Subject:
|
LNP checksum optimizations
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Wed, 31 Jan 2001 17:03:05 GMT
|
Viewed:
|
1611 times
|
| |
|
|
In a previous post I said that some optimizations can be made in LNP. Now
I have the patch ready and uploaded it on sourceforge. Sadly I made it
against a custom version of the kernel, so it may need a manual
intervention to apply (fuzzyness shouldn't help much).
In brief, what I changed.
1. The checksum for outgoing packets is computed while copying data in the
buffer, in one pass.
2. The checksum is computed as 8-bit integer (instead of 16 bits; upper 8
bits were discarded in the original code).
3. The checksum for incoming packets is computed as new bytes are
available, and not after receiving all data. So the computing time is
split between all interrupt calls.
Comments, bug reportings, opinions are welcome!
bye
Bernardo
|
|
1 Message in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
