troubles running on freebsd console

[dankamongmen]

The FreeBSD console sets TERM=vt100 by default. I'm seeing some very strange results there (running in QEMU): we have sixel support, which is nice, but we don't generally succeed in starting, which is bad. Throwing on logging shows that we're soaking core cycling on input. Very strange.

[dankamongmen]

handoff_initial_responses_late:1014:handing off initial responses
block_on_input:2126:blocking on input availability
block_on_input:2183:waiting on 1 fds (ibuf: 0/1024)
tiocgwinsz:1436:Bogus return from TIOCGWINSZ on 3 (0/0)

[dankamongmen]

process_melange:2044:input 0 (22)/22 [0x1b] ( )
block_on_input:2126:blocking on input availability
block_on_input:2130:nonblocking read to check for completion
block_on_input:2183:waiting on 1 fds (ibuf: 22/1024)
block_on_input:2210:poll returned 0
block_on_input:2224:got events: ti
process_melange:2044:input 0 (22)/22 [0x1b] ( )
block_on_input:2126:blocking on input availability
block_on_input:2130:nonblocking read to check for completion
block_on_input:2183:waiting on 1 fds (ibuf: 22/1024)
block_on_input:2210:poll returned 0
block_on_input:2224:got events: ti
process_melange:2044:input 0 (22)/22 [0x1b] ( )
block_on_input:2126:blocking on input availability
block_on_input:2130:nonblocking read to check for completion
block_on_input:2183:waiting on 1 fds (ibuf: 22/1024)
block_on_input:2210:poll returned 0
block_on_input:2224:got events: ti
process_melange:2044:input 0 (22)/22 [0x1b] ( )
block_on_input:2126:blocking on input availability
block_on_input:2130:nonblocking read to check for completion
block_on_input:2183:waiting on 1 fds (ibuf: 22/1024)
block_on_input:2210:poll returned 0
block_on_input:2224:got events: ti
process_melange:2044:input 0 (22)/22 [0x1b] ( )
block_on_input:2126:blocking on input availability
block_on_input:2130:nonblocking read to check for completion
block_on_input:2183:waiting on 1 fds (ibuf: 22/1024)
block_on_input:2210:poll returned 0
block_on_input:2224:got events: ti

[dankamongmen]

we ought presumably be able to use whatever ioctl() or other method vidcontrol(1) employs and get the screen resolution. together with the cell geometry, a division will yield the cell-pixel geometry.