Messages in this thread | | | Date | Fri, 20 Mar 2009 19:32:10 +0100 | From | Ingo Molnar <> | Subject | Re: [Patch 01/11] Introducing generic hardware breakpoint handler interfaces |
| |
* Alan Stern <stern@rowland.harvard.edu> wrote:
> > + /* Check that the virtual address is in the proper range */ > > + if (tsk) { > > + if (!arch_check_va_in_userspace(bp->info.address, tsk)) > > + return -EFAULT; > > + } else { > > + if (!arch_check_va_in_kernelspace(bp->info.address)) > > + return -EFAULT; > > + } > > Roland pointed out that these checks need to take into account the > length of the breakpoint. For example, in > arch_check_va_in_userspace() it isn't sufficient for the start of > the breakpoint region to be a userspace address; the end of the > breakpoint region must also be in userspace.
i pointed it out - but yes, this needs to be fixed.
Ingo
| |