linux-kernel - RE: [PATCH rdma-next 1/1] RDMA/mana_ib: Fix integer overflow during queue creation

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID:
 <PA1PR83MB0662C52043DE3FA1F2EFBCA0B4D12@PA1PR83MB0662.EURPRD83.prod.outlook.com>
Date: Tue, 11 Mar 2025 10:05:47 +0000
From: Konstantin Taranov <kotaranov@...rosoft.com>
To: Long Li <longli@...rosoft.com>, Konstantin Taranov
	<kotaranov@...ux.microsoft.com>, Shiraz Saleem <shirazsaleem@...rosoft.com>,
	"jgg@...pe.ca" <jgg@...pe.ca>, "leon@...nel.org" <leon@...nel.org>
CC: "linux-rdma@...r.kernel.org" <linux-rdma@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH rdma-next 1/1] RDMA/mana_ib: Fix integer overflow during
 queue creation

> > Subject: [PATCH rdma-next 1/1] RDMA/mana_ib: Fix integer overflow
> > during queue creation
> >
> > From: Konstantin Taranov <kotaranov@...rosoft.com>
> >
> > Use size_t instead of u32 in helpers for queue creations to detect
> > overflow of queue size. The queue size cannot exceed size of u32.
> >
> > Fixes: bd4ee700870a ("RDMA/mana_ib: UD/GSI QP creation for kernel")
> > Signed-off-by: Konstantin Taranov <kotaranov@...rosoft.com>
> > ---
> >  drivers/infiniband/hw/mana/cq.c      |  9 +++++----
> >  drivers/infiniband/hw/mana/main.c    | 15 +++++++++++++--
> >  drivers/infiniband/hw/mana/mana_ib.h |  4 ++--
> >  drivers/infiniband/hw/mana/qp.c      | 11 ++++++-----
> >  4 files changed, 26 insertions(+), 13 deletions(-)
> >
> > diff --git a/drivers/infiniband/hw/mana/cq.c
> > b/drivers/infiniband/hw/mana/cq.c index 5c325ef..07b97da 100644
> > --- a/drivers/infiniband/hw/mana/cq.c
> > +++ b/drivers/infiniband/hw/mana/cq.c
> > @@ -18,7 +18,7 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const
> > struct ib_cq_init_attr *attr,
> >  	struct gdma_context *gc;
> >  	bool is_rnic_cq;
> >  	u32 doorbell;
> > -	u32 buf_size;
> > +	size_t buf_size;
> >  	int err;
> >
> >  	mdev = container_of(ibdev, struct mana_ib_dev, ib_dev); @@ -45,7
> > +45,8 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct
> > +ib_cq_init_attr
> > *attr,
> >  		}
> >
> >  		cq->cqe = attr->cqe;
> > -		err = mana_ib_create_queue(mdev, ucmd.buf_addr, cq->cqe
> *
> > COMP_ENTRY_SIZE,
> > +		buf_size = (size_t)cq->cqe * COMP_ENTRY_SIZE;
> > +		err = mana_ib_create_queue(mdev, ucmd.buf_addr, buf_size,
> >  					   &cq->queue);
> >  		if (err) {
> >  			ibdev_dbg(ibdev, "Failed to create queue for create
> cq, %d\n",
> > err); @@ -57,8 +58,8 @@ int mana_ib_create_cq(struct ib_cq *ibcq,
> > const struct ib_cq_init_attr *attr,
> >  		doorbell = mana_ucontext->doorbell;
> >  	} else {
> >  		is_rnic_cq = true;
> > -		buf_size = MANA_PAGE_ALIGN(roundup_pow_of_two(attr-
> >cqe
> > * COMP_ENTRY_SIZE));
> > -		cq->cqe = buf_size / COMP_ENTRY_SIZE;
> > +		cq->cqe = attr->cqe;
> > +		buf_size =
> > MANA_PAGE_ALIGN(roundup_pow_of_two((size_t)attr->cqe *
> > +COMP_ENTRY_SIZE));
> 
> Why not do a check like:
> If (attr->cqe > U32_MAX/COMP_ENTRY_SIZE)
> 	return -EINVAL;
> 
> And you don’t need to check them in mana_ib_create_kernel_queue() and
> mana_ib_create_queue().
> 

Yes, I was initially thinking about the small fix as you proposed and then ended up
adding checks to all paths. As I see the same can happen if a user asks for a large WQ of RC.
I believe a kernel client can also cause this overflow. We plan to add kernel RC soon and,
as far as I understand, a kernel user can also ask to create a large CQ resulting in similar overflow.

- Konstantin