[PATCH net v4] netconsole: Fix race condition in between reader and writer of userdata

[PATCH net v4] netconsole: Fix race condition in between reader and writer of userdata

[Gustavo Luiz Duarte]

The update_userdata() function constructs the complete userdata string
in nt->extradata_complete and updates nt->userdata_length. This data
is then read by write_msg() and write_ext_msg() when sending netconsole
messages. However, update_userdata() was not holding target_list_lock
during this process, allowing concurrent message transmission to read
partially updated userdata.

This race condition could result in netconsole messages containing
incomplete or inconsistent userdata - for example, reading the old
userdata_length with new extradata_complete content, or vice versa,
leading to truncated or corrupted output.

Fix this by acquiring target_list_lock with spin_lock_irqsave() before
updating extradata_complete and userdata_length, and releasing it after
both fields are fully updated. This ensures that readers see a
consistent view of the userdata, preventing corruption during concurrent
access.

The fix aligns with the existing locking pattern used throughout the
netconsole code, where target_list_lock protects access to target
fields including buf[] and msgcounter that are accessed during message
transmission.

Also get rid of the unnecessary variable complete_idx, which makes it
easier to bail out of update_userdata().

Fixes: df03f830d099 ("net: netconsole: cache userdata formatted string in netconsole_target")
Signed-off-by: Gustavo Luiz Duarte <gustavold@gmail.com>
---
This patch fixes a race condition in netconsole's userdata handling
where concurrent message transmission could read partially updated
userdata fields, resulting in corrupted netconsole output.

The patch fixes the issue by ensuring update_userdata() holds
the target_list_lock while updating both extradata_complete and
userdata_length, preventing readers from seeing inconsistent state.

Changes in v4:
- Avoid calling WARN_ON_ONCE() while holding the spinlock.
- Link to v3: https://lore.kernel.org/r/20251027-netconsole-fix-race-v3-1-8d40a67e02d2@meta.com

Changes in v3:
- Drop testcase.
- Link to v2: https://lore.kernel.org/r/20251022-netconsole-fix-race-v2-0-337241338079@meta.com

Changes in v2:
- Added testcase to Makefile.
- Reordered fix and testcase to avoid failure in CI.
- testcase: delay cleanup until child process are killed, plus shellcheck fixes.
- Link to v1: https://lore.kernel.org/all/20251020-netconsole-fix-race-v1-0-b775be30ee8a@gmail.com/
---
 drivers/net/netconsole.c | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/drivers/net/netconsole.c b/drivers/net/netconsole.c
index 194570443493..5d8d0214786c 100644
--- a/drivers/net/netconsole.c
+++ b/drivers/net/netconsole.c
@@ -886,8 +886,11 @@ static ssize_t userdatum_value_show(struct config_item *item, char *buf)
 
 static void update_userdata(struct netconsole_target *nt)
 {
-	int complete_idx = 0, child_count = 0;
 	struct list_head *entry;
+	int child_count = 0;
+	unsigned long flags;
+
+	spin_lock_irqsave(&target_list_lock, flags);
 
 	/* Clear the current string in case the last userdatum was deleted */
 	nt->userdata_length = 0;
@@ -897,8 +900,11 @@ static void update_userdata(struct netconsole_target *nt)
 		struct userdatum *udm_item;
 		struct config_item *item;
 
-		if (WARN_ON_ONCE(child_count >= MAX_EXTRADATA_ITEMS))
-			break;
+		if (child_count >= MAX_EXTRADATA_ITEMS) {
+			spin_unlock_irqrestore(&target_list_lock, flags);
+			WARN_ON_ONCE(1);
+			return;
+		}
 		child_count++;
 
 		item = container_of(entry, struct config_item, ci_entry);
@@ -912,12 +918,11 @@ static void update_userdata(struct netconsole_target *nt)
 		 * one entry length (1/MAX_EXTRADATA_ITEMS long), entry count is
 		 * checked to not exceed MAX items with child_count above
 		 */
-		complete_idx += scnprintf(&nt->extradata_complete[complete_idx],
-					  MAX_EXTRADATA_ENTRY_LEN, " %s=%s\n",
-					  item->ci_name, udm_item->value);
+		nt->userdata_length += scnprintf(&nt->extradata_complete[nt->userdata_length],
+						 MAX_EXTRADATA_ENTRY_LEN, " %s=%s\n",
+						 item->ci_name, udm_item->value);
 	}
-	nt->userdata_length = strnlen(nt->extradata_complete,
-				      sizeof(nt->extradata_complete));
+	spin_unlock_irqrestore(&target_list_lock, flags);
 }
 
 static ssize_t userdatum_value_store(struct config_item *item, const char *buf,

---
base-commit: 84a905290cb4c3d9a71a9e3b2f2e02e031e7512f
change-id: 20251020-netconsole-fix-race-f465f37b57ea

Best regards,
-- 
Gustavo Duarte <gustavold@meta.com>

Re: [PATCH net v4] netconsole: Fix race condition in between reader and writer of userdata

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Tue, 28 Oct 2025 15:06:32 -0700 you wrote:
> The update_userdata() function constructs the complete userdata string
> in nt->extradata_complete and updates nt->userdata_length. This data
> is then read by write_msg() and write_ext_msg() when sending netconsole
> messages. However, update_userdata() was not holding target_list_lock
> during this process, allowing concurrent message transmission to read
> partially updated userdata.
> 
> [...]

Here is the summary with links:
  - [net,v4] netconsole: Fix race condition in between reader and writer of userdata
    https://git.kernel.org/netdev/net/c/00764aa5c9bb

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html