On 12/07/2022 14:55, Yang Yingliang wrote: > This reverts commit 59ebbe40fb51e307032ae7f63b2749fad2d4635a. > > If devm_add_action() fails in devm_add_action_or_reset(), > devm_spi_unregister() will be called, it decreases the > refcount of 'ctlr->dev' to 0, then it will cause uaf in > the drivers that calling spi_put_controller() in error path. Whether a revert is the right fix or not, this is the same conclusion I came to reading your patch for my driver & on that basis: Reviewed-by: Conor Dooley <conor.dooley@xxxxxxxxxxxxx> Seems like the master variant of this is used over 40 times: rg "(?s)devm_spi_register_master.*master_put" drivers/spi --multiline -l > > Fixes: 59ebbe40fb51 ("spi: simplify devm_spi_register_controller") > Signed-off-by: Yang Yingliang <yangyingliang@xxxxxxxxxx> > --- > drivers/spi/spi.c | 19 ++++++++++++++----- > 1 file changed, 14 insertions(+), 5 deletions(-) > > diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c > index dc1a324e3271..ef751ccd65be 100644 > --- a/drivers/spi/spi.c > +++ b/drivers/spi/spi.c > @@ -3130,9 +3130,9 @@ int spi_register_controller(struct spi_controller *ctlr) > } > EXPORT_SYMBOL_GPL(spi_register_controller); > > -static void devm_spi_unregister(void *ctlr) > +static void devm_spi_unregister(struct device *dev, void *res) > { > - spi_unregister_controller(ctlr); > + spi_unregister_controller(*(struct spi_controller **)res); > } > > /** > @@ -3151,13 +3151,22 @@ static void devm_spi_unregister(void *ctlr) > int devm_spi_register_controller(struct device *dev, > struct spi_controller *ctlr) > { > + struct spi_controller **ptr; > int ret; > > + ptr = devres_alloc(devm_spi_unregister, sizeof(*ptr), GFP_KERNEL); > + if (!ptr) > + return -ENOMEM; > + > ret = spi_register_controller(ctlr); > - if (ret) > - return ret; > + if (!ret) { > + *ptr = ctlr; > + devres_add(dev, ptr); > + } else { > + devres_free(ptr); > + } > > - return devm_add_action_or_reset(dev, devm_spi_unregister, ctlr); > + return ret; > } > EXPORT_SYMBOL_GPL(devm_spi_register_controller); >