On Mon, 14 Nov 2005, David Hollis wrote: > Rawhide kernels may have some bits from the netdev git repos, but not > all. As the repos are pulled into Linus' kernel, they'll wind up in > rawhide kernels. It's probably around a 1-4 week turnaround (depending > on what state the patch window is in at the time) for patches to get > into mainline, and thus rawhide. Atually, our problem is more with Fedora kernels and LinuS kernels having dfferent version numbers for different features. Eg, when sk_alloc changed argument order and they swapped a bool for an address location, in LinuS trees that happened in 2.6.12, while that wasn't the case for fedra kernels. We are now seeing similar issues with the many changes in networking, and we can only put in one version check. examples: #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,8) #define NEED_INET_PROTOCOL #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,12) #define HAVE_SOCK_ZAPPED #define NET_26_12_SKALLOC #endif /* see <linux/security.h> */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,13) #define HAVE_SOCK_SECURITY /* skb->nf_debug disappared completely in 2.6.13 */ #define HAVE_SKB_NF_DEBUG #endif /* skb->stamp changed to skb->tstamp in 2.6.14 */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,14) #define HAVE_TSTAMP #define HAVE_INET_SK_SPORT #else #else #define HAVE_SKB_LIST #endif /* it seems 2.6.14 accidentally removed sysctl_ip_default_ttl */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,14) #define HAVE_MISSING_IP_DEFAULT_TTL #endif Making all of this build on Fedora kernels is reall becoming too big of a task. But I don't really have a solution for this either :( Paul -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
