From: Joas Antonio dos Santos <joasantonio108@gmail.com> gtp0_handle_echo_resp() and gtp1u_handle_echo_resp() return the result of genlmsg_multicast_netns() to gtp_encap_recv(). When the netlink notification is delivered that result is 0, and gtp_encap_recv() treats 0 as "the skb has been consumed": switch (ret) { case 1: netdev_dbg(gtp->dev, "pass up to the process\n"); break; case 0: break; case -1: netdev_dbg(gtp->dev, "GTP packet has been dropped\n"); kfree_skb(skb); ret = 0; break; } The echo response handlers only read from the skb and never free it or pass it on, so every echo response that is reported to a listener on the gtp multicast group leaks the skb. Echo responses are handled for any peer when the UDP sockets were created by the kernel, so a remote host can leak memory on the receiver by sending GTP echo responses while a process with CAP_NET_ADMIN is subscribed to the gtp multicast group, for example a daemon that monitors echo responses. The error paths do not leak but have a related problem. -ENOMEM and the error from genlmsg_multicast_netns() (-ESRCH when nobody listens) are not -1, so they are returned unchanged to the UDP stack, which takes a negative value from encap_rcv() as a request to resubmit the packet to the protocol with that number. Return -1 when the notification cannot be built, so that gtp_encap_recv() drops the skb, and consume the skb once the notification has been handed to netlink, whether or not anybody is listening. Found by LLM-assisted source review. Fixes: d33bd757d362 ("gtp: Implement GTP echo request") Cc: stable@vger.kernel.org Signed-off-by: Joas Antonio dos Santos <joasantonio108@gmail.com> --- This issue was found with the help of an LLM-based code review tool (Claude). I checked the code paths by hand against v7.4-rc3 and reviewed the fix myself. It is build-tested only (W=1, arm64); I do not have the hardware to test it at runtime. --- drivers/net/gtp.c | 22 ++++++++++++++-------- 1 file changed, 14 insertions(+), 8 deletions(-) diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c index 1dd32e979cf5..85d1b0c46ecf 100644 --- a/drivers/net/gtp.c +++ b/drivers/net/gtp.c @@ -561,16 +561,19 @@ static int gtp0_handle_echo_resp(struct gtp_dev *gtp, struct sk_buff *skb) msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC); if (!msg) - return -ENOMEM; + return -1; ret = gtp_genl_fill_echo(msg, 0, 0, 0, GTP_CMD_ECHOREQ, echo); if (ret < 0) { nlmsg_free(msg); - return ret; + return -1; } - return genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), - msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), + msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + consume_skb(skb); + + return 0; } static int gtp_proto_to_family(__u16 proto) @@ -748,16 +751,19 @@ static int gtp1u_handle_echo_resp(struct gtp_dev *gtp, struct sk_buff *skb) msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC); if (!msg) - return -ENOMEM; + return -1; ret = gtp_genl_fill_echo(msg, 0, 0, 0, GTP_CMD_ECHOREQ, echo); if (ret < 0) { nlmsg_free(msg); - return ret; + return -1; } - return genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), - msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), + msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + consume_skb(skb); + + return 0; } static int gtp_parse_exthdrs(struct sk_buff *skb, unsigned int *hdrlen) --- base-commit: b3dfbde62cf1981c4638939c159dfd087b7a72e9 change-id: 20261110-gtp-echo-rsp-leak-7a3fac62bd3c Best regards, -- Joas Antonio dos Santos <joasantonio108@gmail.com>
…
+++ b/drivers/net/gtp.c @@ -561,16 +561,19 @@ static int gtp0_handle_echo_resp(struct gtp_dev *gtp, struct sk_buff *skb) … - return genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), - msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), + msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + consume_skb(skb); + + return 0; } …
I suggest to reconsider the error detection a bit more. https://elixir.bootlin.com/linux/v7.3.1/source/drivers/net/gtp.c#L541-L573 https://cwe.mitre.org/data/definitions/252.html Regards, Markus
I suggest to reconsider the error detection a bit more. https://elixir.bootlin.com/linux/v7.3.1/source/drivers/net/gtp.c#L541-L573 https://cwe.mitre.org/data/definitions/252.html
Hi Markus, thanks for looking at the patch. Dropping the return value of genlmsg_multicast_netns() is intentional here. At that point the echo response has been parsed and the notification has been handed to netlink, which owns and frees msg on every path. The only thing left for the handler to do is to release the received skb, and that is the same whether the notification reached a listener or not. The failures that can come back are -ESRCH (nobody is subscribed to the group) and -ENOBUFS (a listener's receive queue is full). Neither is a receive error for the GTP packet, and there is no recovery to attempt. Passing them on is what the current code does, and it is the second problem described in the commit message: gtp_encap_recv() only knows 1, 0 and -1, so any other negative value goes back to the UDP stack, which treats it as a request to resubmit the packet to another protocol. The errors that the handler can act on, nlmsg_new() and gtp_genl_fill_echo() failing, are still detected and now return -1 so that the skb is dropped. Regards, Joas Em ter., 10 de nov. de 2026 às 10:08, Markus Elfring <Markus.Elfring@web.de> escreveu:
…
+++ b/drivers/net/gtp.c @@ -561,16 +561,19 @@ static int gtp0_handle_echo_resp(struct gtp_dev *gtp, struct sk_buff *skb) … - return genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), - msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + genlmsg_multicast_netns(>p_genl_family, dev_net(gtp->dev), + msg, 0, GTP_GENL_MCGRP, GFP_ATOMIC); + consume_skb(skb); + + return 0; } …
I suggest to reconsider the error detection a bit more. https://elixir.bootlin.com/linux/v7.3.1/source/drivers/net/gtp.c#L541-L573 https://cwe.mitre.org/data/definitions/252.html
Regards, Markus
-- Joas Antonio dos Santos Cyber Security Specialist and Ethical Hacker joasantonio108@gmail.com LinkedIn
participants (3)
-
Joas Antonio -
Joas Antonio dos Santos via B4 Relay -
Markus Elfring