07:53 <ema> problem: I've got a unix timestamp represented as a double in a C program, and I'd like to print it with a systemtap userspace probe

07:53 <ema> any way to do that?

10:55 <fche> hi ema

10:56 <fche> hm I think at the moment you'd be stuck with doing some bitwise operations to extract the mantissa part

10:56 <fche> to get at the integer seconds part

10:56 <fche> (shouldn't be more than a shift and a mask)

10:57 <fche> and then you can print it via ctime()

15:36 <lindi-> I'm sure you can write a floating point library in the systemtap language :)

15:36 <irker760> systemtap: smakarov systemtap.git:master * release-4.2-80-g173d3fdb2 / testsuite/systemtap.base/beginenderror.exp translate.cxx: PR10280 fix prev commit: UTS_RELEASE is vacuous, UTS_VERSION is not

15:42 <fche> lindi-, yes indeed - there is a long standing PR for that

15:42 <lindi-> oh, cool

15:42 <fche> but I'd be happy to start really small, like ... extract ieee844 parts

15:42 <fche> (noting we can't do actual FP arithmetic in kernel space)

15:43 <fche> make that ieee754 :)

15:43 <lindi-> just make your stap script output python3 and then pipe that to a python3 interpreter in userland

15:44 <lindi-> maybe json could work too

15:44 <lindi-> if it is just for printing then it doesn't need to be done in the kernel at all

15:44 <fche> doesn't help if you want to operate on the values in situ

15:44 <lindi-> true

15:48 <irker760> systemtap: smakarov systemtap.git:master * release-4.2-81-g15b43276a / translate.cxx: PR10280: fix 'ISO C90 forbids mixed declarations and code'