Hi, Le lundi 01 juin 2026 à 15:32 +0800, Icenowy Zheng a écrit : > Single-core variants of this hardware do not use the work at all, and > the worker function is set to NULL, which leads to warnings when > cancelling the work in release callback. > > Skip the work init/cancel code when the JPEG hardware isn't multi-core. Thanks for sending this patch, thought this specific issue has alrady been fixed upstream, see commit b1845a227fda37b2fe5327df3ca0015d7e290235 Nicolas > > Cc: stable@vger.kernel.org > Fixes: 34c519feef3e ("media: mtk-jpeg: fix use-after-free in release path due > to uncancelled work") > Fixes: d40e95274925 ("media: mtk-jpeg: reconstructs the initialization mode of > worker") > Signed-off-by: Icenowy Zheng > --- >  drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c | 6 ++++-- >  1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c > b/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c > index 8c684756d5fc2..83e54a7ef49c0 100644 > --- a/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c > +++ b/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c > @@ -1160,7 +1160,8 @@ static int mtk_jpeg_open(struct file *file) >   goto free; >   } >   > - INIT_WORK(&ctx->jpeg_work, jpeg->variant->jpeg_worker); > + if (jpeg->variant->multi_core) > + INIT_WORK(&ctx->jpeg_work, jpeg->variant->jpeg_worker); >   INIT_LIST_HEAD(&ctx->dst_done_queue); >   spin_lock_init(&ctx->done_queue_lock); >   v4l2_fh_init(&ctx->fh, vfd); > @@ -1202,7 +1203,8 @@ static int mtk_jpeg_release(struct file *file) >   struct mtk_jpeg_dev *jpeg = video_drvdata(file); >   struct mtk_jpeg_ctx *ctx = mtk_jpeg_file_to_ctx(file); >   > - cancel_work_sync(&ctx->jpeg_work); > + if (jpeg->variant->multi_core) > + cancel_work_sync(&ctx->jpeg_work); >   mutex_lock(&jpeg->lock); >   v4l2_m2m_ctx_release(ctx->fh.m2m_ctx); >   v4l2_ctrl_handler_free(&ctx->ctrl_hdl);