Received-SPF: pass (google.com: domain of linux-kernel+bounces-17272-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Message-ID: <e194200b-18e9-406d-9faf-03b54c1e606c@deltatee.com>
Date: Thu, 4 Jan 2024 14:48:33 -0700
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
To: Christophe JAILLET <christophe.jaillet@wanadoo.fr>,
 Bjorn Helgaas <bhelgaas@google.com>, Eric Dumazet <edumazet@google.com>
Cc: alexis.lothore@bootlin.com, linux-kernel@vger.kernel.org,
 kernel-janitors@vger.kernel.org, linux-pci@vger.kernel.org
References: <02d9ec4a10235def0e764ff1f5be881ba12e16e8.1704397858.git.christophe.jaillet@wanadoo.fr>
Content-Language: en-CA
From: Logan Gunthorpe <logang@deltatee.com>
In-Reply-To: <02d9ec4a10235def0e764ff1f5be881ba12e16e8.1704397858.git.christophe.jaillet@wanadoo.fr>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Subject: Re: [PATCH] PCI/P2PDMA: Fix a sleeping issue in a RCU read section


On 2024-01-04 12:52, Christophe JAILLET wrote:
> It is not allowed to sleep within a RCU read section, so use GFP_ATOMIC
> instead of GFP_KERNEL here.
> 
> Fixes: ae21f835a5bd ("PCI/P2PDMA: Finish RCU conversion of pdev->p2pdma")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>

This makes sense to me. Though, the use of RCU could probably use a
review. Seeing p2pdma is only released through a devm action on the
pdev, I would think it shouldn't be needed if we hold a reference to the
pdev.

Other than that:

Reviewed-by: Logan Gunthorpe <logang@deltatee.com>

Thanks!

Logan