Project

General

Profile

Bug #4395

Internal compiler errors of Aspectator

Added by Evgeny Novikov over 8 years ago. Updated over 8 years ago.

Status:
Closed
Priority:
Normal
Category:
-
Start date:
07/27/2013
Due date:
% Done:

0%

Estimated time:
Detected in build:
svn
Platform:
Published in build:
eec536a

Description

This is a collection of such the issues.


Files

segfault.tar.xz (1.68 MB) segfault.tar.xz Ilya Shchepetkov, 07/31/2013 05:00 PM

Related issues

Blocked by C Instrumentation Framework - Bug #4390: arch/x86/kvm/emulate.c.prepared:912:1: internal compiler error: Segmentation faultClosedEvgeny Novikov07/26/2013

Actions
Blocked by C Instrumentation Framework - Bug #4396: drivers/usb/gadget/f_fs.c:2003:1: internal compiler error: Segmentation faultClosedEvgeny Novikov07/27/2013

Actions
Blocked by C Instrumentation Framework - Bug #4397: name_distr.c.prepared:61:0: ... aspectator: internal compiler error: Segmentation fault (program cc1)ClosedEvgeny Novikov07/27/2013

Actions
#1

Updated by Ilya Shchepetkov over 8 years ago

19 more examples of such errors are in the attached file.

#2

Updated by Evgeny Novikov over 8 years ago

For examples from the attached file on commit df9bb68 of the bug-4395 branch (aspectator submodule) I have:
  • arp_tables.c.macroinstrumented - Ok.
  • cfg80211.c.macroinstrumented - Ok.
  • debug.c.macroinstrumented - Ok.
  • debugfs.c.macroinstrumented - Ok.
  • do_balan.c.macroinstrumented - #4399 (this is a C file of fs/reiserfs/reiserfs.ko module).
  • emulate.c.macroinstrumented - #4398.
  • fs.c.macroinstrumented - strange thing instead of:
    static const char objtree[] = OBJTREE;
    

    there is:
    static const char objtree[] = /home/shchepetkov/1 -3.11-rc1;                                                               
    

    and the same for:
    static const char srctree[] = SRCTREE;
    

    It is interesting but in reference launches almost all modules from the kernel directory aren't verified. So that we didn't see this error before.
  • fw.c.macroinstrumented - Ok.
  • g_ffs.c.macroinstrumented - Ok (it includes f_fs.c that had segfault before #4396).
  • interrupt.c.macroinstrumented - Ok.
  • ip6_tables.c.macroinstrumented - Ok.
  • ip_tables.c.macroinstrumented - Ok.
  • main.c.macroinstrumented - Ok.
  • mmu.c.macroinstrumented - Ok.
  • netdev.c.macroinstrumented - Ok.
  • pcie_bus.c.macroinstrumented - Ok.
  • trace.c.macroinstrumented - Ok.
  • txrx.c.macroinstrumented - Ok.
  • wmi.c.macroinstrumented - Ok.
#3

Updated by Evgeny Novikov over 8 years ago

BTW, some other segfaults (like mentioned in #4397) were not found because of Aspectator was killed due to the memory limit. About 640 CC commands of total 8886 were skipped. So I will raise priority of #372.

#4

Updated by Evgeny Novikov over 8 years ago

  • Status changed from Open to Closed
  • Published in build set to eec536a

Fixed in eec536a. CIF submodule was updated in commit:df9ad40 of LDV tools.

After the fix I gained quite positive changes in verification results for the launch on the whole kernel against 32_7a rule model:
  • 23 Unknown → Safe.
  • 2 Unknown → Unsafe. Both looks to be False positives. So I would like to wait until we will get results during following reference launches on 3.12-rc1 and just then mark up them.

And there is no more Segmentation faults of Aspectator!

Also available in: Atom PDF