From mboxrd@z Thu Jan 1 00:00:00 1970 From: Yann Droneaud Subject: Re: [PATCHv1 4/4] RDMA/cxgb4: add missing padding at end of struct c4iw_alloc_ucontext_resp Date: Thu, 22 May 2014 08:21:05 +0200 Message-ID: <1400739665.21049.5.camel@localhost.localdomain> References: Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Roland Dreier Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Dan Carpenter , Steve Wise List-Id: linux-rdma@vger.kernel.org Hi Roland, Le lundi 05 mai 2014 =C3=A0 19:35 +0200, Yann Droneaud a =C3=A9crit : > i386 ABI disagree with most other ABIs regarding alignment > of data type larger than 4 bytes: on most ABIs a padding must > be added at end of the structures, while it is not required on > i386. >=20 > So for most ABI struct c4iw_alloc_ucontext_resp get implicitly padded > to be aligned on a 8 bytes multiple, while for i386, such padding > is not added. >=20 > Tool pahole could be used to find such implicit padding: >=20 > $ pahole --anon_include \ > --nested_anon_include \ > --recursive \ > --class_name c4iw_alloc_ucontext_resp \ > drivers/infiniband/hw/cxgb4/iw_cxgb4.o >=20 > Then, structure layout can be compared between i386 and x86_64: >=20 > +++ obj-i386/drivers/infiniband/hw/cxgb4/iw_cxgb4.o.pahole.txt 20= 14-03-28 11:43:05.547432195 +0100 > --- obj-x86_64/drivers/infiniband/hw/cxgb4/iw_cxgb4.o.pahole.txt 20= 14-03-28 10:55:10.990133017 +0100 > @@ -2,9 +2,8 @@ struct c4iw_alloc_ucontext_resp { > __u64 status_page_key; /* 0 = 8 */ > __u32 status_page_size; /* 8 = 4 */ >=20 > - /* size: 12, cachelines: 1, members: 2 */ > - /* last cacheline: 12 bytes */ > + /* size: 16, cachelines: 1, members: 2 */ > + /* padding: 4 */ > + /* last cacheline: 16 bytes */ > }; >=20 > This ABI disagreement will make an x86_64 kernel try to write > past the buffer provided by an i386 binary. >=20 > When boundary check will be implemented, the x86_64 kernel will > refuse to write past the i386 userspace provided buffer > and the uverbs will fail. >=20 > If the structure lay in memory on a page boundary and next page > is not mapped, ib_copy_to_udata() will fail and the uverb > will fail. >=20 > Additionally, as reported by Dan Carpenter, without the implicit > padding being properly cleared, an information leak would take > place in most architectures. >=20 > This patch adds an explicit padding to struct c4iw_alloc_ucontext_res= p, > and, like 92b0ca7cb149 ('IB/mlx5: Fix stack info leak in > mlx5_ib_alloc_ucontext()'), makes function c4iw_alloc_ucontext() > not writting this padding field to userspace. This way, x86_64 kernel > will be able to write struct c4iw_alloc_ucontext_resp as expected by > unpatched and patched i386 libcxgb4. >=20 > Link: http://marc.info/?i=3Dcover.1399309513.git.ydroneaud-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org > Link: http://marc.info/?i=3D1395848977.3297.15.camel-bi+AKbBUZKY6gyzm1THtWQ@public.gmane.org= omain > Link: http://marc.info/?i=3D20140328082428.GH25192@mwanda > Fixes: 05eb23893c2c ('cxgb4/iw_cxgb4: Doorbell Drop Avoidance Bug Fix= es') > Reported-by: Yann Droneaud > Reported-by: Dan Carpenter > Signed-off-by: Yann Droneaud I believe this one should go in v3.15-rc7 as it fixes an issue introduced in v3.15-rc1. See=20 http://marc.info/?i=3D20140328082428.GH25192@mwanda http://marc.info/?i=3D20140502235616.GJ4963@mwanda The other patchs could probably wait for v3.16-rc1 for integration in linux-stable. Regards. --=20 Yann Droneaud OPTEYA -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" i= n the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html