diff options
| author | Vladislav Grishenko <themiron@mail.ru> | 2026-07-30 15:48:55 +0500 |
|---|---|---|
| committer | Vladislav Grishenko <themiron@mail.ru> | 2026-08-03 01:06:01 +0500 |
| commit | f2606293c1007aa68a8bcf1e92ae3d64e250eadb (patch) | |
| tree | a804a66f12bf586f3724d959b40dab40e0862244 /drivers/ppposeq/ppposeq.h | |
| parent | 332daf3705c8f6ec2e04e041261f86ba7b8650a6 (diff) | |
| download | accel-ppp-sstp-ppposeq.tar.gz accel-ppp-sstp-ppposeq.zip | |
sstp: add ppposeq transport to avoid userspace HDLC framingsstp-ppposeq
A pty is a byte stream, so the tty flip buffer merges frames written
back to back and sstp has to re-delimit them with async HDLC escaping
and a CRC-16 FCS. On a 1452-byte payload that is ~3600 ns per frame,
most of it spent on the FCS.
PPPOSEQ is a pppox protocol whose socket is the ppp endpoint itself,
so one datagram is one frame and no framing is needed at all. The
same payload takes ~380 ns per frame, about 9 times less. Requires
kernel 2.6.37, the first with PX_MAX_PROTO 3, whose remaining slot
it claims.
Supported kernels are from 2.6.37 to 7.2.
The new ppp-mode option selects the transport; auto, the default,
falls back to async when the module is unavailable, so hosts with
prebuilt kernels are unaffected.
PPP_SYNC is removed, being disabled and unfixable over a pty: frame
boundaries cannot be recovered from the stream, and coalescing cannot
be prevented since frames arrive from the network stack.
Diffstat (limited to 'drivers/ppposeq/ppposeq.h')
| -rw-r--r-- | drivers/ppposeq/ppposeq.h | 34 |
1 files changed, 34 insertions, 0 deletions
diff --git a/drivers/ppposeq/ppposeq.h b/drivers/ppposeq/ppposeq.h new file mode 100644 index 00000000..ecdbdfb1 --- /dev/null +++ b/drivers/ppposeq/ppposeq.h @@ -0,0 +1,34 @@ +#ifndef __PPPOSEQ_H +#define __PPPOSEQ_H + +#include <linux/if_pppox.h> + +/* + * ppposeq - PPP over a SEQPACKET AF_PPPOX socket. + * + * Replaces the pty + ppp_async transport for userspace PPP terminators + * such as sstp. A pty is a byte stream, so frame boundaries are lost in + * the tty flip buffer and have to be rebuilt with HDLC escape+FCS framing. + * Here the socket itself is the ppp endpoint and each datagram carries + * exactly one ppp frame, so no framing is needed on either side. + * + * fd = socket(AF_PPPOX, SOCK_SEQPACKET, PX_PROTO_OSEQ); + * connect(fd, &sa, sizeof(sa)); // registers the channel + * ioctl(fd, PPPIOCGCHAN, &idx); + * chan = open("/dev/ppp"); ioctl(chan, PPPIOCATTCHAN, &idx); + * // frames flow over fd with send()/recv() + * + * Only the protocol number is new; everything else uses the common + * AF_PPPOX and PPPIOC* interfaces. + */ + +#ifndef PX_PROTO_OSEQ +#define PX_PROTO_OSEQ 3 +#endif + +struct sockaddr_ppposeq { + __kernel_sa_family_t sa_family; /* AF_PPPOX */ + unsigned int sa_protocol; /* PX_PROTO_OSEQ */ +} __attribute__((packed)); + +#endif |
