[PATCH 0/2] add missing calls to kfree

[PATCH 0/2] add missing calls to kfree

[Quentin Lambert]

The first patch is a style fix, the second add calls to kfree.
The reason for me thinking that they are necessary is that every other branches
leading to an error return are jumping to the same destination.

---
 drivers/staging/rts5208/ms.c |   18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)
---

[PATCH 1/2] staging: rts5208/ms.c: change the label name to respect the coding style

[Quentin Lambert]

Signed-off-by: Quentin Lambert <lambert.quentin@gmail.com>
---
 drivers/staging/rts5208/ms.c |   12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

--- a/drivers/staging/rts5208/ms.c
+++ b/drivers/staging/rts5208/ms.c
@@ -4313,7 +4313,7 @@ int mg_get_local_EKB(struct scsi_cmnd *s
 	if (retval != STATUS_SUCCESS) {
 		set_sense_type(chip, lun, SENSE_TYPE_MG_KEY_FAIL_NOT_AUTHEN);
 		rtsx_trace(chip);
-		goto GetEKBFinish;
+		goto free_buffer;
 	}
 
 	retval = ms_transfer_data(chip, MS_TM_AUTO_READ, PRO_READ_LONG_DATA,
@@ -4322,7 +4322,7 @@ int mg_get_local_EKB(struct scsi_cmnd *s
 		set_sense_type(chip, lun, SENSE_TYPE_MG_KEY_FAIL_NOT_AUTHEN);
 		rtsx_clear_ms_error(chip);
 		rtsx_trace(chip);
-		goto GetEKBFinish;
+		goto free_buffer;
 	}
 	if (check_ms_err(chip)) {
 		set_sense_type(chip, lun, SENSE_TYPE_MG_KEY_FAIL_NOT_AUTHEN);
@@ -4334,7 +4334,7 @@ int mg_get_local_EKB(struct scsi_cmnd *s
 	bufflen = min_t(int, 1052, scsi_bufflen(srb));
 	rtsx_stor_set_xfer_buf(buf, bufflen, srb);
 
-GetEKBFinish:
+free_buffer:
 	kfree(buf);
 	return retval;
 }
@@ -4566,7 +4566,7 @@ int mg_get_ICV(struct scsi_cmnd *srb, st
 	if (retval != STATUS_SUCCESS) {
 		set_sense_type(chip, lun, SENSE_TYPE_MEDIA_UNRECOVER_READ_ERR);
 		rtsx_trace(chip);
-		goto GetICVFinish;
+		goto free_buffer;
 	}
 
 	retval = ms_transfer_data(chip, MS_TM_AUTO_READ, PRO_READ_LONG_DATA,
@@ -4575,7 +4575,7 @@ int mg_get_ICV(struct scsi_cmnd *srb, st
 		set_sense_type(chip, lun, SENSE_TYPE_MEDIA_UNRECOVER_READ_ERR);
 		rtsx_clear_ms_error(chip);
 		rtsx_trace(chip);
-		goto GetICVFinish;
+		goto free_buffer;
 	}
 	if (check_ms_err(chip)) {
 		set_sense_type(chip, lun, SENSE_TYPE_MEDIA_UNRECOVER_READ_ERR);
@@ -4587,7 +4587,7 @@ int mg_get_ICV(struct scsi_cmnd *srb, st
 	bufflen = min_t(int, 1028, scsi_bufflen(srb));
 	rtsx_stor_set_xfer_buf(buf, bufflen, srb);
 
-GetICVFinish:
+free_buffer:
 	kfree(buf);
 	return retval;
 }

Re: [PATCH 1/2] staging: rts5208/ms.c: change the label name to respect the coding style

[Greg Kroah-Hartman]

On Fri, Aug 12, 2016 at 02:58:34PM +0200, Quentin Lambert wrote:
> Signed-off-by: Quentin Lambert <lambert.quentin@gmail.com>
> ---
>  drivers/staging/rts5208/ms.c |   12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)

I can't take patches without any changelog comments, sorry.

greg k-h

[PATCH 2/2] staging: rts5208/ms.c: add missing releases in mg_get_local_EKB and mg_get_ICV

[Quentin Lambert]

mg_get_local_EKB and mg_get_ICV used to return with an error code before
releasing all resources. This patch add a jump to the appropriate label
ensuring that the resources are properly released before returning.

This issue was found with Hector.

Signed-off-by: Quentin Lambert <lambert.quentin@gmail.com>
---
 drivers/staging/rts5208/ms.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

--- a/drivers/staging/rts5208/ms.c
+++ b/drivers/staging/rts5208/ms.c
@@ -4328,7 +4328,8 @@ int mg_get_local_EKB(struct scsi_cmnd *s
 		set_sense_type(chip, lun, SENSE_TYPE_MG_KEY_FAIL_NOT_AUTHEN);
 		rtsx_clear_ms_error(chip);
 		rtsx_trace(chip);
-		return STATUS_FAIL;
+		retval = STATUS_FAIL;
+		goto free_buffer;
 	}
 
 	bufflen = min_t(int, 1052, scsi_bufflen(srb));
@@ -4581,7 +4582,8 @@ int mg_get_ICV(struct scsi_cmnd *srb, st
 		set_sense_type(chip, lun, SENSE_TYPE_MEDIA_UNRECOVER_READ_ERR);
 		rtsx_clear_ms_error(chip);
 		rtsx_trace(chip);
-		return STATUS_FAIL;
+		retval = STATUS_FAIL;
+		goto free_buffer;
 	}
 
 	bufflen = min_t(int, 1028, scsi_bufflen(srb));