Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] accel/qda: non-zero remote_session_id in qda_fastrpc_invoke_pack() for dsp info - [Medium] accel/qda: destroyed mutex qdev->attr_lock while held in qda_deinit_device() --
commit 8c89a0764b44acd4840198a5f29101cbf6dc706a Author: Ekansh Gupta <[email protected]> accel/qda: Add DSP capability query This commit adds a QDA_QUERY_DSP_CAPABILITIES query type to DRM_IOCTL_QDA_QUERY to fetch DSP capability attributes through the DSP utilities handle and cache them on the device for subsequent queries. > diff --git a/drivers/accel/qda/qda_drv.c b/drivers/accel/qda/qda_drv.c > index 1e72dce769b9a..51d8ef10c6c1a 100644 > --- a/drivers/accel/qda/qda_drv.c > +++ b/drivers/accel/qda/qda_drv.c > @@ -154,6 +154,7 @@ struct qda_dev *qda_alloc_device(struct device *dev) > > INIT_LIST_HEAD(&qdev->cb_devs); > mutex_init(&qdev->import_lock); > + mutex_init(&qdev->attr_lock); > xa_init_flags(&qdev->ctx_xa, XA_FLAGS_ALLOC1); > atomic64_set(&qdev->ctx_seq, 0); > atomic_set(&qdev->session_id_seq, 0); > @@ -181,6 +182,7 @@ void qda_deinit_device(struct qda_dev *qdev) > qda_memory_manager_exit(qdev->iommu_mgr); > xa_destroy(&qdev->ctx_xa); > mutex_destroy(&qdev->import_lock); > + mutex_destroy(&qdev->attr_lock); > } [Severity: Medium] Can mutex_destroy(&qdev->attr_lock) in qda_deinit_device() run while qdev->attr_lock is still held or waited on by qda_get_dsp_capability()? In qda_get_dsp_capability(), qdev->attr_lock is held across qda_fastrpc_invoke(): drivers/accel/qda/qda_ioctl.c:qda_get_dsp_capability() { guard(mutex)(&qdev->attr_lock); if (!qdev->valid_attributes) { err = qda_fastrpc_invoke(QDA_FASTRPC_TYPE_GET_DSP_INFO, qdev->dsp_attributes, session); } Inside qda_fastrpc_invoke() -> qda_fastrpc_invoke_common(), qda_rpmsg_send_msg() calls drm_dev_exit() after rpmsg_send(), and then qda_rpmsg_wait_for_rsp() sleeps in wait_for_completion_interruptible(&ctx->work) while qdev->attr_lock remains held. If the remote processor crashes or the rpmsg device is removed before the DSP replies, qda_rpmsg_remove() runs: drivers/accel/qda/qda_rpmsg.c:qda_rpmsg_remove() { drm_dev_unplug(&qdev->drm_dev); qdev->rpdev = NULL; qda_fastrpc_flush_pending(qdev); qda_cb_unpopulate(qdev); qda_deinit_device(qdev); } Because the waiting thread already exited the drm_dev_enter() section, drm_dev_unplug() does not wait for it. qda_fastrpc_flush_pending() wakes the invocation with complete(&ctx->work) and qda_rpmsg_remove() proceeds directly to qda_deinit_device() -> mutex_destroy(&qdev->attr_lock) before the woken thread in qda_get_dsp_capability() has unlocked qdev->attr_lock. > diff --git a/drivers/accel/qda/qda_fastrpc.c b/drivers/accel/qda/qda_fastrpc.c > index e9688e3ba4048..6044cfad53e96 100644 > --- a/drivers/accel/qda/qda_fastrpc.c > +++ b/drivers/accel/qda/qda_fastrpc.c [ ... ] > @@ -980,6 +987,43 @@ static int qda_fastrpc_prepare_args_create_static(struct > qda_fastrpc_invoke_ctx > return 0; > } > > +/* > + * DSP capability query packed layout: > + * [0] count sizeof(u32), the capacity of the array below IN > + * [1] attrs (QDA_MAX_DSP_ATTRIBUTES - 1) * sizeof(u32) OUT > + * > + * Index 0 of the cached array is reserved, so only the rest is exchanged. > + */ > +static int qda_fastrpc_prepare_args_get_dsp_info(struct > qda_fastrpc_invoke_ctx *ctx) > +{ > + const u32 attr_count = QDA_MAX_DSP_ATTRIBUTES - 1; > + struct qda_gem_obj *kernel_gem; > + int err; > + > + const size_t off_attrs = sizeof(u32); > + const size_t total_size = off_attrs + attr_count * sizeof(u32); > + > + ctx->sc = QDA_FASTRPC_SCALARS(0, 1, 1); > + ctx->handle = QDA_FASTRPC_DSP_UTILITIES_HANDLE; [Severity: High] When ctx->handle is set to QDA_FASTRPC_DSP_UTILITIES_HANDLE here, should qda_fastrpc_invoke_pack() also set msg->fastrpc.remote_session_id to 0? Currently, qda_fastrpc_invoke_pack() only checks for QDA_FASTRPC_INIT_HANDLE: drivers/accel/qda/qda_fastrpc.c:qda_fastrpc_invoke_pack() { /* The session has no process yet to identify while one is being created */ if (ctx->handle == QDA_FASTRPC_INIT_HANDLE) msg->fastrpc.remote_session_id = 0; else msg->fastrpc.remote_session_id = ctx->session->remote_session_id; } When userspace issues DRM_IOCTL_QDA_QUERY with QDA_QUERY_DSP_CAPABILITIES before creating a DSP process on the session, qda_fastrpc_invoke_pack() will pack the session's non-zero remote_session_id into the RPMsg wire header for a stateless utilities call that has no remote process context (whereas drivers/misc/fastrpc.c passes kernel = true in fastrpc_get_info_from_dsp() so fastrpc_invoke_send() sets msg->client_id = 0). -- Sashiko AI review ยท https://sashiko.dev/#/patchset/[email protected]?part=23
