[Bug uprobes/34301] user string copy fault -14...near 'user_string_n' at /usr/share/systemtap/tapset/uconversions.stp:128:10

[fche at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=34301

--- Comment #4 from Frank Ch. Eigler <fche at redhat dot com> ---
Did some closer looking at your trace and your script, and on the bright side,
we can rule out DWARF related problems after all, since these probes &
parameters are all sys/sdt.h markers.  Parameter locations are communicated
by gcc via ELF notes, and should be 100% fresh tasty goodness.  Systemtap
has no discretion or skill in the matter: gcc spells out the registers
holding each parameter, and stap takes it at its word, no heuristics.

Interestingly, that bad pointer looks like a piece of ASCII text ("8-IAny" LE).
This kinda smells like a possible compiler bug.  Would have to look at a
disassembly of the .so file around the appropriate marker nop instructions
(if we only knew which one it was - the stap error doesn't say!).  So a bit
of an impasse.

For your purposes, --suppress-handler-errors or --skip-badvars are
legitimate options.

Opened bug #34320 to improve the context given with the error message so
one has a chance at investigating deeper.

-- 
You are receiving this mail because:
You are the assignee for the bug.