3 ms·

16 bit signed int overflow brings air traffic control to standstill

2 points by madaxe_again 13y ago

kotnik 13y ago
"The agency said it had now added more flight-processing memory to the computer system." int32_t to the rescue!
jameskilton 13y ago
Would love to see the source that describes the problem in more detail but if the "16-bit signed" is true... Why would altitude ever be negative? Just a thought.
- zimpenfish 13y ago
  Low level flight over the Dead Sea (422m below sea level) could be negative, I guess.
  - Turing_Machine 13y ago
    Furnace Creek Airport in Death Valley is at -211 feet (-64 m). http://www.nps.gov/deva/planyourvisit/airports.htm http://www.nps.gov/deva/planyourvisit/airports.htm
- zimpenfish 13y ago
  http://www.youtube.com/watch?v=vQu3EcPRxRA http://www.youtube.com/watch?v=vQu3EcPRxRA "USAFE C-130J Practice negative Altitude Landing by the Dead Sea"
dalke 13y ago
That BBC article says nothing about signed int overflow. I believe the title of this link is pure conjecture by the submitter. In any case, max signed 16-bit integer is 32767. It's unsigned 16-bit integer to get to 65536, so this title is wrong. This topic has come up a couple of times already here on HN, including https://news.ycombinator.com/item?id=7691583 https://news.ycombinator.com/item?id=7691583 and https://news.ycombinator.com/item?id=7694121 https://news.ycombinator.com/item?id=7694121 . The report is that the U-2 was flying at 60K feet, not 65K which might trigger the problem. One commenter pointed to a page which point to http://www.barryschiff.com/high_flight.htm http://www.barryschiff.com/high_flight.htm which says "The U-2's Mode C transponder, however, can indicate no higher than FL600." Mode C can go slightly higher, but "commercially available Mode C transponders cannot currently report an altitude greater than 62,000 feet" says https://www.faa.gov/about/office_org/headquarters_offices/ast/licenses_permits/sub_orbital_rockets/newregs/media/EP_FR.pdf https://www.faa.gov/about/office_org/headquarters_offices/as... . So of the limited information we do know, the U-2 should not have triggered an overflow, assuming that such a bug existed. BTW, according to http://www.airsport-corp.com/modec.htm http://www.airsport-corp.com/modec.htm , the altitude encoder data has a higher range: -1200 to 126,700 ft. One of the reason for negative altitude is of course places below sea level, but another is that altitude when measured via barometric pressure depends on the weather. High pressure days give a lower measured altitude.

2 points by madaxe_again 13y ago

"The agency said it had now added more flight-processing memory to the computer system." int32_t to the rescue!

Would love to see the source that describes the problem in more detail but if the "16-bit signed" is true... Why would altitude ever be negative? Just a thought.

zimpenfish 13y ago
Low level flight over the Dead Sea (422m below sea level) could be negative, I guess.
- Turing_Machine 13y ago
  Furnace Creek Airport in Death Valley is at -211 feet (-64 m). http://www.nps.gov/deva/planyourvisit/airports.htm http://www.nps.gov/deva/planyourvisit/airports.htm
zimpenfish 13y ago
http://www.youtube.com/watch?v=vQu3EcPRxRA http://www.youtube.com/watch?v=vQu3EcPRxRA "USAFE C-130J Practice negative Altitude Landing by the Dead Sea"

dalke 13y ago

That BBC article says nothing about signed int overflow. I believe the title of this link is pure conjecture by the submitter. In any case, max signed 16-bit integer is 32767. It's unsigned 16-bit integer to get to 65536, so this title is wrong. This topic has come up a couple of times already here on HN, including https://news.ycombinator.com/item?id=7691583 https://news.ycombinator.com/item?id=7691583 and https://news.ycombinator.com/item?id=7694121 https://news.ycombinator.com/item?id=7694121 . The report is that the U-2 was flying at 60K feet, not 65K which might trigger the problem. One commenter pointed to a page which point to http://www.barryschiff.com/high_flight.htm http://www.barryschiff.com/high_flight.htm which says "The U-2's Mode C transponder, however, can indicate no higher than FL600." Mode C can go slightly higher, but "commercially available Mode C transponders cannot currently report an altitude greater than 62,000 feet" says https://www.faa.gov/about/office_org/headquarters_offices/ast/licenses_permits/sub_orbital_rockets/newregs/media/EP_FR.pdf https://www.faa.gov/about/office_org/headquarters_offices/as... . So of the limited information we do know, the U-2 should not have triggered an overflow, assuming that such a bug existed. BTW, according to http://www.airsport-corp.com/modec.htm http://www.airsport-corp.com/modec.htm , the altitude encoder data has a higher range: -1200 to 126,700 ft. One of the reason for negative altitude is of course places below sea level, but another is that altitude when measured via barometric pressure depends on the weather. High pressure days give a lower measured altitude.