gfortran formatted read incorrectly reads beyond end of record when record length is < io,format length. Demo: ian@ian:~/source/test> cat ftest.f95 program ftest integer i character*1 b(10) open (77, file='input_file') read (77, '(10A1)') b print *, b end program ftest ian@ian:~/source/test> cat input_file Line 1 Line 2 Line 3 Line 4 ian@ian:~/source/test> gfortran -v Using built-in specs. Configured with: ../gcc/configure --enable-languages=c,f95 --prefix=/usr/work/2005020/irun Thread model: posix gcc version 4.0.0 20050120 (experimental) ian@ian:~/source/test> gfortran -o ftest ftest.f95 ian@ian:~/source/test> ./ftest Line 1 Lin
This looks promising. I'll do a full check later. Thomas --- transfer.c.orig 2005-01-31 18:03:12.000000000 +0100 +++ transfer.c 2005-01-31 18:04:00.000000000 +0100 @@ -150,6 +150,14 @@ else p = base = data; + /* If we have seen the end of the record already, we just + * return blanks. + */ + if (sf_seen_eor) { + memset(base,' ',*length); + return base; + } + memset(base,'\0',*length); current_unit->bytes_left = options.default_recl; @@ -1222,8 +1230,11 @@ case FORMATTED_SEQUENTIAL: length = 1; /* sf_read has already terminated input because of an '\n' */ - if (sf_seen_eor) - break; + if (sf_seen_eor) + { + sf_seen_eor = 0; + break; + } do {
Patch here: http://gcc.gnu.org/ml/gcc-patches/2005-01/msg02295.html
The patch does not handle advance="NO" correctly. I'll need to do something else. Thomas
Replying to myself here... After having thought of this, I think the patch should go in because it fixes a g77 regression, and advance= isn't among the things that g77 supports. Thomas
Updated patch: http://gcc.gnu.org/ml/gcc-patches/2005-03/msg00566.html
Updated patch: http://gcc.gnu.org/ml/gcc-patches/2005-03/msg00729.html
Subject: Bug 19568 CVSROOT: /cvs/gcc Module name: gcc Changes by: tkoenig@gcc.gnu.org 2005-04-10 08:35:40 Modified files: libgfortran : ChangeLog libgfortran/io : transfer.c gcc/testsuite : ChangeLog Added files: gcc/testsuite/gfortran.dg: eor_handling_1.f90 eor_handling_2.f90 eor_handling_3.f90 eor_handling_4.f90 eor_handling_5.f90 noadv_size.f90 pad_no.f90 Log message: 2005-04-10 Thomas Koenig <Thomas.Koenig@online.de> PR libfortran/17992 PR libfortran/19568 PR libfortran/19595 PR libfortran/20005 PR libfortran/20092 PR libfortran/20131 PR libfortran/20138 PR libfortran/20661 PR libfortran/20744 * io/transfer.c (top level): eor_condition: New static variable. (read_sf): Remove unnecessary zeroing of buffer (there is enough information in its length). Return a string of length 0 (to be padded by caller) if EOR was seen previously. Remove erroneous special casing of EOR for standard input. Set eor_condition for non-advancing I/O if an end of line was detected. Increment ioparm.size if necessary. (formatted_transfer): Skip the function if there is an EOR condition. (data_transfer_init): Initialize eor_condition to zero (false). (next_record_r): Clear sf_seen_eor if a \n has been seen already. (finalize_transfer): If there is an EOR condition, raise the error. 2005-04-10 Thomas Koenig <Thomas.Koenig@online.de> * eor_handling_1.f90: New test case. * eor_handling_2.f90: New test case. * eor_handling_3.f90: New test case. * eor_handling_4.f90: New test case. * eor_handling_5.f90: New test case. * noadv_size.f90: New test case. * pad_no.f90: New test case. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/ChangeLog.diff?cvsroot=gcc&r1=1.188&r2=1.189 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/io/transfer.c.diff?cvsroot=gcc&r1=1.35&r2=1.36 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.5320&r2=1.5321 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_1.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_2.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_3.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_4.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_5.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/noadv_size.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/pad_no.f90.diff?cvsroot=gcc&r1=NONE&r2=1.1
Fixed in 4.1.0, waiting for 4.0 to reopen.
Subject: Bug 19568 CVSROOT: /cvs/gcc Module name: gcc Branch: gcc-4_0-branch Changes by: tkoenig@gcc.gnu.org 2005-05-03 19:39:36 Modified files: libgfortran : ChangeLog libgfortran/io : transfer.c gcc/testsuite : ChangeLog Added files: gcc/testsuite/gfortran.dg: eor_handling_1.f90 eor_handling_2.f90 eor_handling_3.f90 eor_handling_4.f90 eor_handling_5.f90 noadv_size.f90 pad_no.f90 Log message: 2005-05-03 Thomas Koenig <Thomas.Koenig@online.de> Backport from mainline: PR libfortran/17992 PR libfortran/19568 PR libfortran/19595 PR libfortran/20005 PR libfortran/20092 PR libfortran/20131 PR libfortran/20661 PR libfortran/20744 * io/transfer.c (top level): eor_condition: New static variable. (read_sf): Remove unnecessary zeroing of buffer (there is enough information in its length). Return a string of length 0 (to be padded by caller) if EOR was seen previously. Remove erroneous special casing of EOR for standard input. Set eor_condition for non-advancing I/O if an end of line was detected. Increment ioparm.size if necessary. (formatted_transfer): Skip the function if there is an EOR condition. (data_transfer_init): Initialize eor_condition to zero (false). (next_record_r): Clear sf_seen_eor if a \n has been seen already. (finalize_transfer): If there is an EOR condition, raise the error. 2005-05-03 Thomas Koenig <Thomas.Koenig@online.de> Backport from mainline: * eor_handling_1.f90: New test case. * eor_handling_2.f90: New test case. * eor_handling_3.f90: New test case. * eor_handling_4.f90: New test case. * eor_handling_5.f90: New test case. * noadv_size.f90: New test case. * pad_no.f90: New test case. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.163.2.22&r2=1.163.2.23 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/io/transfer.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.32.2.2&r2=1.32.2.3 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.5084.2.156&r2=1.5084.2.157 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_1.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_2.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_3.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_4.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/eor_handling_5.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/noadv_size.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/pad_no.f90.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.6.1
Patch committed to 4.0.1. Fixed.