Re: [RFC] module: init-failure path can free a module with live try_module_get() users

[Mahanta Jambigi]

On 18/09/26 6:52 pm, David Laight wrote:
> On Mon, 24 Aug 2026 11:43:46 +0530
> Mahanta Jambigi <mjambigi@xxxxxxxxxxxxx> wrote:
> 
>> Hi Luis, Petr, Daniel, Sami, Aaron,
>>
>> I'm writing to ask about what looks like a generic module-init failure
>> lifetime problem in the module loader. I ran into it while working on
>> the SMC networking module (net/smc/), but after several patch
>> iterations, it seems the root issue may belong in kernel/module/main.c
>> rather than in SMC itself. I'd appreciate your guidance on whether this
>> reading is correct, and if so, what fix direction would be preferred.
>>
>> THE ISSUE IN do_init_module()
>> =============================
>>
>> include/linux/module.h has a long-standing FIXME in module_is_live():
>>
>>     /* FIXME: It'd be nice to isolate modules during init, too, so they
>>        aren't used before they (may) fail. But presently too much code
>>        (IDE & SCSI) require entry into the module during init. */
>>     static inline bool module_is_live(struct module *mod)
>>     {
>>         return mod->state != MODULE_STATE_GOING;
>>     }
>>
>> Because MODULE_STATE_COMING is not MODULE_STATE_GOING, try_module_get()
>> can succeed once a module's __init is executing. If __init makes the
>> module externally reachable partway through and then later fails, the
>> failure path in do_init_module() appears to do:
> 
> It is rather worse that that.
> If sock_create() auto-loads a module (eg sctp) then nothing stops a second
> sock_create() entering the protocol code before the initialisation completes.
> That can be hit by two separate applications, I hit it from an out of tree
> kernel module and avoided the problem by putting a mutex() around the
> sock_create() call.
> 
> It might help by letting try_module_get(THIS_MODULE) always succeed
> while blocking other requests until initialisation completes.
> The code making the call must own a reference (otherwise the code could
> just disappear), and that reference stops the module being unloaded.
> That would let the initialisation code grab extra references (eg for
> a worker thread) without allowing other codes paths enter the
> part-initialised driver.

Thanks David — you're right that the race is broader. This patch
addresses only the UAF on the __init failure path: once we set
MODULE_STATE_GOING and call synchronize_rcu(), new callers see GOING and
fail; we then drain existing refs before free_module().

The concurrent-init race you describe — two callers entering
MODULE_STATE_COMING simultaneously during a successful init — is not
addressed here and would require changes to try_module_get() itself, as
you suggest. That is the long-standing FIXME in module_is_live() and is
a separate, larger change.