From: Bartosz Golaszewski <[email protected]>
When allocating an array of elements, users should check for
multiplication overflow or preferably use one of the provided helpers
like: devm_kmalloc_array().
This provides devm_krealloc_array() for users who want to reallocate
managed arrays.
Signed-off-by: Bartosz Golaszewski <[email protected]>
---
include/linux/device.h | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/include/linux/device.h b/include/linux/device.h
index 5ed101be7b2e..e77203faea55 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -226,6 +226,17 @@ static inline void *devm_kmalloc_array(struct device *dev,
return devm_kmalloc(dev, bytes, flags);
}
+static inline void *
+devm_krealloc_array(struct device *dev, void *ptr, size_t new_n,
+ size_t new_size, gfp_t gfp)
+{
+ size_t bytes;
+
+ if (unlikely(check_mul_overflow(new_n, new_size, &bytes)))
+ return NULL;
+
+ return devm_krealloc(dev, ptr, bytes, gfp);
+}
static inline void *devm_kcalloc(struct device *dev,
size_t n, size_t size, gfp_t flags)
{
--
2.29.1
On Mon, 2 Nov 2020 15:22:25 +0100
Bartosz Golaszewski <[email protected]> wrote:
> From: Bartosz Golaszewski <[email protected]>
>
> When allocating an array of elements, users should check for
> multiplication overflow or preferably use one of the provided helpers
> like: devm_kmalloc_array().
>
> This provides devm_krealloc_array() for users who want to reallocate
> managed arrays.
>
> Signed-off-by: Bartosz Golaszewski <[email protected]>
+CC Greg KH.
As this is going into a very generic place I'd like a relevant ack.
That file is a bit of a wild west for acks, but Greg seems most
appropriate person.
So Greg, any comments on this one?
Thanks,
Jonathan
> ---
> include/linux/device.h | 11 +++++++++++
> 1 file changed, 11 insertions(+)
>
> diff --git a/include/linux/device.h b/include/linux/device.h
> index 5ed101be7b2e..e77203faea55 100644
> --- a/include/linux/device.h
> +++ b/include/linux/device.h
> @@ -226,6 +226,17 @@ static inline void *devm_kmalloc_array(struct device *dev,
>
> return devm_kmalloc(dev, bytes, flags);
> }
> +static inline void *
> +devm_krealloc_array(struct device *dev, void *ptr, size_t new_n,
> + size_t new_size, gfp_t gfp)
> +{
> + size_t bytes;
> +
> + if (unlikely(check_mul_overflow(new_n, new_size, &bytes)))
> + return NULL;
> +
> + return devm_krealloc(dev, ptr, bytes, gfp);
> +}
> static inline void *devm_kcalloc(struct device *dev,
> size_t n, size_t size, gfp_t flags)
> {
On Sat, Nov 14, 2020 at 03:46:41PM +0000, Jonathan Cameron wrote:
> On Mon, 2 Nov 2020 15:22:25 +0100
> Bartosz Golaszewski <[email protected]> wrote:
>
> > From: Bartosz Golaszewski <[email protected]>
> >
> > When allocating an array of elements, users should check for
> > multiplication overflow or preferably use one of the provided helpers
> > like: devm_kmalloc_array().
> >
> > This provides devm_krealloc_array() for users who want to reallocate
> > managed arrays.
> >
> > Signed-off-by: Bartosz Golaszewski <[email protected]>
>
> +CC Greg KH.
>
> As this is going into a very generic place I'd like a relevant ack.
> That file is a bit of a wild west for acks, but Greg seems most
> appropriate person.
>
> So Greg, any comments on this one?
As there is only 1 user of this function in the patch series, you don't
save any extra code space here, I don't think this is worth it.
We are seeing less and less gains from these new devm_* additions, and
only more confusion and problems with them. So perhaps don't add this?
I don't think it is needed.
thanks,
greg k-h
On Sat, Nov 14, 2020 at 5:16 PM Greg KH <[email protected]> wrote:
>
> On Sat, Nov 14, 2020 at 03:46:41PM +0000, Jonathan Cameron wrote:
> > On Mon, 2 Nov 2020 15:22:25 +0100
> > Bartosz Golaszewski <[email protected]> wrote:
> >
> > > From: Bartosz Golaszewski <[email protected]>
> > >
> > > When allocating an array of elements, users should check for
> > > multiplication overflow or preferably use one of the provided helpers
> > > like: devm_kmalloc_array().
> > >
> > > This provides devm_krealloc_array() for users who want to reallocate
> > > managed arrays.
> > >
> > > Signed-off-by: Bartosz Golaszewski <[email protected]>
> >
> > +CC Greg KH.
> >
> > As this is going into a very generic place I'd like a relevant ack.
> > That file is a bit of a wild west for acks, but Greg seems most
> > appropriate person.
> >
> > So Greg, any comments on this one?
>
> As there is only 1 user of this function in the patch series, you don't
> save any extra code space here, I don't think this is worth it.
>
It's worth it in that the overflow check before allocation doesn't
seem to belong in a driver IMO but is a general check that should live
in common code.
> We are seeing less and less gains from these new devm_* additions, and
> only more confusion and problems with them. So perhaps don't add this?
> I don't think it is needed.
>
I think you're referring to the discussion on
devm_platform_ioremap_resource()? I would argue that consolidation of
common operations in helpers is rarely a bad thing but it's a
discussion for another thread.
I'm not too attached to this patch - if you think this should be
dropped then fine, but I don't see how the name devm_krealloc_array()
can confuse anyone.
Bartosz
On Mon, Nov 16, 2020 at 11:18 AM Bartosz Golaszewski
<[email protected]> wrote:
>
> On Sat, Nov 14, 2020 at 5:16 PM Greg KH <[email protected]> wrote:
> >
> > On Sat, Nov 14, 2020 at 03:46:41PM +0000, Jonathan Cameron wrote:
> > > On Mon, 2 Nov 2020 15:22:25 +0100
> > > Bartosz Golaszewski <[email protected]> wrote:
> > >
> > > > From: Bartosz Golaszewski <[email protected]>
> > > >
> > > > When allocating an array of elements, users should check for
> > > > multiplication overflow or preferably use one of the provided helpers
> > > > like: devm_kmalloc_array().
> > > >
> > > > This provides devm_krealloc_array() for users who want to reallocate
> > > > managed arrays.
> > > >
> > > > Signed-off-by: Bartosz Golaszewski <[email protected]>
> > >
> > > +CC Greg KH.
> > >
> > > As this is going into a very generic place I'd like a relevant ack.
> > > That file is a bit of a wild west for acks, but Greg seems most
> > > appropriate person.
> > >
> > > So Greg, any comments on this one?
> >
> > As there is only 1 user of this function in the patch series, you don't
> > save any extra code space here, I don't think this is worth it.
> >
>
> It's worth it in that the overflow check before allocation doesn't
> seem to belong in a driver IMO but is a general check that should live
> in common code.
>
> > We are seeing less and less gains from these new devm_* additions, and
> > only more confusion and problems with them. So perhaps don't add this?
> > I don't think it is needed.
> >
>
> I think you're referring to the discussion on
> devm_platform_ioremap_resource()? I would argue that consolidation of
> common operations in helpers is rarely a bad thing but it's a
> discussion for another thread.
>
> I'm not too attached to this patch - if you think this should be
> dropped then fine, but I don't see how the name devm_krealloc_array()
> can confuse anyone.
>
Greg: what's the final call on this?
Bartosz
On Mon, 23 Nov 2020 12:38:26 +0100
Bartosz Golaszewski <[email protected]> wrote:
> On Mon, Nov 16, 2020 at 11:18 AM Bartosz Golaszewski
> <[email protected]> wrote:
> >
> > On Sat, Nov 14, 2020 at 5:16 PM Greg KH <[email protected]> wrote:
> > >
> > > On Sat, Nov 14, 2020 at 03:46:41PM +0000, Jonathan Cameron wrote:
> > > > On Mon, 2 Nov 2020 15:22:25 +0100
> > > > Bartosz Golaszewski <[email protected]> wrote:
> > > >
> > > > > From: Bartosz Golaszewski <[email protected]>
> > > > >
> > > > > When allocating an array of elements, users should check for
> > > > > multiplication overflow or preferably use one of the provided helpers
> > > > > like: devm_kmalloc_array().
> > > > >
> > > > > This provides devm_krealloc_array() for users who want to reallocate
> > > > > managed arrays.
> > > > >
> > > > > Signed-off-by: Bartosz Golaszewski <[email protected]>
> > > >
> > > > +CC Greg KH.
> > > >
> > > > As this is going into a very generic place I'd like a relevant ack.
> > > > That file is a bit of a wild west for acks, but Greg seems most
> > > > appropriate person.
> > > >
> > > > So Greg, any comments on this one?
> > >
> > > As there is only 1 user of this function in the patch series, you don't
> > > save any extra code space here, I don't think this is worth it.
> > >
> >
> > It's worth it in that the overflow check before allocation doesn't
> > seem to belong in a driver IMO but is a general check that should live
> > in common code.
> >
> > > We are seeing less and less gains from these new devm_* additions, and
> > > only more confusion and problems with them. So perhaps don't add this?
> > > I don't think it is needed.
> > >
> >
> > I think you're referring to the discussion on
> > devm_platform_ioremap_resource()? I would argue that consolidation of
> > common operations in helpers is rarely a bad thing but it's a
> > discussion for another thread.
> >
> > I'm not too attached to this patch - if you think this should be
> > dropped then fine, but I don't see how the name devm_krealloc_array()
> > can confuse anyone.
> >
>
> Greg: what's the final call on this?
Reroll the series without this patch. If it turns out to be a good idea
in the long run we can always bring it back, but for now it's blocking
the rest of the series.
Thanks,
Jonathan
>
> Bartosz