On Mon, May 14, 2018 at 08:01:32PM +0000, Kalderon, Michal wrote: > From: linux-rdma-owner@xxxxxxxxxxxxxxx <linux-rdma-owner@xxxxxxxxxxxxxxx> on behalf of Jason Gunthorpe <jgg@xxxxxxxxxxxx> > > > > > > 1 file changed, 28 insertions(+), 31 deletions(-) > > > > > > > > Please don't leak kernel pointers [1], and use %pK instead of %x [2]. > > > > > > > > [1] https://lwn.net/Articles/735589/ > > > > [2] Documentation/core-api/printk-formats.rst > > > > > > > > Thanks > > > > Thanks for pointing these out. I was not familiar with this > > > convention. However, the addresses Which are printed below are the > > > physical doorbell addresses which don't > > > I think PCI physical addresses are expected to be protected as well. > > Not sure what can be done though with this physical address, as it is > an address into the devices doorbell -> any attempt to write data that is > not in the doorbell format will be dropped by device. It obscures information about the kernel memory layout which is the point of all of this. If you want to report the doorbell address for debugging then report it as an offset from the start of the device's bar. > It's important for debugging to see the actual doorbell > address. (and it seems %pK gives a hashed value and not actual value > like %lx) I can send a v2 to change the vma addresses to be printed > with %pk although they are printed in other driver (mlx) with %lx as > well. I'm sure Leon will fix mlx drivers if he is made aware of where they are doing this.. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
