CVE-2026-81005

In the Linux kernel, the following vulnerability has been resolved:

ipmi: si: Fix NULL pointer dereference after failed registration

try smi init() allocates new smi->si sm and later calls ipmi register smi mod(), which maps to ipmi add smi().

During ipmi add smi(), the upper IPMI message handler obtains the initial BMC device information through bmc get device id(). This can fail if the BMC does not return a successful response to the Get Device ID command.

When the BMC returns a nonzero completion code, the device-id helper retries the command and eventually returns -EIO if the device ID still cannot be fetched.

On this failure path, ipmi add smi() logs "Unable to get the device id" and goes to out err started, where it invokes the lower driver's shutdown callback. try smi init() then logs the returned registration failure:

ipmi si IPI0001:00: IPMI message handler: Unable to get the device id: -5 ipmi si IPI0001:00: Unable to register device: error -5

For ipmi si, the shutdown callback is shutdown smi(), which cleans up the SI state machine data, frees smi info->si sm, and sets smi info->si sm and smi info->intf to NULL.

However, intf->in shutdown is not set on this failed-registration rollback path. Therefore, the asynchronous redo bmc reg work item can still retry BMC device-id probing after the lower driver has already cleared its SI state machine data. In the observed case, that retry path reached start next msg(), which passed the NULL smi info->si sm pointer to the selected KCS state machine handler:

BUG: unable to handle kernel NULL pointer dereference at 0000000000000000 Workqueue: events redo bmc reg [ipmi msghandler] RIP: start kcs transaction+0x2c/0x190 [ipmi si] Call Trace: start next msg+0x50/0x80 [ipmi si] check start timer thread.part.9+0x3b/0x50 [ipmi si] sender+0x69/0x80 [ipmi si] i ipmi request+0x2ac/0x9d0 [ipmi msghandler] get device id.isra.29+0xaa/0x180 [ipmi msghandler] bmc get device id+0xef/0x950 [ipmi msghandler] redo bmc reg+0x52/0x60 [ipmi msghandler] process one work+0x1a7/0x360

Set intf->in shutdown on the out err started path before invoking the lower driver's shutdown callback. This prevents later redo bmc reg retries from using an interface whose lower driver state has been cleaned up, and applies the same shutdown state to other IPMI interfaces as well.