[Bug translator/11376] New: process(pid).statement(addr).absolute bugs + fix

["jkenisto at us dot ibm dot com" <sourceware-bugzilla at sourceware dot org>]

stap's original support for uprobes --
  process(<pid>).statement(<addr>).absolute[.return]
-- doesn't work for me unless I apply the following patch:

----- cut here ----
--- tapsets.cxx.orig	2010-03-12 16:33:54.084773884 -0800
+++ tapsets.cxx	2010-03-13 13:26:25.523494988 -0800
@@ -4812,13 +4812,15 @@
           // only the same fields as we're about to emit.
           s.op->line() << " .finder={";
           if (p->pid != 0)
-            s.op->line() << " .pid=" << p->pid;
+            s.op->line() << " .pid=" << p->pid << ",";
           else if (p->section == ".absolute") // proxy for ET_EXEC -> exec()'d
program
             {
               s.op->line() << " .procname=" << lex_cast_qstring(p->module) << ",";
               s.op->line() << " .callback=&stap_uprobe_process_found,";
             }
-          if (p->section != ".absolute") // ET_DYN 
+          if (p->section == "")  // .statement(addr).absolute
+            s.op->line() << " .callback=&stap_uprobe_process_found,";
+          else if (p->section != ".absolute") // ET_DYN 
             {
 	      if (p->has_library && p->sdt_semaphore_addr != 0)
 		s.op->line() << " .procname=\"" << p->path << "\", ";
----- cut here ----

The missing comma (line 4815) yields a compilation error in phase 4.

Failure to consider null section names (line 4821) apparently yields
code that never calls register_uprobe().

I don't know if these fixes are really the correct ones, but they work
for me: probing 32-bit (cc -m32) programs on my x86_64 Fedora 12 system,
running the weekly stap snapshot from March 6, 2010.

It's kind of an obscure feature, but it can come in handy when doing
exhaustive tests of uprobes on the x86 instruction set (see probe*.awk
attachments to PR #11249) or probing a program with a symbol table but
no DWARF.

-- 
           Summary: process(pid).statement(addr).absolute bugs + fix
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: translator
        AssignedTo: systemtap at sources dot redhat dot com
        ReportedBy: jkenisto at us dot ibm dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=11376

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