2019-05-24 03:14:58

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);
}

---

2019-06-05 06:06:51

by Jiri Slaby

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

On 24. 05. 19, 5:12, Gen Zhang 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;

What's the purpose of checking NO_FAIL allocations?

thanks,
--
js
suse labs

2019-06-05 12:24:02

by Mike Snitzer

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

On Wed, Jun 05 2019 at 2:05am -0400,
Jiri Slaby <[email protected]> wrote:

> On 24. 05. 19, 5:12, Gen Zhang 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;
>
> What's the purpose of checking NO_FAIL allocations?

There isn't, that was already pointed out in a different thread for this
same patch (think patch was posted twice):
https://www.redhat.com/archives/dm-devel/2019-May/msg00124.html

Mike

2019-06-05 15:01:47

by Gen Zhang

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

On Wed, Jun 05, 2019 at 08:21:59AM -0400, Mike Snitzer wrote:
> On Wed, Jun 05 2019 at 2:05am -0400,
> Jiri Slaby <[email protected]> wrote:
>
> > On 24. 05. 19, 5:12, Gen Zhang 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;
> >
> > What's the purpose of checking NO_FAIL allocations?
>
> There isn't, that was already pointed out in a different thread for this
> same patch (think patch was posted twice):
> https://www.redhat.com/archives/dm-devel/2019-May/msg00124.html
>
> Mike
Thanks for your reply. The first thread is not replied for a period, so
the second one is posted. But I don't know why Jiri replied to the first
thread.

Thanks
Gen