Netdev List
 help / color / mirror / Atom feed
From: Florian Fainelli <florian.fainelli@broadcom.com>
To: netdev-bot+sashiko@kernel.org
Cc: netdev@vger.kernel.org, rafal@milecki.pl,
	bcm-kernel-feedback-list@broadcom.com, andrew+netdev@lunn.ch,
	davem@davemloft.net, edumazet@kernel.org, kuba@kernel.org,
	pabeni@redhat.com, robh@kernel.org, krzk+dt@kernel.org,
	conor+dt@kernel.org, horms@kernel.org, corbet@lwn.net,
	jonas.gorski@gmail.com, olteanv@gmail.com, pablo@netfilter.org,
	fw@strlen.de, phil@nwl.cc, devicetree@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-doc@vger.kernel.org,
	netfilter-devel@vger.kernel.org, coreteam@netfilter.org
Subject: Re: [PATCH net-next 1/9] net: ethernet: bgmac: validate RX length against the DMA packet capacity
Date: Tue, 17 Nov 2026 15:43:55 -0800	[thread overview]
Message-ID: <5cc0d365-3821-4dab-8015-22cbc94be1e0@broadcom.com> (raw)
In-Reply-To: <179495815587.3136.10288920701294421777@kernel.org>

On 11/17/26 15:29, 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 (1) · 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] In bgmac_dma_rx_read()
>    (drivers/net/ethernet/broadcom/bgmac.c:463), the patch now treats…
> - [Low] The commit message is accurate but leaves out the user-visible
>    impact of the bugs this patch fixes in bgmac_dma_rx_read().
> 
> Pre-existing issues:
> - [High] This problem wasn't introduced by this patch, but it is High
>    severity, and the new check does not close it.
> 
> --
> 
> net: ethernet: bgmac: validate RX length against the DMA packet capacity
> 
> This patch changes the upper bound of the RX length check in
> bgmac_dma_rx_read() from BGMAC_RX_ALLOC_SIZE to BGMAC_RX_MAX_FRAME_SIZE.
> It also rejects lengths shorter than ETH_FCS_LEN. Invalid lengths are
> reported through a rate-limited netdev_err() and counted in rx_length_errors
> and rx_errors.
> 
>> BGMAC_RX_ALLOC_SIZE includes skb_shared_info and alignment space beyond
>> the DMA receive buffer. An invalid completion length can pass that check
>> and reach skb_put() with more data than the skb tailroom permits.
>>
>> Limit received frames to BGMAC_RX_MAX_FRAME_SIZE and reject lengths too
>> short for the FCS subtraction.
> 
> [Severity: Low]
> This isn't a bug, but could the commit message state the user-visible
> impact of what is being fixed?
> 
> For len < ETH_FCS_LEN, the u16 subtraction "len -= ETH_FCS_LEN" wraps to
> 65532..65535. The skb_put() that follows then ends in skb_over_panic() and
> BUG().
> 
> Lengths between the skb tailroom limit (about 1602 bytes with
> NET_SKB_PAD=64) and BGMAC_RX_ALLOC_SIZE also reach skb_over_panic().
> 
> Lengths from 1537 up to that tailroom limit got through skb_put(). They
> delivered bytes past the 1536-byte DMA-mapped frame area. Those bytes come
> from the non-zeroed netdev_alloc_frag() buffer, and the device never wrote
> them.
> 
> The message names the tailroom and FCS cases. It does not say that they
> end in a kernel BUG(), and it does not mention the stale data delivery.
> 
>> diff --git a/drivers/net/ethernet/broadcom/bgmac.c b/drivers/net/ethernet/broadcom/bgmac.c
>> index 9ef239bc487d7..1c502bd64d7df 100644
>> --- a/drivers/net/ethernet/broadcom/bgmac.c
>> +++ b/drivers/net/ethernet/broadcom/bgmac.c
>> @@ -457,10 +457,14 @@ static int bgmac_dma_rx_read(struct bgmac *bgmac, struct bgmac_dma_ring *ring,
>>   				break;
>>   			}
>>   
>> -			if (len > BGMAC_RX_ALLOC_SIZE) {
>> +			/* The allocation also contains headroom and skb_shared_info;
>> +			 * only the mapped packet area may be supplied by the DMA.
>> +			 */
>> +			if (unlikely(len > BGMAC_RX_MAX_FRAME_SIZE || len < ETH_FCS_LEN)) {
>>   				if (net_ratelimit())
>> -					netdev_err(bgmac->net_dev, "Found oversized packet at slot %d, DMA issue!\n",
>> -						   ring->start);
>> +					netdev_err(bgmac->net_dev,
>> +						   "Invalid RX length %u at slot %d (flags %#x)\n",
>> +						   len, ring->start, flags);
> 
> [Severity: Medium]
> Does this new bound drop frames that the driver itself tells the MAC to
> accept, and whose MTU it advertises?
> 
> bgmac_chip_init() sets the MAC max frame length to 1550:
> 
> 	bgmac_umac_write(bgmac, UMAC_MAX_FRAME_LEN, 32 + ETHER_MAX_LEN);
> 
> bgmac_change_mtu() sets it to 32 + mtu, which can be as large as 1564:
> 
> 	bgmac_umac_write(bgmac, UMAC_MAX_FRAME_LEN, 32 + mtu);
> 
> bgmac_enet_probe() sets max_mtu without subtracting ETH_HLEN or VLAN_HLEN:
> 
> 	net_dev->max_mtu = BGMAC_RX_MAX_FRAME_SIZE - ETH_FCS_LEN;
> 
> The driver accepts any MTU from 1519 to 1532. At those MTUs a full-sized
> frame can be up to 14 + 1532 + 4 = 1550 bytes. Such a frame is now always
> dropped, logged as an invalid RX length, and counted in rx_length_errors.
> 
> Before this patch these frames were delivered with stale tail bytes, so
> dropping them is safer. However, the new comment and the commit message
> assume that len > BGMAC_RX_MAX_FRAME_SIZE can only come from a DMA error.
> 
> Should max_mtu and/or UMAC_MAX_FRAME_LEN also be clamped to the
> per-descriptor capacity? One option for max_mtu is
> BGMAC_RX_MAX_FRAME_SIZE - ETH_HLEN - ETH_FCS_LEN, possibly minus VLAN_HLEN.
> Could the commit message also mention that these frames are now dropped?
> 
> [Severity: High]
> This is a pre-existing issue, and the new check does not appear to close
> it. Can a remote sender get stale buffer contents delivered as a received
> frame through an overflow-continue slot?
> 
> bgmac_dma_rx_ring_enable() turns on overflow-continue:
> 
> 	ctl |= BGMAC_DMA_RX_OVERFLOW_CONT;
> 
> Each descriptor holds only BGMAC_RX_BUF_SIZE, which is 30 + 1536 bytes.
> The MAC accepts frames up to 1550 bytes by default, and up to 1564 bytes
> after an MTU change.
> 
> With overflow-continue, the DMA64 engine writes the rest of an oversized
> frame as raw data at offset 0 of the next descriptor's buffer. That buffer
> has no RX status header. brcmsmac's dma_rx() drives the same engine with
> the same 0x400 bit, and it handles these continuation buffers explicitly:
> 
> 	resid = len - (di->rxbufsize - di->rxoffset);
> 	...
> 	while ((resid > 0) && (p = _dma_getnextrxp(di, false))) {
> 		pkt_len = min_t(uint, resid, di->rxbufsize);
> 
> bgmac_dma_rx_read() treats every slot as an independent frame. The new
> check drops the first slot when len > 1536. The next slot is then parsed
> as a fresh frame:
> 
> 	len = le16_to_cpu(rx->len);
> 	flags = le16_to_cpu(rx->flags);
> 
> In that slot, rx->len and rx->flags are frame bytes 1536..1539, which the
> sender controls. A value in [4, 1536] passes both the poison check and the
> new bound. build_skb() and skb_put() then deliver up to about 1532 bytes
> starting at buffer offset 30.
> 
> DMA wrote only the first (frame_len - 1536) bytes of that buffer in this
> round, which is at most 28 bytes. The rest is stale netdev_alloc_frag()
> memory, because bgmac_dma_rx_skb_for_slot() only writes the poison header.
> The resulting skb goes to eth_type_trans() and napi_gro_receive(). From
> there it can be processed locally, bridged, or forwarded.
> 
> The old BGMAC_RX_ALLOC_SIZE check let through an even wider range of
> lengths. This patch narrows that range but does not remove it.
> 
> Could multi-descriptor frames be tracked so the continuation slots are
> dropped too? Alternatively, could UMAC_MAX_FRAME_LEN and max_mtu be capped
> at the single-buffer capacity?
> 
>>   				put_page(virt_to_head_page(buf));
>>   				bgmac->net_dev->stats.rx_length_errors++;
>>   				bgmac->net_dev->stats.rx_errors++;
> 

[Medium] RX bound vs. MAC max frame length / max_mtu

Valid. The driver programmed UMAC_MAX_FRAME_LEN to 32 + ETHER_MAX_LEN 
(or 32 + mtu), and max_mtu ignored ETH_HLEN, so the MAC could accept 
frames larger than one RX descriptor can hold, which the new check then
drops. In v2, UMAC_MAX_FRAME_LEN is clamped to BGMAC_RX_MAX_FRAME_SIZE 
and max_mtu becomes BGMAC_RX_MAX_FRAME_SIZE - ETH_HLEN - VLAN_HLEN -
ETH_FCS_LEN (1514). MTUs 1515..1532 are no longer accepted; frames of
that size were already delivered with stale tail bytes. The commit
message now says so.

[High, pre-existing] overflow-continue continuation slots

The analysis is right: with BGMAC_DMA_RX_OVERFLOW_CONT, the tail of an
oversized frame lands at offset 0 of the next descriptor, which has no 
RX header, and was previously parsed as a new frame. Rather than 
tracking multi-descriptor frames, v2 caps the MAC maximum frame length 
at the single-buffer capacity (the clamp above), so oversized frames are 
dropped by the MAC and never spill into a following descriptor.

[Low] commit message

Fixed in v2: it now describes the skb_over_panic()/BUG() cases (len <
ETH_FCS_LEN and len above the tailroom) and the delivery of
uninitialized netdev_alloc_frag() bytes.

pw-bot: cr
-- 
Florian

  reply	other threads:[~2026-11-17 23:44 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-16 22:25 [PATCH net-next 0/9] net: bgmac: add Northstar IPv4 TCP/UDP NAT flowtable offload Florian Fainelli
2026-11-16 22:25 ` [PATCH net-next 1/9] net: ethernet: bgmac: validate RX length against the DMA packet capacity Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-17 23:43     ` Florian Fainelli [this message]
2026-11-16 22:25 ` [PATCH net-next 2/9] netfilter: flowtable: reject non-host IPv4 packets before lookup Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-17 23:51     ` Florian Fainelli
2026-11-16 22:25 ` [PATCH net-next 3/9] netfilter: flowtable: retain offload ownership until deletion completes Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-17 23:57     ` Florian Fainelli
2026-11-16 22:25 ` [PATCH net-next 4/9] dt-bindings: net: describe the Northstar flow accelerator Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-16 22:25 ` [PATCH net-next 5/9] net: dsa: add Broadcom FA metadata and IPv4 exception paths Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-16 22:25 ` [PATCH net-next 6/9] net: flow_offload: allow drivers to validate and invalidate cached paths Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-16 22:25 ` [PATCH net-next 7/9] net: bgmac: add Northstar IPv4 TCP/UDP NAT flowtable offload Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-16 22:25 ` [PATCH net-next 8/9] net: add KUnit coverage for FA and flowtable path validation Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-16 22:25 ` [PATCH net-next 9/9] docs: net: document Northstar FA setup and NAT qualification Florian Fainelli
2026-11-17 23:29   ` netdev-bot+sashiko
2026-11-16 22:38 ` [PATCH net-next 0/9] net: bgmac: add Northstar IPv4 TCP/UDP NAT flowtable offload netdev-bot+sinfo
2026-11-16 22:45   ` Florian Fainelli

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5cc0d365-3821-4dab-8015-22cbc94be1e0@broadcom.com \
    --to=florian.fainelli@broadcom.com \
    --cc=andrew+netdev@lunn.ch \
    --cc=bcm-kernel-feedback-list@broadcom.com \
    --cc=conor+dt@kernel.org \
    --cc=corbet@lwn.net \
    --cc=coreteam@netfilter.org \
    --cc=davem@davemloft.net \
    --cc=devicetree@vger.kernel.org \
    --cc=edumazet@kernel.org \
    --cc=fw@strlen.de \
    --cc=horms@kernel.org \
    --cc=jonas.gorski@gmail.com \
    --cc=krzk+dt@kernel.org \
    --cc=kuba@kernel.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev-bot+sashiko@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=netfilter-devel@vger.kernel.org \
    --cc=olteanv@gmail.com \
    --cc=pabeni@redhat.com \
    --cc=pablo@netfilter.org \
    --cc=phil@nwl.cc \
    --cc=rafal@milecki.pl \
    --cc=robh@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox