Messages in this thread | | | Date | Mon, 27 Nov 2000 11:30:48 -0800 | From | "H. Peter Anvin" <> | Subject | Re: How to transfer memory from PCI memory directly to user space safely and portable? |
| |
Philipp Rumpf wrote: > > On Mon, Nov 27, 2000 at 10:36:34AM -0800, H. Peter Anvin wrote: > > Followup to: <20001126151120.V2272@parcelfarce.linux.theplanet.co.uk> > > By author: Philipp Rumpf <prumpf@parcelfarce.linux.theplanet.co.uk> > > In newsgroup: linux.dev.kernel > > > > > > I hope count isn't provided by userspace here ? > > > > > > > 1. What happens if the user space memory is swapped to disk? Will > > > > verify_area() make sure that the memory is in physical RAM when it returns, > > > > or will it return -EFAULT, or will something even worse happen? > > > > > > On i386, you'll sleep implicitly waiting for the page fault to be handled; in > > > the generic case, anything could happen. > > > > > > > That doesn't sound right. I would expect it to wait for the page to > > be brought in on any and all architectures, otherwise it seems rather > > impossible to write portable Linux kernel code. > > The code in question was > memcpy_fromio(user_space_dst, iobase, count); > > Assuming user_space_dst is a userspace pointer, what I said is true; on some > architectures we will be dereferencing random pointers in kernel space, and > we won't get -EFAULT right on any architecture. > > Or did I miss something ? >
Yes, the post you responded to was talking about verify_area() [which is, admittedly, obsolete.]
-hpa
-- <hpa@transmeta.com> at work, <hpa@zytor.com> in private! "Unix gives you enough rope to shoot yourself in the foot." http://www.zytor.com/~hpa/puzzle.txt - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
| |