Messages in this thread | | | Date | Mon, 05 Jan 2015 09:05:41 -0500 | From | Peter Hurley <> | Subject | Re: [PATCH v2 2/3] serial: 8250_early: prepare for dynamic BASE_BAUD |
| |
On 01/05/2015 08:52 AM, Vineet Gupta wrote: > On Monday 05 January 2015 07:04 PM, Peter Hurley wrote: >>> Also what do u think of the early IER optimization ? >> Seems ok except the commit changelog is wrong: IER is disabled twice per _line_. >> How did you profile it? It's ok if the answer is "seemed faster"; I'm just >> curious. > > Looking at the code yes this would be per line - but I vaguely remember seeing it > for every char - this was when debugging a systemC model for 8250. Its been a while... > I'd added prints in the model whenever IER would wiggle and it was just being > written with 0's again and again when in the early polling based printing mode.
Maybe that was while running kgdb through serial8250_put_poll_char(), which is per-char.
Regards, Peter Hurley
| |