[Bug translator/32999] handle PIE executables with segments where file-offset is not vaddr

fche at redhat dot com sourceware-bugzilla@sourceware.org
Sat Jun 7 01:26:20 GMT 2025


https://sourceware.org/bugzilla/show_bug.cgi?id=32999

--- Comment #6 from Frank Ch. Eigler <fche at redhat dot com> ---
In the case of the _obstack_newchunk, the calculations appear to be correct.
The inode-uprobes system wants a file-offset based address for breakpoints,
i.e., 0x42400, not a virtual address based address.

  { .finder={  .purpose="inode-uprobes", .callback=&stapiu_process_found,
.build_id_len=0, .procname="/usr/bin/cpp", }, .module_name="/usr/bin/cpp",
.offset=(loff_t)0x42400ULL, .probe=(&stap_probes[0]), },

A gdb breakpoint on the same function would be virtual address based:

(gdb) break _obstack_newchunk
Downloading source file
/usr/src/debug/gcc-15.1.1-2.fc42.x86_64/obj-x86_64-redhat-linux/gcc/../libiberty/../../libiberty/obstack.c
Breakpoint 1 at 0x442400: file ../libiberty/../../libiberty/obstack.c, line
189.                                                                            

Buildid verification seems to be working just fine too.

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


More information about the Systemtap mailing list