Date: Mon, 14 Feb 2000 04:15:41 +1100 (EST) From: Bruce Evans <bde@zeta.org.au> To: "Jose M. Alcaide" <jose@we.lc.ehu.es> Cc: freebsd-hardware@FreeBSD.ORG Subject: Re: [fix and RFC] interrupt-level buffer overflows with Xircom PCMCIAmodem Message-ID: <Pine.BSF.4.21.0002140408080.4381-100000@alphplex.bde.org> In-Reply-To: <38A5E94B.397E77BC@we.lc.ehu.es>
index | next in thread | previous in thread | raw e-mail
On Sun, 13 Feb 2000, Jose M. Alcaide wrote: > Bruce Evans wrote: > > > > On Sat, 12 Feb 2000, Jose M. Alcaide wrote: > > > > > Now, the conclusions: perhaps RS_IBUFSIZE=256 is too low for some devices > > > (notably internal modems); it's only a theory. Or perhaps some devices > > > > Too large for devices with broken 16550 emulation. > > Uh? Are you saying that I should decrement RS_IBUFSIZE instead of > incrementing it? I could test this, but I don't see how a shorter buffer > could help to avoid buffer overruns... Oops, I meant "too low". > > RTS emulation is probably also broken for there to be a problem. Ideally, > > the sender should check its version of RTS (i.e., CTS) before sending > > each character. For bug for bug compatibility with 16550 senders, the > > sender only needs to check it every 16th character. > > I could insert some "probes" in sio.c to see what happens with RTS/CTS, > but I am not a 16550 expert, so I need some advice :-) Maybe you aren't actually using RTS/CTS. RTS is mostly for the termios line discipline. For slip and ppp, it is only deasserted when the interrupt-level buffer is more than 3/4 full, not when buffers in higher layers are almost full. However that should be enough to stop an emulator. Bruce To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the messagehome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0002140408080.4381-100000>