From: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
To: Qian Cai <cai@lca.pw>
Cc: vyasevich@gmail.com, nhorman@tuxdriver.com,
linux-sctp@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] net/sctp: fix GCC8+ -Wpacked-not-aligned warnings
Date: Sun, 28 Jul 2019 16:41:22 -0300 [thread overview]
Message-ID: <20190728194122.GN6204@localhost.localdomain> (raw)
In-Reply-To: <4A42819C-32E5-4869-8A3B-0D4F9DF1B53C@lca.pw>
On Sun, Jul 28, 2019 at 01:05:25PM -0400, Qian Cai wrote:
>
>
> > On Jul 26, 2019, at 5:30 PM, Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> wrote:
> >
> > On Fri, Jul 26, 2019 at 04:57:39PM -0400, Qian Cai wrote:
> >> There are a lot of those warnings with GCC8+ 64bit,
> >>
> >> In file included from ./include/linux/sctp.h:42,
> >> from net/core/skbuff.c:47:
> >> ./include/uapi/linux/sctp.h:395:1: warning: alignment 4 of 'struct
> >> sctp_paddr_change' is less than 8 [-Wpacked-not-aligned]
> >> } __attribute__((packed, aligned(4)));
> >> ^
> >> ./include/uapi/linux/sctp.h:728:1: warning: alignment 4 of 'struct
> >> sctp_setpeerprim' is less than 8 [-Wpacked-not-aligned]
> >> } __attribute__((packed, aligned(4)));
> >> ^
> >> ./include/uapi/linux/sctp.h:727:26: warning: 'sspp_addr' offset 4 in
> >> 'struct sctp_setpeerprim' isn't aligned to 8 [-Wpacked-not-aligned]
> >> struct sockaddr_storage sspp_addr;
> >> ^~~~~~~~~
> >> ./include/uapi/linux/sctp.h:741:1: warning: alignment 4 of 'struct
> >> sctp_prim' is less than 8 [-Wpacked-not-aligned]
> >> } __attribute__((packed, aligned(4)));
> >> ^
> >> ./include/uapi/linux/sctp.h:740:26: warning: 'ssp_addr' offset 4 in
> >> 'struct sctp_prim' isn't aligned to 8 [-Wpacked-not-aligned]
> >> struct sockaddr_storage ssp_addr;
> >> ^~~~~~~~
> >> ./include/uapi/linux/sctp.h:792:1: warning: alignment 4 of 'struct
> >> sctp_paddrparams' is less than 8 [-Wpacked-not-aligned]
> >> } __attribute__((packed, aligned(4)));
> >> ^
> >> ./include/uapi/linux/sctp.h:784:26: warning: 'spp_address' offset 4 in
> >> 'struct sctp_paddrparams' isn't aligned to 8 [-Wpacked-not-aligned]
> >> struct sockaddr_storage spp_address;
> >> ^~~~~~~~~~~
> >> ./include/uapi/linux/sctp.h:905:1: warning: alignment 4 of 'struct
> >> sctp_paddrinfo' is less than 8 [-Wpacked-not-aligned]
> >> } __attribute__((packed, aligned(4)));
> >> ^
> >> ./include/uapi/linux/sctp.h:899:26: warning: 'spinfo_address' offset 4
> >> in 'struct sctp_paddrinfo' isn't aligned to 8 [-Wpacked-not-aligned]
> >> struct sockaddr_storage spinfo_address;
> >> ^~~~~~~~~~~~~~
> >> Fix them by aligning the structures and fields to 8 bytes instead.
> >>
> >> Signed-off-by: Qian Cai <cai@lca.pw>
> >> ---
> >> include/uapi/linux/sctp.h | 18 +++++++++---------
> >> 1 file changed, 9 insertions(+), 9 deletions(-)
> >>
> >> diff --git a/include/uapi/linux/sctp.h b/include/uapi/linux/sctp.h
> >> index b8f2c4d56532..e7bd71cde882 100644
> >> --- a/include/uapi/linux/sctp.h
> >> +++ b/include/uapi/linux/sctp.h
> >
> > These changes gotta be more careful, if possible at all. As is, it's breaking UAPI.
>
> Could you please elaborate how this breaks userspace? I read through all the information
> I can find about UAPI and still have no clue. All I can see if that some field alignments changed
> which are expected, and it still take on 3 cachelines which should not hurt the performance.
For example on the struct I pointed below, if an application is
compiled using the old headers, sctp_paddrparams->spp_hbinterval is at
offset 132. But with this patch, it will change to 136. Then with a
kernel with this patch, it will look for the field at the wrong place.
>
> >
> > -before
> > +after patch
> >
> > struct sctp_paddrparams {
> > sctp_assoc_t spp_assoc_id; /* 0 4 */
> > - struct __kernel_sockaddr_storage spp_address __attribute__((__aligned__(1))); /* 4 128 */
> > - /* --- cacheline 2 boundary (128 bytes) was 4 bytes ago --- */
> > - __u32 spp_hbinterval; /* 132 4 */
> > - __u16 spp_pathmaxrxt; /* 136 2 */
> > - __u32 spp_pathmtu; /* 138 4 */
> > - __u32 spp_sackdelay; /* 142 4 */
> > - __u32 spp_flags; /* 146 4 */
> > - __u32 spp_ipv6_flowlabel; /* 150 4 */
> > - __u8 spp_dscp; /* 154 1 */
> >
> > - /* size: 156, cachelines: 3, members: 9 */
> > + /* XXX 4 bytes hole, try to pack */
> > +
> > + struct __kernel_sockaddr_storage spp_address __attribute__((__aligned__(8))); /* 8 128 */
> > + /* --- cacheline 2 boundary (128 bytes) was 8 bytes ago --- */
> > + __u32 spp_hbinterval; /* 136 4 */
> > + __u16 spp_pathmaxrxt; /* 140 2 */
> > + __u32 spp_pathmtu; /* 142 4 */
> > + __u32 spp_sackdelay; /* 146 4 */
> > + __u32 spp_flags; /* 150 4 */
> > + __u32 spp_ipv6_flowlabel; /* 154 4 */
> > + __u8 spp_dscp; /* 158 1 */
> > +
> > + /* size: 160, cachelines: 3, members: 9 */
> > + /* sum members: 155, holes: 1, sum holes: 4 */
> > /* padding: 1 */
> > - /* forced alignments: 1 */
> > - /* last cacheline: 28 bytes */
> > -} __attribute__((__aligned__(4)));
> > + /* forced alignments: 1, forced holes: 1, sum forced holes: 4 */
> > + /* last cacheline: 32 bytes */
> > +} __attribute__((__aligned__(8)));
> >
> >
> >> @@ -392,7 +392,7 @@ struct sctp_paddr_change {
> >> int spc_state;
> >> int spc_error;
> >> sctp_assoc_t spc_assoc_id;
> >> -} __attribute__((packed, aligned(4)));
> >> +} __attribute__((packed, aligned(8)));
> >>
> >> /*
> >> * spc_state: 32 bits (signed integer)
> >> @@ -724,8 +724,8 @@ struct sctp_assocparams {
> >> */
> >> struct sctp_setpeerprim {
> >> sctp_assoc_t sspp_assoc_id;
> >> - struct sockaddr_storage sspp_addr;
> >> -} __attribute__((packed, aligned(4)));
> >> + struct sockaddr_storage sspp_addr __attribute__((aligned(8)));
> >> +} __attribute__((packed, aligned(8)));
> >>
> >> /*
> >> * 7.1.10 Set Primary Address (SCTP_PRIMARY_ADDR)
> >> @@ -737,8 +737,8 @@ struct sctp_setpeerprim {
> >> */
> >> struct sctp_prim {
> >> sctp_assoc_t ssp_assoc_id;
> >> - struct sockaddr_storage ssp_addr;
> >> -} __attribute__((packed, aligned(4)));
> >> + struct sockaddr_storage ssp_addr __attribute__((aligned(8)));
> >> +} __attribute__((packed, aligned(8)));
> >>
> >> /* For backward compatibility use, define the old name too */
> >> #define sctp_setprim sctp_prim
> >> @@ -781,7 +781,7 @@ enum sctp_spp_flags {
> >>
> >> struct sctp_paddrparams {
> >> sctp_assoc_t spp_assoc_id;
> >> - struct sockaddr_storage spp_address;
> >> + struct sockaddr_storage spp_address __attribute__((aligned(8)));
> >> __u32 spp_hbinterval;
> >> __u16 spp_pathmaxrxt;
> >> __u32 spp_pathmtu;
> >> @@ -789,7 +789,7 @@ struct sctp_paddrparams {
> >> __u32 spp_flags;
> >> __u32 spp_ipv6_flowlabel;
> >> __u8 spp_dscp;
> >> -} __attribute__((packed, aligned(4)));
> >> +} __attribute__((packed, aligned(8)));
> >>
> >> /*
> >> * 7.1.18. Add a chunk that must be authenticated (SCTP_AUTH_CHUNK)
> >> @@ -896,13 +896,13 @@ struct sctp_stream_value {
> >> */
> >> struct sctp_paddrinfo {
> >> sctp_assoc_t spinfo_assoc_id;
> >> - struct sockaddr_storage spinfo_address;
> >> + struct sockaddr_storage spinfo_address __attribute__((aligned(8)));
> >> __s32 spinfo_state;
> >> __u32 spinfo_cwnd;
> >> __u32 spinfo_srtt;
> >> __u32 spinfo_rto;
> >> __u32 spinfo_mtu;
> >> -} __attribute__((packed, aligned(4)));
> >> +} __attribute__((packed, aligned(8)));
> >>
> >> /* Peer addresses's state. */
> >> /* UNKNOWN: Peer address passed by the upper layer in sendmsg or connect[x]
> >> --
> >> 1.8.3.1
>
next prev parent reply other threads:[~2019-07-28 19:41 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-07-26 20:57 [PATCH] net/sctp: fix GCC8+ -Wpacked-not-aligned warnings Qian Cai
2019-07-26 21:30 ` Marcelo Ricardo Leitner
2019-07-28 17:05 ` Qian Cai
2019-07-28 19:41 ` Marcelo Ricardo Leitner [this message]
2019-07-29 10:39 ` David Laight
2019-07-29 15:16 ` Qian Cai
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=20190728194122.GN6204@localhost.localdomain \
--to=marcelo.leitner@gmail.com \
--cc=cai@lca.pw \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-sctp@vger.kernel.org \
--cc=nhorman@tuxdriver.com \
--cc=vyasevich@gmail.com \
/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).