[PATCH 0/2] meson-nand: two small memory related fixes

[PATCH 0/2] meson-nand: two small memory related fixes

[Martin Blumenstingl]

While trying to add support for older Meson SoCs to the meson-nand
driver I was experiencing a crash in meson_nfc_read_buf(). While trying
to find out why that happened I inspected the code in that function and
found that there's:
- a missing check on the return value of a kzalloc() call
- a potential memory leak in it

Both fixes have nothing to do with my original crash (for which I'll
open a separate thread).


Martin Blumenstingl (2):
  mtd: rawnand: meson: add missing ENOMEM check in meson_nfc_read_buf()
  mtd: rawnand: meson: fix a potential memory leak in meson_nfc_read_buf

 drivers/mtd/nand/raw/meson_nand.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

-- 
2.21.0

[PATCH 1/2] mtd: rawnand: meson: add missing ENOMEM check in meson_nfc_read_buf()

[Martin Blumenstingl]

kzalloc() can return NULL if memory could not be allocated. Check the
return value of the kzalloc() call in meson_nfc_read_buf() to make it
consistent with other memory allocations within the meson_nand driver.

Fixes: 8fae856c53500a ("mtd: rawnand: meson: add support for Amlogic NAND flash controller")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
---
 drivers/mtd/nand/raw/meson_nand.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/mtd/nand/raw/meson_nand.c b/drivers/mtd/nand/raw/meson_nand.c
index 3e8aa71407b5..a1d8506b61c7 100644
--- a/drivers/mtd/nand/raw/meson_nand.c
+++ b/drivers/mtd/nand/raw/meson_nand.c
@@ -528,6 +528,9 @@ static int meson_nfc_read_buf(struct nand_chip *nand, u8 *buf, int len)
 	u8 *info;
 
 	info = kzalloc(PER_INFO_BYTE, GFP_KERNEL);
+	if (!info)
+		return -ENOMEM;
+
 	ret = meson_nfc_dma_buffer_setup(nand, buf, len, info,
 					 PER_INFO_BYTE, DMA_FROM_DEVICE);
 	if (ret)
-- 
2.21.0

[PATCH 2/2] mtd: rawnand: meson: fix a potential memory leak in meson_nfc_read_buf

[Martin Blumenstingl]

meson_nfc_dma_buffer_setup() is called with the "info" buffer which is
allocated a few lines before using kzalloc(). If
meson_nfc_dma_buffer_setup() fails we need to free the allocated "info"
buffer instead of only freeing it upon success.

Fixes: 8fae856c53500a ("mtd: rawnand: meson: add support for Amlogic NAND flash controller")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
---
 drivers/mtd/nand/raw/meson_nand.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/mtd/nand/raw/meson_nand.c b/drivers/mtd/nand/raw/meson_nand.c
index a1d8506b61c7..38db4fd61459 100644
--- a/drivers/mtd/nand/raw/meson_nand.c
+++ b/drivers/mtd/nand/raw/meson_nand.c
@@ -534,7 +534,7 @@ static int meson_nfc_read_buf(struct nand_chip *nand, u8 *buf, int len)
 	ret = meson_nfc_dma_buffer_setup(nand, buf, len, info,
 					 PER_INFO_BYTE, DMA_FROM_DEVICE);
 	if (ret)
-		return ret;
+		goto out;
 
 	cmd = NFC_CMD_N2M | (len & GENMASK(5, 0));
 	writel(cmd, nfc->reg_base + NFC_REG_CMD);
@@ -542,6 +542,8 @@ static int meson_nfc_read_buf(struct nand_chip *nand, u8 *buf, int len)
 	meson_nfc_drain_cmd(nfc);
 	meson_nfc_wait_cmd_finish(nfc, 1000);
 	meson_nfc_dma_buffer_release(nand, len, PER_INFO_BYTE, DMA_FROM_DEVICE);
+
+out:
 	kfree(info);
 
 	return ret;
-- 
2.21.0

Re: [PATCH 1/2] mtd: rawnand: meson: add missing ENOMEM check in meson_nfc_read_buf()

[Liang Yang]

Hello Martin,

On 2019/3/2 1:38, Martin Blumenstingl wrote:
> kzalloc() can return NULL if memory could not be allocated. Check the
> return value of the kzalloc() call in meson_nfc_read_buf() to make it
> consistent with other memory allocations within the meson_nand driver.
> 
> Fixes: 8fae856c53500a ("mtd: rawnand: meson: add support for Amlogic NAND flash controller")
> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
> ---
>   drivers/mtd/nand/raw/meson_nand.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/mtd/nand/raw/meson_nand.c b/drivers/mtd/nand/raw/meson_nand.c
> index 3e8aa71407b5..a1d8506b61c7 100644
> --- a/drivers/mtd/nand/raw/meson_nand.c
> +++ b/drivers/mtd/nand/raw/meson_nand.c
> @@ -528,6 +528,9 @@ static int meson_nfc_read_buf(struct nand_chip *nand, u8 *buf, int len)
>   	u8 *info;
>   
>   	info = kzalloc(PER_INFO_BYTE, GFP_KERNEL);
> +	if (!info)
> +		return -ENOMEM;
> +

Thank you very much. it is really good to me.
Acked-by: Liang Yang <liang.yang@amlogic.com>

>   	ret = meson_nfc_dma_buffer_setup(nand, buf, len, info,
>   					 PER_INFO_BYTE, DMA_FROM_DEVICE);
>   	if (ret)
> 

Re: [PATCH 2/2] mtd: rawnand: meson: fix a potential memory leak in meson_nfc_read_buf

[Liang Yang]

Hello Martin,

Thank you very much.

On 2019/3/2 1:38, Martin Blumenstingl wrote:
> meson_nfc_dma_buffer_setup() is called with the "info" buffer which is
> allocated a few lines before using kzalloc(). If
> meson_nfc_dma_buffer_setup() fails we need to free the allocated "info"
> buffer instead of only freeing it upon success.
> 
> Fixes: 8fae856c53500a ("mtd: rawnand: meson: add support for Amlogic NAND flash controller")
> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
> ---
>   drivers/mtd/nand/raw/meson_nand.c | 4 +++-
>   1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mtd/nand/raw/meson_nand.c b/drivers/mtd/nand/raw/meson_nand.c
> index a1d8506b61c7..38db4fd61459 100644
> --- a/drivers/mtd/nand/raw/meson_nand.c
> +++ b/drivers/mtd/nand/raw/meson_nand.c
> @@ -534,7 +534,7 @@ static int meson_nfc_read_buf(struct nand_chip *nand, u8 *buf, int len)
>   	ret = meson_nfc_dma_buffer_setup(nand, buf, len, info,
>   					 PER_INFO_BYTE, DMA_FROM_DEVICE);
>   	if (ret)
> -		return ret;
> +		goto out;
>   Looks good to me.
Acked-by: Liang Yang <liang.yang@amlogic.com>

>   	cmd = NFC_CMD_N2M | (len & GENMASK(5, 0));
>   	writel(cmd, nfc->reg_base + NFC_REG_CMD);
> @@ -542,6 +542,8 @@ static int meson_nfc_read_buf(struct nand_chip *nand, u8 *buf, int len)
>   	meson_nfc_drain_cmd(nfc);
>   	meson_nfc_wait_cmd_finish(nfc, 1000);
>   	meson_nfc_dma_buffer_release(nand, len, PER_INFO_BYTE, DMA_FROM_DEVICE);
> +
> +out:
Looks good to me.
Acked-by: Liang Yang <liang.yang@amlogic.com>
>   	kfree(info);
>   
>   	return ret;
>