net: davinci_emac: fix oops caused by uninitialized ndev->dev

Message ID 1400580697-24994-1-git-send-email-nsekhar@ti.com
State New
Headers show

Commit Message

Sekhar Nori May 20, 2014, 10:11 a.m.
Commit e194312854edc22a2faf1931b3c0608fe20cb969 (drivers: net:
davinci_cpdma: Convert kzalloc() to devm_kzalloc()) triggered
a bug in emac_probe() wherein dev member of net_device is used
for devres allocations even before it is initialized.

This patch fixes that by using the struct device in platform_device
instead.

While at it, use &pdev->dev consistently for console messages instead
of using ndev->dev for just one case and remove an unnecessary line
continuation.

Reported-by: Kevin Hilman <khilman@linaro.org>
Helped-by: George Cherian <george.cherian@ti.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
---
This patch fixes a bug in linux-next so it can wait for
v3.16

 drivers/net/ethernet/ti/davinci_emac.c |    6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

Comments

Mugunthan V N May 20, 2014, 10:14 a.m. | #1
On Tuesday 20 May 2014 03:41 PM, Sekhar Nori wrote:
> Commit e194312854edc22a2faf1931b3c0608fe20cb969 (drivers: net:
> davinci_cpdma: Convert kzalloc() to devm_kzalloc()) triggered
> a bug in emac_probe() wherein dev member of net_device is used
> for devres allocations even before it is initialized.
>
> This patch fixes that by using the struct device in platform_device
> instead.
>
> While at it, use &pdev->dev consistently for console messages instead
> of using ndev->dev for just one case and remove an unnecessary line
> continuation.
>
> Reported-by: Kevin Hilman <khilman@linaro.org>
> Helped-by: George Cherian <george.cherian@ti.com>
> Signed-off-by: Sekhar Nori <nsekhar@ti.com>
> ---

Acked-by: Mugunthan V N <mugunthanvnm@ti.com>

Regards
Mugunthan V N
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Lad, Prabhakar May 20, 2014, 4:46 p.m. | #2
On Tue, May 20, 2014 at 3:41 PM, Sekhar Nori <nsekhar@ti.com> wrote:
> Commit e194312854edc22a2faf1931b3c0608fe20cb969 (drivers: net:
> davinci_cpdma: Convert kzalloc() to devm_kzalloc()) triggered
> a bug in emac_probe() wherein dev member of net_device is used
> for devres allocations even before it is initialized.
>
> This patch fixes that by using the struct device in platform_device
> instead.
>
> While at it, use &pdev->dev consistently for console messages instead
> of using ndev->dev for just one case and remove an unnecessary line
> continuation.
>
> Reported-by: Kevin Hilman <khilman@linaro.org>
> Helped-by: George Cherian <george.cherian@ti.com>
> Signed-off-by: Sekhar Nori <nsekhar@ti.com>

Tested-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>

Regards,
--Prabhakar Lad

> ---
> This patch fixes a bug in linux-next so it can wait for
> v3.16
>
>  drivers/net/ethernet/ti/davinci_emac.c |    6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/ethernet/ti/davinci_emac.c b/drivers/net/ethernet/ti/davinci_emac.c
> index e76eae5..f32d730 100644
> --- a/drivers/net/ethernet/ti/davinci_emac.c
> +++ b/drivers/net/ethernet/ti/davinci_emac.c
> @@ -1865,7 +1865,6 @@ static int davinci_emac_probe(struct platform_device *pdev)
>         struct emac_priv *priv;
>         unsigned long hw_ram_addr;
>         struct emac_platform_data *pdata;
> -       struct device *emac_dev;
>         struct cpdma_params dma_params;
>         struct clk *emac_clk;
>         unsigned long emac_bus_frequency;
> @@ -1911,7 +1910,6 @@ static int davinci_emac_probe(struct platform_device *pdev)
>         priv->coal_intvl = 0;
>         priv->bus_freq_mhz = (u32)(emac_bus_frequency / 1000000);
>
> -       emac_dev = &ndev->dev;
>         /* Get EMAC platform data */
>         res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>         priv->emac_base_phys = res->start + pdata->ctrl_reg_offset;
> @@ -1930,7 +1928,7 @@ static int davinci_emac_probe(struct platform_device *pdev)
>                 hw_ram_addr = (u32 __force)res->start + pdata->ctrl_ram_offset;
>
>         memset(&dma_params, 0, sizeof(dma_params));
> -       dma_params.dev                  = emac_dev;
> +       dma_params.dev                  = &pdev->dev;
>         dma_params.dmaregs              = priv->emac_base;
>         dma_params.rxthresh             = priv->emac_base + 0x120;
>         dma_params.rxfree               = priv->emac_base + 0x140;
> @@ -1994,7 +1992,7 @@ static int davinci_emac_probe(struct platform_device *pdev)
>
>
>         if (netif_msg_probe(priv)) {
> -               dev_notice(emac_dev, "DaVinci EMAC Probe found device "\
> +               dev_notice(&pdev->dev, "DaVinci EMAC Probe found device "
>                            "(regs: %p, irq: %d)\n",
>                            (void *)priv->emac_base_phys, ndev->irq);
>         }
> --
> 1.7.10.1
>
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Kevin Hilman May 21, 2014, 11:39 p.m. | #3
Sekhar Nori <nsekhar@ti.com> writes:

> Commit e194312854edc22a2faf1931b3c0608fe20cb969 (drivers: net:
> davinci_cpdma: Convert kzalloc() to devm_kzalloc()) triggered
> a bug in emac_probe() wherein dev member of net_device is used
> for devres allocations even before it is initialized.
>
> This patch fixes that by using the struct device in platform_device
> instead.
>
> While at it, use &pdev->dev consistently for console messages instead
> of using ndev->dev for just one case and remove an unnecessary line
> continuation.
>
> Reported-by: Kevin Hilman <khilman@linaro.org>
> Helped-by: George Cherian <george.cherian@ti.com>
> Signed-off-by: Sekhar Nori <nsekhar@ti.com>

Tested-by: Kevin Hilman <khilman@linaro.org>

I verified this patch fixes the boot on dm365-evm (legacy boot) and
da850-evm (DT boot).

Thanks,

Kevin
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller May 22, 2014, 7:22 p.m. | #4
From: Sekhar Nori <nsekhar@ti.com>
Date: Tue, 20 May 2014 15:41:37 +0530

> Commit e194312854edc22a2faf1931b3c0608fe20cb969 (drivers: net:
> davinci_cpdma: Convert kzalloc() to devm_kzalloc()) triggered
> a bug in emac_probe() wherein dev member of net_device is used
> for devres allocations even before it is initialized.
> 
> This patch fixes that by using the struct device in platform_device
> instead.
> 
> While at it, use &pdev->dev consistently for console messages instead
> of using ndev->dev for just one case and remove an unnecessary line
> continuation.
> 
> Reported-by: Kevin Hilman <khilman@linaro.org>
> Helped-by: George Cherian <george.cherian@ti.com>
> Signed-off-by: Sekhar Nori <nsekhar@ti.com>
> ---
> This patch fixes a bug in linux-next so it can wait for
> v3.16

Applied to net-next, thanks.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/drivers/net/ethernet/ti/davinci_emac.c b/drivers/net/ethernet/ti/davinci_emac.c
index e76eae5..f32d730 100644
--- a/drivers/net/ethernet/ti/davinci_emac.c
+++ b/drivers/net/ethernet/ti/davinci_emac.c
@@ -1865,7 +1865,6 @@  static int davinci_emac_probe(struct platform_device *pdev)
 	struct emac_priv *priv;
 	unsigned long hw_ram_addr;
 	struct emac_platform_data *pdata;
-	struct device *emac_dev;
 	struct cpdma_params dma_params;
 	struct clk *emac_clk;
 	unsigned long emac_bus_frequency;
@@ -1911,7 +1910,6 @@  static int davinci_emac_probe(struct platform_device *pdev)
 	priv->coal_intvl = 0;
 	priv->bus_freq_mhz = (u32)(emac_bus_frequency / 1000000);
 
-	emac_dev = &ndev->dev;
 	/* Get EMAC platform data */
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 	priv->emac_base_phys = res->start + pdata->ctrl_reg_offset;
@@ -1930,7 +1928,7 @@  static int davinci_emac_probe(struct platform_device *pdev)
 		hw_ram_addr = (u32 __force)res->start + pdata->ctrl_ram_offset;
 
 	memset(&dma_params, 0, sizeof(dma_params));
-	dma_params.dev			= emac_dev;
+	dma_params.dev			= &pdev->dev;
 	dma_params.dmaregs		= priv->emac_base;
 	dma_params.rxthresh		= priv->emac_base + 0x120;
 	dma_params.rxfree		= priv->emac_base + 0x140;
@@ -1994,7 +1992,7 @@  static int davinci_emac_probe(struct platform_device *pdev)
 
 
 	if (netif_msg_probe(priv)) {
-		dev_notice(emac_dev, "DaVinci EMAC Probe found device "\
+		dev_notice(&pdev->dev, "DaVinci EMAC Probe found device "
 			   "(regs: %p, irq: %d)\n",
 			   (void *)priv->emac_base_phys, ndev->irq);
 	}