| Message ID | 20200513201454.258111-1-christophe.jaillet@wanadoo.fr (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 25ca180ad380a0c7286442a922e7fbcc6a9f6083 |
| Delegated to: | Kalle Valo |
| Headers | show |
| Series | ath11k: Fix some resource leaks in error path in 'ath11k_thermal_register()' | expand |
On 2020-05-13 13:14, Christophe JAILLET wrote: > If 'thermal_cooling_device_register()' fails, we must undo what has > been > allocated so far. So we must go to 'err_thermal_destroy' instead of > returning directly > > In case of error in 'ath11k_thermal_register()', the previous > 'thermal_cooling_device_register()' call must also be undone. Move the > 'ar->thermal.cdev = cdev' a few lines above in order for this to be > done > in 'ath11k_thermal_unregister()' which is called in the error handling > path. > > Fixes: 2a63bbca06b2 ("ath11k: add thermal cooling device support") > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > --- > I'm not 100% confident with this patch. > > - When calling 'ath11k_thermal_unregister()', we try to release some > resources that have not been allocated yet. I don't know if it can be > an > issue or not. > - I think that we should propagate the error code, instead of forcing > -EINVAL. > Good catch. -Rajkumar
Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote: > If 'thermal_cooling_device_register()' fails, we must undo what has been > allocated so far. So we must go to 'err_thermal_destroy' instead of > returning directly > > In case of error in 'ath11k_thermal_register()', the previous > 'thermal_cooling_device_register()' call must also be undone. Move the > 'ar->thermal.cdev = cdev' a few lines above in order for this to be done > in 'ath11k_thermal_unregister()' which is called in the error handling > path. > > Fixes: 2a63bbca06b2 ("ath11k: add thermal cooling device support") > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Patch applied to ath-next branch of ath.git, thanks. 25ca180ad380 ath11k: Fix some resource leaks in error path in 'ath11k_thermal_register()'
diff --git a/drivers/net/wireless/ath/ath11k/thermal.c b/drivers/net/wireless/ath/ath11k/thermal.c index 259dddbda2c7..5a7e150c621b 100644 --- a/drivers/net/wireless/ath/ath11k/thermal.c +++ b/drivers/net/wireless/ath/ath11k/thermal.c @@ -174,9 +174,12 @@ int ath11k_thermal_register(struct ath11k_base *sc) if (IS_ERR(cdev)) { ath11k_err(sc, "failed to setup thermal device result: %ld\n", PTR_ERR(cdev)); - return -EINVAL; + ret = -EINVAL; + goto err_thermal_destroy; } + ar->thermal.cdev = cdev; + ret = sysfs_create_link(&ar->hw->wiphy->dev.kobj, &cdev->device.kobj, "cooling_device"); if (ret) { @@ -184,7 +187,6 @@ int ath11k_thermal_register(struct ath11k_base *sc) goto err_thermal_destroy; } - ar->thermal.cdev = cdev; if (!IS_REACHABLE(CONFIG_HWMON)) return 0;
If 'thermal_cooling_device_register()' fails, we must undo what has been allocated so far. So we must go to 'err_thermal_destroy' instead of returning directly In case of error in 'ath11k_thermal_register()', the previous 'thermal_cooling_device_register()' call must also be undone. Move the 'ar->thermal.cdev = cdev' a few lines above in order for this to be done in 'ath11k_thermal_unregister()' which is called in the error handling path. Fixes: 2a63bbca06b2 ("ath11k: add thermal cooling device support") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> --- I'm not 100% confident with this patch. - When calling 'ath11k_thermal_unregister()', we try to release some resources that have not been allocated yet. I don't know if it can be an issue or not. - I think that we should propagate the error code, instead of forcing -EINVAL. --- drivers/net/wireless/ath/ath11k/thermal.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)