[lkml]   [2019]   [Feb]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH] x86: uaccess: fix regression in unsafe_get_user

> On Feb 16, 2019, at 2:50 PM, Andy Lutomirski <> wrote:
>>> On Feb 16, 2019, at 12:18 PM, Thomas Gleixner <> wrote:
>>>> On Sat, 16 Feb 2019, Jann Horn wrote:
>>>> On Sat, Feb 16, 2019 at 12:59 AM <> wrote:
>>>> When extracting an initramfs, a filename may be near an allocation boundary.
>>>> Should that happen, strncopy_from_user will invoke unsafe_get_user which
>>>> may cross the allocation boundary. Should that happen, unsafe_get_user will
>>>> trigger a page fault, and strncopy_from_user would then bailout to
>>>> byte_at_a_time behavior.
>>>> unsafe_get_user is unsafe by nature, and rely on pagefault to detect boundaries.
>>>> After 9da3f2b74054 ("x86/fault: BUG() when uaccess helpers fault on kernel addresses")
>>>> it may no longer rely on pagefault as the new page fault handler would
>>>> trigger a BUG().
>>>> This commit allows unsafe_get_user to explicitly trigger pagefaults and
>>>> handle them directly with the error target label.
>>> Oof. So basically the init code is full of things that just call
>>> syscalls instead of using VFS functions (which don't actually exist
>>> for everything), and the VFS syscalls use getname_flags(), which uses
>>> strncpy_from_user(), which can access out-of-bounds pages on
>>> architectures that set CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS, and
>>> that in summary means that all the init code is potentially prone to
>>> tripping over this?
>> Not all init code. It should be only the initramfs decompression.
>>> I don't particularly like this approach to fixing it, but I also don't
>>> have any better ideas, so I guess unless someone else has a bright
>>> idea, this patch might have to go in.
>> So we know that this happens in the context of decompress() which calls
>> flush_buffer() for every chunk. flush_buffer() gets the start_address and
>> the length. We also know that the fault can only happen within:
>> start_address <= fault_address < start_address + length + 8;
>> So something like the untested workaround below should cover the initramfs
>> oddity and avoid to weaken the protection for all other cases.
> What is the actual problem? We’re not actually demand-faulting this data, are we? Are we just overrunning the buffer because the from_user helpers are too clever? Can we fix it for real by having the fancy helpers do *aligned* loads so that they don’t overrun the buffer? Heck, this might be faster, too.

Indeed. I would argue that the current code is a bug even in the normal case. If I lay out my user address space so that I have f,o,o,o,\0 at the end of a page and I have non-side-effect-free memory after it (MMIO, userfaultfd, etc), then passing a pointer to that “fooo” string to a syscall should *not* overrun the buffer.

If I have some time this evening, I’ll see if I can whip up a credible fix.
 \ /
  Last update: 2019-02-16 23:57    [W:0.074 / U:3.064 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site