26 Aug
2014
26 Aug
'14
3:17 a.m.
On Aug 25, 2014, at 2:14 PM, Lyndon Nerenberg <lyndon(a)orthanc.ca> wrote:
There was the size of the serial chip's hardware
buffer (before kicking the device driver with a hardware interrupt), and the driver itself
kicking the kernel via a software interrupt when the driver's buffers were getting
full.
What mattered was how the driver handed off the data from the bottom end of the driver
code to the kernel. Some drivers were better than others. In most cases (of the serial
ports), the culprit was the device driver.
--lyndon