes. In a separate test, frames with bad UDP checksums were already
in the RX ring when offload was turned back on. When NAPI processed
them, invalid datagrams reached the UDP socket in both the IPv4 and
IPv6 tests. The driver used the new offload setting for descriptors
filled under the old one. Changing the order of the register and
software flag updates would not fix that case.
>
> When "ethtool -K ethX rx on" runs, stmmac_set_features() publishes the
> software flag before the MMIO write that re-enables the engine:
>
> stmmac_set_features() {
> ...
> if (features & NETIF_F_RXCSUM)
> priv->hw->rx_csum = priv->plat->rx_coe;
> else
> priv->hw->rx_csum = 0;
> ...
> stmmac_rx_ipc(priv, priv->hw);
> ...
> }
>
> NAPI is not disabled, RX DMA is not stopped and the ring is not drained.
> Meanwhile, stmmac_rx() may run on another CPU and read the flag
> directly:
>
> int coe = priv->hw->rx_csum;
> ...
> if (unlikely(!coe) || !stmmac_has_ip_ethertype(skb) ||
> (status & csum_none))
> skb_checksum_none_assert(skb);
> else
> skb->ip_summed = CHECKSUM_UNNECESSARY;
>
> sun8i uses normal descriptors. ndesc_get_rx_status() never returns
> csum_none, so nothing in the descriptor says whether the hardware
> actually checked a given frame.
>
> Any unchecked frame still in the RX ring would then be passed up as
> CHECKSUM_UNNECESSARY. That covers frames received before the writel()
> in sun8i_dwmac_rx_ipc_enable() and frames received just after it.
> Before this patch, the bit was never cleared, so this state could not
> happen on sun8i.
>
> Swapping the two writes would narrow the window, but it would not cover
> frames already in the ring. Would RX need to be quiesced or drained
> across this transition?
>
> > writel(v, ioaddr + EMAC_RX_CTL0);
> >
> > return 1;
>
> --
> Sashiko AI review · https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20261115-submit-sun8i-emac-rx-checksum-v1-v1-1-f186d127b39f%40gmail.com
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-16 0:58 [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting James Hilliard
` (2 preceding siblings ...)
2026-11-18 4:00 ` netdev-bot+sashiko
@ 2026-11-19 22:48 ` Jakub Kicinski
2026-11-19 23:32 ` James Hilliard
3 siblings, 1 reply; 10+ messages in thread
From: Jakub Kicinski @ 2026-11-19 22:48 UTC (permalink / raw)
To: James Hilliard
Cc: netdev, Maxime Chevallier, Andrew Lunn, David S. Miller,
Eric Dumazet, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, LABBE Corentin,
linux-arm-kernel, linux-sunxi, linux-kernel
On Sun, 15 Nov 2026 17:58:26 -0700 James Hilliard wrote:
> stmmac_set_features() clears hw->rx_csum when RX checksum offload is
> disabled, but sun8i_dwmac_rx_ipc_enable() unconditionally enables the
> hardware checksum engine. The MAC therefore keeps dropping packets with
> invalid IP or transport checksums before they reach the host, even after
> ethtool reports RX checksumming as disabled.
Rx checksumming only controls whether the _good_ packet indication /
offload is communicated up the stack. The device should never drop
packets due to a bad L4 checksum. Checksumming L4 is not much of a DoS
vector. Oleksij was fixing this in other stmmac variants. I _think_ you
should always clear this bit if it leads to packet drops.
(FCS is fine, bad FCS should only be passed up if RX_ALL is set,
but sounds like your device has all checksums controlled by one bit?
in which case we prefer to see all and let driver discard the bad FCS)
If I completely misunderstand things we should at least get a respin
to update the cover letter. The commit message talks about L4 offload,
and rx_csum is L4 AFAIU but the bit is called CRC which deserves a
mention somewhere IMVHO.
--
pw-bot: cr
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [PATCH net] net: stmmac: dwmac-sun8i: Honor the RX checksum offload setting
2026-11-19 22:48 ` Jakub Kicinski
@ 2026-11-19 23:32 ` James Hilliard
0 siblings, 0 replies; 10+ messages in thread
From: James Hilliard @ 2026-11-19 23:32 UTC (permalink / raw)
To: Jakub Kicinski
Cc: netdev, Maxime Chevallier, Andrew Lunn, David S. Miller,
Eric Dumazet, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, LABBE Corentin,
linux-arm-kernel, linux-sunxi, linux-kernel
On Thu, Nov 19, 2026 at 3:48 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Sun, 15 Nov 2026 17:58:26 -0700 James Hilliard wrote:
> > stmmac_set_features() clears hw->rx_csum when RX checksum offload is
> > disabled, but sun8i_dwmac_rx_ipc_enable() unconditionally enables the
> > hardware checksum engine. The MAC therefore keeps dropping packets with
> > invalid IP or transport checksums before they reach the host, even after
> > ethtool reports RX checksumming as disabled.
>
> Rx checksumming only controls whether the _good_ packet indication /
> offload is communicated up the stack. The device should never drop
> packets due to a bad L4 checksum. Checksumming L4 is not much of a DoS
> vector. Oleksij was fixing this in other stmmac variants. I _think_ you
> should always clear this bit if it leads to packet drops.
The case I hit was `rx off`: on H616, ethtool reported RX checksumming
off, but the driver still set EMAC_RX_CTL0 bit 27. Clearing it let
packets with a bad UDP checksum reach the stack, where they failed
software validation. With `rx on`, the driver sets bit 27 both before
and after my patch, so the drops in that mode predate the patch.
H616 also has a separate control for forwarding error frames to DMA:
RX_ERR_FRM (EMAC_RX_CTL1 bit 3). With bit 27 set and normal filtering,
none of 100 packets with a bad UDP checksum reached an RX descriptor.
Setting RX_ERR_FRM let all 100 through to DMA in both the IPv4 and
IPv6 tests. The descriptors had the payload checksum error and error
summary bits set, but not the CRC error bit. The driver discarded
them on the error summary.
> (FCS is fine, bad FCS should only be passed up if RX_ALL is set,
> but sounds like your device has all checksums controlled by one bit?
> in which case we prefer to see all and let driver discard the bad FCS)
Ethernet FCS checking stays active on H616 with bit 27 clear. With
normal filtering, bad-FCS frames produced no RX descriptor in either
setting. With error forwarding enabled, all 100 bad-FCS frames had
RDES0_CRC_ERROR set in each setting, and the driver discarded them.
> If I completely misunderstand things we should at least get a respin
> to update the cover letter. The commit message talks about L4 offload,
> and rx_csum is L4 AFAIU but the bit is called CRC which deserves a
> mention somewhere IMVHO.
The H616 manual calls bit 27 CHECK_CRC and mentions the IPv4 header
checksum. What I should have made clearer is that the bit also affects
transport checksum checking, but clearing it doesn't disable Ethernet
FCS detection.
Would it make sense to address the `rx on` drops in a separate
follow-up along the lines of Oleksij's dwmac4 fix [0]? On sun8i that
would mean enabling error forwarding to DMA and passing packets whose
only error is an IP or transport checksum up as CHECKSUM_NONE, while
still discarding frames with a bad FCS or other fatal RX errors.
If we leave the checker enabled in both modes, `rx off` would only
control whether the driver reports a checksum as verified. In that
form the follow-up would replace the conditional bit-27 write rather
than depend on this patch.
[0] https://lore.kernel.org/r/20250818090217.2789521-4-o.rempel@pengutronix.de/
> --
> pw-bot: cr
^ permalink raw reply [flat|nested] 10+ messages in thread