2019-11-21 19:43:00

by Navid Emamdoost

[permalink] [raw]
Subject: [PATCH] EDAC: Fix memory leak in i5100_init_one

In the implementation of i5100_init_one(), the allocated mci is not
correctly released in cases of failure. The release via edac_mc_free()
should be reached by label bail_einj and subsequents.

Fixes: 52608ba20546 ("i5100_edac: probe for device 19 function 0")
Signed-off-by: Navid Emamdoost <[email protected]>
---
drivers/edac/i5100_edac.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/edac/i5100_edac.c b/drivers/edac/i5100_edac.c
index 251f2b692785..710fabd5fba1 100644
--- a/drivers/edac/i5100_edac.c
+++ b/drivers/edac/i5100_edac.c
@@ -1149,12 +1149,12 @@ static int i5100_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
bail_scrub:
priv->scrub_enable = 0;
cancel_delayed_work_sync(&(priv->i5100_scrubbing));
- edac_mc_free(mci);

bail_disable_einj:
pci_disable_device(einj);

bail_einj:
+ edac_mc_free(mci);
pci_dev_put(einj);

bail_disable_ch1:
--
2.17.1


2019-11-21 20:42:12

by Robert Richter

[permalink] [raw]
Subject: Re: [PATCH] EDAC: Fix memory leak in i5100_init_one

On 21.11.19 13:40:41, Navid Emamdoost wrote:
> In the implementation of i5100_init_one(), the allocated mci is not
> correctly released in cases of failure. The release via edac_mc_free()
> should be reached by label bail_einj and subsequents.
>
> Fixes: 52608ba20546 ("i5100_edac: probe for device 19 function 0")
> Signed-off-by: Navid Emamdoost <[email protected]>
> ---
> drivers/edac/i5100_edac.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/edac/i5100_edac.c b/drivers/edac/i5100_edac.c
> index 251f2b692785..710fabd5fba1 100644
> --- a/drivers/edac/i5100_edac.c
> +++ b/drivers/edac/i5100_edac.c
> @@ -1149,12 +1149,12 @@ static int i5100_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
> bail_scrub:
> priv->scrub_enable = 0;
> cancel_delayed_work_sync(&(priv->i5100_scrubbing));
> - edac_mc_free(mci);
>
> bail_disable_einj:
> pci_disable_device(einj);
>
> bail_einj:
> + edac_mc_free(mci);
> pci_dev_put(einj);

The whole error path looks broken:

bail_disable_einj: may *not* call pci_disable_device().

bail_einj: may not call pci_dev_put()

i5100_setup_debugfs() may fail.

etc.

All needs to be fixed. Please review the error path.

Thanks,

-Robert

>
> bail_disable_ch1:
> --
> 2.17.1
>