This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug tapsets/18649] int_arg() misbehaves on x86[_64] for 32-bit uprobe in binary having debuginfo


https://sourceware.org/bugzilla/show_bug.cgi?id=18649

--- Comment #14 from Martin Cermak <mcermak at redhat dot com> ---
Created attachment 8565
  --> https://sourceware.org/bugzilla/attachment.cgi?id=8565&action=edit
updated patch

(In reply to Josh Stone from comment #13)
> ENABLE_PROLOGUES is not quite right.  The former true/false values should
> now be always/auto to keep the same behavior -- there was no never.

Ouch, you are right.  Fixed.

> It appears you removed prologues from usage() entirely, instead of adding
> your new switch to the output.  Why?

Fixed.

> The manpage explanation of "auto" is odd: "enable only if possible"?  As if
> "always" will instead do the impossible? :)  The real logic behind auto is a
> heuristic for when we think it will be useful, in userspace with low quality
> location info.  Not sure of a short blurb  -- maybe just "enabled by
> heuristic".

:) Yup, updated. 

> The staprun.h change doesn't belong at all.  I assume that was copying
> color_mode, but prologues are only a concern to the translator.

Yes, I've been following Jonathan's color_mode implementation.  When looking
over the patch before attaching it, I felt this is somewhat weird, but
unfortunately left it there.  Fixed.

> Have you confirmed that --prologue-searching=never solves your original
> problem?

Sure.  I did that before and re-tested now again with attached updated patch.

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


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]