Bug #12711: Darwin doesn't show C backtrace correctly if iSIGSEGV is received when IP is in userland

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 2.4.0dev (2016-08-18 trunk 55955) [x86_64-darwin15]
• Backport: 2.1: REQUIRED, 2.2: DONE, 2.3: DONE

[naruse (Yui NARUSE)]

Current Ruby can show C backtrace on the following case

Process.kill :SEGV, $$

But can't on the following:

require"fiddle"
Fiddle.dlunwrap(100).class

[naruse (Yui NARUSE)]

• Status changed from Open to Closed

Applied in changeset r56030.

---

• vm_dump.c (backtrace): use rip in the saved context for the case
  the SIGSEGV is received when the process is in userland.
  Note that ip in the stack should be used if the signal is received
  when it is in kernel (when it is calling syscall) [Bug #12711]

[nagachika (Tomoyuki Chikanaga)]

• Backport changed from 2.1: REQUIRED, 2.2: REQUIRED, 2.3: REQUIRED to 2.1: REQUIRED, 2.2: REQUIRED, 2.3: DONE

ruby_2_3 r56257 merged revision(s) 56030,56035.

[usa (Usaku NAKAMURA)]

• Backport changed from 2.1: REQUIRED, 2.2: REQUIRED, 2.3: DONE to 2.1: REQUIRED, 2.2: DONE, 2.3: DONE

[wanabe (_ wanabe)]

• Related to Bug #13566: A process freezes at the beginning of C level backtrace when a certain SEGV is occurred added