ERROR::PASS2(7stap)

💖 error::pass2 — systemtap pass-2 errors 💖

okay bestie, welcome to the error::pass2 page from the 7stap manual section 🎀 this is your lil guide to everything that can go sideways in pass 2, aka the elaboration stage. no cap.

✨ NAME ✨

error::pass2 — systemtap pass-2 errors, and like… all the ways they can happen 🫶

🌸 DESCRIPTION 🌸

so here's the tea: errors that pop up during pass 2 (elaboration) can have a whole variety of causes. not one single villain, it's giving ensemble cast 😭. the common types include:

• missing debuginfo — your script needs debuginfo to resolve a probe point, but it literally could not find any 🥲. peep error::dwarf(7stap) and warning::debuginfo(7stap) for more deets.
• unavailable probe point classes — some types of probe points are only available on certain system versions, architectures, and configurations 💅. for example, user-space process.* probes may require utrace or uprobes capability in the kernel for this architecture. it's giving "not on this machine, babes."
• unavailable probe points — some probe points may be individually unavailable even when their class is totally fine 🙈. for example, kprobe.function("foobar") may fail if function foobar does not exist in the kernel any more. debugging or symbol data may be absent for some types of .function or .statement probes, so check for availability of debuginfo. try the stap-prep program to download possibly-required debuginfo 👑. use a wildcard parameter such as stap -l 'kprobe.function("*foo*")' to locate still-existing variants ✨. and use ! or ? probe point suffixes to denote optional / preferred-alternatives, so the working parts of a script can continue — iconique, honestly.
• typos — bestie. there might be a spelling error in the probe point name ("sycsall" vs. "syscall" 💀). wildcard probes may not find a match at all in the tapsets. recheck the names using stap -l PROBEPOINT. another common mistake is to use the . operator instead of the correct -> when dereferencing context variable subfields or pointers: $foo->bar->baz even if in C one would say foo->bar.baz. not me making that mistake… never 😭.
• unavailable context variables — systemtap scripts often wish to refer to variables from the context of the probed programs using $variable notation 🍓. these variables may not always be available, depending on versions of the compiler, debugging/optimization flags used, architecture, etc. use stap -L PROBEPOINT to list available context variables for given probes. use the @defined() expression to test for the resolvability of a context variable expression. consider using the stap --skip-badvars option to silently replace misbehaving context variable expressions with zero 👀. and experiment with the stap --prologue-searching option ✨.
• module cache inconsistencies — occasionally, the systemtap module cache ($HOME/.systemtap/cache) might contain obsolete information from a prior system configuration/version, and produce false results as systemtap attempts to reuse it 🫠. retrying with stap --poison-cache ... forces new information to be generated 💅.

note: this should not happen and likely represents a systemtap bug. please report it 🙏 — you'd be doing the whole community a solid, fr fr.

🔍 GATHERING MORE INFORMATION 🔍

increasing the verbosity of pass-2 with an option such as --vp 02 can help pinpoint the problem 🔥. turn that dial up, bestie.

🌈 SEE ALSO 🌈

• stap(1)
• stap-prep(1)
• stapprobes(3stap)
• probe::*(3stap)
• error::dwarf(7stap)
• error::inode-uprobes(7stap)
• warning::debuginfo(7stap)
• error::reporting(7stap)

and that's the whole page, bestie 💜 that was error::pass2 straight outta the 7stap manual — go forth, debug your probes, and slay 👑✨