** Description changed:

  SRU Justification:
  
  [ Impact ]
  
   * The preferred way of doing stack unwinding on Linux on Z is via dwarf call 
frame information.
  In absence of a dwarf unwinder (as in the Linux kernel) a stack chain can be 
maintained at runtime in addition to the dwarf unwinding information.
  
   * This allows for simple backtrace implementations, but imposes a small
  runtime overhead. For this to work, all code that might be part of
  backtrace must be built with the -mbackchain GCC option.
  
   * The -fno-omit-framepointer switch is neither necessary nor helpful in this 
context.
    Having a (soft/simulated) frame pointer does not improve backtraces at all 
on IBM Z.
  
   * However, forcing a frame pointer via the -fno-omit-frame-pointer
  option negatively affects performance for multiple reasons: extra
  prologue/epilogue overhead and fewer shrink-wrapping opportunities.
  
   * Given -fno-omit-frame-pointer does not provide any improvements
  (backtraces or otherwise) and only reduces performance, -fno-omit-frame-
  pointers should not be used on IBM Z.
  
   * So we are facing here a performance penalty without any gain - on
  this particular platform.
  
   * And sometimes (in rare cases like LP#2060108) frame pointers may even
  lead to failed builds.
  
  [ Test Plan ]
  
   * Due to the above description of the impact and rationale,
     this pragmatic approach for testing is given:
  
   * Build the affected packages where frame-pointers should be reverted
     using the updated dpkg package (that incl. the modified build defaults)
     on (or for) this particular platform.
  
   * Now frame-pointer usage be checked in the following different ways:
  
   * 1) For the ease of use (and thanks to Julian Klode), there is this python
        test script available that allows to verify a binary in regard to
        frame pointers:
        https://gist.github.com/julian-klode/85e55553f85c410a1b856a93dce77208
-       
https://gist.githubusercontent.com/julian-klode/85e55553f85c410a1b856a93dce77208/raw/488b8509e6f23fe48f917961fe711b285dcb2e28/dwprod.py
+       
https://gist.githubusercontent.com/julian-klode/85e55553f85c410a1b856a93dce77208/raw/488b8509e6f23fe48f917961fe711b285dcb2e28/dwprod.py
+       requires python3-pyelftools
  
   * 2) Another more manual way is to verify based on debug symbols like this:
        - find and install the ddeb package
        - maybe extract the  file (e.g. unzstd)
        - use 'readelf -wi'
        - and grep for 'DW_AT_produce' (build options)
        - look for entries regarding frame-pointer
        The output may look similar to this:
        readelf -wi 
./usr/lib/debug/lib/modules/6.8.0-38-generic/kernel/arch/s390/crypto/aes_s390.ko
 | grep DW_AT_produce
            <23>   DW_AT_producer    : (indirect string, offset: 0x7d): GNU AS 
2.42
            <129>   DW_AT_producer    : (indirect string, offset: 0x3eef): GNU 
C11 13.2.0 -m64 -mpacked-stack -mbackchain -msoft-float -march=z13 -mtune=z16 
-mindirect-branch=thunk-extern -mfunction-return=thunk-extern 
-mindirect-branch-table -mrecord-mcount -mnop-mcount -mfentry -mzarch -g 
-gdwarf-5 -O2 -std=gnu11 -p -fshort-wchar -funsigned-char -fno-common 
-fno-strict-aliasing -fno-asynchronous-unwind-tables 
-fno-delete-null-pointer-checks -fno-allow-store-data-races 
-fno-stack-protector -ftrivial-auto-var-init=zero -fno-stack-clash-protection 
-fzero-call-used-regs=used-gpr -fno-inline-functions-called-once 
-falign-functions=8 -fstrict-flex-arrays=3 -fno-strict-overflow 
-fstack-check=no -fconserve-stack -fsanitize=bounds-strict -fsanitize=shift 
-fsanitize=bool -fsanitize=enum -fPIC
  
   * 3) And maybe watching the build messages / log for the build options that
        were used (but that is probably not sufficient - it's better to inspect
        the output.)
  
  [ Where problems could occur ]
  
   * The dpkg modifications could have been done erroneously.
     A dpkg test build and/or builds of other packages with the modified dpkg
     version in place would show this.
  
   * The settings in dpkg might be overwritten by other settings/packages.
     Tests like above, would show this.
  
   * One may think there could be issues in an environment where some packages
     have frame-pointer enabled and other don't.
     This is fine and was confirmed by IBM toolchain team and ours
     (as well as by a longer running <weeks> test system,
      with FP disabled in kernel, that showed no issues - like expected).
  
  [ Other Info ]
  
   * These changes were implemented during the opening of the oracular series.
     The very same changes are backported to 24.04 LTS.
  
   * These only affect the ppc64el and s390x architectures,
     for other architectures it's a no-change upload.
  
   * We didn't see any fallout for these changes during the development
     on the oracular series, and therefore don't expect any fallout or
     regressions in 24.04 LTS either.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2064538

Title:
  Revert back frame pointers for s390x (remove -fno-omit-frame-pointer
  but use -mbackchain)

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-z-systems/+bug/2064538/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to