7 Replies Latest reply on Mar 28, 2006 6:35 PM by 807575

    Segmentation fault (core dumped)

    807575
      Hello all
      i have strange strange problem when im running application i compiled from source
      ( apache server in this case but it apache forum i dont geting any info ) when i run it after while it crashed and gives me core dump
      now its ok in normal situation i will load the core dump and see some hints about the problem
      but in this case i can get any clues... here is what i get when i load the core in dbx
      can some one please point me and give me some clues about the error i dont know from where it comes
      and what function causing it from this information .

      t@8 (l@8) terminated by signal SEGV (no mapping at the fault address)
      0xffffffffffffffff: <bad address 0xffffffffffffffff>
      (dbx) where
      current thread: t@8
      [1] 0xfeecb824(0x12be44, 0xfc57bf1c, 0x4, 0x12bd30, 0x0, 0x0), at 0xfeecb823
      [2] 0xfe25ab90(0x12bd30, 0xfc57bf80, 0x0, 0xfffffff8, 0x0, 0xfc57bfa9), at 0xfe25ab8f
      [3] 0xfeec5010(0x12bd54, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfeec500f
      (dbx) threads
      t@1 a l@1 ?() LWP suspended in elf_new_lm()
      t@3 b l@3 ?() LWP suspended in doorreturn()
      o> t@8 a l@8 ?() signal SIGSEGV in ?()
      t@9 b l@9 ?() LWP suspended in _poll()
      (dbx)

      Thanks allot
        • 1. Re: Segmentation fault (core dumped)
          807575
          ok just little aadon the same scenario but bit different error :
          @9 (l@9) signal SEGV (no mapping at the fault address) in (unknown) at 0xfeb1c498
          0xfeb1c498: <bad address 0xfeb1c498>

          (dbx) where

          =>[1] 0xfeb1c498(0x124f80, 0xfecbb488, 0xfecce951, 0x4b0, 0x0, 0x0), at 0xfeb1c497

          [2] std::moneypunct<char,false>::pos_format(0x12a538, 0xfecf6cb4, 0xfe981f38, 0x0, 0x20, 0xfe981f38), at 0xfe982a40

          [3] 0xfeb289b0(0xe1bc8, 0x0, 0xfeb28894, 0xfeb28894, 0xfeb28894, 0xe1bc8), at 0xfeb289af

          [4] 0xfeb53930(0xe1bc8, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfeb5392f



          how can i track down who is calling the std::moneypunct<char,false>::pos_format.... function
          • 2. Re: Segmentation fault (core dumped)
            807575
            Apparently you are trying to debug a stripped executable, since the only visible symbols are the external references int external libraries (like std::moneypunct in libCstd.so.1).

            Re-build the program using -g and don't strip anything. You will then be able to see what is going on by using the debugger.
            • 3. Re: Segmentation fault (core dumped)
              807575
              Hello and thanks for the fast reply !!
              im really polling my hair here .. i did compile my application with -g , this for sure
              because when i debug my application from its entry point i can see the source code , define break points and so on . my case here is :
              after i compiled the application ( apache server in my case ) and i run it , it crashed and gives me core dump that .
              1. i have no idea who cause it i have no information in the debugger , the information i see in the stack dont gives me any clues.
              2. is there any tip or trick to catch this crash ?
              3. how can i know if the executable if striped or not?
              thanks allot
              • 4. Re: Segmentation fault (core dumped)
                paul_floyd
                1. i have no idea who cause it i have no information
                in the debugger , the information i see in the stack
                dont gives me any clues.
                None at all? It's not unusual that the very top of the stack will be in library code that does not have any symbolic information. You should still be able to see your code a few levels below that.
                2. is there any tip or trick to catch this crash ?
                Can you run the app in the debugger?
                3. how can i know if the executable if striped or
                not?
                Use 'file'

                Paul
                • 5. Re: Segmentation fault (core dumped)
                  807575
                  Hello and thanks for the fast reply
                  so .. the continue saga of it .. I removed some dynamic links the apache defiantly don�t use and pointed directly to the one its use
                  and now when I load the apache server throw the Sunstudio and run it from there it is running fine without crashes,
                  but when I try to run the apache as stand alone outside the debugger I getting again core dump in this form:

                  t@9 (l@9) terminated by signal ILL (illegal opcode)
                  0xfeb1be78: J453_FieldInfo+0x0004: unimp 0x0
                  (dbx) where
                  current thread: t@9
                  [1] 0xfeb1be78(0x124f80, 0xfecbae68, 0xfecce331, 0x4b0, 0x0, 0x0), at 0xfeb1be77
                  [2] OX509ParseCertificate(0x12a558, 0xfecf659c, 0xfe981918, 0x0, 0x20, 0xfe981918), at 0xfe982420
                  [3] 0xfeb28390(0xe1bc8, 0x0, 0xfeb28274, 0xfeb28274, 0xfeb28274, 0xe1bc8), at 0xfeb2838f
                  [4] 0xfeb53310(0xe1bc8, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfeb5330f
                  (dbx) threads
                  t@1 a l@1 ?() LWP suspended in soaccept()
                  t@3 b l@3 ?() LWP suspended in doorreturn()
                  o> t@9 b l@9 ?() signal SIGILL in ?()
                  t@12 a l@12 start()   sleep on 0x1f0478  in  _lwp_park()
                  t@13 b l@13 door_create_func() LWP suspended in doorreturn()
                  t@14 a l@14 start()   LWP suspended in  poll()
                  t@15 a l@15 start()   sleep on 0x1fa6f0  in  _lwp_park()
                  t@16 a l@16 start()   LWP suspended in  poll()
                  t@17 a l@17 start()   LWP suspended in  poll()
                  t@18 a l@18 start()   LWP suspended in  poll()
                  t@19 b l@19 Execute() LWP suspended in _poll()
                  (dbx)

                  whats is going on here ?
                  I cant understand, and worse I have no additional info only this in the stack. please help
                  • 6. Re: Segmentation fault (core dumped)
                    807575
                    so .. the continue saga of it .. I removed some
                    dynamic links the apache defiantly don�t use and
                    pointed directly to the one its use
                    and now when I load the apache server throw the
                    Sunstudio and run it from there it is running fine
                    without crashes,
                    Can you explain what you mean by removing "some dynamic links the apache defiantly don�t use"? What exactly did you do?

                    That is, did you remove dynamic links in the file system so that libraries formerly listed in a directory were no longer there? Or do you mean that you removed -l or -L options on the CC or ld command line to eliminate dynamic libraries from the link?
                    • 7. Re: Segmentation fault (core dumped)
                      807575
                      sorry what i meant was , i removed shred objects the apache server dont need for its funconalty
                      and i only left the one he use . you see the apache has bench of *.so files he loads when he start up as plugins but he dont need them all only need the one you define him

                      home i made my self clear now , thanks for the help