Re: [PATCH v4 4/4] scsi: core: Rework the code for dropping the LLD module reference

From: Bart Van Assche <bvanassche@acm.org>
To: John Garry <john.garry@huawei.com>,
	"Martin K . Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org, Christoph Hellwig <hch@lst.de>,
	Ming Lei <ming.lei@redhat.com>, Hannes Reinecke <hare@suse.de>,
	Mike Christie <michael.christie@oracle.com>,
	Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>,
	"James E.J. Bottomley" <jejb@linux.ibm.com>
Subject: Re: [PATCH v4 4/4] scsi: core: Rework the code for dropping the LLD module reference
Date: Wed, 14 Sep 2022 10:54:42 -0700	[thread overview]
Message-ID: <b62a8571-b1cf-fc09-3dae-3c75c06bbb32@acm.org> (raw)
In-Reply-To: <011da034-b67a-c232-ebe0-d6d7d802247f@huawei.com>

On 9/14/22 02:52, John Garry wrote:
> On 13/09/2022 20:57, Bart Van Assche wrote:
>> -    mod = sdev->host->hostt->module;
>> +    mod = sdev->drop_module_ref ? sdev->host->hostt->module : NULL;
> 
> I suppose that this works.
> 
> My reservation is that there were some concerns of current module 
> referencing solution, so may not be better to build directly on it:
> 
> https://lore.kernel.org/linux-scsi/Ynt0aFMX+z%2FUhGJ2@infradead.org/

Hi John,

The above link points at a reply from Christoph Hellwig with the 
suggestion to use __module_get(). Implementing that suggestion would 
require to modify the kernel module implementation. My conclusion from 
reading the code in kernel/module/main.c is that increasing the module 
reference count after unloading has started currently has no effect. I 
have not found any code in the delete_module() system call 
implementation that waits for the reference count to drop to zero after 
try_stop_module() succeeded. I will look into this.

Thanks,

Bart.