From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dan Carpenter Date: Sat, 12 Jan 2019 10:51:05 +0000 Subject: Re: [PATCH] powerpc/powernv/npu: Allocate enough memory in pnv_try_setup_npu_table_group() Message-Id: <20190112105105.GP1718@kadam> List-Id: References: <20190109102328.GB5476@kadam> <20190112003035.GA10427@350D> <20190112054426.GF1743@kadam> <20190112073441.GJ14180@gate.crashing.org> In-Reply-To: <20190112073441.GJ14180@gate.crashing.org> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Segher Boessenkool Cc: Alexey Kardashevskiy , Alistair Popple , Mark Hairgrove , kernel-janitors@vger.kernel.org, Paul Mackerras , linuxppc-dev@lists.ozlabs.org On Sat, Jan 12, 2019 at 01:34:42AM -0600, Segher Boessenkool wrote: > On Sat, Jan 12, 2019 at 08:44:26AM +0300, Dan Carpenter wrote: > > On Sat, Jan 12, 2019 at 11:30:35AM +1100, Balbir Singh wrote: > > > On Wed, Jan 09, 2019 at 01:23:29PM +0300, Dan Carpenter wrote: > > > > There is a typo so we accidentally allocate enough memory for a pointer > > > > when we wanted to allocate enough for a struct. > > > > > > > > Fixes: 0bd971676e68 ("powerpc/powernv/npu: Add compound IOMMU groups") > > > > Signed-off-by: Dan Carpenter > > > > --- > > > > arch/powerpc/platforms/powernv/npu-dma.c | 2 +- > > > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > > > > > diff --git a/arch/powerpc/platforms/powernv/npu-dma.c b/arch/powerpc/platforms/powernv/npu-dma.c > > > > index d7f742ed48ba..3f58c7dbd581 100644 > > > > --- a/arch/powerpc/platforms/powernv/npu-dma.c > > > > +++ b/arch/powerpc/platforms/powernv/npu-dma.c > > > > @@ -564,7 +564,7 @@ struct iommu_table_group *pnv_try_setup_npu_table_group(struct pnv_ioda_pe *pe) > > > > } > > > > } else { > > > > /* Create a group for 1 GPU and attached NPUs for POWER8 */ > > > > - pe->npucomp = kzalloc(sizeof(pe->npucomp), GFP_KERNEL); > > > > + pe->npucomp = kzalloc(sizeof(*pe->npucomp), GFP_KERNEL); > > > > > > To avoid these in the future, I wonder if instead of sizeof(pe->npucomp), we insist on > > > sizeof structure > > > > > > pe->npucomp = kzalloc(sizeof(struct npucomp), GFP_KERNEL); > > > > > > > The latest kernel fashion is sizeof(*ptr). It can go wrong either way. > > I don't have strong feelings about it. These sorts of bugs don't last > > long because they're caught in testing or with static analysis. > > And it is easy to see someone forgot the * in "sizeof *ptr", and with > experience it will just automatically look wrong if it is forgotten; but > it isn't obvious at all if the wrong struct is used, which cannot happen > with the *ptr form, but happens frequently with the "sizeof(struct x)" > form. It doesn't happen very frequently. I look for this with Smatch. The results I see are when the first few members of a struct are a header and the actual size can vary. hdr = alloc(sizeof(struct larger_struct)); regards, dan carpenter