From: Krzysztof Kozlowski krzysztof.kozlowski@linaro.org
[ Upstream commit c73d19f89cb03c43abbbfa3b9caa1b8fc719764c ]
Device can be unbound, so driver must also release memory for the wakeup source.
Signed-off-by: Krzysztof Kozlowski krzysztof.kozlowski@linaro.org Link: https://lore.kernel.org/r/20250406202730.55096-1-krzysztof.kozlowski@linaro.... Signed-off-by: Sebastian Reichel sebastian.reichel@collabora.com Signed-off-by: Sasha Levin sashal@kernel.org ---
Perfect! This confirms the pattern. Now let me analyze the commit: **YES** This commit should be backported to stable kernel trees. Here's my extensive analysis: ## Justification for Backporting ### 1. **Bug Fix Nature** The commit fixes a resource leak bug where wakeup sources are not properly cleaned up when the device is unbound. This is a clear bug fix that affects system resource management. ### 2. **Matching Pattern with Similar Commits** This commit follows the exact same pattern as the similar commits marked "YES" for backporting: - **Similar Commit #1 (gpio: mpc8xxx)** and **Similar Commit #3 (gpio: zynq)** both have identical commit messages and fix the same type of wakeup source leak by adding `device_init_wakeup(&pdev->dev, 0)` in the remove function. - All these commits are from the same author (Krzysztof Kozlowski) and have `Cc: stable@vger.kernel.org` in their commit messages, indicating explicit intention for stable tree inclusion. ### 3. **Code Analysis** The code change is minimal and surgical: - **Single line addition**: `device_init_wakeup(&ucb->dev, 0);` in the `collie_bat_remove()` function - **Proper cleanup ordering**: The line is added at the beginning of the remove function, which is the correct place for resource cleanup - **Symmetric operation**: The probe function calls `device_init_wakeup(&ucb->dev, 1)` at line 414, and now the remove function properly disables it with `device_init_wakeup(&ucb->dev, 0)` ### 4. **Resource Leak Prevention** From examining the kernel tree, I can see that: - The probe function enables wakeup with `device_init_wakeup(&ucb->dev, 1)` - Without this fix, when the driver is unbound, the wakeup source memory is leaked - This follows the established pattern seen in other drivers like gpio-zynq.c where `device_init_wakeup(&pdev->dev, 0)` is called in remove functions ### 5. **Low Risk Assessment** - **Minimal scope**: Only affects the collie battery driver cleanup path - **No functional changes**: Doesn't change driver behavior, only fixes resource cleanup - **No architectural changes**: Single line addition for proper resource management - **No new features**: Pure bug fix ### 6. **Stable Tree Criteria Compliance** - ✅ **Important bug fix**: Prevents resource leaks - ✅ **Minimal risk**: Single line, well-understood API call - ✅ **No new features**: Only fixes cleanup - ✅ **Small and contained**: Confined to one driver's remove function - ✅ **Clear side effects**: Only positive effect of proper resource cleanup ### 7. **Explicit Stable Tree Intent** The commit message doesn't include `Cc: stable@vger.kernel.org`, but the similar commits #1 and #3 from the same author fixing identical issues do have this tag, suggesting this is part of a series addressing wakeup source leaks across multiple drivers. The fix is essential for proper resource management and follows established kernel patterns for wakeup source cleanup during device unbinding.
drivers/power/supply/collie_battery.c | 1 + 1 file changed, 1 insertion(+)
diff --git a/drivers/power/supply/collie_battery.c b/drivers/power/supply/collie_battery.c index 68390bd1004f0..3daf7befc0bf6 100644 --- a/drivers/power/supply/collie_battery.c +++ b/drivers/power/supply/collie_battery.c @@ -440,6 +440,7 @@ static int collie_bat_probe(struct ucb1x00_dev *dev)
static void collie_bat_remove(struct ucb1x00_dev *dev) { + device_init_wakeup(&ucb->dev, 0); free_irq(gpiod_to_irq(collie_bat_main.gpio_full), &collie_bat_main); power_supply_unregister(collie_bat_bu.psy); power_supply_unregister(collie_bat_main.psy);