fche changed the topic of #systemtap to: http://sourceware.org/systemtap; email systemtap@sourceware.org if answers here not timely, conversations may be logged
<irker354> systemtap: fche systemtap.git:master * release-4.3-107-g41040e6cd / testsuite/systemtap.stress/current.stp: testsuite: current.stp module("*") defang
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Ping timeout: 240 seconds]
lijunlong has quit [Ping timeout: 240 seconds]
lijunlong has joined #systemtap
khaled has quit [Quit: Konversation terminated!]
sscox has joined #systemtap
hpt has joined #systemtap
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
irker354 has quit [Quit: transmission timeout]
derek0883 has quit [Remote host closed the connection]
orivej has joined #systemtap
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
_whitelogger has joined #systemtap
khaled_ has joined #systemtap
khaled_ has quit [Quit: Konversation terminated!]
khaled_ has joined #systemtap
orivej has quit [Ping timeout: 265 seconds]
mjw has joined #systemtap
sscox has quit [Ping timeout: 240 seconds]
hpt has quit [Ping timeout: 258 seconds]
tromey has joined #systemtap
amerey has joined #systemtap
<ema> hey! I'm testing systemtap 4.3 with linux 5.9.1, preparing a new upload for debian
<ema> I found that 4c5bed543b9b8c4b9cc544762afc9810938f88fa needs to be backported
<ema> now I'm wondering if there's anything else missing because my usual "test" command does not seem to work anymore:
<ema> stap -e 'probe process("/bin/ls").function("format_user") { printf("format_user(uid=%d)\n", $u) }'
<ema> the probe point cannot be found
<ema> however, if I specify the source file name in the function() match, that works:
<ema> stap -e 'probe process("/bin/ls").function("format_user@src/ls.c") { printf("format_user(uid=%d)\n", $u) }'
<ema> does this ring any bells?
orivej has joined #systemtap
sscox has joined #systemtap
<fche> hmmmm those should both work
<ema> I see that this is not true for all functions
<ema> for example:
<ema> stap -L 'process("/bin/ls").function("*")' | grep xnumto
<ema> process("/bin/ls").function("xnumtoumax@lib/xdectoint.c:36") $n_str:char const* $base:int $min:uintmax_t $max:uintmax_t $suffixes:char const* $err:char const* $err_exit:int $tnum:uintmax_t
<ema> stap -ve 'probe process("/bin/ls").function("xnumtoumax") { println(pp()) }'
<ema> this works fine ^
<fche> a lone function name can sometimes be resolved via minisymtab in the executables, vs. requiring debuginfo
<fche> not sure that explains everything tho
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Ping timeout: 240 seconds]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Read error: Connection reset by peer]
derek0883 has joined #systemtap
irker464 has joined #systemtap
<irker464> systemtap: alizhang systemtap.git:azhang/pr13838 * release-4.3-155-g355b8cba8 / java/Makefile.in: fix java/Makefile.in
<irker464> systemtap: alizhang systemtap.git:azhang/pr13838 * release-4.3-156-gaed3356af / java/Makefile.in: fix java/Makefile.in to keep it same to master branch
<kerneltoast> fche, what's the oldest kernel stap officially supports?
przemoc has quit [Read error: Connection reset by peer]
przemoc has joined #systemtap
derek0883 has quit [Remote host closed the connection]
sscox has quit [Ping timeout: 260 seconds]
derek0883 has joined #systemtap
<fche> RHEL6 I believe, so 2.6.32+
khaled_ has quit [Ping timeout: 246 seconds]
<kerneltoast> wow it's older than some people on the internet
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
<fche> we <3 customers who run old stable software on old stable hardware
<kerneltoast> the debian way of life
<fche> and RHEL :)
tromey has quit [Quit: ERC (IRC client for Emacs 27.1.50)]
<kerneltoast> fche, you like reviewing patches?
<fche> yo dawg
<fche> what have you for us this time
<kerneltoast> (i left out a hunk for task_finder.c oops, but who uses task_finder.c anyway?)
<fche> ok, what is this supposed to fix?
<fche> um starting over to read from the top, never mind
<kerneltoast> hah, i added a commit message this time!
<fche> and I initially just scrolled past, bad fche, bad bad fche
<fche> so the effect is to move __stp_get_mm_path outside the task_lock ?
<kerneltoast> yeah so we can acquire the mmap sem
<kerneltoast> with sleeping allowed
<fche> is that atomic_inc_not_zero (&mm->mm_users) a Proper way of doing this - protecting the pointer across the task lock being released?
<kerneltoast> yeah, it's also called mmget_not_zero() in newer kernels
<kerneltoast> there are two refcounts in the mm struct though
<kerneltoast> one for the mm pointer itself, and one for the mm mappings
<kerneltoast> mm_users is for the mm mappings, not the mm pointer
<kerneltoast> i think if you have a reference to mm_users, the mm pointer won't disappear
derek0883 has quit [Remote host closed the connection]
<kerneltoast> yeah a valid mm_users reference (i.e. you don't increment it when it's zero) is all that's needed
<fche> has running this through the testsuites shown any noticeable improvement?
* fche must head out for dinner, bck in a bit
<kerneltoast> fche, it fixes a really annoying bug I've been working on for a little over a month
<kerneltoast> where stapio is hung because its target process never gets attached to
<kerneltoast> I'm running the big testsuite on it now
derek0883 has joined #systemtap
khaled_ has joined #systemtap
sscox has joined #systemtap
<irker464> systemtap: sultan systemtap.git:azhang/pr13838 * release-4.3-106-gb0cccb3cf / runtime/stp_utrace.c: stp_utrace: replace task_utrace_lock with non-blocking RCU read locks
<fche> aha
<irker464> systemtap: fche systemtap.git:azhang/pr13838 * release-4.3-107-g41040e6cd / testsuite/systemtap.stress/current.stp: testsuite: current.stp module("*") defang
<fche> that's familiar
<irker464> systemtap: alizhang systemtap.git:azhang/pr13838 * release-4.3-108-g804129be1 / : PR13838: Added basic floating point support to systemtap
<irker464> systemtap: alizhang systemtap.git:azhang/pr13838 * release-4.3-109-gc6018acd4 / runtime/softfloat.c: PR13838: Fixed rebase merging mistakes, delete duplicates f64_to_i64 and missing countLeadingZeros64 in runtime/softfloat.c
<irker464> systemtap: alizhang systemtap.git:azhang/pr13838 * release-4.3-162-g7ce7e3811 / : PR13838: fixed format and space for softfloat.* and files under softfloat/
tux3 has quit [*.net *.split]
eichiro has quit [*.net *.split]
eichiro has joined #systemtap
tux3 has joined #systemtap
<kerneltoast> fche, after talking to agentzh, it seems like trying to sleep in the task finder is a pain (my patch sleeps inside an rcu read lock, oops!) and we should just error out instead
<fche> ok
<kerneltoast> so the new approach will be a oneliner to error out when we cannot attach to the specified PID
<fche> I am allow
<irker464> systemtap: amerey systemtap.git:PR26015 * release-4.3-52-gaf08cd479 / : Make syscall arg writing conditional on _IS_SREG_ARCH
amerey has quit [Quit: Leaving]
<fche> makes sense
<fche> what's the form of the error message when this hits?
<kerneltoast> ERROR: Unable to get path (error -2) for pid 12580
<kerneltoast> ERROR: probe process("/usr/local/openresty/nginx/sbin/nginx").end registration error (rc -2)
<fche> aha
<kerneltoast> i guess it's the first patch i've shown you that makes sense :P
<fche> heh
<fche> fche is over very little brain
<fche> mmm what's the correct quote
<fche> fche is a bear of very little brain
<kerneltoast> one of my professors would say he had very little cache when asked if remembered something
<fche> yeah they probably are very overcommitted with memory/contexts
orivej has quit [Ping timeout: 265 seconds]
<kerneltoast> oof, one of the reviews for him on ratemyprofessor says "Avoid like the covid"
<fche> heh, ironic as college age students rarely need to fear the bug
<kerneltoast> hey i don't wanna get sick
<fche> yeah yeah no one does
<kerneltoast> task finder is going to leave me immunocompromised
<fche> OH NO
<kerneltoast> soon *i* will be returning ENOENT
<kerneltoast> guessing you're not going to whip me to run the full testsuite for this one?
<fche> nah, ship it
<kerneltoast> !!!
<kerneltoast> agentzh