Le 25/09/2023 à 20:46, Jeff Johnson a écrit :
On 9/23/2023 11:57 PM, Christophe JAILLET wrote:
If an error occurs and channel_detector_exit() is called, it relies on
entries of the 'detectors' array to be NULL.
Otherwise, it may access to un-initialized memory.
Fix it and initialize the memory, as what was done before the commit in
Fixes.
Fixes: a063b650ce5d ("ath: dfs_pattern_detector: Avoid open coded
arithmetic in memory allocation")
Signed-off-by: Christophe JAILLET
<christophe.jaillet-39ZsbGIQGT5GWvitb5QawA@xxxxxxxxxxxxxxxx>
---
Patch #1/2 is a fix, for for wireless.
Patch #2/2 is for wireless-next I guess, but depnds on #1
Not sure if we can mix different target in the same serie. Let me know.
BTW, sorry for messing up things with a063b650ce5d :(
---
drivers/net/wireless/ath/dfs_pattern_detector.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/wireless/ath/dfs_pattern_detector.c
b/drivers/net/wireless/ath/dfs_pattern_detector.c
index 27f4d74a41c8..2788a1b06c17 100644
--- a/drivers/net/wireless/ath/dfs_pattern_detector.c
+++ b/drivers/net/wireless/ath/dfs_pattern_detector.c
@@ -206,7 +206,7 @@ channel_detector_create(struct
dfs_pattern_detector *dpd, u16 freq)
INIT_LIST_HEAD(&cd->head);
cd->freq = freq;
- cd->detectors = kmalloc_array(dpd->num_radar_types,
+ cd->detectors = kcalloc(dpd->num_radar_types,
sizeof(*cd->detectors), GFP_ATOMIC);
nit: align descendant on (
Agreed, but as the code is removed in patch 2/2, I thought that having a
smaller diff was a better option.
Let me know if I should resend the serie.
CJ
if (cd->detectors == NULL)
goto fail;
