On Thu, Oct 26, 2017 at 03:54:05PM +0200, Jan Kara wrote: > On Wed 25-10-17 23:23:18, weiping zhang wrote: > > we should make sure parents directory exist, and then create dir or > > files under that. > > > > Signed-off-by: weiping zhang <zhangweiping@xxxxxxxxxxxxxxx> > > OK, this looks reasonable to me but instead of instead of just leaving > debugfs in half-initialized state, we should rather properly tear it down, > return error from bdi_debug_register() and handle it in > bdi_register_va()... > > At beginning I try to return error code to caller, then I find bdi_register_owner's return value was not checked in device_add_disk, so I think there may have some stories. But now, I found we must check bdi_register_owner in device_add_disk, otherwise bdi may lead some undefined behavior. blk_mq_debugfs_register also has same issue. I will send new patch series to fix these issues. -- weiping -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
