Re: text buffer size in boot

or maybe it is current issue after all. or fix or workaround that can be added

basically i thought what if i add:

----------------------------------------------------------------------
--- /root/files/fbsd/current/src/libexec/rc/rc  2025-07-15 11:59:53.719893000 +0300
+++ rc  2025-10-21 12:50:29.281862000 +0300
@@ -149,4 +149,5 @@

 echo ''
 date
+sleep 10
 exit 0
----------------------------------------------------------------------

this fully fixed everything. seems like i don't need full 10s, but <1s. eg too fast for human to notice

it seems like something is writing text in 4kb chunks and i'm assuming serial isn't taking data that fast. then maybe init goes multiuser and data in buffer is lost

i recall tests of fbsd being able to boot really fast in some vm platforms. or maybe things there never wrote more than 4kb out

i actually never tested what if init happens really fast. but usually it doesn't, and it actually starts things up slowly one at time. i vaguely recall that init can also do parallel launch

i'm wondering if this is also happening on other console types, including bios and efi, but it's never a problem since they output text short and slow

then comes my initscript and blasts ~600 bytes out fast, doesn't wait, exits, and it expects it all to reach console

but apparently init (?) has problems here

again, it's all there if system is booted up and syslog has recorded boot log (i have all.log as well as console.log configured). it only affects whatever physical console there is

so is this a bug that only affects some hw?