David Miller <davem@davemloft.net> writes:It's not about extensions, syncppp is just unmaintainable. It's implementation of two protocols - Cisco HDLC and (sync) PPP intermixed together. What would be needed is a) splitting them into two files (including protocol selection in drivers - somehow), b) replacing existing broken syncppp state machines distributed all over the place with a single sane state machine, c) perhaps adding IPv6, people want it, d) packet flow from hw drivers would need to be upgraded as well (*_type_trans() and fast path for IP/IPv6 and perhaps IPX, though I don't know anyone ever using IPX with sync PPP). I guess then you'd get generic HDLC. I think it's easier to a) simply remove syncppp.c, b) apply my patch with new PPP, c) blindly port the old drivers to generic HDLC, d) hope for the best, with a bit of luck nobody would notice. Alternative c) simply remove the old drivers (d) unchanged). I don't want the breakage resulting from either action thus I think syncppp.c should stay there (basically unchanged, though we could at last remove those variables used only by the original FreeBSD code in 1994). If the new implementation is my patch or if it uses generic PPP is debatable. I think simplicity is a good thing, the interface to generic PPP would be bigger than the whole sync-only PPP. But I'm not against using generic PPP, it's (AFAICS) clean and *working* and *maintained* (and I'm not worried about backward compatibility and certainly not about users' visions). -- Krzysztof Halasa --
| Linus Torvalds | Linux 2.6.25-rc2 |
| Robin Lee Powell | NFS hang + umount -f: better behaviour requested. |
| David Woodhouse | Re: [GIT *] Allow request_firmware() to be satisfied from in-kernel, use it in mor... |
| Greg KH | [GIT PATCH] driver core patches against 2.6.24 |
git: | |
| Miles Bader | way to automatically add untracked files? |
| Jan-Benedict Glaw | Errors GITtifying GCC and Binutils |
| Andreas Hildebrandt | CVS-$Id:$ replacement in git? |
| Alexander Gladysh | [Q] Encrypted GIT? |
| Alex Thurlow | Router performance on OpenBSD and OpenBGPD |
| Karel Kulhavy | lookup option in /etc/resolv.conf ignored |
| Richard Stallman | Real men don't attack straw men |
| Sunnz | Can OpenBSD do what BusyBox does? |
| Julius Volz | [PATCH RFC 02/24] IPVS: Add genetlink interface implementation |
| Lennart Sorensen | Re: [PATCH 1/2] IPV4: remove addresses and routes when carrier is lost |
| Jussi Kivilinna | [PATCH 04/14] [rndis_host] Halt device if rndis_bind fails. |
| Lennert Buytenhek | Re: using software TSO on non-TSO capable netdevices |
