netdev - Re: [PATCH RFC net] net: hibmcge: Release irq resources on error and device tear-down

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20241205205511.GF2581@kernel.org>
Date: Thu, 5 Dec 2024 20:55:11 +0000
From: Simon Horman <horms@...nel.org>
To: Jijie Shao <shaojijie@...wei.com>, Jian Shen <shenjian15@...wei.com>,
	Salil Mehta <salil.mehta@...wei.com>
Cc: Andrew Lunn <andrew+netdev@...n.ch>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
	netdev@...r.kernel.org
Subject: Re: [PATCH RFC net] net: hibmcge: Release irq resources on error and
 device tear-down

On Thu, Dec 05, 2024 at 05:05:23PM +0000, Simon Horman wrote:
> This patch addresses two problems related to leaked resources allocated
> by hbg_irq_init().
> 
> 1. On error release allocated resources
> 2. Otherwise, release the allocated irq vector on device tear-down
>    by setting-up a devres to do so.
> 
> Found by inspection.
> Compile tested only.
> 
> Fixes: 4d089035fa19 ("net: hibmcge: Add interrupt supported in this module")
> Signed-off-by: Simon Horman <horms@...nel.org>

Sorry for the noise, but on reflection I realise that the devm_free_irq()
portion of my patch, which is most of it, is not necessary as the
allocations are made using devm_request_irq().  And the driver seems to
rely on failure during init resulting in device tear-down, at which point
devres will take care of freeing the IRQs.

But I don't see where the IRQ vectors are freed, either on error in
hbg_irq_init or device tear-down. I think the following, somewhat smaller
patch, would be sufficient to address that.

diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c
index 25dd25f096fe..44294453d0e4 100644
--- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c
+++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c
@@ -83,6 +83,11 @@ static irqreturn_t hbg_irq_handle(int irq_num, void *p)
 static const char *irq_names_map[HBG_VECTOR_NUM] = { "tx", "rx",
 						     "err", "mdio" };
 
+static void hbg_free_irq_vectors(void *data)
+{
+	pci_free_irq_vectors(data);
+}
+
 int hbg_irq_init(struct hbg_priv *priv)
 {
 	struct hbg_vector *vectors = &priv->vectors;
@@ -96,6 +101,13 @@ int hbg_irq_init(struct hbg_priv *priv)
 	if (ret < 0)
 		return dev_err_probe(dev, ret, "failed to allocate vectors\n");
 
+	ret = devm_add_action_or_reset(dev, hbg_free_irq_vectors, priv->pdev);
+	if (ret) {
+		pci_free_irq_vectors(priv->pdev);
+		return dev_err_probe(dev, ret,
+				     "failed to add devres to free vectors\n");
+	}
+
 	if (ret != HBG_VECTOR_NUM)
 		return dev_err_probe(dev, -EINVAL,
 				     "requested %u MSI, but allocated %d MSI\n",
-- 
2.45.2