|
From: <joh...@gm...> - 2006-09-12 10:23:26
|
Hi, I am a KDE developer. When I run any KDE4 program (even with a totally clean build) it segment faults. When I run this in GDB it tells me that the crash is in QImage/QIcon. I tried to valgrind one of the programs (all kde 4 programs crash) valgrind crashed. I recently changed by video card to an old Matrox G400 video card. I run X using the vesa driver. I have no XV. I think that this might be relevant. Valgrind gave the following errors: --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:24 [hundreds and hundreds of these] --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:32 [just a few of these] --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:48 [ couple of these] --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:7 [ 1 of these] Then assertion failure: valgrind: m_debuginfo/storage.c:311 (vgModuleLocal_addDiCfSI): Assertion 'cfsi->len > 0 && cfsi->len < 2000000' failed. ==11381== at 0x3801750E: report_and_quit (m_libcassert.c:136) ==11381== by 0x38017871: vgPlain_assert_fail (m_libcassert.c:200) ==11381== by 0x380588CC: vgModuleLocal_addDiCfSI (storage.c:311) ==11381== by 0x38062180: run_CF_instructions (readdwarf.c:2338) ==11381== by 0x38062A9A: vgModuleLocal_read_callframe_info_dwarf2 (readdwarf.c:2699) ==11381== by 0x3805ADE7: vgModuleLocal_read_elf_debug_info (readelf.c:1206) ==11381== by 0x3803477C: vgPlain_di_notify_mmap (debuginfo.c:177) ==11381== by 0x380434F5: vgModuleLocal_generic_PRE_sys_mmap (syswrap-generic.c:1874) ==11381== by 0x380538C5: vgSysWrap_amd64_linux_sys_mmap_before (syswrap-amd64-linux.c:944) ==11381== by 0x38047AC3: vgPlain_client_syscall (syswrap-main.c:719) ==11381== by 0x38036DEF: vgPlain_scheduler (scheduler.c:721) ==11381== by 0x38050C59: run_a_thread_NORETURN (syswrap-linux.c:87) sched status: running_tid=1 Thread 1: status = VgTs_Runnable ==11381== at 0x40146BA: (within /lib/ld-2.4.so) ==11381== by 0x4006309: (within /lib/ld-2.4.so) ==11381== by 0x40081D5: (within /lib/ld-2.4.so) ==11381== by 0x400B45C: (within /lib/ld-2.4.so) ==11381== by 0x400D1E5: (within /lib/ld-2.4.so) ==11381== by 0x400BB2B: (within /lib/ld-2.4.so) ==11381== by 0x40032AF: (within /lib/ld-2.4.so) ==11381== by 0x40137DA: (within /lib/ld-2.4.so) ==11381== by 0x40012C5: (within /lib/ld-2.4.so) ==11381== by 0x4000A77: (within /lib/ld-2.4.so) Btw, I just want to say that I really love using valgrind. Thank you all for this really wonderful tool JohnFlux |
|
From: Julian S. <js...@ac...> - 2006-09-12 10:41:47
|
Hmm, ok, we'll need a bunch more details. What you have is an assertion failure in Valgrind, not a segfault. What version of Valgrind is this? I assume you are running an AMD64 machine. What OS and version is it? What version of gcc/g++ are you using? What gcc flags did you build KDE with? Valgrind has asserted whilst reading debugging info from one of the shared objects making up your KDE build. It would be helpful if you could re-run Valgrind on your app, but with the additional flag "-v", and then send the entire log of what it prints out. J On Tuesday 12 September 2006 11:23, joh...@gm... wrote: > Hi, > I am a KDE developer. When I run any KDE4 program (even with a > totally clean build) it segment faults. When I run this in GDB it > tells me that the crash is in QImage/QIcon. > I tried to valgrind one of the programs (all kde 4 programs crash) > valgrind crashed. > > I recently changed by video card to an old Matrox G400 video card. > I run X using the vesa driver. I have no XV. I think that this might > be relevant. > > Valgrind gave the following errors: > > --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:24 > [hundreds and hundreds of these] > --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:32 [just a > few of these] > --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:48 [ couple of > these] --11381-- DWARF2 CFI reader: unhandled CFI instruction 0:7 [ 1 of > these] > > Then assertion failure: > > valgrind: m_debuginfo/storage.c:311 (vgModuleLocal_addDiCfSI): > Assertion 'cfsi->len > 0 && cfsi->len < 2000000' failed. > ==11381== at 0x3801750E: report_and_quit (m_libcassert.c:136) > ==11381== by 0x38017871: vgPlain_assert_fail (m_libcassert.c:200) > ==11381== by 0x380588CC: vgModuleLocal_addDiCfSI (storage.c:311) > ==11381== by 0x38062180: run_CF_instructions (readdwarf.c:2338) > ==11381== by 0x38062A9A: vgModuleLocal_read_callframe_info_dwarf2 > (readdwarf.c:2699) > ==11381== by 0x3805ADE7: vgModuleLocal_read_elf_debug_info > (readelf.c:1206) ==11381== by 0x3803477C: vgPlain_di_notify_mmap > (debuginfo.c:177) ==11381== by 0x380434F5: > vgModuleLocal_generic_PRE_sys_mmap > (syswrap-generic.c:1874) > ==11381== by 0x380538C5: vgSysWrap_amd64_linux_sys_mmap_before > (syswrap-amd64-linux.c:944) > ==11381== by 0x38047AC3: vgPlain_client_syscall (syswrap-main.c:719) > ==11381== by 0x38036DEF: vgPlain_scheduler (scheduler.c:721) > ==11381== by 0x38050C59: run_a_thread_NORETURN (syswrap-linux.c:87) > > sched status: > running_tid=1 > > Thread 1: status = VgTs_Runnable > ==11381== at 0x40146BA: (within /lib/ld-2.4.so) > ==11381== by 0x4006309: (within /lib/ld-2.4.so) > ==11381== by 0x40081D5: (within /lib/ld-2.4.so) > ==11381== by 0x400B45C: (within /lib/ld-2.4.so) > ==11381== by 0x400D1E5: (within /lib/ld-2.4.so) > ==11381== by 0x400BB2B: (within /lib/ld-2.4.so) > ==11381== by 0x40032AF: (within /lib/ld-2.4.so) > ==11381== by 0x40137DA: (within /lib/ld-2.4.so) > ==11381== by 0x40012C5: (within /lib/ld-2.4.so) > ==11381== by 0x4000A77: (within /lib/ld-2.4.so) > > > Btw, I just want to say that I really love using valgrind. Thank you > all for this really wonderful tool > > > JohnFlux > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job > easier Download IBM WebSphere Application Server v.1.0.1 based on Apache > Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Valgrind-users mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-users |
|
From: Julian S. <js...@ac...> - 2006-09-12 23:30:00
|
Ok, it bombed reading info from /home/kdedev/src/kde/build/qt-copy/lib/libQtGui.so.4.2.0 So can you send me that file? J On Tuesday 12 September 2006 18:56, joh...@gm... wrote: > Hi Julian, > > Thanks for responding so quickly. I've tried to answer the questions: > > What version of Valgrind is this? > > $ valgrind --version > valgrind-3.2.0-Debian > > > I assume you are running an AMD64 machine. What OS and version is it? > > kubuntu, edgy (which is the newest unstable) > It is the x86_64 version indeed, so 64bit version of everything > > > What version of gcc/g++ are you using? > > $ gcc --version > gcc (GCC) 4.1.2 20060903 (prerelease) (Ubuntu 4.1.1-13ubuntu1) > > > What gcc flags did you build KDE with? > > I don't know how to check this.. I use cmake sorry. My configure flags > are: > > configure-flags --enable-sendfile --enable-mitshm > > > Valgrind has asserted whilst reading debugging info from one of the > > shared objects making up your KDE build. It would be helpful if you > > could re-run Valgrind on your app, but with the additional flag "-v", > > and then send the entire log of what it prints out. > > Okay but I've had to zip it. It's 190KB even zipped so I hope I don't > annoy everyone. > > John |
|
From: Julian S. <js...@ac...> - 2006-09-13 10:46:42
|
Using the libQtGui.so.4.2.0 you sent, I can reproduce the failure using 3.2.0. Whereas the upcoming 3.2.1 does not crash, which is good. I'm more concerned about zillions of these --3382-- DWARF2 CFI reader: unhandled CFI instruction 0:24 The DWARF3 Draft 8 spec doesn't what instruction number 24 is, only 0 through 23. Tom, any ideas? J On Wednesday 13 September 2006 01:19, you wrote: > Thank you for helping me! > > I just want to mention that I can run the QT4 demo with no problems, > and that I have rebuilt from scratch qt4 twice with no change. None > of the kde4 programs worked. > > John > > On 13/09/06, Julian Seward <js...@ac...> wrote: > > Ok, it bombed reading info from > > /home/kdedev/src/kde/build/qt-copy/lib/libQtGui.so.4.2.0 > > > > So can you send me that file? > > > > J > > > > On Tuesday 12 September 2006 18:56, joh...@gm... wrote: > > > Hi Julian, > > > > > > Thanks for responding so quickly. I've tried to answer the questions: > > > > What version of Valgrind is this? > > > > > > $ valgrind --version > > > valgrind-3.2.0-Debian > > > > > > > I assume you are running an AMD64 machine. What OS and version is > > > > it? > > > > > > kubuntu, edgy (which is the newest unstable) > > > It is the x86_64 version indeed, so 64bit version of everything > > > > > > > What version of gcc/g++ are you using? > > > > > > $ gcc --version > > > gcc (GCC) 4.1.2 20060903 (prerelease) (Ubuntu 4.1.1-13ubuntu1) > > > > > > > What gcc flags did you build KDE with? > > > > > > I don't know how to check this.. I use cmake sorry. My configure flags > > > are: > > > > > > configure-flags --enable-sendfile --enable-mitshm > > > > > > > Valgrind has asserted whilst reading debugging info from one of the > > > > shared objects making up your KDE build. It would be helpful if you > > > > could re-run Valgrind on your app, but with the additional flag "-v", > > > > and then send the entire log of what it prints out. > > > > > > Okay but I've had to zip it. It's 190KB even zipped so I hope I don't > > > annoy everyone. > > > > > > John |
|
From: Tom H. <to...@co...> - 2006-09-13 10:59:21
|
In message <200...@ac...>
Julian Seward <js...@ac...> wrote:
> Using the libQtGui.so.4.2.0 you sent, I can reproduce the failure using
> 3.2.0. Whereas the upcoming 3.2.1 does not crash, which is good.
>
> I'm more concerned about zillions of these
> --3382-- DWARF2 CFI reader: unhandled CFI instruction 0:24
>
> The DWARF3 Draft 8 spec doesn't what instruction number 24 is,
> only 0 through 23. Tom, any ideas?
Not really... The final version of DWARF3 is out now (get it from
dwarf.freestandards.org) and that only seems to define 0 to 22.
I guess the next question is, what versions of gcc and binutils were
used to build that library?
Tom
--
Tom Hughes (to...@co...)
http://www.compton.nu/
|
|
From: <joh...@gm...> - 2006-09-13 13:40:15
|
Hi Tom, $ apt-cache policy binutils binutils: Installed: 2.17-1ubuntu1 $ gcc --version gcc (GCC) 4.1.2 20060903 (prerelease) (Ubuntu 4.1.1-13ubuntu1) But i also have gcc 4.0 and gcc 3.4 installed. cmake should be using the 4.1.2 version right? I'm not sure how to check sorry. John On 13/09/06, Tom Hughes <to...@co...> wrote: > In message <200...@ac...> > Julian Seward <js...@ac...> wrote: > > > Using the libQtGui.so.4.2.0 you sent, I can reproduce the failure using > > 3.2.0. Whereas the upcoming 3.2.1 does not crash, which is good. > > > > I'm more concerned about zillions of these > > --3382-- DWARF2 CFI reader: unhandled CFI instruction 0:24 > > > > The DWARF3 Draft 8 spec doesn't what instruction number 24 is, > > only 0 through 23. Tom, any ideas? > > Not really... The final version of DWARF3 is out now (get it from > dwarf.freestandards.org) and that only seems to define 0 to 22. > > I guess the next question is, what versions of gcc and binutils were > used to build that library? > > Tom > > -- > Tom Hughes (to...@co...) > http://www.compton.nu/ > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job > easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Valgrind-users mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-users > |
|
From: Tom H. <to...@co...> - 2006-09-14 09:21:52
|
In message <43d...@ma...>
joh...@gm... wrote:
> Hi Tom,
> $ apt-cache policy binutils
> binutils:
> Installed: 2.17-1ubuntu1
> $ gcc --version
> gcc (GCC) 4.1.2 20060903 (prerelease) (Ubuntu 4.1.1-13ubuntu1)
>
>
> But i also have gcc 4.0 and gcc 3.4 installed. cmake should be using
> the 4.1.2 version right? I'm not sure how to check sorry.
I've had a look at the copy of the library you sent me now and I have
to say that it does look seriously bogus.
Trying to read it with libdwarf (via dwarfdump) also fails although
for a different reason - libddwarf's main complaint is that the
instructions in the FDE go beyond the length declared in the header.
If I disable that error in libdwarf then it appears to be able to
decode all the instructions without finding any unexpected opcodes
but I'm not 100% convinced that it is coping.
Equally the length mismatches may be throwing valgrind off, although
it doesn't look like it should from a quick look at the code.
Either way, it looks like the eh_frame information in the library is
horribly broken, which is most likely gcc's fault.
Tom
--
Tom Hughes (to...@co...)
http://www.compton.nu/
|