| Message ID | 1389077339-12814-3-git-send-email-fan.du@windriver.com |
|---|---|
| State | Rejected, archived |
| Delegated to: | David Miller |
| Headers | show |
On Tue, 2014-01-07 at 14:48 +0800, Fan Du wrote: > Otherwise 64bits kernel has sizeof(struct xfrm_usersa_info) 224 bytes, > while 32bits compiled iproute2 see the same structure as 220 bytes, which > leading deficit xfrm sa, in turn broken IPsec connectivity. > > Fix this by packing the structure. > > Signed-off-by: Fan Du <fan.du@windriver.com> > --- > include/uapi/linux/xfrm.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/include/uapi/linux/xfrm.h b/include/uapi/linux/xfrm.h > index 470bfae..61460c4 100644 > --- a/include/uapi/linux/xfrm.h > +++ b/include/uapi/linux/xfrm.h > @@ -366,7 +366,7 @@ struct xfrm_usersa_info { > #define XFRM_STATE_AF_UNSPEC 32 > #define XFRM_STATE_ALIGN4 64 > #define XFRM_STATE_ESN 128 > -}; > +} __attribute__((packed)); > > #define XFRM_SA_XFLAG_DONT_ENCAP_DSCP 1 > That change will make access to the structure very slow on some architectures, and I suspect it will cause other compatibility problems. I think the right thing to do is to reduce the minimum length of the structure in the netlink policy so that padding at the end is not required. (It looks like all field offsets will be the same on all 32/64-bit architecture pairs and there is only a differing amount of padding at the end of the structure for 32/64-bit alignment.) Ben.
On 2014年01月09日 04:33, Ben Hutchings wrote: > On Tue, 2014-01-07 at 14:48 +0800, Fan Du wrote: >> Otherwise 64bits kernel has sizeof(struct xfrm_usersa_info) 224 bytes, >> while 32bits compiled iproute2 see the same structure as 220 bytes, which >> leading deficit xfrm sa, in turn broken IPsec connectivity. >> >> Fix this by packing the structure. >> >> Signed-off-by: Fan Du<fan.du@windriver.com> >> --- >> include/uapi/linux/xfrm.h | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/include/uapi/linux/xfrm.h b/include/uapi/linux/xfrm.h >> index 470bfae..61460c4 100644 >> --- a/include/uapi/linux/xfrm.h >> +++ b/include/uapi/linux/xfrm.h >> @@ -366,7 +366,7 @@ struct xfrm_usersa_info { >> #define XFRM_STATE_AF_UNSPEC 32 >> #define XFRM_STATE_ALIGN4 64 >> #define XFRM_STATE_ESN 128 >> -}; >> +} __attribute__((packed)); >> >> #define XFRM_SA_XFLAG_DONT_ENCAP_DSCP 1 >> > > That change will make access to the structure very slow on some > architectures, and I suspect it will cause other compatibility problems. > > I think the right thing to do is to reduce the minimum length of the > structure in the netlink policy so that padding at the end is not > required. Could you please be more specific about this? Thanks. I'm afraid we can only rearrange structure member order to reduce size on 64bits, alas that's not feasible here :( (It looks like all field offsets will be the same on all > 32/64-bit architecture pairs and there is only a differing amount of > padding at the end of the structure for 32/64-bit alignment.) > > Ben. >
diff --git a/include/uapi/linux/xfrm.h b/include/uapi/linux/xfrm.h index 470bfae..61460c4 100644 --- a/include/uapi/linux/xfrm.h +++ b/include/uapi/linux/xfrm.h @@ -366,7 +366,7 @@ struct xfrm_usersa_info { #define XFRM_STATE_AF_UNSPEC 32 #define XFRM_STATE_ALIGN4 64 #define XFRM_STATE_ESN 128 -}; +} __attribute__((packed)); #define XFRM_SA_XFLAG_DONT_ENCAP_DSCP 1
Otherwise 64bits kernel has sizeof(struct xfrm_usersa_info) 224 bytes, while 32bits compiled iproute2 see the same structure as 220 bytes, which leading deficit xfrm sa, in turn broken IPsec connectivity. Fix this by packing the structure. Signed-off-by: Fan Du <fan.du@windriver.com> --- include/uapi/linux/xfrm.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)