2021-06-02 02:04:51

by Jarmo Tiitto

[permalink] [raw]
Subject: [PATCH 1/1] pgo: Fix allocate_node() handling of non-vmlinux nodes.

Currently allocate_node() will reserve nodes even if *p
doesn't point into __llvm_prf_data_start - __llvm_prf_data_end
range.

Fix it by checking if p points into vmlinux range
and otherwise return NULL.

Signed-off-by: Jarmo Tiitto <[email protected]>
---
kernel/pgo/instrument.c | 4 ++++
1 file changed, 4 insertions(+)

diff --git a/kernel/pgo/instrument.c b/kernel/pgo/instrument.c
index 0e07ee1b17d9..9bca535dfa91 100644
--- a/kernel/pgo/instrument.c
+++ b/kernel/pgo/instrument.c
@@ -55,6 +55,10 @@ void prf_unlock(unsigned long flags)
static struct llvm_prf_value_node *allocate_node(struct llvm_prf_data *p,
u32 index, u64 value)
{
+ /* check if p points into vmlinux. If not, don't allocate. */
+ if (p < __llvm_prf_data_start || p >= __llvm_prf_data_end)
+ return NULL;
+
if (&__llvm_prf_vnds_start[current_node + 1] >= __llvm_prf_vnds_end)
return NULL; /* Out of nodes */

--
2.31.1


2021-06-02 17:45:30

by Kees Cook

[permalink] [raw]
Subject: Re: [PATCH 1/1] pgo: Fix allocate_node() handling of non-vmlinux nodes.

On Wed, Jun 02, 2021 at 03:57:04AM +0300, Jarmo Tiitto wrote:
> Currently allocate_node() will reserve nodes even if *p
> doesn't point into __llvm_prf_data_start - __llvm_prf_data_end
> range.
>
> Fix it by checking if p points into vmlinux range
> and otherwise return NULL.
>
> Signed-off-by: Jarmo Tiitto <[email protected]>
> ---
> kernel/pgo/instrument.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/kernel/pgo/instrument.c b/kernel/pgo/instrument.c
> index 0e07ee1b17d9..9bca535dfa91 100644
> --- a/kernel/pgo/instrument.c
> +++ b/kernel/pgo/instrument.c
> @@ -55,6 +55,10 @@ void prf_unlock(unsigned long flags)
> static struct llvm_prf_value_node *allocate_node(struct llvm_prf_data *p,
> u32 index, u64 value)
> {
> + /* check if p points into vmlinux. If not, don't allocate. */
> + if (p < __llvm_prf_data_start || p >= __llvm_prf_data_end)
> + return NULL;

This should be a tighter check (struct llvm_prf_data has size, so just
checking for p < __llvm_prf_data_end isn't sufficient. I recommend using
the memory_contains() helper.

And I think this should be louder as it's entirely unexpected right
now. Perhaps:

if (WARN_ON_ONCE(!memory_contains(__llvm_prf_data_start,
__llvm_prf_data_end,
p, sizeof(*p))))
return NULL;

> +
> if (&__llvm_prf_vnds_start[current_node + 1] >= __llvm_prf_vnds_end)
> return NULL; /* Out of nodes */
>
> --
> 2.31.1
>

--
Kees Cook

2021-06-02 18:57:03

by Jarmo Tiitto

[permalink] [raw]
Subject: Re: [PATCH 1/1] pgo: Fix allocate_node() handling of non-vmlinux nodes.

Kees Cook wrote keskiviikkona 2. kesäkuuta 2021 20.41.28 EEST:
> On Wed, Jun 02, 2021 at 03:57:04AM +0300, Jarmo Tiitto wrote:
> > Currently allocate_node() will reserve nodes even if *p
> > doesn't point into __llvm_prf_data_start - __llvm_prf_data_end
> > range.
> >
> > Fix it by checking if p points into vmlinux range
> > and otherwise return NULL.
> >
> > Signed-off-by: Jarmo Tiitto <[email protected]>
> > ---
> > kernel/pgo/instrument.c | 4 ++++
> > 1 file changed, 4 insertions(+)
> >
> > diff --git a/kernel/pgo/instrument.c b/kernel/pgo/instrument.c
> > index 0e07ee1b17d9..9bca535dfa91 100644
> > --- a/kernel/pgo/instrument.c
> > +++ b/kernel/pgo/instrument.c
> > @@ -55,6 +55,10 @@ void prf_unlock(unsigned long flags)
> > static struct llvm_prf_value_node *allocate_node(struct llvm_prf_data *p,
> > u32 index, u64 value)
> > {
> > + /* check if p points into vmlinux. If not, don't allocate. */
> > + if (p < __llvm_prf_data_start || p >= __llvm_prf_data_end)
> > + return NULL;
>
> This should be a tighter check (struct llvm_prf_data has size, so just
> checking for p < __llvm_prf_data_end isn't sufficient. I recommend using
> the memory_contains() helper.
>
> And I think this should be louder as it's entirely unexpected right
> now. Perhaps:
>
> if (WARN_ON_ONCE(!memory_contains(__llvm_prf_data_start,
> __llvm_prf_data_end,
> p, sizeof(*p))))
> return NULL;
>
> > +
> > if (&__llvm_prf_vnds_start[current_node + 1] >= __llvm_prf_vnds_end)
> > return NULL; /* Out of nodes */
> >
> > --
> > 2.31.1
> >
>
> --
> Kees Cook
>


Well, if you do that the WARN_ON_ONCE() will always trigger, unless CONFIG_MODULES=n ????

This is because 'struct llvm_prf_data *p' argument is expected
(at least I think so) to point into __llvm_prf_data section in vmlinux
and also into each module's own __llvm_prf_data section.

So in the end the compiler supplied pointer is likely correct,
but the current v9 PGO patch attempts to reserve all vnodes
from vmlinux __llvm_prf_vnds section, instead of respective module section.

I think it would be normal to ignore pointers here,
until my module PGO machinery is ready.

But I agree on using memory_contains() for checking if p is within bounds.
I will follow on with v2 of this patch. :-)

-Jarmo