| Message ID | cf020b6a04b3a9d7f08750927b1d100f63ff4689.1638445455.git.mchehab+huawei@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | media: ipu3: don't use recursion at the Kernel | expand |
Hi Mauro On 02/12/2021 11:44, Mauro Carvalho Chehab wrote: > The Kernel stack is too small. Doing recursions there is a very > bad idea, as, if something gets wrong, it could lead to data > corruption. TIL - I'll bear that in mind in the future, thanks. > So, re-implement cio2_check_fwnode_graph() to avoid > recursion. > > Compile-tested only. > > Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org> FWIW: Reviewed-by: Daniel Scally <djrscally@gmail.com> Tested-by: Daniel Scally <djrscally@gmail.com> > --- > drivers/media/pci/intel/ipu3/ipu3-cio2-main.c | 17 ++++++++--------- > 1 file changed, 8 insertions(+), 9 deletions(-) > > diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c > index 356ea966cf8d..8e4f250a8b56 100644 > --- a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c > +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c > @@ -1691,16 +1691,15 @@ static int cio2_check_fwnode_graph(struct fwnode_handle *fwnode) > { > struct fwnode_handle *endpoint; > > - if (IS_ERR_OR_NULL(fwnode)) > - return -EINVAL; > - > - endpoint = fwnode_graph_get_next_endpoint(fwnode, NULL); > - if (endpoint) { > - fwnode_handle_put(endpoint); > - return 0; > + while (!IS_ERR_OR_NULL(fwnode)) { > + endpoint = fwnode_graph_get_next_endpoint(fwnode, NULL); > + if (endpoint) { > + fwnode_handle_put(endpoint); > + return 0; > + } > + fwnode = fwnode->secondary; > } > - > - return cio2_check_fwnode_graph(fwnode->secondary); > + return -EINVAL; > } > > /**************** PCI interface ****************/
Hi Mauro, On Thu, Dec 02, 2021 at 12:44:26PM +0100, Mauro Carvalho Chehab wrote: > The Kernel stack is too small. Doing recursions there is a very > bad idea, as, if something gets wrong, it could lead to data > corruption. So, re-implement cio2_check_fwnode_graph() to avoid > recursion. Any decent compiler should be able to optimise out tail recursion. But is checking the secondary pointer even needed these days? fwnode_graph_get_next_endpoint() does that already, it's not something drivers should be required to do.
diff --git a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c index 356ea966cf8d..8e4f250a8b56 100644 --- a/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c +++ b/drivers/media/pci/intel/ipu3/ipu3-cio2-main.c @@ -1691,16 +1691,15 @@ static int cio2_check_fwnode_graph(struct fwnode_handle *fwnode) { struct fwnode_handle *endpoint; - if (IS_ERR_OR_NULL(fwnode)) - return -EINVAL; - - endpoint = fwnode_graph_get_next_endpoint(fwnode, NULL); - if (endpoint) { - fwnode_handle_put(endpoint); - return 0; + while (!IS_ERR_OR_NULL(fwnode)) { + endpoint = fwnode_graph_get_next_endpoint(fwnode, NULL); + if (endpoint) { + fwnode_handle_put(endpoint); + return 0; + } + fwnode = fwnode->secondary; } - - return cio2_check_fwnode_graph(fwnode->secondary); + return -EINVAL; } /**************** PCI interface ****************/
The Kernel stack is too small. Doing recursions there is a very bad idea, as, if something gets wrong, it could lead to data corruption. So, re-implement cio2_check_fwnode_graph() to avoid recursion. Compile-tested only. Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org> --- drivers/media/pci/intel/ipu3/ipu3-cio2-main.c | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-)