Re: blktests nvme 041,042 leak memory

[Hannes Reinecke <hare@suse.de>]

On 6/3/24 10:52, Maurizio Lombardi wrote:
> po 3. 6. 2024 v 10:37 odesílatel Hannes Reinecke <hare@suse.de> napsal:
>>
>> On 5/31/24 16:37, Maurizio Lombardi wrote:
>> Hmm. Don't really like it.
>> We took great pains of moving then entire DH-CHAP code into its own
>> file, and this patch reverts it and is causing the DH-CHAP code to be
>> littered all over the place.
> 
> This was just an example, I'm not asking to merge it.
> 
> I just had to find a way to make it possible to call nvme_auth_stop() even
> when nvme_auth_init() has never been called without crashing the kernel,
> moving INIT_WORK was the easiest way.
> 
Like this?

diff --git a/drivers/nvme/host/auth.c b/drivers/nvme/host/auth.c
index a264b3ae078b..668cc4d891d7 100644
--- a/drivers/nvme/host/auth.c
+++ b/drivers/nvme/host/auth.c
@@ -983,7 +983,8 @@ EXPORT_SYMBOL_GPL(nvme_auth_init_ctrl);

  void nvme_auth_stop(struct nvme_ctrl *ctrl)
  {
-       cancel_work_sync(&ctrl->dhchap_auth_work);
+       if (ctrl->dhchap_ctxs)
+               cancel_work_sync(&ctrl->dhchap_auth_work);
  }
  EXPORT_SYMBOL_GPL(nvme_auth_stop);

Cheers,

Hannes