On 5/10/2017 1:18 PM, Stephen Smalley wrote: > On Tue, 2017-05-09 at 23:50 +0300, Dan Jurgens wrote: >> From: Daniel Jurgens <dani...@mellanox.com> >> >> >> +#ifdef DARWIN >> + memcpy(&newc->u.ibpkey.subnet_prefix[0], >> &subnet_prefix.s6_addr[0], >> + sizeof(newc->u.ibpkey.subnet_prefix)); >> +#else >> + memcpy(&newc->u.ibpkey.subnet_prefix[0], >> &subnet_prefix.s6_addr32[0], >> + sizeof(newc->u.ibpkey.subnet_prefix)); >> +#endif > We can just always use s6_addr instead of s6_addr32 and drop the > #ifdef. Just pushed a commit to fix that elsewhere. Also we switched > from #ifdef DARWIN to __APPLE__ a while ago, but that won't matter once > you drop the #ifdef altogether. OK > >> @@ -722,10 +728,11 @@ extern int >> policydb_set_target_platform(policydb_t *p, int platform); >> #define POLICYDB_VERSION_CONSTRAINT_NAMES 29 >> #define POLICYDB_VERSION_XEN_DEVICETREE 30 /* Xen- >> specific */ >> #define POLICYDB_VERSION_XPERMS_IOCTL 30 /* Linux-specific */ >> +#define POLICYDB_VERSION_INFINIBAND 31 > This is Linux-specific too. I'll add a similar comment. > >> >> /* Range of policy versions we understand*/ >> #define POLICYDB_VERSION_MIN POLICYDB_VERSION_BASE >> -#define POLICYDB_VERSION_MAX POLICYDB_VERSION_XPERMS_IOCTL >> +#define POLICYDB_VERSION_MAX POLICYDB_VERSION_INFINIBAND >> >> /* Module versions and specific changes*/ >> #define MOD_POLICYDB_VERSION_BASE 4 >> @@ -743,10 +750,11 @@ extern int >> policydb_set_target_platform(policydb_t *p, int platform); >> #define MOD_POLICYDB_VERSION_TUNABLE_SEP 14 >> #define MOD_POLICYDB_VERSION_NEW_OBJECT_DEFAULTS 15 >> #define MOD_POLICYDB_VERSION_DEFAULT_TYPE 16 >> -#define MOD_POLICYDB_VERSION_CONSTRAINT_NAMES 17 >> +#define MOD_POLICYDB_VERSION_CONSTRAINT_NAMES 17 >> +#define MOD_POLICYDB_VERSION_INFINIBAND 18 >> >> #define MOD_POLICYDB_VERSION_MIN MOD_POLICYDB_VERSION_BASE >> -#define MOD_POLICYDB_VERSION_MAX >> MOD_POLICYDB_VERSION_CONSTRAINT_NAMES >> +#define MOD_POLICYDB_VERSION_MAX MOD_POLICYDB_VERSION_INFINIBAND >> >> #define POLICYDB_CONFIG_MLS 1 > Hmmm...we never introduced a binary module version for xperms, since > the only user is presently Android and they don't use binary modules > and in general we'd like to get rid of binary modules altogether and > switch entirely to source modules (either .te modules with a te2cil > converter or cil modules). But I guess you probably want to support > this in the interim for convenient usage within existing Fedora/RHEL > policies. > Yes, we want to pull this back into RHEL once it's available upstream.
Thank you for your quick review. I'll continue going through your comments on the other patches and post a v1 after giving some more time for others to comment as well.
