Messages in this thread | | | Date | Wed, 2 Dec 1998 13:54:46 -0500 | From | "Theodore Y. Ts'o" <> | Subject | Re: Hi speed serials for 2.1.130 |
| |
Date: Wed, 2 Dec 1998 12:14:28 +0100 From: Pavel Machek <pavel@bug.ucw.cz>
This is adapted patch from mizuhara@st.rim.or.jp, gigo@yk.rim.or.jp, ytakeuch@po.iijnet.or.jp. It alows you to use speeds up-to 921kbit/second on modern chipsets. I've ported this patch to 2.1.130 and de-dosified it quite a bit. PLEASE TEST.
Pavel, I know you've probably done a lot of cleanup already, but this patch doesn't pass the cleanliness test.
In particular, I'm really suspicious of the large table of magic numbers, with no explanation of what they do. There also seem to suspicious magic numbers that include the COM 1/2/3/4 port addresses. Is this some kind of PNP code that's getting stuffed into the serial driver?
Given that the only change to the actual guts of the serial operation is to set the base_baud rate, I'd much, much, much prefer to see as much of this either (a) placed into a user mode program, such as setserial, or (b) into a separate loadable module which configures the hardware, and then unloads itself.
Stashing this huge amount of code into unpageable kernel space doesn't seem like a particularly good long-term architecture, and the way the code is written, I'm really concerned about it impacting the long-term maintainability of the serial dirver.
- Ted
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |