Quantcast

Strange crash in numpy.linalg.lstsq.

classic Classic list List threaded Threaded
13 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Strange crash in numpy.linalg.lstsq.

Jonathan Taylor-3
Hi,

I am getting a strange crash in numpy.linalg.lstsq.  I have put the code that causes the crash along with two data files on my website at:

http://www.cs.toronto.edu/~jtaylor/crash/

I would be interested to know if this bug can be duplicated and/or if anyone has any suggestions as to why:

import numpy as np
A = np.load('A.npy')
b = np.load('b.npy')
rc = np.linalg.lstsq(A,b)

produces:

*** glibc detected *** /usr/bin/python: free(): invalid next size (normal): 0x091793c0 ***
======= Backtrace: =========
/lib/tls/i686/cmov/libc.so.6[0xb7dc7a85]
/lib/tls/i686/cmov/libc.so.6(cfree+0x90)[0xb7dcb4f0]
/u/jtaylor/lib/python2.5/site-packages/numpy/core/multiarray.so[0xb795403e]
/usr/bin/python[0x811247a]
/usr/bin/python(PyEval_EvalCodeEx+0x323)[0x80cae33]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalCode+0x57)[0x80cb347]
/usr/bin/python(PyRun_FileExFlags+0xf8)[0x80ea818]
/usr/bin/python[0x80c1f5a]
/usr/bin/python(PyObject_Call+0x27)[0x805cb97]
/usr/bin/python(PyEval_EvalFrameEx+0x4064)[0x80c7e04]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python[0x8113696]
/usr/bin/python(PyObject_Call+0x27)[0x805cb97]
/usr/bin/python(PyEval_EvalFrameEx+0x4064)[0x80c7e04]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python[0x8113696]
/usr/bin/python(PyObject_Call+0x27)[0x805cb97]
/usr/bin/python[0x8062bfb]
/usr/bin/python(PyObject_Call+0x27)[0x805cb97]
/usr/bin/python(PyEval_EvalFrameEx+0x3d07)[0x80c7aa7]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalFrameEx+0x5945)[0x80c96e5]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x6d09)[0x80caaa9]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalFrameEx+0x5945)[0x80c96e5]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
/usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
/usr/bin/python(PyEval_EvalCode+0x57)[0x80cb347]
/usr/bin/python(PyRun_FileExFlags+0xf8)[0x80ea818]
/usr/bin/python(PyRun_SimpleFileExFlags+0x199)[0x80eaab9]
/usr/bin/python(Py_Main+0xa35)[0x8059335]
/usr/bin/python(main+0x22)[0x80587f2]
/lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe0)[0xb7d72450]
/usr/bin/python[0x8058761]
======= Memory map: ========
08048000-08140000 r-xp 00000000 08:06 83501      /usr/bin/python2.5
08140000-08165000 rw-p 000f7000 08:06 83501      /usr/bin/python2.5
08165000-0919a000 rw-p 08165000 00:00 0          [heap]
b5200000-b5221000 rw-p b5200000 00:00 0
b5221000-b5300000 ---p b5221000 00:00 0
b53fc000-b5499000 r-xp 00000000 00:1a 552170     /h/44/jtaylor/lib/python2.5/site-packages/Cython/Compiler/Parsing.so
b5499000-b54a2000 rw-p 0009d000 00:1a 552170     /h/44/jtaylor/lib/python2.5/site-packages/Cython/Compiler/Parsing.so
b54a2000-b5624000 rw-p b54a2000 00:00 0
b5624000-b568f000 r-xp 00000000 00:1a 553542     /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_backend_agg.so
b568f000-b5691000 rw-p 0006a000 00:1a 553542     /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_backend_agg.so
b5691000-b56f6000 r-xp 00000000 08:06 90831      /usr/lib/python2.5/lib-dynload/unicodedata.so
b56f6000-b5705000 rw-p 00065000 08:06 90831      /usr/lib/python2.5/lib-dynload/unicodedata.so
b5705000-b5725000 r-xp 00000000 00:1a 553545     /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_tkagg.so
b5725000-b5726000 rw-p 00020000 00:1a 553545     /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_tkagg.so
b5726000-b5727000 ---p b5726000 00:00 0
b5727000-b5f27000 rwxp b5727000 00:00 0
b5f27000-b5f3e000 r-xp 00000000 08:06 85532      /usr/lib/libxcb.so.1.0.0
b5f3e000-b5f3f000 rw-p 00016000 08:06 85532      /usr/lib/libxcb.so.1.0.0
b5f3f000-b5f53000 r-xp 00000000 08:06 1187870    /lib/tls/i686/cmov/libnsl-2.7.so
b5f53000-b5f55000 rw-p 00013000 08:06 1187870    /lib/tls/i686/cmov/libnsl-2.7.so
b5f55000-b5f57000 rw-p b5f55000 00:00 0
b5f57000-b603b000 r-xp 00000000 08:06 85536      /usr/lib/libX11.so.6.2.0
b603b000-b603e000 rw-p 000e4000 08:06 85536      /usr/lib/libX11.so.6.2.0
b603e000-b60e7000 r-xp 00000000 08:06 85098      /usr/lib/libtcl8.4.so.0
b60e7000-b60f1000 rw-p 000a8000 08:06 85098      /usr/lib/libtcl8.4.so.0
b60f1000-b60f2000 rw-p b60f1000 00:00 0
b60f2000-b61c4000 r-xp 00000000 08:06 85102      /usr/lib/libtk8.4.so.0
b61c4000-b61cf000 rw-p 000d2000 08:06 85102      /usr/lib/libtk8.4.so.0
b61cf000-b61d0000 rw-p b61cf000 00:00 0
b61d0000-b62a8000 r-xp 00000000 08:06 85103      /usr/lib/libBLT.2.4.so.8.4
b62a8000-b62b9000 rw-p 000d8000 08:06 85103      /usr/lib/libBLT.2.4.so.8.4
b62b9000-b62ba000 rw-p b62b9000 00:00 0
b62ba000-b62dc000 r-xp 00000000 08:06 180469     /usr/lib/libpng12.so.0.15.0
b62dc000-b62dd000 rw-p 00022000 08:06 180469     /usr/lib/libpng12.so.0.15.0
b62f5000-b62f6000 rw-p b62f5000 00:00 0
b62f6000-b631d000 r-xp 00000000 00:1a 553544     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_png.so
b631d000-b631e000 rw-p 00027000 00:1a 553544     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_png.so
b631e000-b6367000 r-xp 00000000 00:1a 553543     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_image.so
b6367000-b6369000 rw-p 00049000 00:1a 553543     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_image.so
b6369000-b63d3000 r-xp 00000000 08:06 83795      /usr/lib/libfreetype.so.6.3.16
b63d3000-b63d6000 rw-p 0006a000 08:06 83795      /usr/lib/libfreetype.so.6.3.16
b63d6000-b6424000 r-xp 00000000 00:1a 553535     /h/44/jtaylor/build/matplotlib/lib/matplotlib/ft2font.so
b6424000-b6427000 rw-p 0004e000 00:1a 553535     /h/44/jtaylor/build/matplotlib/lib/matplotlib/ft2font.so
b6427000-b650f000 r-xp 00000000 08:06 88506      /usr/lib/libstdc++.so.6.0.9
b650f000-b6512000 r--p 000e8000 08:06 88506      /usr/lib/libstdc++.so.6.0.9
b6512000-b6514000 rw-p 000eb000 08:06 88506      /usr/lib/libstdc++.so.6.0.9
b6514000-b651a000 rw-p b6514000 00:00 0
b651e000-b6528000 r-xp 00000000 08:06 313979     /usr/lib/python2.5/lib-dynload/_tkinter.so
b6528000-b6529000 rw-p 0000a000 08:06 313979     /usr/lib/python2.5/lib-dynload/_tkinter.so
b6529000-b652d000 r-xp 00000000 08:06 90832      /usr/lib/python2.5/lib-dynload/zlib.so
b652d000-b652e000 rw-p 00004000 08:06 90832      /usr/lib/python2.5/lib-dynload/zlib.so
b652e000-b6532000 r-xp 00000000 00:1a 553538     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_cntr.so
b6532000-b6533000 rw-p 00004000 00:1a 553538     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_cntr.so
b6533000-b6577000 r-xp 00000000 00:1a 553541     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_path.so
b6577000-b6578000 rw-p 00044000 00:1a 553541     /h/44/jtaylor/build/matplotlib/lib/matplotlib/_path.so
b6578000-b6587000 r-xp 00000000 08:06 92419      /usr/lib/python2.5/lib-dynload/datetime.so
b6587000-b658a000 rw-p 0000e000 08:06 92419      /usr/lib/python2.5/lib-dynload/datetime.so
b658a000-b65b9000 r-xp 00000000 00:1a 532925     /h/44/jtaylor/lib/python2.5/site-packages/numpy/random/mtrand.so
b65b9000-b65cb000 rw-p 0002e000 00:1a 532925     /h/44/jtaylor/lib/python2.5/site-packages/numpy/random/mtrand.so
b65cb000-b6923000 r-xp 00000000 08:06 517267     /usr/lib/atlas/libblas.so.3.0
b6923000-b6927000 rw-p 00358000 08:06 517267     /usr/lib/atlas/libblas.so.3.0
b6927000-b6e6f000 r-xp 00000000 08:06 517268     /usr/lib/atlas/liblapack.so.3.0
b6e6f000-b6e72000 rw-p 00548000 08:06 517268     /usr/lib/atlas/liblapack.so.3.0
b6e72000-b6f76000 rw-p b6e72000 00:00 0
b6f76000-b6f7a000 r-xp 00000000 08:06 85530      /usr/lib/libXdmcp.so.6.0.0
b6f7a000-b6f7b000 rw-p 00003000 08:06 85530      /usr/lib/libXdmcp.so.6.0.0
b6f7b000-b6f7f000 r-xp 00000000 08:06 92437      /usr/lib/python2.5/lib-dynload/_csv.so
b6f7f000-b6f81000 rw-p 00004000 08:06 92437      /usr/lib/python2.5/lib-dynload/_csv.so
b6f81000-b6f84000 r-xp 00000000 08:06 92427      /usr/lib/python2.5/lib-dynload/_locale.so
b6f84000-b6f85000 rw-p 00003000 08:06 92427      /usr/lib/python2.5/lib-dynload/_locale.so
b6f85000-b6f8e000 r-xp 00000000 00:1a 533113     /h/44/jtaylor/lib/python2.5/site-packages/numpy/fft/fftpack_lite.so
b6f8e000-b6f8f000 rw-p 00008000 00:1a 533113     /h/44/jtaylor/lib/python2.5/site-packages/numpy/fft/fftpack_lite.so
b6f8f000-b6fad000 r-xp 00000000 00:1a 533028     /h/44/jtaylor/lib/python2.5/site-packages/numpy/core/scalarmath.so
b6fad000-b6fae000 rw-p 0001e000 00:1a 533028     /h/44/jtaylor/lib/python2.Aborted


_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

josef.pktd
On Mon, Aug 17, 2009 at 1:42 PM, Jonathan
Taylor<[hidden email]> wrote:

> Hi,
>
> I am getting a strange crash in numpy.linalg.lstsq.  I have put the code
> that causes the crash along with two data files on my website at:
>
> http://www.cs.toronto.edu/~jtaylor/crash/
>
> I would be interested to know if this bug can be duplicated and/or if anyone
> has any suggestions as to why:
>
> import numpy as np
> A = np.load('A.npy')
> b = np.load('b.npy')
> rc = np.linalg.lstsq(A,b)
>
> produces:
>
> *** glibc detected *** /usr/bin/python: free(): invalid next size (normal):
> 0x091793c0 ***
> ======= Backtrace: =========
> /lib/tls/i686/cmov/libc.so.6[0xb7dc7a85]
> /lib/tls/i686/cmov/libc.so.6(cfree+0x90)[0xb7dcb4f0]
> /u/jtaylor/lib/python2.5/site-packages/numpy/core/multiarray.so[0xb795403e]
> /usr/bin/python[0x811247a]
> /usr/bin/python(PyEval_EvalCodeEx+0x323)[0x80cae33]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalCode+0x57)[0x80cb347]
> /usr/bin/python(PyRun_FileExFlags+0xf8)[0x80ea818]
> /usr/bin/python[0x80c1f5a]
> /usr/bin/python(PyObject_Call+0x27)[0x805cb97]
> /usr/bin/python(PyEval_EvalFrameEx+0x4064)[0x80c7e04]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python[0x8113696]
> /usr/bin/python(PyObject_Call+0x27)[0x805cb97]
> /usr/bin/python(PyEval_EvalFrameEx+0x4064)[0x80c7e04]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python[0x8113696]
> /usr/bin/python(PyObject_Call+0x27)[0x805cb97]
> /usr/bin/python[0x8062bfb]
> /usr/bin/python(PyObject_Call+0x27)[0x805cb97]
> /usr/bin/python(PyEval_EvalFrameEx+0x3d07)[0x80c7aa7]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalFrameEx+0x5945)[0x80c96e5]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x6d09)[0x80caaa9]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalFrameEx+0x5945)[0x80c96e5]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalFrameEx+0x565e)[0x80c93fe]
> /usr/bin/python(PyEval_EvalCodeEx+0x6e7)[0x80cb1f7]
> /usr/bin/python(PyEval_EvalCode+0x57)[0x80cb347]
> /usr/bin/python(PyRun_FileExFlags+0xf8)[0x80ea818]
> /usr/bin/python(PyRun_SimpleFileExFlags+0x199)[0x80eaab9]
> /usr/bin/python(Py_Main+0xa35)[0x8059335]
> /usr/bin/python(main+0x22)[0x80587f2]
> /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe0)[0xb7d72450]
> /usr/bin/python[0x8058761]
> ======= Memory map: ========
> 08048000-08140000 r-xp 00000000 08:06 83501      /usr/bin/python2.5
> 08140000-08165000 rw-p 000f7000 08:06 83501      /usr/bin/python2.5
> 08165000-0919a000 rw-p 08165000 00:00 0          [heap]
> b5200000-b5221000 rw-p b5200000 00:00 0
> b5221000-b5300000 ---p b5221000 00:00 0
> b53fc000-b5499000 r-xp 00000000 00:1a 552170
> /h/44/jtaylor/lib/python2.5/site-packages/Cython/Compiler/Parsing.so
> b5499000-b54a2000 rw-p 0009d000 00:1a 552170
> /h/44/jtaylor/lib/python2.5/site-packages/Cython/Compiler/Parsing.so
> b54a2000-b5624000 rw-p b54a2000 00:00 0
> b5624000-b568f000 r-xp 00000000 00:1a 553542
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_backend_agg.so
> b568f000-b5691000 rw-p 0006a000 00:1a 553542
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_backend_agg.so
> b5691000-b56f6000 r-xp 00000000 08:06 90831
> /usr/lib/python2.5/lib-dynload/unicodedata.so
> b56f6000-b5705000 rw-p 00065000 08:06 90831
> /usr/lib/python2.5/lib-dynload/unicodedata.so
> b5705000-b5725000 r-xp 00000000 00:1a 553545
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_tkagg.so
> b5725000-b5726000 rw-p 00020000 00:1a 553545
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/backends/_tkagg.so
> b5726000-b5727000 ---p b5726000 00:00 0
> b5727000-b5f27000 rwxp b5727000 00:00 0
> b5f27000-b5f3e000 r-xp 00000000 08:06 85532      /usr/lib/libxcb.so.1.0.0
> b5f3e000-b5f3f000 rw-p 00016000 08:06 85532      /usr/lib/libxcb.so.1.0.0
> b5f3f000-b5f53000 r-xp 00000000 08:06 1187870
> /lib/tls/i686/cmov/libnsl-2.7.so
> b5f53000-b5f55000 rw-p 00013000 08:06 1187870
> /lib/tls/i686/cmov/libnsl-2.7.so
> b5f55000-b5f57000 rw-p b5f55000 00:00 0
> b5f57000-b603b000 r-xp 00000000 08:06 85536      /usr/lib/libX11.so.6.2.0
> b603b000-b603e000 rw-p 000e4000 08:06 85536      /usr/lib/libX11.so.6.2.0
> b603e000-b60e7000 r-xp 00000000 08:06 85098      /usr/lib/libtcl8.4.so.0
> b60e7000-b60f1000 rw-p 000a8000 08:06 85098      /usr/lib/libtcl8.4.so.0
> b60f1000-b60f2000 rw-p b60f1000 00:00 0
> b60f2000-b61c4000 r-xp 00000000 08:06 85102      /usr/lib/libtk8.4.so.0
> b61c4000-b61cf000 rw-p 000d2000 08:06 85102      /usr/lib/libtk8.4.so.0
> b61cf000-b61d0000 rw-p b61cf000 00:00 0
> b61d0000-b62a8000 r-xp 00000000 08:06 85103      /usr/lib/libBLT.2.4.so.8.4
> b62a8000-b62b9000 rw-p 000d8000 08:06 85103      /usr/lib/libBLT.2.4.so.8.4
> b62b9000-b62ba000 rw-p b62b9000 00:00 0
> b62ba000-b62dc000 r-xp 00000000 08:06 180469     /usr/lib/libpng12.so.0.15.0
> b62dc000-b62dd000 rw-p 00022000 08:06 180469     /usr/lib/libpng12.so.0.15.0
> b62f5000-b62f6000 rw-p b62f5000 00:00 0
> b62f6000-b631d000 r-xp 00000000 00:1a 553544
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_png.so
> b631d000-b631e000 rw-p 00027000 00:1a 553544
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_png.so
> b631e000-b6367000 r-xp 00000000 00:1a 553543
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_image.so
> b6367000-b6369000 rw-p 00049000 00:1a 553543
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_image.so
> b6369000-b63d3000 r-xp 00000000 08:06 83795
> /usr/lib/libfreetype.so.6.3.16
> b63d3000-b63d6000 rw-p 0006a000 08:06 83795
> /usr/lib/libfreetype.so.6.3.16
> b63d6000-b6424000 r-xp 00000000 00:1a 553535
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/ft2font.so
> b6424000-b6427000 rw-p 0004e000 00:1a 553535
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/ft2font.so
> b6427000-b650f000 r-xp 00000000 08:06 88506      /usr/lib/libstdc++.so.6.0.9
> b650f000-b6512000 r--p 000e8000 08:06 88506      /usr/lib/libstdc++.so.6.0.9
> b6512000-b6514000 rw-p 000eb000 08:06 88506      /usr/lib/libstdc++.so.6.0.9
> b6514000-b651a000 rw-p b6514000 00:00 0
> b651e000-b6528000 r-xp 00000000 08:06 313979
> /usr/lib/python2.5/lib-dynload/_tkinter.so
> b6528000-b6529000 rw-p 0000a000 08:06 313979
> /usr/lib/python2.5/lib-dynload/_tkinter.so
> b6529000-b652d000 r-xp 00000000 08:06 90832
> /usr/lib/python2.5/lib-dynload/zlib.so
> b652d000-b652e000 rw-p 00004000 08:06 90832
> /usr/lib/python2.5/lib-dynload/zlib.so
> b652e000-b6532000 r-xp 00000000 00:1a 553538
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_cntr.so
> b6532000-b6533000 rw-p 00004000 00:1a 553538
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_cntr.so
> b6533000-b6577000 r-xp 00000000 00:1a 553541
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_path.so
> b6577000-b6578000 rw-p 00044000 00:1a 553541
> /h/44/jtaylor/build/matplotlib/lib/matplotlib/_path.so
> b6578000-b6587000 r-xp 00000000 08:06 92419
> /usr/lib/python2.5/lib-dynload/datetime.so
> b6587000-b658a000 rw-p 0000e000 08:06 92419
> /usr/lib/python2.5/lib-dynload/datetime.so
> b658a000-b65b9000 r-xp 00000000 00:1a 532925
> /h/44/jtaylor/lib/python2.5/site-packages/numpy/random/mtrand.so
> b65b9000-b65cb000 rw-p 0002e000 00:1a 532925
> /h/44/jtaylor/lib/python2.5/site-packages/numpy/random/mtrand.so
> b65cb000-b6923000 r-xp 00000000 08:06 517267
> /usr/lib/atlas/libblas.so.3.0
> b6923000-b6927000 rw-p 00358000 08:06 517267
> /usr/lib/atlas/libblas.so.3.0
> b6927000-b6e6f000 r-xp 00000000 08:06 517268
> /usr/lib/atlas/liblapack.so.3.0
> b6e6f000-b6e72000 rw-p 00548000 08:06 517268
> /usr/lib/atlas/liblapack.so.3.0
> b6e72000-b6f76000 rw-p b6e72000 00:00 0
> b6f76000-b6f7a000 r-xp 00000000 08:06 85530      /usr/lib/libXdmcp.so.6.0.0
> b6f7a000-b6f7b000 rw-p 00003000 08:06 85530      /usr/lib/libXdmcp.so.6.0.0
> b6f7b000-b6f7f000 r-xp 00000000 08:06 92437
> /usr/lib/python2.5/lib-dynload/_csv.so
> b6f7f000-b6f81000 rw-p 00004000 08:06 92437
> /usr/lib/python2.5/lib-dynload/_csv.so
> b6f81000-b6f84000 r-xp 00000000 08:06 92427
> /usr/lib/python2.5/lib-dynload/_locale.so
> b6f84000-b6f85000 rw-p 00003000 08:06 92427
> /usr/lib/python2.5/lib-dynload/_locale.so
> b6f85000-b6f8e000 r-xp 00000000 00:1a 533113
> /h/44/jtaylor/lib/python2.5/site-packages/numpy/fft/fftpack_lite.so
> b6f8e000-b6f8f000 rw-p 00008000 00:1a 533113
> /h/44/jtaylor/lib/python2.5/site-packages/numpy/fft/fftpack_lite.so
> b6f8f000-b6fad000 r-xp 00000000 00:1a 533028
> /h/44/jtaylor/lib/python2.5/site-packages/numpy/core/scalarmath.so
> b6fad000-b6fae000 rw-p 0001e000 00:1a 533028
> /h/44/jtaylor/lib/python2.Aborted
>
>
> _______________________________________________
> NumPy-Discussion mailing list
> [hidden email]
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
>

no problem here, with official Windows numpy

Josef

>>> np.version.version
'1.3.0'

>python -i why_crash.py
>>> print rc
(array([ -5.23462841,  -4.85584394,  -2.99233015,  -7.54676368,
       -10.15455332,   7.074554  ,   8.4043877 ,   2.79661679,
         3.41336578,   5.29202285,   2.70716181,  15.53449435,
         9.34557621,  17.32209602,  18.16994838, -50.98017437,
       -50.96547959, -17.51283078,   7.68637678,  13.53704022,
        20.66278929, -23.64368007,  -4.70942583,  18.2568222 ,
         2.45709374,  12.97156815,  15.47026211, -44.93348725,
         7.7558192 , -39.13996758,   1.20214959,  23.61872159,
       -20.21198664,  -7.87137325,  -4.20255668, -45.24948722,
        12.49507108,  24.22157348,  23.46404032,  18.62294373,
       -26.31401828,  24.35842929, -37.5578372 ,  18.24079679,
        28.90693972, -40.40246853,  23.85976491,  11.70965078,
        17.38628028,   6.14989021,   0.19683346,  11.57781284,
        -6.70961655, -21.98525308, -11.30257635,  31.16804751,
         5.08794164,   0.26279222, -27.78390652, -26.3151511 ,
        14.89172102,  29.02572416, -10.84227516,   3.20577699,
       -34.73738042,  24.90588989,  37.92166034, -30.30146211,
        37.28852751, -16.03146259, -30.87415056, -33.02832669,
       -21.63514384,  11.15711455,  10.43855884,  -7.08345237,
        31.50460928, -28.64336727, -12.32269443, -24.59112645,
        41.71351395, -29.85091349,  -4.07409268,   0.82708638,
        14.67839587,  41.58165228, -29.44030397,  31.13279856,
       -28.46626932,  31.21863319, -30.50159697,  -6.26718832,
       -26.41654876,  -2.42547434,  44.00738912, -10.94028372,
        -0.65862359, -25.08227995, -26.04263867,  13.25529043,
        -7.41115206,  36.11891076,  47.22737694,  23.39250661,
       -16.59126536,  37.75596345,  12.59698144,   9.15952276,
       -22.0567611 , -27.79573887, -30.57535286,  28.71831817,
       -21.38243352,  19.30944773,  49.81583705, -19.59172648]), array([ 1063.81
458595]), 116, array([ 10.77032961,   3.02162267,   3.02054405,   3.0010756 ,
         2.96191492,   2.94807426,   2.94230063,   2.93906657,
         2.92832506,   2.91399677,   2.88159001,   2.86294336,
         2.85790349,   2.84497487,   2.82744239,   2.81275744,
         2.78836986,   2.77119523,   2.76422221,   2.75861982,
         2.75015801,   2.72908307,   2.68445243,   2.67800314,
         2.666536  ,   2.65671856,   2.64826304,   2.63879427,
         2.6296631 ,   2.60120053,   2.59118748,   2.58256916,
         2.57264941,   2.56585886,   2.53898947,   2.53365513,
         2.52103196,   2.49959127,   2.47968021,   2.46456052,
         2.46068247,   2.44924031,   2.43199483,   2.41963211,
         2.41515001,   2.40937849,   2.39016287,   2.3762653 ,
         2.35560428,   2.34357138,   2.3260469 ,   2.30884773,
         2.29027418,   2.27944481,   2.27465575,   2.25660949,
         2.21410648,   2.20263598,   2.1791073 ,   2.15789688,
         2.14225592,   2.13043072,   2.09846149,   2.07491627,
         2.06112946,   2.04336228,   2.02056257,   1.99107297,
         1.98856298,   1.97039638,   1.9575191 ,   1.93587212,
         1.91997992,   1.85665009,   1.84338407,   1.79610228,
         1.79328928,   1.78429932,   1.74123465,   1.7241243 ,
         1.7010803 ,   1.64746663,   1.62765943,   1.62303706,
         1.61800823,   1.60531761,   1.52425119,   1.50620662,
         1.485018  ,   1.45765932,   1.40861388,   1.39268607,
         1.3483904 ,   1.32025766,   1.31350522,   1.28517948,
         1.25950863,   1.23770526,   1.18665953,   1.15504454,
         1.14088912,   1.11336858,   1.01682096,   0.9791356 ,
         0.93161774,   0.90834728,   0.8611552 ,   0.82261935,
         0.79141265,   0.64055544,   0.60890393,   0.58578707,
         0.4948037 ,   0.38776132,   0.35580931,   0.20854201]))
>>>
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Matthew Brett
In reply to this post by Jonathan Taylor-3
Hi Jonathan,

> http://www.cs.toronto.edu/~jtaylor/crash/
>
> I would be interested to know if this bug can be duplicated and/or if anyone
> has any suggestions as to why:
>
> import numpy as np
> A = np.load('A.npy')
> b = np.load('b.npy')
> rc = np.linalg.lstsq(A,b)
>
> produces:
>
> *** glibc detected *** /usr/bin/python: free(): invalid next size (normal):
> 0x091793c0 ***

I just tried it on 4 ubuntu machines, and one Fedora 11 machine,  in
various states of numpy-ness (including recent SVN) with no crash.
What versions of stuff do you have over there?

See you,

Matthew
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Charles R Harris
In reply to this post by Jonathan Taylor-3


On Mon, Aug 17, 2009 at 11:42 AM, Jonathan Taylor <[hidden email]> wrote:
Hi,

I am getting a strange crash in numpy.linalg.lstsq.  I have put the code that causes the crash along with two data files on my website at:

http://www.cs.toronto.edu/~jtaylor/crash/

I would be interested to know if this bug can be duplicated and/or if anyone has any suggestions as to why:

Usually these problems are due to ATLAS. If you are using ATLAS, what is your OS/distribution? What hardware are you running on? Did you build ATLAS yourself?

Chuck



_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Jonathan Taylor-3
Hi,

I am using a computer that is administered.  It is an intel Ubuntu box
and came with an ATLAS compiled.  I thus compiled my own numpy1.3.0
against that ATLAS.

I was thinking about recompiling ATLAS myself.  This machine only has
g77 and not gfortran on it.  Will that still work?

Thanks,
Jonathan.

On Mon, Aug 17, 2009 at 3:12 PM, Charles R Harris
<[hidden email]> wrote:

>
>
> On Mon, Aug 17, 2009 at 11:42 AM, Jonathan Taylor <[hidden email]> wrote:
>>
>> Hi,
>>
>> I am getting a strange crash in numpy.linalg.lstsq.  I have put the code that causes the crash along with two data files on my website at:
>>
>> http://www.cs.toronto.edu/~jtaylor/crash/
>>
>> I would be interested to know if this bug can be duplicated and/or if anyone has any suggestions as to why:
>
> Usually these problems are due to ATLAS. If you are using ATLAS, what is your OS/distribution? What hardware are you running on? Did you build ATLAS yourself?
>
> Chuck
>
>
>
> _______________________________________________
> NumPy-Discussion mailing list
> [hidden email]
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Charles R Harris


On Mon, Aug 17, 2009 at 1:43 PM, Jonathan Taylor <[hidden email]> wrote:
Hi,

I am using a computer that is administered.  It is an intel Ubuntu box
and came with an ATLAS compiled.  I thus compiled my own numpy1.3.0
against that ATLAS.

I was thinking about recompiling ATLAS myself.  This machine only has
g77 and not gfortran on it.  Will that still work?

As long as everything is consistent it should. Ubuntu has had some issues with ATLAS and I suspect that is what you are seeing. David Cournapeau could tell you more.

Chuck


_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Jonathan Taylor-3
I compiled lapack and atlas from scratch using g77 but now
numpy.test() hangs when I try to use any numpy functionality.  I think
I saw someone else write about this.  Is this a common problem?

Thanks,
Jonathan.

On Mon, Aug 17, 2009 at 4:38 PM, Charles R
Harris<[hidden email]> wrote:

>
>
> On Mon, Aug 17, 2009 at 1:43 PM, Jonathan Taylor
> <[hidden email]> wrote:
>>
>> Hi,
>>
>> I am using a computer that is administered.  It is an intel Ubuntu box
>> and came with an ATLAS compiled.  I thus compiled my own numpy1.3.0
>> against that ATLAS.
>>
>> I was thinking about recompiling ATLAS myself.  This machine only has
>> g77 and not gfortran on it.  Will that still work?
>
> As long as everything is consistent it should. Ubuntu has had some issues
> with ATLAS and I suspect that is what you are seeing. David Cournapeau could
> tell you more.
>
> Chuck
>
>
> _______________________________________________
> NumPy-Discussion mailing list
> [hidden email]
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
>
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Keith Goodman
On Mon, Aug 17, 2009 at 1:50 PM, Jonathan
Taylor<[hidden email]> wrote:
> I compiled lapack and atlas from scratch using g77 but now
> numpy.test() hangs when I try to use any numpy functionality.  I think
> I saw someone else write about this.  Is this a common problem?

Yes, it seems common. I know of 4 recent ATLAS builds (including mine
but not yours) that have failed on 32 bit systems. The recent
successes I have seen (including mine) have been on 64 bit systems.
But maybe 32/64 bit has nothing to do with it. I am sure there are
many 32 bit systems running a self-compiled ATLAS just fine.

Oh, you crash on any numpy functionality? I only crashed on ATLAS type problems.
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Jonathan Taylor-3
Yes... ATLAS type problems like matrix multiplication.

Is there some alternative to get a working numpy going?  How might I
go about compiling numpy without ATLAS?  I really got to get at least
something working temporarily.

Thanks,
Jon.

On Mon, Aug 17, 2009 at 5:03 PM, Keith Goodman<[hidden email]> wrote:

> On Mon, Aug 17, 2009 at 1:50 PM, Jonathan
> Taylor<[hidden email]> wrote:
>> I compiled lapack and atlas from scratch using g77 but now
>> numpy.test() hangs when I try to use any numpy functionality.  I think
>> I saw someone else write about this.  Is this a common problem?
>
> Yes, it seems common. I know of 4 recent ATLAS builds (including mine
> but not yours) that have failed on 32 bit systems. The recent
> successes I have seen (including mine) have been on 64 bit systems.
> But maybe 32/64 bit has nothing to do with it. I am sure there are
> many 32 bit systems running a self-compiled ATLAS just fine.
>
> Oh, you crash on any numpy functionality? I only crashed on ATLAS type problems.
> _______________________________________________
> NumPy-Discussion mailing list
> [hidden email]
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Keith Goodman
On Mon, Aug 17, 2009 at 2:13 PM, Jonathan
Taylor<[hidden email]> wrote:
> Is there some alternative to get a working numpy going?  How might I
> go about compiling numpy without ATLAS?  I really got to get at least
> something working temporarily.

Just build numpy again but skip the ATLAS steps.
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Jonathan Taylor-3
It seems to automatically detect it though.  Specifically
lapack_lite.so always seems to reference libatlas.

On Mon, Aug 17, 2009 at 5:21 PM, Keith Goodman<[hidden email]> wrote:

> On Mon, Aug 17, 2009 at 2:13 PM, Jonathan
> Taylor<[hidden email]> wrote:
>> Is there some alternative to get a working numpy going?  How might I
>> go about compiling numpy without ATLAS?  I really got to get at least
>> something working temporarily.
>
> Just build numpy again but skip the ATLAS steps.
> _______________________________________________
> NumPy-Discussion mailing list
> [hidden email]
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Keith Goodman
On Mon, Aug 17, 2009 at 2:27 PM, Jonathan
Taylor<[hidden email]> wrote:

> It seems to automatically detect it though.  Specifically
> lapack_lite.so always seems to reference libatlas.
>
> On Mon, Aug 17, 2009 at 5:21 PM, Keith Goodman<[hidden email]> wrote:
>> On Mon, Aug 17, 2009 at 2:13 PM, Jonathan
>> Taylor<[hidden email]> wrote:
>>> Is there some alternative to get a working numpy going?  How might I
>>> go about compiling numpy without ATLAS?  I really got to get at least
>>> something working temporarily.
>>
>> Just build numpy again but skip the ATLAS steps.

Yes, sorry. The only way I've tried doing it is uninstalling the
unbuntu ATLAS binary. I don't know how to ignore it if it is
installed.
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange crash in numpy.linalg.lstsq.

Jonathan Taylor-3
Right... So I was able to get everything working finally.  I am not
100% sure how or why it works though so I am going to outline what I
did here for reference.

I first tried just using LAPACK 3.1.1 (since it seemed set up for g77
instead of gfortran which I do not have).  I compiled this to yield
the associated lapack and blas libraries using the fortran compiler
settings detailed on the scipy install web page.  This actually gave
me a numpy with the same problems (hanging on numpy.test()).  Thus I
realized the problem was LAPACK and not with ATLAS.

Eventually I got numpy with LAPACK to work when I used the default
settings in the example config file of LAPACK instead of using the
suggested settings on the numpy web site.  I then compiled ATLAS and
this worked as well.  It still seems a little bit weird that these
settings can break the software though.

Thanks for the suggestions and I hope this helps someone.
Jonathan.

On Mon, Aug 17, 2009 at 5:34 PM, Keith Goodman<[hidden email]> wrote:

> On Mon, Aug 17, 2009 at 2:27 PM, Jonathan
> Taylor<[hidden email]> wrote:
>> It seems to automatically detect it though.  Specifically
>> lapack_lite.so always seems to reference libatlas.
>>
>> On Mon, Aug 17, 2009 at 5:21 PM, Keith Goodman<[hidden email]> wrote:
>>> On Mon, Aug 17, 2009 at 2:13 PM, Jonathan
>>> Taylor<[hidden email]> wrote:
>>>> Is there some alternative to get a working numpy going?  How might I
>>>> go about compiling numpy without ATLAS?  I really got to get at least
>>>> something working temporarily.
>>>
>>> Just build numpy again but skip the ATLAS steps.
>
> Yes, sorry. The only way I've tried doing it is uninstalling the
> unbuntu ATLAS binary. I don't know how to ignore it if it is
> installed.
> _______________________________________________
> NumPy-Discussion mailing list
> [hidden email]
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
_______________________________________________
NumPy-Discussion mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
Loading...