On Tue, Oct 22, 2019 at 4:48 PM Colin King <colin.king@xxxxxxxxxxxxx> wrote:This patch has nothing to do with the memleak fix.
From: Colin Ian King <colin.king@xxxxxxxxxxxxx>
Currently the kfree of pointer qfprom_cdata is kfreeing an
error value that has been cast to a pointer rather than a
valid address. Fix this by removing the kfree.
Hmm, we just added this to other places[1] as a fix for mem leaks
using the nvmem api. >
I think we need to fix up the qfprom_read wrapper. Srini?
[1] https://lore.kernel.org/stable/20191010083616.685532154@xxxxxxxxxxxxxxxxxxx/
Fixes: 95ededc17e4e ("thermal: qcom: tsens-v1: Add support for MSM8956 and MSM8976")
Signed-off-by: Colin Ian King <colin.king@xxxxxxxxxxxxx>
---
drivers/thermal/qcom/tsens-v1.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/thermal/qcom/tsens-v1.c b/drivers/thermal/qcom/tsens-v1.c
index 2d1077b64887..bd2ddb684a45 100644
--- a/drivers/thermal/qcom/tsens-v1.c
+++ b/drivers/thermal/qcom/tsens-v1.c
@@ -240,10 +240,8 @@ static int calibrate_8976(struct tsens_priv *priv)
u32 *qfprom_cdata;
qfprom_cdata = (u32 *)qfprom_read(priv->dev, "calib");
- if (IS_ERR(qfprom_cdata)) {
- kfree(qfprom_cdata);
+ if (IS_ERR(qfprom_cdata))
return PTR_ERR(qfprom_cdata);
- }
mode = (qfprom_cdata[4] & MSM8976_CAL_SEL_MASK);
dev_dbg(priv->dev, "calibration mode is %d\n", mode);
--
2.20.1