2018-07-17 10:51:18

by Corentin Labbe

Subject: [PATCH] Revert "ata: ahci_platform: convert kcalloc to devm_kcalloc"

Since ahci_platform_put_resources() use target_pwrs after "devm_" freed
it, we cannot use devm_kcalloc for allocating target_pwrs.

This reverts commit bd0038b1b4f499d814d8f33a55b1df5ea6cf3b85.

Reported-by: Mikko Perttunen <[email protected]>
Signed-off-by: Corentin Labbe <[email protected]>
---
drivers/ata/libahci_platform.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
index fe8939e161ea..2ceebaf2ed56 100644
--- a/drivers/ata/libahci_platform.c
+++ b/drivers/ata/libahci_platform.c
@@ -271,6 +271,8 @@ static void ahci_platform_put_resources(struct device *dev, void *res)
for (c = 0; c < hpriv->nports; c++)
if (hpriv->target_pwrs && hpriv->target_pwrs[c])
regulator_put(hpriv->target_pwrs[c]);
+
+ kfree(hpriv->target_pwrs);
}

static int ahci_platform_get_phy(struct ahci_host_priv *hpriv, u32 port,
@@ -406,7 +408,11 @@ struct ahci_host_priv *ahci_platform_get_resources(struct platform_device *pdev)
rc = -ENOMEM;
goto err_out;
}
- hpriv->target_pwrs = devm_kcalloc(dev, hpriv->nports, sizeof(*hpriv->target_pwrs), GFP_KERNEL);
+ /*
+ * We cannot use devm_ here, since ahci_platform_put_resources() uses
+ * target_pwrs after devm_ have freed memory
+ */
+ hpriv->target_pwrs = kcalloc(hpriv->nports, sizeof(*hpriv->target_pwrs), GFP_KERNEL);
if (!hpriv->target_pwrs) {
rc = -ENOMEM;
goto err_out;
--
2.16.4

2018-07-17 10:59:28

by Mikko Perttunen

Subject: Re: [PATCH] Revert "ata: ahci_platform: convert kcalloc to devm_kcalloc"

Thanks!

Reviewed-by: Mikko Perttunen <[email protected]>

On 17.07.2018 13:49, Corentin Labbe wrote:
> Since ahci_platform_put_resources() use target_pwrs after "devm_" freed
> it, we cannot use devm_kcalloc for allocating target_pwrs.
>
> This reverts commit bd0038b1b4f499d814d8f33a55b1df5ea6cf3b85.
>
> Reported-by: Mikko Perttunen <[email protected]>
> Signed-off-by: Corentin Labbe <[email protected]>
> ---
> drivers/ata/libahci_platform.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/ata/libahci_platform.c b/drivers/ata/libahci_platform.c
> index fe8939e161ea..2ceebaf2ed56 100644
> --- a/drivers/ata/libahci_platform.c
> +++ b/drivers/ata/libahci_platform.c
> @@ -271,6 +271,8 @@ static void ahci_platform_put_resources(struct device *dev, void *res)
> for (c = 0; c < hpriv->nports; c++)
> if (hpriv->target_pwrs && hpriv->target_pwrs[c])
> regulator_put(hpriv->target_pwrs[c]);
> +
> + kfree(hpriv->target_pwrs);
> }
>
> static int ahci_platform_get_phy(struct ahci_host_priv *hpriv, u32 port,
> @@ -406,7 +408,11 @@ struct ahci_host_priv *ahci_platform_get_resources(struct platform_device *pdev)
> rc = -ENOMEM;
> goto err_out;
> }
> - hpriv->target_pwrs = devm_kcalloc(dev, hpriv->nports, sizeof(*hpriv->target_pwrs), GFP_KERNEL);
> + /*
> + * We cannot use devm_ here, since ahci_platform_put_resources() uses
> + * target_pwrs after devm_ have freed memory
> + */
> + hpriv->target_pwrs = kcalloc(hpriv->nports, sizeof(*hpriv->target_pwrs), GFP_KERNEL);
> if (!hpriv->target_pwrs) {
> rc = -ENOMEM;
> goto err_out;
>

2018-07-18 06:31:43

by Marcel Ziswiler

Subject: Re: [PATCH] Revert "ata: ahci_platform: convert kcalloc to devm_kcalloc"

On Tue, 2018-07-17 at 13:57 +0300, Mikko Perttunen wrote:
> Thanks!

Yes, that fixes Apalis TK1 crashing as of late as well.

> Reviewed-by: Mikko Perttunen <[email protected]>

Tested-by: Marcel Ziswiler <[email protected]>

> On 17.07.2018 13:49, Corentin Labbe wrote:
> > Since ahci_platform_put_resources() use target_pwrs after "devm_"
> > freed
> > it, we cannot use devm_kcalloc for allocating target_pwrs.
> >
> > This reverts commit bd0038b1b4f499d814d8f33a55b1df5ea6cf3b85.
> >
> > Reported-by: Mikko Perttunen <[email protected]>
> > Signed-off-by: Corentin Labbe <[email protected]>
> > ---
> > drivers/ata/libahci_platform.c | 8 +++++++-
> > 1 file changed, 7 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/ata/libahci_platform.c
> > b/drivers/ata/libahci_platform.c
> > index fe8939e161ea..2ceebaf2ed56 100644
> > --- a/drivers/ata/libahci_platform.c
> > +++ b/drivers/ata/libahci_platform.c
> > @@ -271,6 +271,8 @@ static void ahci_platform_put_resources(struct
> > device *dev, void *res)
> > for (c = 0; c < hpriv->nports; c++)
> > if (hpriv->target_pwrs && hpriv->target_pwrs[c])
> > regulator_put(hpriv->target_pwrs[c]);
> > +
> > + kfree(hpriv->target_pwrs);
> > }
> >
> > static int ahci_platform_get_phy(struct ahci_host_priv *hpriv,
> > u32 port,
> > @@ -406,7 +408,11 @@ struct ahci_host_priv
> > *ahci_platform_get_resources(struct platform_device *pdev)
> > rc = -ENOMEM;
> > goto err_out;
> > }
> > - hpriv->target_pwrs = devm_kcalloc(dev, hpriv->nports,
> > sizeof(*hpriv->target_pwrs), GFP_KERNEL);
> > + /*
> > + * We cannot use devm_ here, since
> > ahci_platform_put_resources() uses
> > + * target_pwrs after devm_ have freed memory
> > + */
> > + hpriv->target_pwrs = kcalloc(hpriv->nports, sizeof(*hpriv-
> > >target_pwrs), GFP_KERNEL);
> > if (!hpriv->target_pwrs) {
> > rc = -ENOMEM;
> > goto err_out;

2018-07-24 15:24:45

by Jon Hunter

Subject: Re: [PATCH] Revert "ata: ahci_platform: convert kcalloc to devm_kcalloc"

On 18/07/18 07:30, Marcel Ziswiler wrote:
> On Tue, 2018-07-17 at 13:57 +0300, Mikko Perttunen wrote:
>> Thanks!
>
> Yes, that fixes Apalis TK1 crashing as of late as well.
>
>> Reviewed-by: Mikko Perttunen <[email protected]>
>
> Tested-by: Marcel Ziswiler <[email protected]>

Tested-by: Jon Hunter <[email protected]>

Can someone pick this up? Without this some Tegra boards cannot boot.

Thanks!
Jon

--
nvpublic

2018-07-24 16:49:07

by Tejun Heo

Subject: Re: [PATCH] Revert "ata: ahci_platform: convert kcalloc to devm_kcalloc"

On Tue, Jul 17, 2018 at 10:49:35AM +0000, Corentin Labbe wrote:
> Since ahci_platform_put_resources() use target_pwrs after "devm_" freed
> it, we cannot use devm_kcalloc for allocating target_pwrs.
>
> This reverts commit bd0038b1b4f499d814d8f33a55b1df5ea6cf3b85.
>
> Reported-by: Mikko Perttunen <[email protected]>
> Signed-off-by: Corentin Labbe <[email protected]>

Applied to libata/for-4.19. Sorry about the delay.

Thanks.

--
tejun