jcs
/subtext
/amendments
/serial.c
Amendments on January 23, 2024
jcs made amendment
579
2 months ago
Amendments on November 27, 2023
serial: Nullify session on init, catch init string errors
This allows the init string to not have ATQ0, which we now do by
default so if the init string is later changed and produces an
ERROR response, we will log it.
Also nullify the serial node session on init, which will happen
after forced hangup. Otherwise we may have been keeping the
association of the serial node's session and disconnecting it
on forced hangup, even if that session was later reassigned to
another node.
Amendments on November 15, 2023
jcs made amendment
544
5 months ago
Amendments on November 9, 2023
*: Do config changes on the fly, restarting modules as needed
Annotate each config option with which component it needs restarted
when it changes and do those after saving from the sysop menu. This
also provides the ability to log each changed config option with its
old and new values.
Amendments on November 8, 2023
serial: Make number of rings before answering a configurable setting
Defaults to 1 as before, but can be increased to allow caller id
strings to come through the logger.
Amendments on May 4, 2023
jcs made amendment
507
11 months ago
Amendments on March 15, 2023
serial: Add error checking, session buf canaries to try to find bug
Something is overwriting things in the session and trashing ibuflen
or obuflen, so add canaries in between struct members to try to find
where they get trashed.
Amendments on March 8, 2023
jcs made amendment
378
about 1 year ago
Amendments on February 23, 2023
jcs made amendment
309
about 1 year ago
Amendments on February 5, 2023
*: Minor fixes found by cppcheck and scan-build
Mostly unused variables
jcs made amendment
299
about 1 year ago
Amendments on November 7, 2022
serial: Fix maximum input length calculation when reading serial data
This looks like it was copied from somewhere else since 'n' was never
getting used to affect 'len', so len could be a very large number of
outstanding bytes from the serial port (such as during a ZModem
transfer) which would then cause FSRead to write too much data to
ibuf, trashing the session object and possibly other malloc'd data.
Take the smaller of the serial port outstanding count, and the number
of bytes left in ibuf, when reading from the serial port.
Also enable printing of serial port errors since it's useful to see
overruns when the port speed is too high.
Amendments on October 26, 2022
serial: Add failsafe to avoid sending more than sizeof(obuf)
Some bug is causing obuflen to be more than sizeof(obuf), maybe
somewhere in the ZModem code.
Amendments on October 5, 2022
serial: Switch to async writes, fully reset driver between calls
Doing a synchronous write with PBWrite can cause the entire system
to block if the remote modem has sent an XOFF. Switch to an async
write and then poll for its completion, or timeout and kill the IO
if it's beyond our timeout. We busy loop during polling, but if it's
beyond 30 ticks we uthread_yield.
Also, if the remote has sent an XOFF, when we reset the modem, it
won't accept our new commands even after an ATZ and a SerReset.
Between calls, fully reset the driver by closing it and doing
OpenDriver again.
Amendments on October 1, 2022
logger: There will only be one, stop passing object around
This also helps at shutdown to know when we can no longer log.
jcs made amendment
259
about 1 year ago
Amendments on August 14, 2022
logger: Add buffered logging, support logging without updating window
During telnet negotiation, many lines are logged in quick succession
which means we spend a lot of time updating the log TextEdit. Just
buffer these logs and spit them out all at once upon login or
disconnection, so the user's login prompt is immediately responsive.
Rename session_log to session_logf while I'm in here, to reflect that
it's a format-string function.
Amendments on August 13, 2022
jcs made amendment
242
about 1 year ago
Amendments on August 3, 2022
serial: Rewrite to allocate session on RING, then manually answer
If we can't allocate a session, just don't answer.
Amendments on July 16, 2022
serial: If session_create fails, hang up
Maybe we should not put the modem in auto-answer and require manual
ATA, but then only do that if we can establish a session first. That
way we don't waste time negotiating the modem just to hang up.
1
2