2019-05-27 00:54:55

by Gen Zhang

Subject: [PATCH] dm-region-hash: fix a missing-check bug in __rh_alloc()

In function __rh_alloc(), the pointer nreg is allocated a memory space
via kmalloc(). And it is used in the following codes. However, when
there is a memory allocation error, kmalloc() fails. Thus null pointer
dereference may happen. And it will cause the kernel to crash. Therefore,
we should check the return value and handle the error.
Further, in __rh_find(), we should also check the return value and
handle the error.

Signed-off-by: Gen Zhang <[email protected]>
---
diff --git a/drivers/md/dm-region-hash.c b/drivers/md/dm-region-hash.c
index 1f76045..2fa1641 100644
--- a/drivers/md/dm-region-hash.c
+++ b/drivers/md/dm-region-hash.c
@@ -290,8 +290,11 @@ static struct dm_region *__rh_alloc(struct dm_region_hash *rh, region_t region)
struct dm_region *reg, *nreg;

nreg = mempool_alloc(&rh->region_pool, GFP_ATOMIC);
- if (unlikely(!nreg))
+ if (unlikely(!nreg)) {
nreg = kmalloc(sizeof(*nreg), GFP_NOIO | __GFP_NOFAIL);
+ if (!nreg)
+ return NULL;
+ }

nreg->state = rh->log->type->in_sync(rh->log, region, 1) ?
DM_RH_CLEAN : DM_RH_NOSYNC;
@@ -329,6 +332,8 @@ static struct dm_region *__rh_find(struct dm_region_hash *rh, region_t region)
if (!reg) {
read_unlock(&rh->hash_lock);
reg = __rh_alloc(rh, region);
+ if (!reg)
+ return NULL;
read_lock(&rh->hash_lock);
}

---

2019-05-27 01:52:56

by Mike Snitzer

Subject: Re: dm-region-hash: fix a missing-check bug in __rh_alloc()

On Sun, May 26 2019 at 8:50pm -0400,
Gen Zhang <[email protected]> wrote:

> In function __rh_alloc(), the pointer nreg is allocated a memory space
> via kmalloc(). And it is used in the following codes. However, when
> there is a memory allocation error, kmalloc() fails. Thus null pointer
> dereference may happen. And it will cause the kernel to crash. Therefore,
> we should check the return value and handle the error.
> Further, in __rh_find(), we should also check the return value and
> handle the error.
>
> Signed-off-by: Gen Zhang <[email protected]>
> ---
> diff --git a/drivers/md/dm-region-hash.c b/drivers/md/dm-region-hash.c
> index 1f76045..2fa1641 100644
> --- a/drivers/md/dm-region-hash.c
> +++ b/drivers/md/dm-region-hash.c
> @@ -290,8 +290,11 @@ static struct dm_region *__rh_alloc(struct dm_region_hash *rh, region_t region)
> struct dm_region *reg, *nreg;
>
> nreg = mempool_alloc(&rh->region_pool, GFP_ATOMIC);
> - if (unlikely(!nreg))
> + if (unlikely(!nreg)) {
> nreg = kmalloc(sizeof(*nreg), GFP_NOIO | __GFP_NOFAIL);
> + if (!nreg)
> + return NULL;
> + }
>
> nreg->state = rh->log->type->in_sync(rh->log, region, 1) ?
> DM_RH_CLEAN : DM_RH_NOSYNC;

This patch isn't needed. __GFP_NOFAIL means the allocation won't fail.

And there are many other users of __GFP_NOFAIL that don't check for
failure.

Mike

2019-05-27 03:14:31

by Gen Zhang

Subject: Re: dm-region-hash: fix a missing-check bug in __rh_alloc()

On Sun, May 26, 2019 at 09:49:14PM -0400, Mike Snitzer wrote:
> On Sun, May 26 2019 at 8:50pm -0400,
> Gen Zhang <[email protected]> wrote:
>
> > In function __rh_alloc(), the pointer nreg is allocated a memory space
> > via kmalloc(). And it is used in the following codes. However, when
> > there is a memory allocation error, kmalloc() fails. Thus null pointer
> > dereference may happen. And it will cause the kernel to crash. Therefore,
> > we should check the return value and handle the error.
> > Further, in __rh_find(), we should also check the return value and
> > handle the error.
> >
> > Signed-off-by: Gen Zhang <[email protected]>
> > ---
> > diff --git a/drivers/md/dm-region-hash.c b/drivers/md/dm-region-hash.c
> > index 1f76045..2fa1641 100644
> > --- a/drivers/md/dm-region-hash.c
> > +++ b/drivers/md/dm-region-hash.c
> > @@ -290,8 +290,11 @@ static struct dm_region *__rh_alloc(struct dm_region_hash *rh, region_t region)
> > struct dm_region *reg, *nreg;
> >
> > nreg = mempool_alloc(&rh->region_pool, GFP_ATOMIC);
> > - if (unlikely(!nreg))
> > + if (unlikely(!nreg)) {
> > nreg = kmalloc(sizeof(*nreg), GFP_NOIO | __GFP_NOFAIL);
> > + if (!nreg)
> > + return NULL;
> > + }
> >
> > nreg->state = rh->log->type->in_sync(rh->log, region, 1) ?
> > DM_RH_CLEAN : DM_RH_NOSYNC;
>
> This patch isn't needed. __GFP_NOFAIL means the allocation won't fail.
>
> And there are many other users of __GFP_NOFAIL that don't check for
> failure.
>
> Mike
Appreciate your reply, Mike.

Thanks
Gen