Re: [PATCH 0/2] mm: make copy_to_kernel_nofault() not fault on user addresses

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 02.09.24 08:31, Omar Sandoval wrote:
On Mon, Sep 02, 2024 at 08:19:33AM +0200, Christophe Leroy wrote:


Le 02/09/2024 à 07:31, Omar Sandoval a écrit :
[Vous ne recevez pas souvent de courriers de osandov@xxxxxxxxxxx. Découvrez pourquoi ceci est important à https://aka.ms/LearnAboutSenderIdentification ]

From: Omar Sandoval <osandov@xxxxxx>

Hi,

I hit a case where copy_to_kernel_nofault() will fault (lol): if the
destination address is in userspace and x86 Supervisor Mode Access
Prevention is enabled. Patch 2 has the details and the fix. Patch 1
renames a helper function so that its use in patch 2 makes more sense.
If the rename is too intrusive, I can drop it.

The name of the function is "copy_to_kernel". If the destination is a user
address, it is not a copy to kernel but a copy to user and you already have
the function copy_to_user() for that. copy_to_user() properly handles SMAP.

I'm not trying to copy to user. I am (well, KDB is) trying to copy to an
arbitrary address, and I want it to return an error instead of crashing
if the address is not a valid kernel address. As far as I can tell, that
is the whole point of copy_to_kernel_nofault().

The thing is that you (well, KDB) triggers something that would be considered a real BUG when triggered from "ordinary" (non-debugging) code.

But now I am confused: "if the destination address is in userspace" does not really make sense in the context of KDB, no?

  [15]kdb> mm 0 1234
[ 94.652476] BUG: kernel NULL pointer dereference, address: 0000000000000000

Why is address 0 in "user space"? "Which" user space?

Isn't the problem here that KDB lets you blindly write to any non-existing memory address?


Likely it should do some proper filtering like we do in fs/proc/kcore.c:

Take a look at the KCORE_RAM case where we make sure the page exists, is online and may be accessed. Only then, we trigger a copy_from_kernel_nofault(). Note that the KCORE_USER is a corner case only for some special thingies on x86 (vsyscall), and can be ignored for our case here.

--
Cheers,

David / dhildenb





[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux