From: Leon Romanovsky <leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
To: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Cc: Leon Romanovsky <leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>,
RDMA mailing list
<linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
Alaa Hleihel <alaa-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>,
Matan Barak <matanb-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>,
Noa Osherovich <noaos-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Subject: [PATCH rdma-rc 03/15] IB/uverbs: Always the attribute size provided by the user
Date: Tue, 13 Feb 2018 12:18:29 +0200 [thread overview]
Message-ID: <20180213101841.20101-4-leon@kernel.org> (raw)
In-Reply-To: <20180213101841.20101-1-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
From: Matan Barak <matanb-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
This fixes several bugs around the copy_to/from user path:
- copy_to used the user provided size of the attribute
and could copy data beyond the end of the kernel buffer into
userspace.
- copy_from didn't know the size of the kernel buffer and
could have left kernel memory unexpectedly un-initialized.
- copy_from did not use the user length to determine if the
attribute data is inlined or not.
Signed-off-by: Matan Barak <matanb-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Leon Romanovsky <leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
drivers/infiniband/core/uverbs_std_types.c | 5 +++--
include/rdma/uverbs_ioctl.h | 35 ++++++++++++++++++++++++------
2 files changed, 31 insertions(+), 9 deletions(-)
diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c
index cab0ac3556eb..c6502c7b7c46 100644
--- a/drivers/infiniband/core/uverbs_std_types.c
+++ b/drivers/infiniband/core/uverbs_std_types.c
@@ -323,7 +323,8 @@ static int uverbs_create_cq_handler(struct ib_device *ib_dev,
cq->res.type = RDMA_RESTRACK_CQ;
rdma_restrack_add(&cq->res);
- ret = uverbs_copy_to(attrs, CREATE_CQ_RESP_CQE, &cq->cqe);
+ ret = uverbs_copy_to(attrs, CREATE_CQ_RESP_CQE, &cq->cqe,
+ sizeof(cq->cqe));
if (ret)
goto err_cq;
@@ -375,7 +376,7 @@ static int uverbs_destroy_cq_handler(struct ib_device *ib_dev,
resp.comp_events_reported = obj->comp_events_reported;
resp.async_events_reported = obj->async_events_reported;
- return uverbs_copy_to(attrs, DESTROY_CQ_RESP, &resp);
+ return uverbs_copy_to(attrs, DESTROY_CQ_RESP, &resp, sizeof(resp));
}
static DECLARE_UVERBS_METHOD(
diff --git a/include/rdma/uverbs_ioctl.h b/include/rdma/uverbs_ioctl.h
index 6da44079aa58..32cb14703914 100644
--- a/include/rdma/uverbs_ioctl.h
+++ b/include/rdma/uverbs_ioctl.h
@@ -351,29 +351,50 @@ static inline const struct uverbs_attr *uverbs_attr_get(const struct uverbs_attr
}
static inline int uverbs_copy_to(const struct uverbs_attr_bundle *attrs_bundle,
- size_t idx, const void *from)
+ size_t idx, const void *from, size_t size)
{
const struct uverbs_attr *attr = uverbs_attr_get(attrs_bundle, idx);
u16 flags;
+ size_t min_size;
if (IS_ERR(attr))
return PTR_ERR(attr);
+ min_size = min_t(size_t, attr->ptr_attr.len, size);
+ if (copy_to_user(attr->ptr_attr.ptr, from, min_size))
+ return -EFAULT;
+
flags = attr->ptr_attr.flags | UVERBS_ATTR_F_VALID_OUTPUT;
- return (!copy_to_user(attr->ptr_attr.ptr, from, attr->ptr_attr.len) &&
- !put_user(flags, &attr->uattr->flags)) ? 0 : -EFAULT;
+ if (put_user(flags, &attr->uattr->flags))
+ return -EFAULT;
+
+ return 0;
}
-static inline int _uverbs_copy_from(void *to, size_t to_size,
+static inline bool uverbs_attr_ptr_is_inline(const struct uverbs_attr *attr)
+{
+ return attr->ptr_attr.len <= sizeof(attr->ptr_attr.data);
+}
+
+static inline int _uverbs_copy_from(void *to,
const struct uverbs_attr_bundle *attrs_bundle,
- size_t idx)
+ size_t idx,
+ size_t size)
{
const struct uverbs_attr *attr = uverbs_attr_get(attrs_bundle, idx);
if (IS_ERR(attr))
return PTR_ERR(attr);
- if (to_size <= sizeof(((struct ib_uverbs_attr *)0)->data))
+ /*
+ * Validation ensures attr->ptr_attr.len >= size. If the caller is
+ * using UVERBS_ATTR_SPEC_F_MIN_SZ then it must call copy_from with
+ * the right size.
+ */
+ if (unlikely(size < attr->ptr_attr.len))
+ return -EINVAL;
+
+ if (uverbs_attr_ptr_is_inline(attr))
memcpy(to, &attr->ptr_attr.data, attr->ptr_attr.len);
else if (copy_from_user(to, attr->ptr_attr.ptr, attr->ptr_attr.len))
return -EFAULT;
@@ -382,7 +403,7 @@ static inline int _uverbs_copy_from(void *to, size_t to_size,
}
#define uverbs_copy_from(to, attrs_bundle, idx) \
- _uverbs_copy_from(to, sizeof(*(to)), attrs_bundle, idx)
+ _uverbs_copy_from(to, attrs_bundle, idx, sizeof(*to))
/* =================================================
* Definitions -> Specs infrastructure
--
2.16.1
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2018-02-13 10:18 UTC|newest]
Thread overview: 35+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-02-13 10:18 [PATCH rdma-rc 00/15] RDMA fixes for v4.16 Leon Romanovsky
[not found] ` <20180213101841.20101-1-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 10:18 ` [PATCH rdma-rc 01/15] IB/ipoib: Do not warn if IPoIB debugfs doesn't exist Leon Romanovsky
[not found] ` <20180213101841.20101-2-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 16:06 ` Dennis Dalessandro
[not found] ` <a95eace1-2e3e-e97a-cbaa-ca58771e5cff-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2018-02-13 18:12 ` Leon Romanovsky
[not found] ` <20180213181205.GV2197-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2018-02-13 18:45 ` Dennis Dalessandro
2018-02-13 10:18 ` [PATCH rdma-rc 02/15] RDMA/restrack: Remove unimplemented XRCD object Leon Romanovsky
[not found] ` <20180213101841.20101-3-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 17:00 ` Dennis Dalessandro
[not found] ` <7e464caf-6875-9232-be9a-31324b03323f-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2018-02-13 18:09 ` Leon Romanovsky
[not found] ` <20180213180956.GU2197-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2018-02-13 18:16 ` Dennis Dalessandro
2018-02-13 10:18 ` Leon Romanovsky [this message]
2018-02-13 10:18 ` [PATCH rdma-rc 04/15] IB/uverbs: Use inline data transfer for UHW_IN Leon Romanovsky
2018-02-13 10:18 ` [PATCH rdma-rc 05/15] IB/uverbs: Use u64_to_user_ptr() not a union Leon Romanovsky
[not found] ` <20180213101841.20101-6-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 16:10 ` Dennis Dalessandro
2018-02-13 10:18 ` [PATCH rdma-rc 06/15] IB/uverbs: Fix method merging in uverbs_ioctl_merge Leon Romanovsky
2018-02-13 10:18 ` [PATCH rdma-rc 07/15] IB/uverbs: Use __aligned_u64 for uapi headers Leon Romanovsky
[not found] ` <20180213101841.20101-8-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 16:12 ` Dennis Dalessandro
2018-02-13 10:18 ` [PATCH rdma-rc 08/15] IB/uverbs: Add ioctl support for 32bit processes Leon Romanovsky
[not found] ` <20180213101841.20101-9-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 16:56 ` Dennis Dalessandro
[not found] ` <f77173f4-703b-b5ec-06ad-24263805251d-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2018-02-13 17:16 ` Jason Gunthorpe
[not found] ` <20180213171632.GI4499-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2018-02-14 11:31 ` Dennis Dalessandro
2018-02-13 10:18 ` [PATCH rdma-rc 09/15] IB/uverbs: Fix possible oops with duplicate ioctl attributes Leon Romanovsky
2018-02-13 10:18 ` [PATCH rdma-rc 10/15] IB/uverbs: Hold the uobj write lock after allocate Leon Romanovsky
2018-02-13 10:18 ` [PATCH rdma-rc 11/15] RDMA/uverbs: Protect from races between lookup and destroy of uobjects Leon Romanovsky
2018-02-13 10:18 ` [PATCH rdma-rc 12/15] IB/uverbs: Tidy lockdep_check Leon Romanovsky
[not found] ` <20180213101841.20101-13-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 17:10 ` Dennis Dalessandro
[not found] ` <1120b3ff-8cb4-d661-60b1-e1f7656840fd-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2018-02-13 17:23 ` Jason Gunthorpe
2018-02-13 10:18 ` [PATCH rdma-rc 13/15] IB/uverbs: Tidy uverbs_uobject_add Leon Romanovsky
[not found] ` <20180213101841.20101-14-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 17:09 ` Dennis Dalessandro
[not found] ` <c43bec28-0437-961e-fe65-55886973b6da-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2018-02-13 17:20 ` Jason Gunthorpe
[not found] ` <20180213172041.GJ4499-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2018-02-13 18:44 ` Dennis Dalessandro
2018-02-13 10:18 ` [PATCH rdma-rc 14/15] IB/uverbs: Fix unbalanced unlock on error path for rdma_explicit_destroy Leon Romanovsky
[not found] ` <20180213101841.20101-15-leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
2018-02-13 17:11 ` Dennis Dalessandro
2018-02-13 10:18 ` [PATCH rdma-rc 15/15] RDMA/uverbs: Protect from command mask overflow Leon Romanovsky
2018-02-15 22:26 ` [PATCH rdma-rc 00/15] RDMA fixes for v4.16 Jason Gunthorpe
2018-02-15 22:30 ` Jason Gunthorpe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180213101841.20101-4-leon@kernel.org \
--to=leon-dgejt+ai2ygdnm+yrofe0a@public.gmane.org \
--cc=alaa-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org \
--cc=dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
--cc=jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=matanb-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org \
--cc=noaos-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).