mt_probe() allocates td->haptic with devm_kzalloc(), tying its lifetime to the HID device's driver unbind. hid_haptic_init() then stores the pointer in ff->private and installs hid_haptic_destroy() as the force-feedback destroy callback. When the HID device is removed while a process still holds an evdev fd, devres frees td->haptic at unbind time. hid_haptic_destroy() runs later from input_dev_release() and dereferences freed memory. The input core then calls kfree(ff->private) on the same pointer, double-freeing it. The existing get_device()/put_device() pair in init/destroy pins the struct hid_device but does not keep its devres allocations alive, since devres runs at driver unbind, not at the final device kref put. Replace devm_kzalloc() with plain kzalloc() so the haptic struct survives driver unbind. The input core's input_ff_destroy() already calls kfree(ff->private) after the destroy callback, so hid_haptic_destroy() must not free the struct itself -- it only needs to tear down the sub-allocations it owns. On the non-haptic path in mt_probe(), replace devm_kfree() with kfree(). On the error paths before hid_hw_start(), free the struct explicitly since ownership has not yet transferred to the input core. On hid_hw_start() failure, only free the struct when the haptic subsystem was not initialized (td->is_haptic_touchpad is false); otherwise the input core's teardown already freed it. Fixes: 8d0bf7908b5a ("HID: multitouch: add haptic multitouch support") Cc: stable@vger.kernel.org Signed-off-by: Aldo Ariel Panzardo --- v3: drop the kfree(haptic) added to hid_haptic_destroy() in v1/v2 -- input_ff_destroy() already calls kfree(ff->private) after the destroy callback, so the explicit kfree was a guaranteed double free (found by Sashiko AI review). Also handle the hid_hw_start() failure path: free td->haptic only when the haptic subsystem was not initialized (!td->is_haptic_touchpad), since otherwise the input core's teardown already freed it. v2: do not free td->haptic on hid_hw_start() failure (found by Sashiko AI review). v1: https://lore.kernel.org/linux-input/20261009031207.3233206-1-qwe.aldo@gmail.com/ drivers/hid/hid-multitouch.c | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) diff --git a/drivers/hid/hid-multitouch.c b/drivers/hid/hid-multitouch.c index 4e19a0c4d..9cbe61832 100644 --- a/drivers/hid/hid-multitouch.c +++ b/drivers/hid/hid-multitouch.c @@ -2132,7 +2132,7 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id) dev_err(&hdev->dev, "cannot allocate multitouch data\n"); return -ENOMEM; } - td->haptic = devm_kzalloc(&hdev->dev, sizeof(*(td->haptic)), GFP_KERNEL); + td->haptic = kzalloc(sizeof(*(td->haptic)), GFP_KERNEL); if (!td->haptic) return -ENOMEM; @@ -2181,12 +2181,14 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id) ret = hid_parse(hdev); if (ret != 0) - return ret; + goto err_free_haptic; if (mtclass->name == MT_CLS_APPLE_TOUCHBAR && !hid_find_field(hdev, HID_INPUT_REPORT, - HID_DG_TOUCHPAD, HID_DG_TRANSDUCER_INDEX)) - return -ENODEV; + HID_DG_TOUCHPAD, HID_DG_TRANSDUCER_INDEX)) { + ret = -ENODEV; + goto err_free_haptic; + } if (mtclass->quirks & MT_QUIRK_FIX_CONST_CONTACT_ID) mt_fix_const_fields(hdev, HID_DG_CONTACTID); @@ -2195,8 +2197,11 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id) hdev->quirks |= HID_QUIRK_NOGET; ret = hid_hw_start(hdev, HID_CONNECT_DEFAULT); - if (ret) + if (ret) { + if (!td->is_haptic_touchpad) + kfree(td->haptic); return ret; + } ret = sysfs_create_group(&hdev->dev.kobj, &mt_attribute_group); if (ret) @@ -2206,9 +2211,13 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id) mt_set_modes(hdev, HID_LATENCY_NORMAL, TOUCHPAD_REPORT_ALL); if (!td->is_haptic_touchpad) - devm_kfree(&hdev->dev, td->haptic); + kfree(td->haptic); return 0; + +err_free_haptic: + kfree(td->haptic); + return ret; } static int mt_suspend(struct hid_device *hdev, pm_message_t state) -- 2.43.0