Messages in this thread | | | From | tenthumbs@cybernex ... | Subject | Re: "Exception at ..." with 2.1.9 and 2.1.10 kernels | Date | Wed, 20 Nov 1996 01:20:02 GMT |
| |
On Mon, 18 Nov 1996 00:53:36 +0100 (MET), you wrote:
> > > Sorry, I'm still not buying that. Getname comes from linux/fs/namei.c. In > > the code, it says that it is copying from user-space to kernel-space for > > "efficiency." Exactly how is this a user problem? User code doesn't know > > about kernel space. > > But the kernel knows about userspace. > > > I still say something's goofy. > > Goofy lusercode. For example look into sys_open, the open(2) syscall and > look what would happen if you pass bad parameters, for example a NULL > pointer: the kernel passes this pointer to getname(). > > Or even easier, write a program that passes a NULL pointer > to open(2)/read(2)/write(2) etc. What happens? You get one of these > nice messages. > > Btw, some days ago someone posted a long and quite good article about > this topic which describes in detail how the new way to access user space > works. > > Ralf > You're right. I'm wrong. Sorry about the militancy.
| |