Re: optimization/10087: optimizer produces wrong code when indexing malloc'ed 2D array

Re: optimization/10087: optimizer produces wrong code when indexing malloc'ed 2D array

[bangerth]

Synopsis: optimizer produces wrong code when indexing malloc'ed 2D array

State-Changed-From-To: open->feedback
State-Changed-By: bangerth
State-Changed-When: Fri Mar 14 22:38:03 2003
State-Changed-Why:
    The uuencoded file you pasted isn't ok, the last line doesn't
    start with an "M". Can you please send me the file in private
    mail so that I can attach it to the report?
    
    For the future, please use the file attachment box in
    the form, rather than pasting files into one of the
    fields. This way, they are much simpler to retrieve.
    
    Thanks
      Wolfgang

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10087

Re: optimization/10087: optimizer produces wrong code when indexing malloc'ed 2D array

[Peter van Hoof]

The following reply was made to PR optimization/10087; it has been noted by GNATS.

From: Peter van Hoof <p.van-hoof@qub.ac.uk>
To: p.van-hoof@qub.ac.uk, gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org, 
    nobody@gcc.gnu.org, gcc-prs@gcc.gnu.org
Cc:  
Subject: Re: optimization/10087: optimizer produces wrong code when indexing
         malloc'ed 2D array
Date: Sat, 15 Mar 2003 00:54:54 +0000

 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10087
 
 On my ultrasparc the following options are sufficient to create the bug:
 
 scooby>	gcc -O bug.c
 scooby>	a.out
 result 1.000000e+00
 scooby>	gcc -O -fstrength-reduce -fgcse  bug.c
 scooby>	a.out
 result 1.039778e-312
 

Re: optimization/10087: optimizer produces wrong code when indexing malloc'ed 2D array

[bangerth]

Synopsis: optimizer produces wrong code when indexing malloc'ed 2D array

State-Changed-From-To: feedback->open
State-Changed-By: bangerth
State-Changed-When: Fri Mar 14 23:27:36 2003
State-Changed-Why:
    Attachment received

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10087

Re: optimization/10087: optimizer produces wrong code when indexing malloc'ed 2D array

[Peter van Hoof]

The following reply was made to PR optimization/10087; it has been noted by GNATS.

From: Peter van Hoof <p.van-hoof@qub.ac.uk>
To: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, 
    nobody@gcc.gnu.org, p.van-hoof@qub.ac.uk, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: optimization/10087: optimizer produces wrong code when indexing
         malloc'ed 2D array
Date: Fri, 14 Mar 2003 23:24:00 +0000

 This is a multi-part message in MIME format.
 --------------030002000303010400060907
 Content-Type: text/plain; charset=us-ascii; format=flowed
 Content-Transfer-Encoding: 7bit
 
 This must be a bug in my uuencode... Here is the attachment.
 
 --------------030002000303010400060907
 Content-Type: application/octet-stream;
  name="bug.i.bz2"
 Content-Transfer-Encoding: base64
 Content-Disposition: attachment;
  filename="bug.i.bz2"
 
 QlpoOTFBWSZTWUZk2ZUABn//gHw0BARbf///cyYeDr////pgEt94Bc7h9sA+972PeX3Nh9zI
 Pr7u9PHoeezfUg1u3y0cgp9O7em+e4K+8G9Dd7g4SKE0in6RianqPAj0p+qPUPKZDZQDJkB6
 jajQCU0ICE00Ep5U80KNME0GaQaDQxAGmEEgqKeVNPKeUNHqANAA0ADQAAAAJNJERBNUxoZR
 tT1PaiGjQAYg0BiYgyYRKEmUeqfqNQzKPUaA0NAZAA0BoaNGgCJIgIAmjQAlM1NGg0NANNAA
 ADEQdVkQSJAQ8ff2K8fP6wD39cMJ8LZ5agQh9nb70C2hA9pIshTWYFEWCMBf/8YcGpRKrBQY
 qRViKIFZWQVBJ7kSnptRRYYNL7vRODm1UMnTkyIMRVylEQUtopbZWopbRYVCgr1SmW5ZKCwD
 dqoooUtKiiyCigqgoVtpRkFkKUpbaFtUWStqWCNQoiIgJbQtKlRVKKsYAqilH83QvVFMVUch
 iCriI9kQ6QRDfxnT47jnhSp0ekU6uVO1DrEvBgxE3E6mLUpNrTazvKZ4rFsPScWmVWrtet6S
 XtSKTJOC8TUeYrM2cxaLToN4Ayi1ol06jqNKN+Ix/brhvbY3be5lA3FxjKKLe3Hg1gKZLmFU
 uN1mhlpvjZ/HmwDfNpbCjBVEYUpRjticbd6aahhtKrCqHNflHWavDg4U9Un0fBJD8EgetSEo
 sAQD9sG+VWL7VYqgsVGSJAlFQKomhAWIm+QT7HG38WxTT3NnLdgkXkFPk4bK9DrsjGBKiNFm
 5ZbHcuRjmfCDlGRxwkeDIghK8RnoENCBc3kR2NN2Q0rJIx1OnOikMmDJoM4psALTq64uty84
 WRcVqHKQoyRH8gX2DkvzgG51k0gRBBSHLKwaS+yydZNmznuwtp4+K0O9DvJ3y8unup07N9pv
 tIrUuFoMtqsXLmJQ7t6DXQPpQmPO7bvKtpylZ1iwZOLTXjx9zRoQ4gWFPInWLOJC9VQ4CZbZ
 KUaqpqZrESYazIVIdfCyGDEDaM9MuaHCV2kJapKFVgh20zEwNMA7J0k1xQ0jOMkY93T/dNKf
 drEb22DJhA18KbXPPk4nuKI8ChUgqmtwAIHiHgibNkOdpIFmAGATxRxGBHxAOz8bZ9Ttu+W8
 eKry5Tva1FaPu85KIL/nS9j5k2NfWrevaoocdnRlPAkejMveZh5p9jsZJvu63WmE1DuSdDH3
 iCyPnVOc4nouej9Nxr+2xj4z5fn/bvshz6luhIId9HmECbXw/d8+jhWHLAWKQrWFYVlYic0K
 MabRZHJMvKFGC1nBaCtbzQkrZGgUaAVkSA2PzDS3UZ7moZkBvaI2xMDfSVBRZFiIIwUO2yrw
 zyef+GzQikkZCMjpjWt/IBO8BIAnYLwXqJ+UBhH9V1QmBhcNwHpxp5eTqVTzeimHqoq81q8U
 JpSR5B6FYiFMPA6dMgmZoq1IitHCU7LDTwmHDxNpw4qbdutygXXh2dj8ZfR8QXhb6i9S1VPY
 rlP044YgJtfkATeAnkgaKwIwRIKQS670dAE4AJZsHkAlLmAllDmJFQLBCEkkiEiUwUughHbw
 13anC5YuAka+tPnvRV5eBUB6HClrarG3Hbu4aubXotpzKyjWKnIcWStbkvkyIb3lsoRKIbLj
 ShkLwZidIdcyrmp/Y9PW1DTny0yod+IqMG5z7AMdsNp0LHybO2yZ5HqnTJSG7Dly5W+OKatl
 NTZOyemysOvTQQXA3gJw1AS+QCQ1jIfP1Ur3ASLpHKHOIC7tNvi+vjTnWSWbLlbs1k1Dt0HA
 aa874NKQqCMEDvtbSS1XSsul2f0e4L3Tz9fhQ8Q/R4gkyiILofJlmOEub8348IOznSx0wYw8
 xKUnVZl73pciGIgSK99gNb1y61u2hTDxnQSDUBiZx6R0zduvfv+Q4G/kPVPAzOIDs44JB8Ty
 veU8uUPj44BrzLjiynBqlk5bzqrhBxtymdH9jHUVkcNVbu3ZQq8UaEdbPFdZrk9Wx949PJab
 bLWiz+JAQlxEOuE6EJITY3yJhTZTbsNbxsWcQGhpKvaYIIgIxTKEiWfb1yy+l7ZJbrXboa8s
 zsw2FyMdD2H7ow9BxrJnqmAfmk7EgXS1+7IbX957n/B7RtDbaIEISQWIUZOUAxAX6AL+sQDz
 /PWC3cYBzJqrtw/DDymnWayjOtYU5/XyKuz2WUGuVBSu3Ah90EaGhtCbAbQ762OGFz2c9MwW
 py5lHZ/sNJhKw93UfNthxYlj6vP5J0EJCZRykqVrVIZISM6tvYDcZuTUzlAkltE8x1qc9Xmq
 YCPaF9pM5eCnKbTVJhISBkmSGS5bxkwzG+uhngjHXpzx2oaTVxjKVS8nfNktjkks4mm2ZZas
 Q3d01EHszDXlos7SiYZyURApSbWaGK9KtAgLUvMspZ7v7lrGZqaF5RME99ykcPew+qpo/KnP
 eXe191Wfli1Zrn1c3tj26oUR1ONb6V2sUWVV6FTk9IK2wRGU/A9eFjUNA1mbjdio9teR3EkK
 LnXWCGNjm4C13GOIdd/0rUKEGiX4TbVjfoGgaPEAoo2IGYCQY0Myoayu7gwwuGvEopaN5MgE
 wtiY5gEtq3cQuvAzZ8CqJ4frxEQ8IiIoFyRvm1PIAY7+rKF4MPTZLpjLpxSV03374fbbpyn9
 Yf9iuWG+ILlmE3iAx5AJSUAmBjIR3yJswATaetrWHzAE8nd6+kD4IqCwBYKCJFGMJPIa32Ht
 hoC4rkE89uSAvb4mu7K9XOmAPF4cTaaLCREWCiEQFiqzu48ISTmBxy9z7B65A2CIRGTTAKDB
 vrNKwLQPPvWCF2E5UAESizaLUWoSFct/YbiF3I9/D+OWx2JxQF7jgKrtDeAYGVtudtwgedY7
 7Np5pYMns0rkk8LnINknnQ1AkOEyijqgLRn3VnzsXrGVXvpQmyJ8jGRjIJCFa2tWnNRdndEK
 4hs3I4pCOCFkUsKrhcJ9F2AA03D2FWCeTXWBsQaPQA4PDvzKo4Mybi9HHWlNdu6KI4Z10z37
 89EMTKsEAtSnmaGnsCegncyiKsFVdWk6DwQTXCtoHhE44ryxA1QOTkogWiEhdErEpFJ4Gw9T
 tFV17/QvC3UliJRI27nqU0kxpTK4aB6/opxpdIqnKMFGCtW8RlISaC1G4pneXvKQM7CRfUFF
 q0mBWZY1SOBVAJXQsKW2FipUQeQAnA0zvN0TW5LVaxDVDNq2ItQDNcKlkqUe2gAc3JVTA01B
 L9AI+BhXrkFj/kNsfoefwAE2Vm+CjXnSHTuNDiFT6LZ8scMdcL4YgGv1fIc3MDfEB718ptMA
 FdVO/l/O/kgL8a4bQJnMoe3jQDIyEiQiPxa5GQZBkqB4iCKRRTvkrNeoMxEWC9BqqqTx74aP
 Zga2vb5SAb9RPArIBNNC2JFaR4C6BiZFIfN3CNPrH2/xT7h7+G/4ec1IyTKlOZC9yngGByqy
 XwLAdl9AC09VIImRC0MS2foNzaAl/kcuaUSX1wA0Q264537QE9QE+3j3YuXI1F3I6788abEE
 tMgdgCd3tw1OMosEZnEpLZemAnpFUkFIebL7nHqrFUVBEVmGRYseczux48JaroCxrsQwsUho
 AenJ2OxLHKXC3hgGR3tbk1gCSKIelspOko7czFSVuuoZJCRkmANf3ENWeEARSVvbuuzlj7On
 JZ4oCzO1iKSOMaNqKd5gN7sAWRRxgsGjLdnbOGMyweBM0UwBXJAxRBkBTakIMisGLSK4JSmR
 q0btcS4iQiIxGIqwmtbO83BZBlKFHe5JA6Ga4Dbx32+YEO3MBOGzhtA26Nlexm1AGiAJLNew
 3uSMJYrWMBEMuBUo4i9pyY6J8XmBF4RwGnKPJMOXTjoWhwCbk20dwCVh7tlYKr1G9IBoVeZz
 K6PZqHIf8AGPjY0OJfxm4VNmWUlLRVhLaKKRYr3aMOw0dUXGebAzEhQau7y4o0YGCCnOlERI
 y6RAQ1oiJSaXk7gCAYoSE2DTqzA2px2nllOPf9We6tnKmWVULAJpLbEERtnLwrIDeCbIWFW9
 xtwU1/7kZGMJeVvpqNSoyVETqoB5gJiltfb2WaadoZEJM3PU8uO+xp39504zXaVpo+FBraGv
 w9VwwD3mPKGQQsUUYULakm23mWoNUUqgtvpfsRJHttYubw4k47M9s5Yguz5QE5AJ0eshAkJk
 ZaIIVXnXI1gixTIDChlptiaIjIEgkjhnw3qg70GjPBE3EHo0XLBaNSwYGIbO+uwBMMbbFAOl
 4YByqI2G0sL0CGFQiNFEMg3UhEge20dMmsbLEiCsRgmkatdAWq0ti4xGVUtsYKQwSXQGqUMJ
 qEjAyMdFFKjKUNMN7qTfpPLl3xwcVbZQpRUUFEVVFUEQ6ZxFFjwBDYFVAVYKoCu5A9KKVYys
 loDIhhAXSSJgQC54d5cDbnVmQzpCHTNEmpxgM0rfKbIknEIYC5iK8Xrff+ERHZqa7QgJSFlB
 1PF1e0vofL2fH1555nr2+p3gN6pJHE9IYEPElpbE7u4uOASkOhSpv5cuvRwEO01bYxLGwLEC
 SFu+yRVaNSg2bLpaGohNQhj1fR4wn0Ue0ahUn6bYoiH502gbGLNC4mYE0hsSWwI5C44JYjSX
 Py2bK6wckG2KMDQbmuXEU/wdROd0dKXFEVvBF7ytcJKHZTOW3yBZqkHTTSgEt6zLl6WLDIwW
 DGECEGGkrd5JutNKnpwsPNBDL0eXH4KGPHB0PN3SEZNwHXa0WeyQifMVVUansYOfDImBzh0J
 KCjt85a1EkS6ETXM07dbqanw544K+5FUYWjCixbMQEzRAhvAZhJNdsWQILGd55hmYKQWB9ou
 5IpwoSCMybMq
 --------------030002000303010400060907--
 

optimization/10087: optimizer produces wrong code when indexing malloc'ed 2D array

[p.van-hoof]

>Number:         10087
>Category:       optimization
>Synopsis:       optimizer produces wrong code when indexing malloc'ed 2D array
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Fri Mar 14 22:16:01 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Peter van Hoof
>Release:        3.3 20030310 (prerelease)
>Organization:
Queen's University of Belfast
>Environment:
System: SunOS scooby 5.9 Generic_112233-04 sun4u sparc SUNW,Sun-Blade-100
Architecture: sun4

	
host: sparc-sun-solaris2.9
build: sparc-sun-solaris2.9
target: sparc-sun-solaris2.9
configured with: /scratch/temp/ss/gcc-20030310/configure --prefix=/opt/local/gcc330
>Description:
In the attached code a two-dimensional array of structs is malloc'ed,
then the first field of that struct is initialized to 1 for all
array members. Finally the minimum value of all these fields is
determined by looping over the array again. This should of course
return 1, but the actual result is 1.039778e-312.

This code runs OK with gcc 3.2.2
>How-To-Repeat:

scooby>	gcc -v -save-temps -O2 bug.c
Reading specs from /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/specs
Configured with: /scratch/temp/ss/gcc-20030310/configure --prefix=/opt/local/gcc330
Thread model: posix
gcc version 3.3 20030310 (prerelease)
 /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/cc1 -E -quiet -v -D__GNUC__=3 -D__GNUC_MINOR__=3 -D__GNUC_PATCHLEVEL__=0 -Dsparc -D__sparc__ -D__sparc -D__GCC_NEW_VARARGS__ -Acpu=sparc -Amachine=sparc bug.c -O2 bug.i
ignoring nonexistent directory "/opt/local/gcc330/sparc-sun-solaris2.9/include"
#include "..." search starts here:
#include <...> search starts here:
 /usr/local/include
 /opt/local/gcc330/include
 /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/include
 /usr/include
End of search list.
 /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/cc1 -fpreprocessed bug.i -quiet -dumpbase bug.c -auxbase bug -O2 -version -o bug.s
GNU C version 3.3 20030310 (prerelease) (sparc-sun-solaris2.9)
	compiled by GNU C version 3.3 20030310 (prerelease).
GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072
 /usr/ccs/bin/as -V -Qy -s -o bug.o bug.s
/usr/ccs/bin/as: Sun WorkShop 6 update 2 Compiler Common 6.2 Solaris_9_CBE 2001/04/02
 /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/collect2 -V -Y P,/usr/ccs/lib:/usr/lib -Qy /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/crt1.o /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/crti.o /usr/ccs/lib/values-Xa.o /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/crtbegin.o -L/opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3 -L/usr/ccs/bin -L/usr/ccs/lib -L/opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/../../.. bug.o -lgcc -lgcc_eh -lc -lgcc -lgcc_eh -lc /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/crtend.o /opt/local/gcc330/lib/gcc-lib/sparc-sun-solaris2.9/3.3/crtn.o
ld: Software Generation Utilities - Solaris Link Editors: 5.9-1.373
scooby>	a.out 
result 1.039778e-312

This is the input:

begin 644 bug.i.bz2
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;9A)-=L60(+&=YYAF8*06!]HNY(IPH2",R;,J
 
end

>Fix:
Using a lower level of optimization works around the bug.
>Release-Note:
>Audit-Trail:
>Unformatted: