|
From: Bart V. A. <bar...@gm...> - 2007-11-11 13:25:50
|
Hello, Is any other Valgrind developer using openSUSE 10.3 ? Many regression tests that run fine during the nightly build fail on openSUSE 10.3. I don't know whether this is due to changes in Valgrind or due to changes in gcc, binutils or glibc. $ svn update Fetching external item into 'VEX' External at revision 1793. At revision 7146. $ ./autogen.sh && ./configure && make -s && make -s check && make -s regtest ... == 350 tests, 66 stderr failures, 50 stdout failures, 0 post failures == ... $ cat memcheck/tests/writev.stderr.diff 3c3 < at 0x........: writev (in /...libc...) --- > at 0x........: (within /...libc...) 9c9 < at 0x........: writev (in /...libc...) --- > at 0x........: (within /...libc...) 15c15 < at 0x........: readv (in /...libc...) --- > at 0x........: (within /...libc...) |
|
From: Julian S. <js...@ac...> - 2007-11-11 20:16:17
|
> Is any other Valgrind developer using openSUSE 10.3 ? Many regression > tests that run fine during the nightly build fail on openSUSE 10.3. > == 350 tests, 66 stderr failures, 50 stdout failures, 0 post failures == On openSUSE 10.3 running on x86_64 I get this: == 350 tests, 12 stderr failures, 2 stdout failures, 0 post failures == memcheck/tests/pointer-trace (stderr) memcheck/tests/vcpu_fnfns (stdout) memcheck/tests/writev (stderr) memcheck/tests/x86/scalar (stderr) memcheck/tests/x86/scalar_exit_group (stderr) memcheck/tests/x86/scalar_supp (stderr) massif/tests/culling1 (stderr) massif/tests/culling2 (stderr) massif/tests/deep-C (stderr) massif/tests/peak2 (stderr) massif/tests/realloc (stderr) none/tests/mremap (stderr) none/tests/mremap2 (stdout) helgrind/tests/tc22_exit_w_lock (stderr) J |
|
From: Nicholas N. <nj...@cs...> - 2007-11-11 21:22:24
|
On Sun, 11 Nov 2007, Bart Van Assche wrote: > Is any other Valgrind developer using openSUSE 10.3 ? Many regression > tests that run fine during the nightly build fail on openSUSE 10.3. I > don't know whether this is due to changes in Valgrind or due to > changes in gcc, binutils or glibc. > > > $ svn update > > Fetching external item into 'VEX' > External at revision 1793. > > At revision 7146. > > $ ./autogen.sh && ./configure && make -s && make -s check && make -s regtest > ... > == 350 tests, 66 stderr failures, 50 stdout failures, 0 post failures == > ... > > $ cat memcheck/tests/writev.stderr.diff > 3c3 > < at 0x........: writev (in /...libc...) > --- >> at 0x........: (within /...libc...) > 9c9 > < at 0x........: writev (in /...libc...) > --- >> at 0x........: (within /...libc...) > 15c15 > < at 0x........: readv (in /...libc...) > --- >> at 0x........: (within /...libc...) For that test, at least, there are three .exp* files, and the .exp2 one looks like it should match. What does writev.stderr.diff2 look like? Nick |
|
From: Bart V. A. <bar...@gm...> - 2007-11-12 18:59:29
|
On Nov 11, 2007 10:22 PM, Nicholas Nethercote <nj...@cs...> wrote: > On Sun, 11 Nov 2007, Bart Van Assche wrote: > > > Is any other Valgrind developer using openSUSE 10.3 ? Many regression > > tests that run fine during the nightly build fail on openSUSE 10.3. I > > don't know whether this is due to changes in Valgrind or due to > > changes in gcc, binutils or glibc. > > > > > > $ svn update > > > > Fetching external item into 'VEX' > > External at revision 1793. > > > > At revision 7146. > > > > $ ./autogen.sh && ./configure && make -s && make -s check && make -s > regtest > > ... > > == 350 tests, 66 stderr failures, 50 stdout failures, 0 post failures == > > ... > > > > $ cat memcheck/tests/writev.stderr.diff > > 3c3 > > < at 0x........: writev (in /...libc...) > > --- > >> at 0x........: (within /...libc...) > > 9c9 > > < at 0x........: writev (in /...libc...) > > --- > >> at 0x........: (within /...libc...) > > 15c15 > > < at 0x........: readv (in /...libc...) > > --- > >> at 0x........: (within /...libc...) > > For that test, at least, there are three .exp* files, and the .exp2 one > looks like it should match. What does writev.stderr.diff2 look like? > > Nick > Below you can find the contents of all writev.*.diff* files on my system (for SVN revisions 7153 / 1793). I'm really puzzled by this -- any help in solving this is certainly appreciated. head -n 999 memcheck/tests/writev.*.diff* ==> memcheck/tests/writev.stderr.diff <== 3c3 < at 0x........: writev (in /...libc...) --- > at 0x........: (within /...libc...) 9c9 < at 0x........: writev (in /...libc...) --- > at 0x........: (within /...libc...) 15c15 < at 0x........: readv (in /...libc...) --- > at 0x........: (within /...libc...) ==> memcheck/tests/writev.stderr.diff2 <== 15c15 < at 0x........: readv (in /...libc...) --- > at 0x........: (within /...libc...) ==> memcheck/tests/writev.stderr.diff3 <== 3c3 < at 0x........: do_writev (in /...libc...) --- > at 0x........: (within /...libc...) 9c9 < at 0x........: do_writev (in /...libc...) --- > at 0x........: (within /...libc...) 15c15 < at 0x........: do_readv (in /...libc...) --- > at 0x........: (within /...libc...) |
|
From: Nicholas N. <nj...@cs...> - 2007-11-12 21:14:33
|
On Mon, 12 Nov 2007, Bart Van Assche wrote: > Below you can find the contents of all writev.*.diff* files on my system > (for SVN revisions 7153 / 1793). I'm really puzzled by this -- any help in > solving this is certainly appreciated. This is just one of those annoying cases where different systems give slightly different stack traces. We're already handling three different cases, but your system is slightly different again. Probably the best thing to do is add a .exp4 file. The other test cases that failed -- did they fail for similar reasons? How many of them failed? Nick > head -n 999 memcheck/tests/writev.*.diff* > ==> memcheck/tests/writev.stderr.diff <== > 3c3 > < at 0x........: writev (in /...libc...) > --- >> at 0x........: (within /...libc...) > 9c9 > < at 0x........: writev (in /...libc...) > --- >> at 0x........: (within /...libc...) > 15c15 > < at 0x........: readv (in /...libc...) > --- >> at 0x........: (within /...libc...) > > ==> memcheck/tests/writev.stderr.diff2 <== > 15c15 > < at 0x........: readv (in /...libc...) > --- >> at 0x........: (within /...libc...) > > ==> memcheck/tests/writev.stderr.diff3 <== > 3c3 > < at 0x........: do_writev (in /...libc...) > --- >> at 0x........: (within /...libc...) > 9c9 > < at 0x........: do_writev (in /...libc...) > --- >> at 0x........: (within /...libc...) > 15c15 > < at 0x........: do_readv (in /...libc...) > --- >> at 0x........: (within /...libc...) > |
|
From: Julian S. <js...@ac...> - 2007-11-13 00:02:43
|
On Monday 12 November 2007 22:14, Nicholas Nethercote wrote: > On Mon, 12 Nov 2007, Bart Van Assche wrote: > > Below you can find the contents of all writev.*.diff* files on my system > > (for SVN revisions 7153 / 1793). I'm really puzzled by this -- any help > > in solving this is certainly appreciated. > > This is just one of those annoying cases where different systems give > slightly different stack traces. We're already handling three different > cases, but your system is slightly different again. Probably the best > thing to do is add a .exp4 file. This one has failed for ages for me on openSUSE 10.2. Perhaps some of the existing .exp files can never happen. Bart, what is the total list of failed tests? I don't think memcheck/tests/writev deserves special attention. One reason for mass failure of tests on a 64-bit machine is the absence of a fully working 32-bit build environment; is that the case? J |
|
From: Bart V. A. <bar...@gm...> - 2007-11-16 17:59:38
|
On Nov 13, 2007 1:01 AM, Julian Seward <js...@ac...> wrote: > > This one has failed for ages for me on openSUSE 10.2. Perhaps some of > the existing .exp files can never happen. > > Bart, what is the total list of failed tests? I don't think > memcheck/tests/writev deserves special attention. One reason for > mass failure of tests on a 64-bit machine is the absence of a > fully working 32-bit build environment; is that the case? Thanks for the hint. After having ensured that I can again build 32-bit executables, the regression tests now work fine: # make -s distclean && svn update && ./autogen.sh && ./configure && make -s && make -s check regtest ... == 351 tests, 6 stderr failures, 2 stdout failures, 0 post failures == memcheck/tests/pointer-trace (stderr) memcheck/tests/vcpu_fnfns (stdout) memcheck/tests/writev (stderr) memcheck/tests/x86/scalar (stderr) memcheck/tests/x86/scalar_exit_group (stderr) memcheck/tests/x86/scalar_supp (stderr) none/tests/mremap (stderr) none/tests/mremap2 (stdout) Regards, Bart Van Assche. |