Buelmann: new implementation for ceiling Since my first functional fix in the ceiling computation in commit ceecabfddb57, I noticed that the computation used a linear search, that became rather computational expensive after that commit. The simple question is: why not a binary search? So, this commit implements the binary search. But there is a long story attached to this. Comparing ceiling results from hwOS and this OSTC4 code were very different. Basically, the original OSTC4 algorithm computed the ceiling using the same GFlow to GFhigh slope, in such a way, that the ceiling was in sync with the presented deco stops, where the hwOS code presents a GFhigh based ceiling. This said, it is more logical when the OSTC4 and hwOS code give similar results. This new recursive algorithm gives very similar results for the ceiling compared to hwOS. To be complete here, the Buelmann ceiling is the depth to which you can ascend, so that the leading tissue reaches GFhigh. This also explains why the deepest deco stop is normally deeper than the ceiling (unless one dives with GF like 80/80). The code implemented here is rather straightforward recursion. Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>

#! openocd -f
# Define the prob used:
source [find interface/stlink-v2.cfg]
set WORKAREASIZE 0x8000
transport select hla_swd

# Reset options
set ENABLE_LOW_POWER 1
set STOP_WATCHDOG 1
reset_config srst_only srst_nogate connect_assert_srst

# Seelct the right chip
set CHIPNAME stm32f427IIT6
set CONNECT_UNDER_RESET 1
source [find target/stm32f4x.cfg]

# Allow to continue execution after a connection:
init_reset run

puts "Flash banks:"
flash banks

#puts "Reading FIRMWARE..."
# Requires OpenOCD 0.10.0 / released January 2017 ?
#flash dump_image CPU1-DISC-dump.hex 0x00000000 0x8000

#puts "Done."
exit