On 23/12/2020 10:36, Stefano Garzarella wrote: > On Wed, Dec 23, 2020 at 11:27:05AM +0800, Yejune Deng wrote: >> The function io_remove_personalities() is very similar to >> io_unregister_personality(),but the latter has a more reasonable >> return value. >> >> Signed-off-by: Yejune Deng <yejune.deng@xxxxxxxxx> >> --- >> fs/io_uring.c | 25 ++++++------------------- >> 1 file changed, 6 insertions(+), 19 deletions(-) > > The patch LGTM, maybe as an alternative you can leave io_remove_personality() with the interface needed by idr_for_each() and implement io_unregister_personality() calling io_remove_personality() with the right parameters. Right, don't replace sane types with void * just because. Leave well-typed io_unregister_personality() and call it from io_remove_personalities(). Also * idr_for_each() - Iterate through all stored pointers. ... * If @fn returns anything other than %0, the iteration stops and that * value is returned from this function. For io_remove_personality() iod==NULL should not happen because it's under for_each and synchronised, but leave the return value be io_remove_personality(void *, ...) { struct io_ring_ctx *ctx = data; io_unregister_personality(ctx, id); return 0; } -- Pavel Begunkov